Environ. Sci. Technol. 2005, 39, 5661-5673
Identification and Quantification of Aerosol Polar Oxygenated Compounds Bearing Carboxylic or Hydroxyl Groups. 2. Organic Tracer Compounds from Monoterpenes M . J A O U I , * ,† T . E . K L E I N D I E N S T , ‡ M. LEWANDOWSKI,‡ J. H. OFFENBERG,‡ AND E. O. EDNEY‡ Alion Science and Technology, P.O. Box 12313, Research Triangle Park, North Carolina 27709, and U.S. Environmental Protection Agency, Office of Research and Development, National Exposure Research Laboratory, Research Triangle Park, North Carolina 27711
In this study, a comparison is made of polar organic compounds found in the field with those produced in secondary organic aerosol from laboratory irradiations of natural hydrocarbons and oxides of nitrogen (NOx). The field samples comprised atmospheric particulate matter (PM2.5) collected at Research Triangle Park (RTP), NC, during the summer of 2003, and the laboratory samples originated from the photooxidation of the following monoterpenes: R-pinene, β-pinene, and d-limonene. To determine the structural characteristics of the polar compounds, the filter samples were solvent extracted and derivatized using a technique based on single and multistep derivatizations. The resulting compound derivatives were analyzed by GC-MS in the methane-CI and EI modes. In addition to previously reported biogenic oxidation products (pinic acid, pinonic acid, norpinic acid, nopinone, and pinonaldehyde), seven multifunctional organic compounds were found in both field and laboratory samples. These compounds, which are proposed as possible atmospheric tracers for secondary organic aerosol from monoterpenes, were consistent with the following identifications: 3-isopropyl pentanedioic acid; 3-acetyl pentanedioic acid; 3-carboxy heptanedioic acid; 3-acetyl hexanedioic acid; 2-isopropyl1,2-dihydroxybutanol; 4-isopropyl-2,4-dihydroxyhexanol; and 3-(2-hydroxy-ethyl)-2,2-dimethyl-cyclobutane carboxylic acid. Initial attempts have been made to quantify the concentrations of these tracer compounds on the basis of surrogate compound calibrations. The occurrence of these compounds in both laboratory and field measurements suggests that secondary organic aerosol originating from biogenic hydrocarbons are contributing to the regional aerosol burden in the southeastern United States. Several of these compounds also appear to contribute to the global aerosol burden in that they have also been identified in Europe and Brazil.
* Corresponding author phone: (919)541-7728;
[email protected]. † Alion Science and Technology. ‡ U.S. Environmental Protection Agency. 10.1021/es048111b CCC: $30.25 Published on Web 06/28/2005
2005 American Chemical Society
e-mail:
Introduction Particulate matter originating from natural and anthropogenic sources plays an important role in global climate and atmospheric chemistry. Its direct environmental impact involves increasing the scattering and absorption of solar radiation, which leads to visibility degradation and alters the amount of solar radiation that reaches the surface of the earth (1, 2). Indirectly, aerosols affect cloud properties and the hydrological cycle of the climate by acting as cloud condensation nuclei (3). Evidence has accumulated that suggests a relationship between exposure to PM2.5 (particulate matter with an aerodynamic diameter CdO) or (-HCdO) using O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine (PFBHA) and hydroxyl/carboxylic groups (-OH/-COOH) using bis-(trimethylsilyl) trifluoroacetamide (BSTFA) (6-8). However, the method has been shown to suffer from a lack of specificity between alcoholic and acidic hydroxyl groups coexisting in the same molecule. For example, Edney et al. identified several POCs in field samples with the PFBHA + BSTFA double derivatization technique but were unable to provide unambiguous identifications and structures for compounds with more than one hydroxyl or carboxylic acid group (6). Even with this limitation, several compounds were identified that were detected in both ambient PM2.5 samples and laboratory irradiations of the R-pinene/oxides of nitrogen (NOx) system. To address this limitation, a new analytical method has been developed to more completely characterize POCs bearing one or more of the following groups: (-OH), (-COOH), (>CdO), and (-HCdO) (9). This method is based on initially derivatizing (-COOH) groups with methanol in the presence of a relatively strong acid, boron trifluoride (the BF3 technique), and then using BSTFA as a reagent to silylate the remaining (-OH) groups and PFBHA to derivatize (>CdO) or (-HCdO) groups. GC-MS analysis in the positive methane-chemical ionization (CI) and electron ionization (EI) modes is then used for the identification. Once individual compounds have been identified, they can be associated with peaks obtained from the same extract derivatized with BSTFA only. Quantitative analysis is performed by using the silylated chromatogram obtained from EI mode. The objective of the present work was to examine the degree to which individual POCs can serve as tracers for biogenic SOA in ambient PM2.5. Our previous work to identify POCs in PM2.5 suffered from three limitations that subsequent studies are designed to address: (1) Because only one VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5661
TABLE 1. Conditions for the Indoor 14.5 m3 Smog Chamber Experiments for the Photooxidation of Some Monoterpenes in the Presence of NOx and UV Light exp. ID
HC
chamber operation
NO (ppb)
NOx (ppb)
HC (ppmC)
T (°C)
initial RH (%)
ER-131 ER-135 ER-146 ER-147 ER-150 ER-151 ER-163
R-pinene R-pinene d-limonene β-pinene R-pinene R-pinene + d-limonene R-pinene
dynamic dynamic dynamic dynamic dynamic dynamic dynamic
258 263 222 141 103 237 421
263 268 226 146 106 251 450
4.15 4.94 2.88 2.00 3.75 1.75 4.08
25 25 25 25 19 23 24
28 30 30 31 30 30 31
derivatization approach was used (PFBHA + BSTFA), many identifications were ambiguous which increases the possibility of misinterpretations in field samples (6). (2) The techniques focused on qualitative analysis solely. (3) Only a very limited effort was undertaken to determine the precursors of the tracer compounds. The first limitation has been addressed through the development of new analytical techniques to identify and quantify POCs found in laboratory aerosol and in ambient PM2.5. These techniques have already been reported (9). The second and third limitations are addressed in the present work. Field samples for this study were collected during the summer of 2003 at a site in Research Triangle Park, North Carolina (RTP, NC). For the laboratory portion of this work, the photooxidation products from three of the most commonly emitted monoterpenes (R-pinene, β-pinene, and d-limonene) were studied in controlled hydrocarbon-oxides of nitrogen irradiations in a smog chamber.
