Thermal Desorption Mass Spectrometric Analysis of Organic Aerosol

thermal desorption particle beam mass spectrometer. Temperature-programmed thermal desorption of collected aerosol shows that in each reaction two maj...
0 downloads 0 Views 215KB Size
Environ. Sci. Technol. 2000, 34, 2105-2115

Thermal Desorption Mass Spectrometric Analysis of Organic Aerosol Formed from Reactions of 1-Tetradecene and O3 in the Presence of Alcohols and Carboxylic Acids HERBERT J. TOBIAS AND P A U L J . Z I E M A N N * ,† Air Pollution Research Center, University of California, Riverside, California 92521

The chemistry of secondary organic aerosol formation from reactions of 1-tetradecene and O3 in dry air in the presence of excess alcohols and carboxylic acids was investigated in an environmental chamber using a thermal desorption particle beam mass spectrometer. Temperature-programmed thermal desorption of collected aerosol shows that in each reaction two major aerosol products are formed. The more volatile compounds in each pair of products are R-alkoxytridecyl or R-acyloxytridecyl hydroperoxides, which were identified by comparison of mass spectra with those of standard compounds generated by the corresponding liquid-phase ozonolysis reactions. The formation of organic hydroperoxides in the gas and liquid phases is consistent with a mechanism involving reactions of the alcohols and carboxylic acids with stabilized Criegee biradicals generated in the alkene-O3 reaction. The less volatile compounds are R-alkoxy-R′-hydroxyditridecyl or R-acyloxy-R′-hydroxyditridecyl peroxides (peroxyhemiacetals) formed by reactions of the hydroperoxides with tridecanal, which is generated along with formaldehyde during biradical formation. The vapor pressures of these compounds estimated from their desorption temperatures are ∼10-7-10-14 Torr. These types of reactions could play a role in atmospheric aerosol nucleation and growth and provide a mechanism for creating fine-particle organic peroxides, which are currently of interest because of their potential for adversely impacting human health.

Introduction Atmospheric fine particles (diameter < 2.5 µm, PM2.5) influence global climate directly by scattering and absorbing radiation and indirectly through their action as cloud condensation nuclei (1, 2). At the local and regional scale, light scattering by these species is the most important determinant of atmospheric visibility (3). Fine particles also provide most of the aerosol surface area available for heterogeneous chemical reactions, thus impacting atmospheric chemistry (4). For as yet unknown reasons, they also * Corresponding author phone: (909)787-5127; fax: (909)787-5004; e-mail: [email protected]. † Also at the Department of Environmental Sciences and Department of Chemistry. 10.1021/es9907156 CCC: $19.00 Published on Web 04/22/2000

 2000 American Chemical Society

appear to adversely impact human health (5). An important component of fine particles is secondary organic material (6, 7), which is formed in the atmosphere through gas-toparticle conversion processes that include gas-phase and possibly heterogeneous reactions of the oxidants O3, OH, and NO3 with volatile organic compounds (VOCs) of biogenic and anthropogenic origin (8-12). Through these reactions, VOCs gain various functional groups including carbonyl (CO), hydroxy (COH), hydroperoxy (COOH), carboxylic acid (C(O)OH), nitrite (CONO), or nitrate (CONO2) that increase compound polarity and molecular weight and, therefore, decrease volatility. Depending on the gas-phase concentrations of product species and the nature of preexisting aerosol particles (i.e., concentration, size distribution, solid or liquid, organic, aqueous), the low-volatility compounds can either condense onto or into preexisting aerosol or homogeneously nucleate to create new particles (12). The driving force for nucleation and aerosol growth increases as the vapor pressures of aerosol-forming compounds decrease. Current understanding of secondary organic aerosol formation has been developed principally from environmental chamber studies of reactions of single VOCs with single or multiple oxidants. Many of these experiments have only included analyses of VOC reactants and particle size distributions, which are valuable for quantifying aerosol yield (13, 14). In some experiments the chemical composition of particles collected by filtration or impaction have been determined by gas chromatography-mass spectrometry (GC-MS) of solvent-extracted components and have provided important insight into chemical mechanisms of aerosol formation (15-19). However, while this approach yields valuable information, the technique is time-consuming and is prone to sampling artifacts (20, 21). Furthermore, many of the polar and labile compounds formed are not readily amenable to gas chromatography without prior derivatization (17-19). The technique also does not yield the real-time information necessary to follow aerosol formation processes in detail. Therefore, although much is known about the gasphase kinetics of the initial reactions of VOCs and a number of studies have provided information on the volatile products and mechanisms of these reactions (refs 8-10 and references therein), little is known about the identity of the gaseous organic products that undergo nucleation or condensation to form aerosol. As part of a research program focusing on studies of the chemistry of gas-to-particle conversion, we have recently developed a new instrument for real-time particle chemical analysis that should help to provide some of the needed compositional information on secondary organic aerosols. We have demonstrated that this instrument, which we refer to as a thermal desorption particle beam mass spectrometer (TDPBMS), can be used for real-time, quantitative analysis of the components of organic particles, at least within the ∼0.02-0.5 µm size range (22). We have also developed a temperature-programmed TDPBMS technique (TPTD) to aid in compound identification (23). Here we present the results of our first application of TDPBMS to detailed studies of the chemistry of secondary aerosol formation, in which we have analyzed the composition of aerosol particles formed in environmental chamber reactions of 1-tetradecene and O3 in dry air in the presence of excess alcohols and carboxylic acids. Elsewhere in this issue, we describe results of similar studies carried out in humid and dry air in the absence of added acidic organics (24). Gas-phase ozonolysis of ethene in the presence of alcohols, carboxylic acids, and water has been investigated VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2105

FIGURE 1. Schematic diagram of the thermal desorption particle beam mass spectrometer (TDPBMS) and associated apparatus used for aerosol studies. recently using infrared spectroscopy (25-27) and highperformance liquid chromatography (28). It was found that the major products included R-alkoxymethyl and R-acyloxymethyl hydroperoxides that were thought to be formed through reactions of the alcohols and carboxylic acids with stabilized Criegee biradicals, which is the generally accepted mechanism for the more thoroughly studied liquid-phase alkene-O3 reactions (29). In our study, we investigate the importance of these reactions in aerosol formation from ozonolysis of normal alkenes, using 1-tetradecene as a surrogate for this class of compounds because its relatively high molecular weight and terminal double-bond enhance aerosol formation. Although the amount of atmospheric aerosol formed from normal alkenes is apparently relatively small (30), these compounds provide a good starting point for understanding the chemical mechanisms by which alkenes in general participate in atmospheric aerosol nucleation and growth through reactions with O3. The gas- and liquid-phase chemistry of normal alkenes have been more thoroughly studied than that of the cyclic alkenes (especially those of biogenic origin), and as shown here, the liquidphase reactions of normal alkenes provide a simple and efficient means for generating standard hydroperoxide compounds for use in mass spectral identification of environmental chamber aerosol products. The particulate organic hydroperoxides and peroxides formed in these reactions can act as oxidants and are therefore of current interest because of their potential for impacting human health (31).

Experimental Section Materials. Fine chemicals were obtained from Aldrich Chemical, Inc., and were used without further purification. All solvents were HPLC grade, obtained from Fisher. Particle Mass Spectrometric Analysis by TDPBMS and TPTD. A schematic of the TDPBMS is shown in Figure 1. Detailed descriptions of the instrument and its operation for real-time analysis (22) and for TPTD (23) are presented elsewhere. Aerosol is sampled into the TDPBMS through a 100-µm orifice, which maintains the flow at 0.075 L/min and reduces the pressure from atmospheric to 2 Torr. Particles 2106

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 11, 2000

then enter a tube containing a series of aerodynamic lenses (32, 33) that focus the particles into a very narrow, lowdivergence particle beam that transports ∼0.02-0.5-µm particles from atmospheric pressure into the high-vacuum chamber with near-unit efficiency. After exiting the nozzle, the particles pass through two flat-plate skimmers separating three differentially pumped chambers and enter the detection chamber where the pressure is 5 × 10-8 Torr. The vacuum is maintained by turbomolecular pumps that are mounted on each chamber and backed by an oil-free mechanical pump to reduce contaminating organic vapors in the system. Inside the detection chamber, the particles impact on the walls of a V-shaped molybdenum foil that is either resistively heated continuously at 165 ( 3 °C for real-time TDPBMS analysis or cooled to -50 °C by an external liquid nitrogen bath for collection of particles for TPTD analysis. The vaporizer temperature is monitored by an attached thermocouple and during real-time analysis is regulated by a temperature controller. After vaporization, the molecules diffuse into an ionizer where they are impacted by 70-eV electrons, and the resulting ions are mass analyzed in a quadrupole mass spectrometer (Extrel MEXM 500, 1-500 amu mass range) equipped with a conversion dynode/pulse counting detector. Particle analysis by TPTD was carried out on ∼1 µg of aerosol collected in ∼30 min on the cryogenically cooled vaporizer. The sample was desorbed by heating at a ramp rate of ∼1 °C/min for ∼2 h, while mass spectra were continuously recorded. During TPTD, the aerosol components desorb according to their vapor pressures, so mass spectra of individual compounds can be extracted from the time-dependent mass spectra. Although this technique is primarily used to identify major aerosol components, we have demonstrated (23) that for sufficiently large differences in vapor pressures it is possible to obtain mass spectra for compounds present at concentrations at least an order of magnitude less than those of the major components. Because particles are exposed to subsaturated air when sampling from the DMA (calibration particles) and within the TDPBMS vacuum (calibration and chamber particles), volatile compounds may evaporate prior to analysis. Calculations and experiments with compounds of known vapor

pressures (22) indicate that significant evaporation can occur for calibration particles with vapor pressures greater than ∼10-5 Torr but that environmental chamber particles probably need higher vapor pressures since they are exposed to subsaturated conditions for a shorter period during sampling. This is not expected to pose a serious problem for the TDPBMS technique, however, since gas-particle partitioning calculations and measurements (34) indicate that in the ambient atmosphere compounds with vapor pressures greater than ∼10-5 Torr will be present primarily (>90%) in the gas phase. A greater fraction of semivolatile compounds can reside in particles at the higher mass concentrations obtained in environmental chamber experiments, but loss of these compounds during sampling will not lead to artifacts regarding the identity of compounds that would actually reside in ambient particles. Liquid-Phase Ozonolysis of 1-Tetradecene. Liquid-phase ozonolyses of 1-tetradecene with excess alcohols and carboxylic acids were carried out by dissolving 20 µL of 1-tetradecene (92% purity) in either 15 mL of methanol or 2-propanol or 15 mL of acetone containing 1 mL of formic, acetic, heptanoic, or nonanoic acids. Acetone was used as a solvent because it is nonreactive and dissolves formic and acetic acid in sufficient quantities for these experiments. Results obtained using cyclohexane for the heptanoic and nonanoic acid experiments were the same as those obtained with acetone, so there appears to be no effect of acetone on the low-volatility reaction products. Ozone from a Welsbach T-408 O3 generator (∼2% O3/O2) was then bubbled through the solutions at ∼0.15 L/min for 60 s, which was sufficiently long to ozonize nearly all the alkene, without adding excess O3. This method should produce R-alkoxyalkyl and R-acyloxyalkyl hydroperoxides in nearly quantitative yields (35-37). All reacted solutions were atomized in a Collison atomizer using dry, clean air as the carrier gas. Aerosol from the atomizer passes through diffusion driers containing activated charcoal to adsorb the alcohol or acetone solvent, excess carboxylic acids, unreacted 1-tetradecene, and volatile reaction products that evaporate from the particles. The particles are then charged to near-Boltzmann equilibrium (38) as they pass through a radioactive bipolar charger containing 210Po, and the polydisperse, charged aerosol enters a differential mobility analyzer (DMA) (39) for selection of a monodisperse aerosol, which flows into the TDPBMS. The monodisperse particles we used had diameters of 0.2 µm. Gas-Phase Ozonolysis of 1-Tetradecene. The chemistry of secondary organic aerosol formation from reactions of 1-tetradecene and O3 in the presence of either excess alcohol or carboxylic acid was investigated in a series of environmental chamber experiments. Aerosol was generated by reacting 0.2-1 ppmv of 1-tetradecene with 1.5 ppmv of O3 and added alcohol or carboxylic acid in a 7000-L Teflon bag filled with dry, clean air (Aadco pure air generator, 95% of the OH radicals formed in the alkene-O3 reaction (40). For the alcohol reactions, sufficient alcohol was present to serve as a scavenger. The 1-tetradecene, alcohols, carboxylic acids, and cyclohexane were added by evaporating the gently heated liquid from a glass bulb into a flowing clean air stream. In two experiments, ∼7 ppmv of tridecanal was added to the chamber after 4 h, which was long after the ozonolysis reaction was complete (the lifetime of 1-tetradecene due to reaction with 1.5 ppmv of O3 is

calculated to be ∼50 min, using a 1-decene rate constant of 9.3 × 10-18 cm3 molecule-1 s-1 from ref 10). In one experiment, the tridecanal was added by evaporating the heated liquid into a clean air stream, and in the other it was added as aerosol and vapor generated by atomizing a tridecanal/ methanol solution. The particles rapidly evaporate because of the relatively high vapor pressure of tridecanal (∼5 × 10-3 Torr). The latter technique was preferable because much of the low-volatility tridecanoic acid residue that is always present as a contaminant in the aldehyde was removed from the air stream when solution droplets impacted and dried on the walls of the atomizer. In two experiments, ∼10 ppmv of formaldehyde was added to the chamber prior to the start of the ozonolysis reaction. Formaldehyde was obtained by evaporating paraformaldehyde into a glass bulb at a measured pressure and then flushing it into the chamber using clean air. Ozone was added to the chamber last by flowing clean air through a 0.5-L bulb containing ∼2% O3/O2. During all chemical additions, a fan was run to mix the chamber and was then turned off. Aerosol formed by homogeneous nucleation, usually a few minutes after adding O3, and was sampled directly into the TDPBMS through stainless steel tubing inserted into a port in the chamber wall. Particles were either analyzed in real time or by TPTD. It is possible to obtain size-dependent composition information by sampling the aerosol through a DMA and then into the TDPBMS, but this was not done here. In some experiments, aerosol size distributions were measured using a scanning electrical mobility spectrometer (22, 41). Particle concentrations after addition of 1-tetradecene were less than 10/cm3 and then typically reached ∼104-106/cm3 a few minutes after addition of O3. Within about 1 h, relatively constant size distributions were achieved, with average particle diameters of ∼0.2-0.4 µm and mass concentrations of ∼500-2000 µg/m3. Aerosol wall losses were ∼20%/h over the 1-5-h experiments. Detection limits for real-time TDPBMS analysis are ∼0.1-1 µg/m3, so the amount of aerosol formed was much more than is needed for those measurements. The high concentrations were used to reduce sampling times for TPTD and to obtain high signal-to-noise, especially for detection of minor components. Background contributions to mass spectra were negligible during real-time analysis except for m/z 28, 32, 40, and 44 (N2+, O2+, Ar+, and CO2+, respectively), and so only contributions from these masses were subtracted from the mass spectra. Ozone concentrations were measured by drawing chamber air through Teflon tubing into a Dasibi 1003-AH O3 analyzer. After each experiment, the chamber was pumped out and then refilled and flushed until the following day or longer (>10 chamber volumes). In a few cases, it took more than a day of flushing to reduce added carboxylic acid concentrations to levels that did not interfere with subsequent experiments, apparently because of their propensity for sticking to the chamber walls and their high reactivity with stabilized biradicals. Occasionally, particles of DOS (dioctyl sebacate, a low-volatility liquid organic ester) were added to the flushed chamber to ascertain the presence of low-volatility contaminant vapors that might partition into aerosol during chamber studies, but no contaminants were observed.