Experimental Methods Chemicals and Solvents. All chemicals of analytical reagent grade used as reactants and standards were purchased from Aldrich Chemical Co. (Milwaukee, WI). These chemicals were of the highest purity available and were used without further purification. All solvents were obtained from Burdick and Jackson (Muskegon, MI) and were specified as GC2 quality. Derivatization agents used in the derivatizations (BF3methanol, BSTFA with 1% trimethylchlorosilane, and PFBHA) were also obtained from Aldrich. Chamber Procedures. The laboratory experiments were carried out in a rectangular 14.5 m3 smog chamber similar to one described by Kleindienst et al. (10). The chamber is fabricated from stainless steel with interior walls bonded with a 40-µm TFE Teflon coating and equipped with four banks of fluorescent bulbs (40 per bank) that provide radiation distributed over the UV spectrum from 300 to 400 nm. The irradiation source consists of a mixture of UVA-340 bulbs (Q-Panel, Cleveland, OH), which mimic solar radiation between 300 and 340 nm, and standard UV bulbs (GE F40-BL) that provide radiation in the 340-400 nm range (10). During experiments, the light intensity was continuously monitored with an integrating radiometer (Eppley Laboratory, Inc., Newport, RI). The radiation level corresponded to a NO2 photolysis rate of 0.17 min-1. The smog chamber was operated as a continuously stirred tank reactor with an average chamber residence time of 6 h. This dynamic mode operation required the continuous, controlled addition of reactants through a manifold to premix the reactants. Nitric oxide was injected from a high-pressure cylinder using a mass flow controller. Monoterpenes were obtained from the headspace of the cooled liquid by bubbling a controlled flow of air through the liquid. The chamber mixture was maintained at a constant relative humidity by providing vaporized water into the dilution airstream. The amount of water metered into the airstream was regulated with a computer-controlled peristaltic pump. This allowed 5662
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
the relative humidity in the chamber to remain at a constant value despite minor temperature variations (10). Ozone was monitored continuously with a standard UV ozone monitor (Bendix model 8002, Lewisburg, WV). NO and NOY were monitored using a TECO model 42C (Franklin, MA) oxides of nitrogen analyzer. Temperature and relative humidity were measured with an Omega digital thermohygrometer Model RH411. Hydrocarbon samples were collected with a cryogenic trap followed by analysis by GCflame ionization detection (FID) (11). Table 1 gives the initial conditions for the chamber experiments with R-pinene, β-pinene, and d-limonene. Seven sets of photooxidation experiments in the presence of NOx and UV light were performed including R-pinene, β-pinene, d-limonene, and a mixture of R-pinene and d-limonene. The chamber was vented for 2 days with clean air before each experiment. The chamber background (particle and gas phase) was analyzed prior to each experiment, and no carryover from previous experiments or hydrocarbon, ozone, or NOx backgrounds were found (see supplement information). The temperature and relative humidity were kept constant throughout the experiment (Table 1). We used a relatively high concentration of hydrocarbons and NOx to be able to have enough mass collected on filters. This is necessary to be able to identify these reaction products on the basis of the different derivatization techniques used. Gas and particle samples were collected at a flow rate of 16 L min-1 using a sampling train consisting of a 60-cm, XAD-4 coated annular denuder (URG, Inc., Chapel Hill, NC) followed by a 47-mm Zefluor filter (Pall Gelman Laboratory, Ann Arbor, MI). The sampling time ranged from 12 to 24 h. The efficiency of XAD-4 coated denuders was investigated previously and was more than 98% (15). The Zefluor filters were weighed before and after sampling for gravimetric determination of the SOA mass. In each case, filters were equilibrated for 24 h in a temperature/humidity-controlled room before gravimetric measurements were made. After obtaining the SOA mass, the filters were placed in a freezer until extraction for GC-MS analysis. Field Sampling Procedures. The field measurements were taken at a research site in Research Triangle Park, North Carolina, during the summer of 2003. The field site was located within a 0.03 km2 grass-covered clearing surrounded by a mixed deciduous and pine forest. The surrounding area consisted of forest, corporate and government research facilities, a major interstate highway, and the suburban outskirts of Durham, North Carolina. The samplers were located on a wooden platform in the approximate center of the field, 1.5 m above the surface. For this study, PM2.5 samples were taken from eight 24-48 h periods for temperatures and relative humidities given in Table 2. The samples were taken under stagnant weather conditions which are a common summertime occurrence. Hot humid air results when high-pressure systems reside over the central part of the southeastern United States and the predominant air flow in the region is from the southwest. Under these conditions, daytime high temper-
TABLE 2. Atmospheric Conditions and Mass during the 2003 Summer Field Study temp. (°C) low high avg. FS03-176a FS03-209 FS03-216 FS03-230 FS03-237 FS03-239 FS03-245 FS03-253 a
17 21 20 20 18 22 19 13
34 33 30 31 34 34 32 26
27 26 25 26 25 29 26 20
RH range (%) low high 42 59 63 63 54 56 63 47
O3 max (ppm)
mass (µg m-3)
0.109 0.091 0.078 0.067 0.095 0.089 0.043 0.073
44.5 27.1 8.7 14.4 29.6 26.1 6.8 10.3
90 g98 g98 g98 g98 97 96 97
Field sample taken in 2003 on Julian date 176 (June 26, 2003).