Results and Discussion Mechanism of 1-Tetradecene Ozonolysis. The gas- (10) and liquid-phase (29) reactions of alkenes with O3 have been widely studied, and the general reaction mechanisms are reasonably well-established and apparently similar in the two phases. Of particular relevance to the work presented here are recent investigations of gas-phase ozonolysis of ethene (25-28) and liquid-phase reactions of normal alkenes (35-37) carried out in the presence of alcohols and carboxylic acids. Ozonolysis of 1-tetradecene [CH3(CH2)11CHdCH2] in VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2107

the presence of the alcohols and carboxylic acids used here apparently occurs by the following pathways:

The initial step in the reaction involves addition of O3 to the alkene double bond, resulting in the formation of a 1,2,3trioxolane (primary ozonide). The primary ozonide contains a significant amount of internal energy and rapidly decomposes by cleavage of the terminal C-C bond and one of the O-O bonds to form, by pathway 1a, tridecanal [CH3(CH2)11CHO] and a small excited Criegee biradical [C˙ H2OO˙ ]*, and by pathway 1b, formaldehyde [HCHO] and a large excited Criegee biradical [CH3(CH2)11C˙ HOO˙ ]*. In the gas phase, excited biradicals can undergo a number of possible unimolecular reactions that lead to a variety of products (reactions 2 and 5), which for the large biradical include tridecanoic acid [CH3(CH2)11C(O)OH], dodecane, and hydroxycarbonyls generated through a channel that also leads to OH production. Of the potential products of unimolecular biradical reactions, tridecanoic acid and some hydroxycarbonyls may have sufficiently low vapor pressures to partition significantly into aerosol. Excited biradicals can also be stabilized by collisions with other molecules (e.g., N2 or O2 for reactions in air and solvent molecules for solution reactions) (reactions 3 and 6). In the liquid phase, collisional stabilization is apparently the only fate of the excited biradicals because of the high rate of collisions with solvent molecules. The large stabilized biradical CH3(CH2)11C˙ HOO˙ can undergo reactions with species including SO2, CO, NO2, and aldehydes (42), but in the presence of sufficiently high concentrations of compounds containing acidic hydrogens (H-OG), such as alcohols (G ) R ) alkyl group) and carboxylic acids (G ) C(O)R ) acyl group), the primary products are expected to be R-alkoxytridecyl and R-acyloxytridecyl hydroperoxides [CH3(CH2)11CH(OG)OOH] (reaction 4) (26, 29). Unless the alcohol or carboxylic acid is relatively large, reactions of the small stabilized biradical C˙ H2OO˙ lead to products too volatile to form aerosol (reaction 7). Hydroperoxide Standards for TDPBMS Mass Spectral Identification of 1-Tetradecene Ozonolysis Products. To identify hydroperoxides in environmental chamber aerosol by TDPBMS, it was necessary to have mass spectra of hydroperoxide standards that could be compared with the aerosol spectra. Because hydroperoxides are too unstable to be available from commercial suppliers, they were synthesized by liquid-phase ozonolysis using the methods described above. For normal alkenes, this procedure is known to give nearly quantitative yields of hydroperoxides (35-37), which are formed by the mechanism shown above. When the reacted solutions are atomized to form aerosol, the solvent, unreacted 1-tetradecene, alcohols, carboxylic acids, tridecanal, formaldehyde, and products of the reactions of the 2108

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 11, 2000

smaller biradical evaporate completely from the particles because of their high volatility, prior to TDPBMS analysis. TPTD analysis, which is demonstrated below for chamber aerosols, was performed on the aerosol products of the liquidphase ozonolysis of 1-tetradecene in excess methanol, 2-propanol, and heptanoic acid. The results indicated that only the hydroperoxide and a trace of tridecanoic acid were present in our hydroperoxide aerosol standards, therefore allowing us to use the real-time mass spectra of the atomized liquid-phase reaction products as standards for identification of R-alkoxytridecyl and R-acyloxytridecyl hydroperoxides formed in environmental chamber reactions. We did not attempt to prepare standards of the peroxyhemiacetals proposed to be formed in the chamber from reactions of hydroperoxides and aldehydes because of the time, effort, and difficulty involved in developing techniques for synthesis and purification of these potentially explosive compounds. Instead, the presence of peroxyhemiacetals in chamber aerosol is inferred from comparison of their mass spectra with the parent hydroperoxides, known hydroperoxidealdehyde chemistry, and controlled chamber reactions of hydroperoxides with added aldehydes. Aerosol Products Formed from Environmental Chamber Reactions of 1-Tetradecene and O3 in the Presence of Alcohols. Results of TPTD aerosol analyses can be displayed as “mass thermograms”, which are plots of mass spectral signal intensity vs vaporizer temperature for a particular mass to charge ratio, m/z (23). Figure 2 shows mass thermograms obtained from analyses of aerosol collected 0.5 and 4 h after the start of the reaction of 1-tetradecene and O3 in the presence of excess methanol [CH3OH]. The plot for m/z 213 at 0.5 h (Figure 2A) shows a large peak at ∼33 °C and a small peak at ∼69 °C. The two peaks correspond to two different compounds, both of which contain m/z 213 in their mass spectra. The higher vapor pressure compound desorbs at a lower temperature. After 4 h (Figure 2B), the intensity of the 69 °C peak increases dramatically, such that it is larger than the 33 °C peak. Comparison of this plot with the m/z 215 mass thermogram obtained after 4 h (Figure 2C) shows that whereas m/z 213 is significant in both compounds, m/z 215 is primarily characteristic of the low vapor pressure, 69 °C compound. The small shifts in the desorption temperatures of compounds analyzed at 0.5 and 4 h is not significant since the temperature depends slightly on the sample size and temperature ramp rate. Rather than viewing mass thermograms for hundreds of m/z values, it is more convenient to plot the temperatures at which intensity maxima occur (i.e., the “desorption temperatures”) in the mass thermograms obtained for each m/z. Since the intensity maxima are determined by taking derivatives of each mass thermogram, we refer to such a plot as a “differential mass thermogram”. In the differential mass thermogram shown in Figure 2D, which is for the aerosol sample collected after 0.5 h, bands corresponding to three compounds are observed at desorption temperatures of 18, 33, and 69 °C. The small shoulder on the front of the 33 °C peak in Figure 2A is part of the 18 °C band. The mass spectrum of each compound can be obtained by plotting the maximum signal intensity for each m/z in a band, which is defined by a particular temperature interval. The results for the 33 and 69 °C compounds are shown in Figure 3, panels B and C; the real-time mass spectrum of the R-methoxytridecyl hydroperoxide [CH3(CH2)11CH(OCH3)OOH] standard prepared by liquid-phase ozonolysis of 1-tetradecene in methanol is shown in Figure 3A. The real-time mass spectrum of the R-isopropoxytridecyl hydroperoxide [CH3(CH2)11CH(OCH(CH3)2)OOH] standard prepared by liquid-phase ozonolysis of 1-tetradecene in 2-propanol is shown in Figure 3D. Comparison of the mass spectrum of the 18 °C compound (not shown) with a mass spectral standard identifies it as a

FIGURE 2. TPTD mass thermograms of aerosol formed from ozonolysis of 1-tetradecene in air containing excess methanol. Mass thermograms for m/z 213 are for aerosol collected after (A) 0.5 and (B) 4 h, and the mass thermogram for (C) m/z 215 is for aerosol collected after 4 h. The differential mass thermogram (D) is for aerosol collected after 0.5 h. trace amount of tridecanoic acid. The spectrum of the 33 °C compound in Figure 3B matches that in Figure 3A and so is identified as R-methoxytridecyl hydroperoxide (abbreviated hereafter as MTHP). The large m/z 213 peak in the MTHP spectrum corresponds to the loss of HO2 from the molecular ion, the peak at m/z 77 is apparently due to CH(OCH3)OOH+, and the remaining ion series correspond to CnH2n-1+ and CnH2n+1+. These series and ions formed by loss of HO2 have been observed in mass spectra of alkyl hydroperoxides (43). The mass spectrum of R-isopropoxytridecyl hydroperoxide in Figure 3D is similar to that of MTHP in that m/z 241 corresponds to the loss of HO2 from the molecular ion, the peak at m/z 105 is apparently due to CH(OCH(CH3)2)OOH+, and the CnH2n-1+ and CnH2n+1+ series are prominent. In addition, a peak at m/z 199 suggests that hydroperoxides formed by reactions of alcohols larger than methanol exhibit loss of the dehydrated alcohol (e.g., CH2dCHCH3 for reaction with CH3CH(OH)CH3). The TPTD mass spectrum shown in Figure 3C for the compound that desorbs at 69 °C is similar to that of MTHP, although the peaks at m/z 169, 197, 199, 215, and 229 are significantly more intense in the spectrum of the 69 °C compound. Most of this compound appears after the

FIGURE 3. Aerosol TDPBMS mass spectra of products formed from ozonolysis of 1-tetradecene in air or in a solution containing excess methanol or 2-propanol. (A) Real-time mass spectrum of r-methoxytridecyl hydroperoxide (MTHP) from methanol solution reaction. (B) TPTD mass spectrum of 33 °C compound in Figure 2D, r-methoxytridecyl hydroperoxide (MTHP), from methanol/air reaction. (C) TPTD mass spectrum of 69 °C compound in Figure 2D, r-methoxy-r′-hydroxyditridecyl peroxide (MHDTP), from methanol/ air reaction. (D) Real-time mass spectrum of r-isopropoxytridecyl hydroperoxide from 2-propanol solution reaction. 1-tetradecene ozonolysis reaction is essentially complete (1-tetradecene lifetime ∼50 min; see Experimental Section) and so is apparently formed by a secondary reaction of the hydroperoxide and another ozonolysis product. We believe this low vapor pressure compound is a peroxyhemiacetal, R-methoxy-R′-hydroxyditridecyl peroxide [CH3(CH2)11CH(OCH3)OOCH(OH)(CH2)11CH3] (abbreviated hereafter as MHDTP), formed by reaction of MTHP with tridecanal.

This type of reaction is known to occur in solution, and the peroxyhemiacetal products are stable enough for purification by vacuum distillation and storage (44-47). This is in contrast to hemiacetals formed from the analogous reactions of alcohols and aldehydes, which have been identified in solution but are in most cases too unstable to isolate (48). The m/z 169, 197, 199, 215, and 229 ions probably correspond to CH3(CH2)11+, CH3(CH2)11CO+, CH3(CH2)11CHOH+, CH3(CH2)11C(OH)2+, and CH3(CH2)11CH(OCH3)O+, respectively. The large m/z 213 peak in the hydroperoxide and peroxide spectra is due to CH3(CH2)11CH(OCH3)+. Fragmentation pathways leading to similar ions have been observed in dialkyl peroxides (49) and are in accord with established mechanisms (50). VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2109

FIGURE 4. Real-time TDPBMS mass spectra of (A-C) aerosol formed from ozonolysis of 1-tetradecene in air containing excess methanol and (D) a tridecanal aerosol standard. Spectra were taken (A) 50 min and (B) 4.5 h after starting the reaction and (C) 50 min after addition of ∼7 ppmv of tridecanal (5.5 h after starting the reaction). In panels A-C, mass spectra were normalized to the intensity of the m/z 213 peak, and then peaks with m/z e 200 and m/z g 215 were multiplied by 13. The m/z 214 peak is primarily an isotope peak of m/z 213 and so was not multiplied by 13. To gain additional evidence that the low vapor pressure product we observe is the peroxyhemiacetal formed by reaction 8, we carried out the same ozonolysis reaction in excess methanol, but added ∼7 ppmv of tridecanal to the chamber after 5 h. This was ∼25 times the amount of tridecanal formed in the ozonolysis reaction and was added well after the ozonolysis reaction was complete. The realtime mass spectra taken 50 min and 4.5 h (just prior to addition of tridecanal) after the start of the reaction and 50 min after addition of tridecanal are shown in Figure 4. Also shown is a mass spectrum of a tridecanal standard. The m/z 197 and 199 peaks increase slightly in intensity over 4.5 h (Figure 4A,B) and then nearly double after the addition of tridecanal (Figure 4C). The change is not dramatic because these are real-time spectra and the signal is dominated by the presence of MTHP. The m/z 169, 215, and 229 peaks change very little after addition of tridecanal, even though their intensities in the TPTD mass spectrum of the low vapor pressure product (Figure 3C) are comparable to those of m/z 197 and 199. The reason for this is that the TPTD spectrum is obtained at ∼69 °C, and the real-time spectrum is at 165 °C. Higher temperatures lead to more fragmentation of the m/z 169, 215, and 229 ions and a corresponding decrease in their intensities, as we have found by real-time analysis of the aerosol at vaporization temperatures lower than 165 °C (data not shown). By contrast, higher temperatures increase the intensity of m/z 171, leading to an enhanced abundance of this peak in the real-time spectrum (Figure 4C) as compared to the TPTD spectrum (Figure 3C). The increase in the 2110