atures are above 30 °C, with relative humidities as high as 65%. No precipitation events occurred during the sampling period at this site, and influences from clouds were minimal. High relative humidities approaching 100% and listed as g98% (Table 2) were observed during the early morning hours of several sampling periods. Thus, site-specific, meteorological conditions (e.g., clouds, solar intensity) play a minor role in establishing the observed organic composition. To mitigate the possibility of sampling artifacts from gas-phase constituents, all filter collection devices used preceding extractable or nonextractable denuders. Thus, PM2.5 samples from ambient air were collected through (1) a parallel-plate, activated carbon denuder/filter organic sampler for identification of POCs; (2) a parallel-plate, activated carbon denuder/quartz filter sampler for total organic carbon (OC) and elemental carbon (EC); and (3) an inorganic denuder/ Teflon filter sampler for measuring the mass concentration along with NO3-, SO42-, and C2O42-. Sampling procedures follow those described in earlier field measurements at the same site (6). Table 2 summarizes the conditions of the field study. Only organic compounds identified in both field and laboratory samples are reported in this paper. An analysis of the 2003 summer field study providing more extensive chemical composition of PM2.5 is reported elsewhere (12). Only a summary of the overall composition of the PM2.5 field samples is reported here in Table 3. Extractions from Sample Filters. Filters from the chamber experiments were Soxhlet-extracted with methylene chloride for 6 h. The extraction efficiency was from 95 to 100% (15 samples). Filters from field samples were Soxhlet-extracted for 24 h using a 1:1 (v/v) methylene chloride and acetonitrile mixture. The extraction efficiency ranged from 40 to 55% (eight samples), values consistent with the fraction of organicto-total mass in PM2.5 expected under these conditions. Prior to the extraction, 20 µg each of trans-p-menth-6-ene-2,8diol (PMD), bornyl acetate (BA), cis-ketopinic acid (KPA), and d50-tetracosane (TCS) were added as internal and recovery standards. BA and TCS do not undergo reaction with the derivatization reagents; KPA is used as a recovery standard for compounds containing (>CdO), (-HCdO), or (-COOH); PMD serves as a recovery standard for compounds with (-OH) or (-COOH) groups. The extract was then dried and recovered with 2 mL of the original solvent.
Derivatization Procedures. The derivatization procedures used in this study were reported in the first paper of this series (see Table S2 in Supporting Information of that paper), which included a discussion of the optimization and possible artifacts associated with each procedure (9). For this work, sample extracts from field and laboratory experiments were split into four equal parts (0.5 mL each); each part was placed in a 15-mL round-bottom tube. Each extract was concentrated to dryness using ultrapure nitrogen and was derivatized as follows. The first portion was derivatized only with BSTFA. The second portion was derivatized with BF3-methanol, then with PFBHA, and finally with BSTFA. The third portion was derivatized with PFBHA followed by BSTFA. Finally, the last portion was derivatized with BF3-methanol followed by BSTFA to distinguish between compounds containing (-COOH) and (-OH) groups. For derivatizations involving multiple steps, a 2-µL aliquot was analyzed prior to conducting the subsequent step. In some cases, as many as seven chromatograms were obtained for a single sample to aid in the identification or for confirmation purposes, although not all are reported here. Quantitative Analysis of the Target Compounds. For quantitative purposes, only silylation was used because of its previously established high conversion efficiencies (9). No attempt was made to quantify compounds formed from the double derivatization techniques. The companion study found a derivatization efficiency greater than 95% for compounds produced by BSTFA derivatization (9). Blanks were analyzed in parallel with the laboratory and field samples. The blank derivatizations were obtained using the same protocol as for each sample derivatization. The chromatograms from BF3-methanol derivatization of the blanks showed few, if any, artifact peaks. As discussed previously, the PFBHA derivatization is prone to produce large artifact peaks, typically early in the chromatogram, which do not interfere with tracer peaks (9). GC-MS Analysis. Analysis was performed on a ThermoQuest (Austin, TX) GC coupled with an ion trap mass spectrometer (ITMS). The injector, operated in splitless mode, was heated to 270 °C, and aliquots of 2 µL of the underivatized and derivatized samples were injected. Separation was achieved on an RTx-5MS capillary column (Restek, Inc., Bellefonte, PA) 60-m in length with a 0.25-mm i.d. and a 0.25-µm film thickness. The temperature program for the analysis started isothermally at 84 °C for 1 min followed by a temperature ramp of 8 °C min-1 to 200 °C, followed by a 2-min hold, then 10 °C min-1 to 300 °C, and a 15-min hold. Samples were analyzed in positive EI mode and in positive CI mode with methane as the reagent gas over an m/z range between 50 and 1000 u. The ion source, ion trap, and interface temperatures were set to 200, 200, and 300 °C, respectively. In selected cases, tandem mass spectrometry was also used to aid in the identification and structure elucidation of unknown compounds. Collisionally induced dissociation (CID) of a series of compound ions was carried out. Fragmentation of the precursor ion (M+•) or other abundant ions, such as the base ion, was enhanced by CID using helium gas as the collision gas in the ion trap.