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 11, 2000

FIGURE 5. TPTD (A) differential mass thermogram and (B) r-methoxytridecyl-r′-hydroxymethyl peroxide (MTHMP) mass spectrum for aerosol formed from ozonolysis of 1-tetradecene in air containing excess methanol and ∼10 ppmv of formaldehyde. intensities of the m/z 197 and 199 peaks is not due to nonreactive absorption of tridecanal into the particles, as can be seen by comparing the tridecanal mass spectrum in Figure 4D with the mass spectrum in Figure 4C. These results all support the hypothesis that the low vapor pressure product is MHDTP formed by reaction of MTHP with tridecanal. Identification of MHDTP led us to wonder why we do not also detect R-methoxytridecyl-R′-hydroxymethyl peroxide [CH3(CH2)11CH(OCH3)OOCH2OH] (hereafter abbreviated MTHMP) formed by reaction of MTHP with formaldehyde, which is produced during ozonolysis by reaction 1b in concentrations similar to tridecanal (51). To investigate this question, we repeated the chamber reaction with methanol but added ∼10 ppmv of formaldehyde to the chamber prior to starting the reaction. This concentration was ∼30 times higher than the ∼0.3 ppmv concentrations of formaldehyde and tridecanal expected to be formed during ozonolysis. The differential mass thermogram and mass spectra obtained by TPTD of aerosol collected after 4 h of reaction are shown in Figure 5. Two bands appear in the differential mass thermogram (Figure 5A), one at ∼29 °C and one at ∼50 °C, while the band seen previously at 69 °C has disappeared. The mass spectrum of the new 50 °C product (Figure 5B) is similar to that of MHDTP, differing primarily in the relative intensities of some of the major peaks. One important difference is the absence of a peak at m/z 169, which is relatively intense in MHDTP but not in MTHP. This is consistent with formation of MTHMP, since it differs from MHDTP only in that one of the two CH3(CH2)11 groups (mass 169) has been replaced by H. The 50 °C desorption temperature for the formaldehyde reaction product, which is halfway between that of MTHP (29 °C) and the tridecanal reaction product (69 °C), is higher than we originally expected. Since the molecular weight of formaldehyde is only 15% of that of tridecanal, it would seem that the temperature shift should be closer to 6 °C (i.e., 15% of 40 °C) instead of 20 °C. However, the additional decrease in vapor pressure, which results in the higher desorption

FIGURE 6. Aerosol TDPBMS mass spectra of products formed from ozonolysis of 1-tetradecene in air or acetone solution containing excess acetic acid. (A) Real-time mass spectrum of r-hydroperoxytridecyl acetate from acetic acid/acetone solution reaction. TPTD mass spectra of (B) r-hydroperoxytridecyl acetate and (C) r-acetyloxy-r′-hydroxyditridecyl peroxide from acetic acid/air reaction. temperature, is probably due to the terminal position of the hydroxyl group in MTHMP as compared to its central location in MHDTP. By comparison, the temperature required to achieve a pressure of 1 Torr of 1-hexanol and 3-hexanol is 24.5 and 2.5 °C, respectively (52). Results of experiments with excess 2-propanol (not shown) were similar to those with methanol in that the differential mass thermogram contained two compounds with similar mass spectra and the lower vapor pressure compound increased in abundance over time. The higher vapor pressure compound can be identified as R-isopropoxytridecyl hydroperoxide [CH3(CH2)11CH(OCH(CH3)2)OOH] by comparing the mass spectrum with that of the liquid-phase hydroperoxide standard (Figure 3D), and the lower vapor pressure species is then apparently R-isopropoxy-R′-hydroxyditridecyl peroxide [CH3(CH2)11CH(OCH(CH3)2)OOCH(OH)(CH2)11CH3]. Aerosol Products Formed from Environmental Chamber Reactions of 1-Tetradecene and O3 in the Presence of Carboxylic Acids. We also analyzed aerosol formed from ozonolysis of 1-tetradecene in the presence of excess formic [CH(O)OH], acetic [CH3C(O)OH], heptanoic [CH3(CH2)5C(O)OH], and nonanoic [CH3(CH2)7C(O)OH] acids and obtained results similar to those for reactions with alcohols. A nonreactive acetone solvent was used for the liquid-phase reactions, and a high concentration of cyclohexane was added to the chamber to scavenge OH radicals. Cyclohexane and its OH reaction products do not participate in aerosol formation since cyclohexane is nonreactive and the products have concentrations too low to compete with other compounds in reactions with stabilized biradicals. In all cases, two compounds were present in the differential mass thermograms (in addition to a trace of tridecanoic acid) with the lowest vapor pressure compound increasing in abundance over 4 h. Comparison of the mass spectra with those of hydroperoxide standards indicates that the higher vapor pressure compounds in each pair are the R-acyloxytridecyl hydroperoxides [CH3(CH2)11CH(OG)OOH, G ) CHO, C(O)(CH2)xCH3 (x ) 0, 5, 7) for formic, acetic, heptanoic, and

FIGURE 7. Aerosol TDPBMS mass spectra of products formed from ozonolysis of 1-tetradecene in air or acetone solution containing excess heptanoic acid. (A) Real-time mass spectrum of r-hydroperoxytridecyl heptanoate from heptanoic acid/acetone solution reaction. TPTD mass spectra of (B) r-hydroperoxytridecyl heptanoate and (C) r-heptyloxy-r′-hydroxyditridecyl peroxide from heptanoic acid/air reaction. nonanoic acids] formed by reaction 4, and the lower vapor pressure compounds are apparently the corresponding R-acyloxy-R′-hydroxyditridecyl peroxides [CH3(CH2)11CH(OG)OOCH(OH)(CH2)11CH3] formed by reaction of the hydroperoxides with tridecanal. Addition of ∼10 ppmv of formaldehyde prior to starting the ozonolysis reaction in excess acetic acid led to results similar to those observed for the corresponding reaction in methanol. The lowest vapor pressure compound, R-acetyloxy-R′-hydroxyditridecyl peroxide, disappeared from the differential mass thermogram and was replaced by R-acetyloxytridecyl-R′-hydroxymethyl peroxide, which desorbed at a temperature (61 °C) midway between R-acetyloxytridecyl hydroperoxide (45 °C) and R-acetyloxy-R′-hydroxyditridecyl peroxide (79 °C). Mass spectra are shown in Figures 6 and 7 for the aerosol products of the gas-phase ozonolysis reactions with acetic and heptanoic acid and for synthesized hydroperoxide standards. The mass spectra differ slightly from those of the alcohol reaction products. The major peak in the hydroperoxide and corresponding peroxyhemiacetal spectra is due to a very stable acylium ion (50), which for the products of the acetic acid reaction is CH3CO+ (m/z 43) and for reaction with heptanoic acid is CH3(CH2)5CO+ (m/z 113). Each spectrum also has a peak corresponding to the addition of H2O to the acylium ion and prominent peaks at m/z 169, 197, 199, and 215, which are noticeably larger in the peroxides than in the hydroperoxides and larger in the R-acyloxy-R′hydroxyditridecyl peroxides than in the R-alkoxy-R′-hydroxyditridecyl peroxides. Unlike the mass spectra of R-alkoxytridecyl hydroperoxides and R-alkoxy-R′-hydroxyditridecyl peroxides, which exhibit large peaks corresponding to loss of HO2 and CH3(CH2)11CH(OH)OO from the hydroperoxide and peroxide molecular ions, respectively (e.g., m/z 213 in Figure 3), the corresponding peak is quite small in the R-acyloxytridecyl hydroperoxides and almost absent in the R-acyloxy-R′-hydroxyditridecyl peroxide spectra. The VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2111

TABLE 1. Characteristic Ion Fragmentation Pathways Observed for Electron Ionization of Alkoxytridecyl and Acyloxytridecyl Hydroperoxides Alkoxytridecyl Hydroperoxidesa CH3(CH2)11CH(OG)OOH + e- f CH3(CH2)11CH(OG)+ + HO2 + 2ef CH(OG)OOH+ + CH3(CH2)11 + 2ef CH3(CH2)11CHOH+ + HO2 + (G-H) + 2e- (G > CH3) f CnH2n-1+ + neutral fragments + 2ef CnH2n+1+ + neutral fragments + 2eAcyloxytridecyl Hydroperoxidesb CH3(CH2)11CH(OG)OOH + e- f CH3(CH2)11CH(OG)+ + HO2 + 2ef CH3(CH2)11C(OH)2+ + neutral fragments + 2ef CH3(CH2)11CO+ + neutral fragments + 2ef GOH2+ + neutral fragments + 2ef G+ + neutral fragments + 2ef CnH2n-1+ + neutral fragments + 2ef CnH2n+1+ + neutral fragments + 2ea

G ) CH3, CH(CH3)2.

b

G ) CHO, C(O)(CH2)xCH3 (x ) 0, 5, 7).