TABLE 3. Overall Composition of Summer 2003 Field Samples Collected at RTP, NC sample
total mass ug/m3
NO3- ug/m3
SO42- ug/m3
C2O42- ug/m3
NH4+ ug/m3
EC ug/m3
OC ug/m3
FS03-209 FS03-216 FS03-230 FS03-237 FS03-239 FS03-245 FS03-253
28.57 14.29 18.76 33.70 29.31 5.72 11.05
0.02 0.01 0.01 0.02 0.02 0.03 0.08
11.94 3.46 6.66 11.88 9.77 1.45 2.61
0.01 0.04 0.11 0.20 0.18 0.04 0.14
2.32 1.32 2.42 4.10 3.14 0.59 1.15
0.6 0.2 0.4 0.3 0.2 0.2 0.4
3.8 2.5 6.9 6.9 3.6 3.2 6.6
VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5663
Results and Discussion The focus of this study is the identification and quantitative analysis of candidate tracer compounds detected both in the aerosol phase from laboratory irradiations of R-pinene/, β-pinene/, and d-limonene/NOx as well as in the PM2.5 field samples from RTP, NC. These three monoterpene compounds are among the most important natural C10-hydrocarbons emitted in the United States (13). This section presents seven tracer compounds, some previously identified, found from the photooxidation of the three monoterpenes and seen in field samples. Most POCs identified in monoterpene SOA and ambient PM2.5 samples (particularly, the candidate tracers) do not have authentic standards, and their identification must be based on the interpretation of the mass spectra of the derivatized compound. The generalized approach for identification is as follows: (1) For POCs having standards, comparisons were made between the mass spectra of the derivatized sample and the authentic spectra in the CI or EI mode, as well as the chromatographic retention time. (2) For POCs not having standards, the initial identification was made on the basis of fragment peaks in the CI mode that permit an initial determination of the number and identity of functional groups and the molecular weight of the derivative. In addition, confirmatory analyses using multiple derivatizations were conducted on particular samples to clarify the identity of potentially ambiguous groups, for example, distinguishing an alcohol group from an acid group (-OH) (see below). Finally, in many cases, these samples were also run in the EI mode to produce greater fragmentation and additional structural information of the derivatized POC. The analysis of the CI mass spectra of the derivatized POCs involves recognition of characteristic ions associated with a particular derivatization scheme. PFBHA reacts with each nonacidic (CdO) group (ketone and aldehyde) on a POC to form an oxime derivative with a molecular weight MPF given by the relationship MPF ) MW + 195n(CdO), where n(-CdO) is the number of derivatized (>CdO) and (-HCd O) groups on the POC and MW is its molecular weight. The major characteristic ion for the PFBHA derivatives is m/z 181. In CI mode, the base peak for most oximes is M + 1 or 181, while other fragments and adducts include m/z M+• + 29, M+• + 41, M+• - 181, and M+• - 197. BSTFA is used to form trimethylsilyl (TMS) derivatives (hereafter referred to as silylated derivatives) with molecular weight MBS, where MBS ) MW + 72n(OH), where n(OH) is the total number of derivatized (-OH) and (-COOH) groups on the POC. Characteristic ions of the silylated derivatives are m/z 73, 75, 117, 147, and 149. Adduct ions from the derivatives include m/z M+• + 73, M+• + 41, M+• + 29, and M+• + 1; fragment ions include m/z M+• - 15, M+• - 73, M+• - 89, M+• - 117, M+• - 133, and M+• - 207. The reaction of BF3-methanol with carboxylic acids forms methyl esters (hereafter referred to as methylated derivatives) with a molecular weight of the derivative given by MBF ) MW + 14n(COOH), where n(COOH) is the number of derivatized (-COOH) groups on the POC. Characteristic ion fragments for these methylated derivatives are 59, M+• - 59, M+• - 31, and M+• + 1, in addition to weak adducts M+• + 15, M+• + 29, and M+• + 41. Multiple derivatizations result in adducts and fragments that include ions from each derivatization. Typically, three to seven adduct or fragment ions are found for each derivatized POC. Characteristic ions are not as apparent in the EI mode because of the much higher energies involved in ionization. However, the main advantage of using EI is that the greater fragmentation can produce peaks representative of the structural characteristics of the parent compound. For molecules with only PFBHA derivatives, the use of EI results almost invariably in m/z 181 being the base peak and other 5664
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
fragments typically being less than 10% of the intensity of the base peak. In this case, the identification of the molecular weight becomes less certain than with CI. For silylation, a similar situation occurs in which characteristic ion 73 or 149 (for multiple silylated groups) is the base peak; however, the fragment peaks are generally much greater in intensity than in the case with PFBHA, that is, the intensities of most fragment ions are 25-50% of the base ion. For BF3-methanol derivatization, fragment ions are predominantly seen and include m/z 39, 55, M+• - 31, M+• - 59, and M+• - 63, one of which is often the base ion (9). The seven tracer compounds identified from the laboratory and field systems do not have standards, and identifications must be based on the interpretation of mass spectra (9). We have already reported some of these target POCs, but because of limitations in the derivatization technique used, most of the POCs had ambiguous identifications (6). Prior to this, similar, and possibly, identical POCs were also reported by Kuba´tova´ et al. in field samples collected in the Amazon basin (Brazil) and in Gent, Belgium (14). In that case, all mass spectra were obtained in EI mode. Moreover, the origin of their identified compounds could not be established. Figure 1 shows a comparison of total ion chromatograms (TICs) from the field sample FS03-245 with that of the three monoterpene photochemical systems, R-pinene/NOx, β-pinene/NOx, and d-limonene/NOx. The samples compared in this figure were derivatized with BSTFA only. For ease in identification, the candidate tracer compounds have been labeled U-1 to U-7. All of the seven compounds are seen in FS03-245, and virtually all are also present in the other field samples. The figure shows that all seven tracer compounds are produced in the R-pinene/NOx system. By contrast, only five of the seven tracer compounds (U-1 to U-5) are present in the d-limonene/NOx system and six of the tracer compounds (U-2 to U-7) are present in the β-pinene/NOx system. As a means to discriminate between alcoholic and acidic (-OH) groups among the tracers, the top half of Figure 2 shows a mass chromatogram of compounds having acidic groups as found from their methylated derivatives for FS03209. The lower half of Figure 2 shows the same selected ion chromatogram after sample derivatization with PFBHA. This approach is a direct means to show which if any of the tracer compounds contain carbonyl groups. The chromatograms are shown for selected ions m/z 127, 171, 185, 215, 398, and 412. As discussed in detail below, U-5 and U-6 have only nonacidic (-OH) groups and are thus not seen in Figure 2. U-2 and U-4 are shown to have shifted because of formation of their corresponding oximes. Given these introductory observations, a detailed characterization based on the mass spectra and identification of functional groups and the deduced molecular mass is now presented for the tracer compounds. In some cases, where the tracer compounds have been discussed previously (6, 14), an abbreviated discussion appears in the main text with a more detailed discussion in the Supporting Information. In all cases, mass spectra are found in the Supporting Information. Unknown U-1. Compound U-1 was found in both ambient PM2.5 and laboratory samples on the basis of retention time and mass spectral pattern analysis. The spectra from the ambient samples compare well with that of SOA from the R-pinene and d-limonene experiments. The identification of this compound is consistent with that of Kuba´tova´ et al. from the analysis of field samples from Balbina, Brazil, and Gent, Belgium (14). The compound was not reported in the chromatogram examined by Edney et al. (6). From the evidence below and that of Kuba´tova´ et al., U-1 is identified as 3-isopropyl pentanedioic acid.
FIGURE 1. GC-MS chromatograms (TIC:CI) resulting from the BSTFA derivatives of particle phase originating from (a) a 2003 summer field sample (FS03-245), (b) r-pinene/NOx, (c) β-pinene/NOx, and (d) d-limonene/NOx (see Table 1 for conditions). U-1 through U-7 are the seven unknown compounds identified in this study. The mass spectra of selected of U-1 derivatives are shown in Figure S1 (supplementary information) which includes the CI and EI spectra from the methylated derivative (Figure S1a and S1b) and the CI spectra of the silylated derivatives (Figure S1c). The CI spectrum for the methylated derivative corresponds to peak U-1 in Figure 2. The methylated derivative analyzed by CI shows fragments at
m/z 171 (M+• - 31), 139 (M+• - 63), and 111 (M+• - 91) and generally weak adducts at m/z 203 (M+• + 1), 231 (M+• + 29), and 243 (M+• + 41). Fragmentation patterns of the methylated U-1 are consistent with the presence of two carboxylic groups giving a molecular weight of 202 u for the derivatized compound and 174 u for the underivatized compound. VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5665
FIGURE 2. GC-MS total ion chromatograms in CI mode for the particle phase derivatized with BF3-methanol (top) followed by PFBHA (bottom) and showing unknown compounds bearing (-COOH) and (>CO)/(-HCO) groups. The CI mass spectrum of the silylated derivative of U-1 (Figure S1c) shows fragments at m/z 303 (M+• - 15) and 229 (M+• - 89; strong) and adducts at 319 (M+• + 1), 347 (M+• + 29), and 359 (M+• + 41) that are consistent with the presence of two (-OH) groups on U-1, indicating a molecular weight for the silylated compound of 318 u. A comparison of the spectra for the methylated derivative with that of the double derivative, methylation followed by silylation, indicated no shift in retention time for U-1, thus confirming the absence of nonacidic (-OH) groups. In addition, as seen in Figure 2, when the methylated derivative was treated with PFBHA, there again was no change in retention time. Additional structural information for a compound that has a molecular weight of 174 u with two carboxylic acid functional groups is obtained from an interpretation of the EI spectra of the methylated U-1. Scheme 1 shows a proposed mechanistic fragmentation pattern for ions observed for the methylated derivative. The EI spectrum in Figure S1b shows fragment ions at m/z 187, 171, and 160 that originate from the loss of a methyl group (-CH3), a methoxy group (-OCH3), and a (-CH2CHCH3) group, respectively, from the molecular ion. The base peak at m/z 100 (Figure S1b) is likely created by a combined loss of isopropyl (-CH3CHCH3) and (-COOCH3) groups from the molecular ion M+; elimination of (-CH3) leads to the formation of a fragment at m/z 85. The presence of ions at m/z 100 and 160 is consistent with assigning an isopropyl group to the 3-position of the molecule, since Figure 2 demonstrates the absence of an acetyl group. Scheme 1b shows fragmentation of the most relevant EI fragments observed of the silylated U-1 (Figure S1d). Unknown U-2. Compound U-2 was detected in both ambient PM2.5 and the SOA from all three laboratory systems on the basis of retention time and mass fragmentation pattern analysis. This compound appears to have been identified by Kuba´tova´ et al. (14) on the basis of mass spectral analysis of 5666