TABLE 2. Properties of Aerosol Products from Gas-Phase Ozonolysis of 1-Tetradecene in Excess Alcohol or Carboxylic Acid alcohol or acid reactant methanol 2-propanol formic acid acetic acid heptanoic acid nonanoic acid

aerosol reaction product

desorption temp (°C)

estd 25 °C vapor pressure (Torr)a

R-methoxytridecyl hydroperoxide R-methoxytridecyl-R-hydroxymethyl peroxide R-methoxy-R′-hydroxyditridecyl peroxide R-isopropoxytridecyl hydroperoxide R-isopropoxy-R′-hydroxyditridecyl peroxide R-hydroperoxytridecyl formate R-formyloxy-R′-hydroxyditridecyl peroxide R-hydroperoxytridecyl acetate R-acetyloxytridecyl-R′-hydroxymethyl peroxide R-acetyloxy-R′-hydroxyditridecyl peroxide R-hydroperoxytridecyl heptanoate R-heptyloxy-R′-hydroxyditridecyl peroxide R-hydroperoxytridecyl nonanoate R-nonyloxy-R′-hydroxyditridecyl peroxide

33 50 69 34 70 43 77 45 61 79 61 84 66 88

3 × 10-7 8 × 10-10 2 × 10-12 2 × 10-7 2 × 10-12 8 × 10-9 2 × 10-13 4 × 10-9 2 × 10-11 1 × 10-13 2 × 10-11 3 × 10-14 5 × 10-12 1 × 10-14

a

Calculated using correlation in Figure 8 and measured peak desorption temperatures. averages of the m/z 97 and 111 desorption peak areas.

remainder of the spectra are composed primarily of low mass peaks in the CnH2n-1+ and CnH2n+1+ series. The characteristic fragmentation pathways observed for the R-alkoxytridecyl and R-acyloxytridecyl hydroperoxides studied here are summarized in Table 1. They can be readily explained using established mechanisms (50). The same peaks are prominent in the mass spectra of the peroxyhemiacetals formed by reactions with tridecanal, with the major difference being increased relative abundances of m/z 169, 197, 199, and 215. Desorption temperatures and vapor pressures of the compounds identified in this study are listed in Table 2. Vapor pressures were estimated using the vapor pressure-desorption temperature correlation shown in Figure 8, which was generated from previous (23) and unpublished TPTD measurements on organic compounds for which 25 °C vapor pressure data are available (52-54). On the basis of uncertainties in desorption temperatures and scatter in the correlation, the values are probably accurate to within about an order of magnitude. Also shown in Table 2 are estimated concentration ratios of [peroxide]/[hydroperoxide] measured 0.5 and 4 h after the start of the reaction. Values were calculated from the areas of m/z 97 and 111 peaks, which are major peaks common to both hydroperoxides and peroxides. Nature of the Hydroperoxide + Aldehyde f Peroxyhemiacetal Reaction: Gas vs Particle Phase. An important question concerning the reaction by which peroxyhemiacetals are formed from hydroperoxides and aldehydes is whether the reaction occurs in the gas phase, on the particle 2112

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 11, 2000

b

[peroxide]/[hydroperoxide]b 0.5 h 4h

0.22

2.0

0.27

2.5

0.20

1.0

0.15

0.21

0.15

0.25

0.13

0.20

Concentrations determined semiquantiatively using

FIGURE 8. Correlation between compound vapor pressures calculated at 25 °C and peak desorption temperatures measured by TPTD. Data were taken from previous (23) and unpublished TPTD measurements made on monocarboxylic acids (nonanoic, decanoic, tridecanoic, myristic, palmitic, and stearic acids), dicarboxylic acids (glutaric, adipic, and suberic acids), and dioctyl sebacate for which 25 °C vapor pressures are available (52-54).

surface, or in the particle interior. To the best of our knowledge, the answer to this question is not known. In one laboratory study of the ozonolysis of R-pinene in the presence of water vapor (55), hydroxymethyl hydroperoxide [CH2(OH)OOH] and bis(hydroxymethyl) peroxide [CH2(OH)OOCH2OH] were identified in products deposited on the walls of a reaction vessel. The hydroperoxide apparently formed by reaction of the stabilized Criegee biradical CHOO with H2O, similar to reaction 7, and the peroxide by subsequent reaction of the hydroperoxide with formaldehyde similar to reaction 8, but it was not possible to determine where the reaction occurred. Although we also cannot determine the location of the reaction in our experiments, the results indicate that it most probably takes place on particle surfaces. Measurements of the yields of aldehydes from ozonolysis of alkenes (51) indicate that the amounts of tridecanal and formaldehyde formed from the 1-tetradecene reaction should be nearly the same. And hydroperoxides would be expected to react faster in the gas phase with formaldehyde than with tridecanal because of less steric hindrance. Therefore, the major peroxide product of a gas-phase reaction of MTHP and these two aldehydes should be MTHMP rather than the observed MHDTP. The absence of the formaldehyde reaction product in the aerosol cannot be due to its having a higher vapor pressure than the tridecanal reaction product, since gasparticle partitioning calculations (22, 34, 56) indicate that a compound with a vapor pressure of ∼10-9 Torr (estimated value in Table 2) should essentially be present entirely in the particle phase. The observations could be explained by a reaction occurring in the particle interior, since the rate would then depend on the particle-phase concentrations of tridecanal and formaldehyde. According to gas-particle partitioning theory, for vapor pressures of ∼4000 and ∼5 × 10-3 Torr for formaldehyde and tridecanal, respectively (52), and a gas-phase [formaldehyde]/[tridecanal] concentration ratio ∼1, the concentration ratio in the particles should be ∼10-6, which would explain the dominance of MHDTP as a product. However, this mechanism cannot explain the complete disappearance of MHDTP from the aerosol when the formaldehyde concentration is increased by a factor of ∼30. This is because even with a gas-phase [formaldehyde]/ [tridecanal] concentration ratio of 30 instead of 1, the ∼106 times higher vapor pressure of formaldehyde should still lead to a concentration ratio in the particles less than 10-4. To explain the results observed with and without added formaldehyde, we are left to conclude that the reaction occurs at the particle surface. The preferential formation of MTHMP at high formaldehyde concentrations could then be due to increased collisions with surface hydroperoxides, and the preferential formation of MHDTP at approximately equal concentrations of formaldehyde and tridecanal could be attributed to the longer lifetime of tridecanal on the surface because of its lower vapor pressure. Such a mechanism is common in surface chemistry (57) and could reduce the relatively large energy barrier that would be expected for a gas-phase reaction between two stable, closed-shell molecules, perhaps through acid catalysis by the acidic hydroperoxide (44). Additional insight into the peroxide formation reaction can be gained from data in Table 2, which show that the [peroxide]/[hydroperoxide] concentration ratios for the reaction of tridecanal with hydroperoxides formed from reactions with methanol, 2-propanol, and formic acid increase by a factor of ∼5-10 over 4 h, compared to an increase of only ∼50% for hydroperoxides formed from reactions with acetic, heptanoic, and nonanoic acids. This dramatic difference in reaction rates may indicate that the first three hydroperoxides form liquid drops that can react completely with tridecanal by diffusion of interior molecules to the surface, whereas the last three hydroperoxides form

solid particles whose interior molecules do not encounter tridecanal because of diffusion limitations. Although we do not know the melting points of these compounds and have not attempted to purify them because of their potential explosive nature, literature data (44) for the melting points of alkyl hydroperoxides [e.g., CH3(CH2)12CH2OOH, mp ∼30 °C] make this explanation plausible.