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
field samples. This compound and possibly additional isomers (much smaller peaks in the TIC) were detected by Edney et al. (6) in both field and laboratory samples (photooxidation of R-pinene), although the PFBHA + BSTFA derivatization did not allow an unambiguous identification of the (-OH) groups on the compound. Figure 2 now confirms the original identification of Edney et al. (6), which shows the presence of both acidic and acetyl groups. A consistent analysis of the silylated and methylated derivatives is based on assuming a parent molecule that contains two acidic (-OH) groups and a single carbonyl group. Given the considerable discussion of this compound previously (6, 14), additional analyses of the interpretation of the EI and CI spectra from multiple derivatizations are given in the Supporting Information which support the tentative identification of 3-acetyl pentanedioic acid. Unknown U-3. Compound U-3 was detected in the field samples and all three laboratory systems. Figure 1, as well as the TIC, show this compound to be among the most intense of the tracer compounds in the field samples and the biogenic/NOx systems. This compound has been identified as 3-carboxy heptanedioic acid, by Kuba´tova´ et al. (14) in field samples and later by Edney et al. (6) in both laboratory and field samples. Edney et al. (6) noted that their identification was potentially ambiguous with that of a hydroxy dicarboxylic acid or a dihydroxy carboxylic acid since silylation was used. This work confirms the identification of the compound found in the field and especially in the laboratory study. The identification of U-3 is confirmed on the basis of the methylated derivative. Figure 2 shows the presence of acidic groups but no carbonyl groups. Additional evidence is provided by Figure S3, which displays mass spectra of U-3 for the methylated derivative in the CI (a) and EI (b) modes, respectively. Several characteristic fragment and adduct ions
SCHEME 1. Fragmentation for the (a) Methylated and (b) Silylated U-1
are present in these mass spectra, including at m/z 155 (M+• - 91), 187 (M+• - 59), 215 (M+• - 31), 247 (M+• + 1), 275 (M+• + 29), and 287 (M+• + 41) for CI mode (Figure S3a) and at m/z 155 (M+• - 91), 186 (M+• - 60), and 215 (M+• - 31) for EI mode (Figure S3b). The fragmentation pattern indicates a methylated derivative of MW 246 u and a parent compound MW 204, assuming three acid groups. This assumption is consistent with the interpretation of Edney et al. who established the presence of a silylated derivative of MW 420 and an underivatized MW of 204. An attempt to form a methylated/silylated derivative of the compound indicated that U-3 contains only (-COOH) groups, because of a lack of a retention time shift between the single and double derivatization. Scheme 2 shows the mechanistic rationale for the EI fragmentations of the silylated U-3. Given the importance of this compound because of its high abundance in smog chamber and field samples, MS/ MS spectra from the base peak in these chromatograms were examined. Figure 3 shows the MS/MS spectra of the most intense ions observed in EI positive mode for ions at m/z 405 (a), 287 (b), and 213 (c). The mechanistic pathways of ions
at m/z 213, 287, and 315 observed in the MS/MS spectrum of the m/z 405 in Figure 3 are similar to that presented in the lower portion of Scheme 2. The ion at m/z 243 observed in Figures 2a and 2b is more likely to originate via rearrangement and loss of CO2 from m/z 287. However, it is difficult to propose a pathway leading to m/z 149 that was observed in Figure 3b in the full EI scan (Figure S4d), thus, the fragment at m/z 149 is interpreted as [(CH3)2SiH-HOSi(CH3)3)]+•. Unknown U-4. U-4 is readily detected in the silylated chromatogram for the field samples as well as for the three laboratory systems as seen in Figure 1. It is also found as the methylated derivative in Figure 2a and forms an oxime seen by the retention time shift in Figure 2b. This and other evidence suggest that U-4 is in a homologous series with U-2. Additional evidence comes from the mass spectra for the methylated derivatives of U-4 in the CI and EI modes (Figure S4a and S4b, respectively). The methylated derivative confirms the acidic identity of the two (-OH) groups found using the PFBHA + BSTFA derivatization (6). Thus, U-4 is identified as 3-acetyl hexanedioic acid, which while in a VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5667
SCHEME 2. Fragmentation for the Methylated and Silylated U-3
homologous series with U-2 has an intensity in the TIC that is generally lower than U-2 in field and laboratory samples. The compound was also identified in the mentioned field samples by Kuba´tova´ et al. (14). The interpretation of the mass spectra found in the Supporting Information supports this tentative identification. Unknown U-5. Compound U-5 was initially identified by Edney et al. (6) and is now one of the most readily identified tracers for monoterpenes. While the intensity of U-5 in the TIC of field samples is moderate, it has a unique m/z 349 ion, which is the dominant peak in its selected ion chromatogram for field and laboratory samples. The major piece of information about this compound comes from the BSTFA derivatization. In CI mode, the spectrum shows fragment and adduct ions at m/z 365, 349, 275, 393, 405, and 437, interpreted as (M+• + 1), (M+• - 15), (M+• - 89), (M+• + 29), (M+• + 41), and (M+• + 73), respectively (Figure S5a). In EI mode, fragments at 349, 259, and 231, interpreted as (M+• 15), (M+• - 105), and (M+• - 133), are consistent with standard BSTFA fragmentation patterns (9). The most reasonable interpretation of these mass spectra is for a compound having three silylated sites with a derivative MW 364 u and a compound MW of 148 u. The identity of this compound using only the BSTFA derivatizations is ambiguous and, as Edney et al. (6) pointed out, the compound could be carboxyl propanedioic acid or a C5 hydroxy dicarboxylic acid. More5668