Implications for Atmospheric Aerosol Chemistry The results of this study show that the aerosol products formed initially from reactions of 1-tetradecene and O3 in the presence of excess alcohols and carboxylic acids are almost exclusively R-alkoxytridecyl and R-acyloxytridecyl hydroperoxides and that the formation of organic hydroperoxides in the gas and liquid phases is consistent with a mechanism involving reactions of the alcohols and carboxylic acids with stabilized Criegee biradicals generated in the alkene-O3 reaction. These experiments were performed in dry air containing high concentrations of added alcohols and carboxylic acids, but in the ambient atmosphere these species must compete with water vapor in reactions with stabilized biradicals. Although water reacts much more slowly than acidic organics (26, 27), especially carboxylic acids (∼kwater/kacid ∼ 10-4), the much higher ambient concentration of water vapor ([water]/[acid] ∼ 107) is expected to result in the formation of primarily R-hydroxyalkyl hydroperoxides and ∼2-3 orders of magnitude lower concentrations of R-acyloxyalkyl hydroperoxides. Hydroperoxides [including those formed in low NOx environments by OH and NO3 oxidation of hydrocarbons (10)] can subsequently undergo reactions with aldehydes and ketones, apparently on particle surfaces, to form relatively stable peroxyhemiacetals. These stabilized biradical reaction pathways therefore provide a means by which alkenes can be converted into low-volatility compounds that are much more likely than the parent compounds to accumulate in particulate matter. For example, ozonolysis of 1-tetradecene yields tridecanal and tridecanoic acid (vapor pressures ∼3 × 10-2, 5 × 10-3, and 7 × 10-6 Torr, respectively; refs 52 and 53) by reactions not involving stabilized biradicals, whereas the vapor pressures of the hydroperoxide products identified here (Table 2) are lower than that of tridecanoic acid by factors of ∼10-106, and peroxyhemiacetal formation reduces vapor pressures by an additional factor of ∼102-105. These reactions have a number of potentially important consequences. Compounds with vapor pressures less than ∼10-6 Torr partition predominantly into the particle phase and accumulate in the easily respired fine-particle mode (11, 12, 34), therefore providing a route by which low-volatility organic hydroperoxides and the corresponding peroxyhemiacetals can be transported and deposited onto deep-lung surfaces. Because of their oxidizing properties, these compounds are considered to be one of the possible causative agents for the adverse effects of PM2.5 on human health (31). However, since the major products of stabilized biradical reactions are expected to be R-hydroxyalkyl hydroperoxides, R-acyloxyalkyl hydroperoxides are unlikely to contribute significantly to the hydroperoxide particulate mass unless most of the R-hydroxyalkyl hydroperoxides are too volatile to form aerosol. Trace levels of low-volatility R-acyloxyalkyl hydroperoxides could however play a role in atmospheric nucleation. This process leads to the creation of new particles that affect aerosol concentrations and composition and, therefore, the population of cloud condensation nuclei, which impact cloud formation and global climate (1). Nucleation has been observed in terrestrial (58, 59) and marine (60) environments, and in some cases reactions involving biogenic organic compounds, including alkenes (e.g., monoterpenes; ref 59), may be involved. New particle formation thought to be due VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2113

to sulfuric acid nucleation has been observed at sulfuric acid concentrations of 10-6-10-3 ppbv ∼ 10-13-10-10 Torr (58, 60), demonstrating that nucleating species can be present in very low concentrations. The most important requirement is that compounds have low vapor pressures and sufficiently high production rates so that the supersaturation necessary for nucleation can be attained before the vapor is lost by condensation onto preexisting particles. The vapor pressure of sulfuric acid at typical ambient temperatures and relative humidities is on the order of 10-14-10-10 Torr (12). Although most stabilized biradicals formed by alkene ozonolysis will probably react with water vapor, formation of even very low concentrations of R-acyloxyalkyl hydroperoxides from reactions of relatively large biradicals and/or organic acids can lead to compounds having very low vapor pressures that are well below those of the corresponding R-hydroxyalkyl hydroperoxides. Because of the extreme sensitivity of homogeneous nucleation to compound vapor pressures, such low-volatility organic compounds could play a role in atmospheric nucleation. One recent modeling study (61) of secondary organic aerosol formation from ozonolysis of R-pinene in an environmental chamber indicated that nucleation involved a compound having a vapor pressure of ∼10-15 Torr or lower. The identity of the compound was not known, but it was suggested that it might be a secondary ozonide or an anhydride of an acyloxyalkyl hydroperoxide. Peroxyhemiacetals are unlikely to be important in nucleation since our results indicate that they are formed on particle surfaces rather than in the gas phase. However, this reaction could provide a means by which volatile hydroperoxides and aldehydes become incorporated into preexisting particles. Knowledge of these reactions is also important for interpreting the results of environmental chamber measurements of aerosol composition and yield, which are being used for modeling secondary aerosol formation from alkenes in the atmosphere. These experiments are usually performed with initial VOC concentrations on the order of ∼0.1-1 ppmv. Under these conditions, carboxylic acids and aldehydes formed during ozonolysis reach concentrations much higher than typical ambient levels and can potentially compete with water vapor for a significant fraction of stabilized biradicals, leading to aerosol products and yields that are not representative of normal atmospheric processes. Results of TDPBMS studies of the aerosol products of reactions of 1-tetradecene and O3 in humid and dry air, which covers the range of conditions representative of the ambient atmosphere, are described elsewhere in this issue (24). The compounds identified in those reactions differ from those observed here and provide further insight into the chemistry of secondary organic aerosol formation from alkene ozonolysis. As we also demonstrate, the results of these studies could not have been obtained using GC-MS analysis, since the major aerosol products decompose on a GC column.

Acknowledgments We thank the U.S. Environmental Protection Agency, Office of Research and Development [Assistance Agreement R826235-010, Science to Achieve Results (STAR) grant] for generously supporting this research. While this research has been supported by the U.S. Environmental Protection Agency, it has not been subjected to Agency review and, therefore, does not necessarily reflect the views of the Agency, and no official endorsement should be inferred. We also thank Roger Atkinson and Janet Arey for helpful discussions.

Literature Cited (1) Charlson, R. J.; Lovelock, J. E.; Andreae, M. O.; Warren, S. G. Nature 1987, 326, 655. (2) Charlson, R. J.; Schwartz, S. E.; Hales, J. M.; Cess, R. D.; Coakley, J. A.; Hansen, J. E.; Hofmann, D. J. Science 1992, 255, 423. 2114

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 11, 2000

(3) National Research Council (NRC). Protecting Visibility in National Parks and Wilderness Areas; National Academy Press: Washington, DC, 1993. (4) Ravishankara, A. R. Science 1997, 276, 1058. (5) Schwartz, J.; Dockery, D. W.; Neas, L. M. J. Air Waste Manage. Assoc. 1996, 46, 927. (6) Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Atmos. Environ. 1996, 30, 3837. (7) Turpin, B. J.; Huntzicker, J. J. Atmos. Environ. 1995, 29, 3527. (8) Atkinson, R. J. Phys. Chem. Ref. Data 1989, Monogr. 1, 1. (9) Atkinson, R. J. Phys. Chem. Ref. Data 1994, Monogr. 2, 1. (10) Atkinson, R. J. Phys. Chem. Ref. Data 1997, 26, 215. (11) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Chemistry of the Upper and Lower Atmosphere; Academic Press: San Diego, 1999. (12) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics; John Wiley & Sons: New York, 1998. (13) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 2580. (14) Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1997, 31, 1890. (15) Grosjean, D. In Ozone and Other Photochemical Oxidants; National Academy of Sciences: Washington, DC, 1977; Chapter 3, pp 45-125. (16) Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. Atmos. Environ. 1997, 31, 1953. (17) Yu, J.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1998, 32, 2357. (18) Jang, M.; Kamens, R. M. Atmos. Environ. 1998, 33, 459. (19) Yu, J.; Cocker, D. R.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. J. Atmos. Chem. 1999, 34, 207. (20) Ligocki, M. P.; Pankow, J. F. Environ. Sci. Technol. 1989, 23, 75. (21) McDow, S. R.; Huntzicker, J. J. Atmos. Environ. 1990, 24A, 2563. (22) Tobias, H. J.; Kooiman, P. M.; Docherty, K. S.; Ziemann, P. J. Aerosol Sci. Technol., in press. (23) Tobias, H. J.; Ziemann, P. J. Anal. Chem. 1999, 71, 3428. (24) Tobias, H. J.; Docherty, K. S.; Beving, D. E.; Ziemann, P. J. Environ. Sci. Technol. 2000, 34, 2116-2125. (25) Horie, O.; Neeb, P.; Limbach, S.; Moortgat, G. K. Geophys. Res. Lett. 1994, 21, 1523. (26) Neeb, P.; Horie, O.; Moortgat, G. K. Int. J. Chem. Kinet. 1996, 28, 721. (27) Neeb, P.; Horie, O.; Moortgat, G. K. J. Phys. Chem. A 1998, 102, 6778. (28) Wolff, S.; Boddenberg, A.; Thamm, J.; Turner, W. V.; Gab, S. Atmos. Environ. 1997, 2965. (29) Bailey, P. S. Ozonation In Organic Chemistry; Academic Press: New York, 1978; Vol. 1. (30) Pandis, S. N.; Harley, R. A.; Cass, G. R.; Seinfeld, J. H. Atmos. Environ. 1992, 26A, 2269. (31) National Research Council (NRC). Research Priorities for Airborne Particulate Matter. I. Immediate Priorities and a Long-Range Research Portfolio; National Academy Press: Washington, DC, 1998; p 67. (32) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 293. (33) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 314. (34) Harner, T.; Bidleman, T. Environ. Sci. Technol. 1998, 32, 1494. (35) Murai, S.; Sonoda, N.; Tsutsumi, S. Bull. Chem. Soc. Jpn. 1964, 37, 1187. (36) Zelikman, E. S.; Yur’ev, Y. N.; Berezova, L. V.; Tsyskovskii, V. K. J. Org. Chem. USSR (Engl. Transl.) 1971, 7, 641. (37) Pospelov, M. V.; Menyailo, A. T.; Bortyan, T. A.; Ustynyuk, Yu. A.; Petrosyan, V. S. J. Org. Chem. USSR (Engl. Transl.) 1973, 9, 312. (38) Liu, B. Y. H.; Pui, D. Y. H. J. Colloid Interface Sci. 1974, 49, 305. (39) Liu, B. Y. H.; Pui, D. Y. H. J. Colloid Interface Sci. 1974, 47, 155. (40) Atkinson, R.; Aschmann, S. M.; Arey, J.; Shorees, B. J. Geophys. Res. 1992, 97, 6065. (41) Wang, S. C.; Flagan, R. C. Aerosol Sci. Technol. 1990, 13, 230. (42) Hatakeyama, S.; Akimoto, H. Res. Chem. Intermed. 1994, 20, 503. (43) Burgess, A. R.; Lane, R. D. G.; Sen Sharma, D. K. J. Chem. Soc. B 1969, 341. (44) Magelli, O. L.; Sheppard, C. S. In Organic Peroxides; Swern, D., Ed.; John Wiley & Sons: New York, 1970; Vol. 1, pp 1-104. (45) Hiatt, R. In Organic Peroxides; Swern, D., Ed.; John Wiley & Sons: New York, 1971; Vol. 2, p 49. (46) Zelikman, E. S.; Berezova, L. V.; Kutaeva, E. P.; Yur’ev, Y. N. J. Org. Chem. USSR (Engl. Transl.) 1976, 12, 769.

(47) Pospelov, M. V.; Menyailo, A. T.; Kaliko, O. R.; Bortyan, T. A.; Belyaeva, E. S.; Karasev. Y. Z. J. Org. Chem. USSR (Engl. Transl.) 1978, 14, 228. (48) Morrison, R. T.; Boyd, R. N. Organic Chemistry, 3rd ed.; Allyn and Bacon; Boston, 1973; p 641. (49) Fraser, R. T. M.; Paul, N. C.; Phillips, L. J. Chem. Soc. B 1970, 1278. (50) McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed.; University Science Books: Sausalito, CA, 1993. (51) Grosjean, E.; Grosjean, D.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 1038. (52) CRC Handbook of Chemistry and Physics, 63rd ed.; Weast, R. C., Ed.; CRC Press: Boca Raton, FL, 1982; p D-209. (53) Tao, Y.; McMurry, P. H. Environ. Sci. Technol. 1987, 23, 1519. (54) Rader, D. J.; McMurry, P. H.; Smith, S. Aerosol Sci. Technol. 1987, 6, 247. (55) Gab, S.; Hellpointer, E.; Turner, W. V.; Korte, F. Nature 1985, 316, 535.

(56) Pankow, J. F. Atmos. Environ. 1994, 28, 185. (57) Adamson, A. W. Physical Chemistry of Surfaces, 4th ed.; John Wiley & Sons: New York, 1982. (58) Weber, R. J.; Marti, J. J.; McMurry, P. H.; Eisele, F. L.; Tanner, D. J.; Jefferson, A. J. Geophys. Res. 1997, 102, 4375. (59) Kavouras, G. I.; Mihalopoulos, N.; Stephanou, E. G. Nature 1998, 395, 683. (60) Weber, R. J.; McMurry, P. H.; Mauldin, L.; Tanner, D. J.; Eisele, F. L.; Brechtel, F. J.; Kreidenweis, S. M.; Kok, G. L.; Schillawski, R. D.; Baumgardner, D. J. Geophys. Res. 1998, 103, 16385. (61) Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. Environ. Sci. Technol. 1999, 33, 1430.

Received for review June 27, 1999. Revised manuscript received January 24, 2000. Accepted March 7, 2000. ES9907156

VOL. 34, NO. 11, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2115