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
over, the possibility of a C6 dihydroxy carboxylic acid or a C7 triol also had to be explored. These issues were clarified through a series of double derivatizations. Initially, we compared the retention times of the silylated compound from BSTFA derivatization against that from the BSTFA + PFBHA double derivatization. When the silylated U-5 compound from field and laboratory samples was derivatized with PFBHA, neither the elution time or fragmentation pattern changed, which signified the absence of a carbonyl group. Thus, only issues regarding the identity and position of the (-OH) groups remain, which are addressed using methylation and a methylation-silylation combination. The triacid (derivative MW 190) would be most easily recognized in the simple methylation. An examination of field and laboratory samples shows no evidence for such a compound. Finally, an examination of the chromatograms from the methylated samples subjected to silylation shows no evidence for a hydroxyl diacid (MWd ) 248 u) or a dihydroxy monoacid (MWd ) 306). These findings strongly suggest that U-5 is a polyhydroxylated compound, that is, a triol. The positions of the (-OH) groups are explored through use of the EI spectra shown in Figure S5b. Chromatograms of the silylated compound in EI show the presence of one predominant peak at m/z 349. The fragment ion at m/z 321 (M+• - 43) seen in Figure S5b indicates the presence of an
FIGURE 3. Positive mode MS/MS spectra in EI mode of 3-carboxy heptanedioic acid (a) at m/z 405, (b) at m/z 287, and (c) at m/z 213. VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5669
SCHEME 3. Fragmentation for the Silylated U-5
isopropyl group from (M+•) because the PFBHA derivatization shows an absence of (>CO) and (-HCO) groups. Scheme 3 depicts the formation of important fragment ions observed in the silylated form of U-5. As noted, the ion at m/z 247 (M+• - 117) is interpreted as (M+• - ((CH3)3SiO(CH2)2)) and not (M+• - (COOSi(CH3)3)) because of the lack of a carbonyl group. The isopropyl group must be located at position 2 because, in the case of 1-substitution, a fragment ion at m/z 219 would be expected because of the loss of (-CH(CH3)2 + CHO(Si(CH3)3)) from the molecular ion (M+•). Thus, U-5 is tentatively identified as 3-isopropyl-1,2-dihydroxybutanol. Unknown U-6. Compound U-6 is seen in the sample derivatized with BSTFA as shown in Figure 1. Mass spectra for the BSTFA derivative of U-6 in the CI and EI modes for field sample FS03-239 are shown in Figure S6. A strong fragment at m/z 377 (M+• - 15) is present in both modes. In the CI mode, additional fragment and adduct ions are at m/z 303, 393, 421, 433, and 465, corresponding to (M+• - 89), (M+• + 1), (M+• + 29), (M+• + 41), and (M+• + 73), respectively. Similarly, EI fragments at m/z 377, 349, 259, and 287 correspond to (M+• - 15), (M+• - 43), (M+• - 133), and (M+• - 105), respectively, which again is consistent with expected fragmentation patterns for silylated compounds (9). From this evidence, the derivative for U-6 has a molecular weight of 392 u. Given the presence of m/z 149 as the base peak in the EI spectrum, it is clear that U-6 has at a minimum two (-OH) groups. Similar to the analysis for U-5, the silylated sample when subjected to PFBHA derivatization results in a peak with an unchanged retention time and fragmentation pattern, indicating an absence of carbonyl groups for U-6. Thus, the most reasonable interpretation of the silylation 5670
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
data for U-6 is that of a compound with three (-OH) groups and a molecular weight of 176 u. Various combinations of possible hydroxy acids (as noted above for U-5) are now considered by interpreting the mass spectral chromatograms for the methylated and methylatedsilylated derivatives. The possibility of a tricarboxylic acid was investigated by examining the sample treated only by methylation. The chromatogram of methylated derivatives indicates the presence of a minor peak (MWd 218) associated with a tricarboxylic acid derivative. This compound was determined not to be U-6 because of an additional peak in the chromatogram of the silylated derivative with a molecular weight of 392 u rather than its relative peak strength. The methylated derivative is likely to be propane-1,2,3-tricarboxylic acid, which had previously been reported by Kuba´tova´ et al. (14) and had been reported as a possible compound by Edney et al. (6). However, examination of the chromatograms for the biogenic systems did not indicate the presence of this triacid. The methylated-silylated derivatives of the field and laboratory samples were then examined. In no case were derivative compounds detected for a hydroxy-diacid (MW 276 u) or for a dihydroxy carboxylic acid (MW 334 u). This leaves U-6 as a compound with a polyhydroxylated structure (i.e., a triol) with an empirical formula of C9H20O3. An examination of the mass spectrum from the silylated derivative of U-6 (Figure S6a) indicates the presence of (M+• - 43) at m/z 349 (weak), suggesting that U-6 may contain an isopropyl group. Scheme S1 (Supporting Information) provides an interpretation of important fragment ions of the silylated U-6. In both U-5 and U-6, a fragment at m/z (M+• - 59) is explained by the loss of (-CH3) following the loss
TABLE 4. Compounds Identified in Both Ambient and Smog Chamber Photooxidation of r-Pinene, β-Pinene, and d-Limonene in the Presence of NOx and Clean Air
a
Product detected in laboratory irradiations of the listed precursor with NOx.
of (-C3H8) via McLafferty rearrangement (16). Thus, U-6 is tentatively identified as 4-isopropyl-2,4-dihydroxyhexanol. Unknown U-7. Figure S7 shows mass spectra of unknown U-7 for its methylated derivatives in CI (a) and EI (b) mode. The EI spectrum of the methylated U-7 (Figure S7c) shows fragmentation patterns in the low mass region (m/z < 100 u), which are similar to those of reaction products from the oxidation of R-pinene and β-pinene (e.g., pinic acid, pinonic acid, pinonaldehyde, and nopinone) and suggests that U-7 contains a 2,2-dimethyl-cyclobutane ring (15, 17). While U-7 was found from the R-pinene/NOx and β-pinene/NOx irradiations (as well as in the field samples), it was not observed in the irradiated d-limonene/NOx system. This observation is consistent with the lack of a four-membered ring in d-limonene (i.e., 1-methyl-4-isopropenyl cyclohex-1-ene). The CI spectrum of the silylated derivatives in Figure S7c contains an intense peak at m/z 227, which was interpreted as the (M+• - OSi(CH3)3) together with ion (M+• + H) at m/z 317, (M+• + 29) at m/z 345, and (M+• + 41) at m/z 357, and other peaks at m/z 301 and 199 corresponding to the loss of (-CH3) and (-C(O)OSi(CH3)3), respectively, from the molecular ion (M+•). The fragmentation indicates an MW 316 u for its BSTFA derivative and an MW 172 u for the underivatized form, suggesting the compound has two (-OH) groups. An analysis of the mass spectrum of the methylated derivatives of U-7 is consistent with the presence of a single carboxylic acid group. The CI spectrum (Figure S7a) shows fragments at m/z 155 and 127 because of (M+• - 31) and (M+• - 59), respectively, and adducts at m/z 187, 215, and 227 from (M+• + 1), (M+•+ 29), and (M+• + 41), respectively. Fragments at m/z (M+• + 1), (M+• - 15), and (M+• - 31) are evident in the EI spectrum and are consistent with patterns of esters previously reported (9). This analysis gives a derivative molecular weight of 186 u for methylation only.
Moreover, the CI spectrum of the methylated and silylated double derivative indicates ions at m/z 259, 227, 195, and 167, interpreted as (M+• + 1), (M+• - 31), (M+• - 63), and (M+• - 91), respectively. Analysis of these mass spectra indicates that U-7 contains one 2,2-dimethyl-cyclobutane ring in addition to one carboxylic and one hydroxyl group and an MW of 172 u (9). Several compounds with MW 172 da were observed from the ozonolysis of R-pinene or from field samples, including norpinic acid and 9-oxononaoic acid, which were rejected on the basis primarily of the retention times from authentic standards. A structure consistent with this interpretation is 3-(2-hydroxy-ethyl)-2,2-dimethylcyclobutane carboxylic acid. Table 4 summarizes the products identified in both ambient PM2.5 and monoterpenes SOA including structure, systematic nomenclature, compound molecular weight, the major relevant fragments/adduct ions observed from their BSTFA derivatives in CI mode, and the laboratory systems in which the product is found. Mechanism of Product Formation. The degradation pathways leading to compounds U-1 through U-7 identified in both PM2.5 and biogenic SOA have not been reported, and this is the first study linking these compounds to the biogenic organic compounds (BOCs) precursors, R-pinene, β-pinene, and d-limonene. Except for U-7, the acyclic structure of these compounds suggests that the formation is based on a ringopening mechanism. However, a plausible mechanism for their formation on the basis of gas-phase chemistry is difficult to formulate suggesting the possibility of a heterogeneous mechanism. Time series profiles from static chamber photooxidations (result not shown) indicate that the concentrations of the reported compounds correlate with the formation profiles of SOA in the chamber. If secondary or tertiary products are formed in the gas phase, they would rapidly partition to the particle phase because of their low volatility. VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5671
SCHEME 4. Proposed Mechanistic Pathways Leading to U-2 from r-Pinene
Recently, R-campholenal and 8-hydroxymenthen-6-on were reported as reaction products in the gas phase from the oxidation of R-pinene on the basis of experimental and theoretical studies (15, 18). The occurrence of these unsaturated compounds, in addition to aldehydes, keto aldehydes, and carboxyaldehydes, in the gas phase may explain in part the formation of some compounds observed in this study. For example, the degradation pathways leading to the formation of 3-acetyl pentanedioic acid in the R-pinene system are proposed to arise via the reaction of OH radicals with the primary and secondary products R-campholenal and pinalic-4-acid, as given in Scheme 4. R-Campholenal and pinalic-4-acid were identified in this study from an analysis of extractable organic denuder samples (results not presented) with mechanistic pathways that have already been reported (7, 15). For example, the OH addition to the R-campholenal double bond in the presence of O2, NO, or RO2 could lead to alkoxy radical formation, which could then decompose via a ring-opening reaction to β-hydroxyalkoxy radicals, as shown in Scheme 4 (top). After elimination of CH3COOH and in the presence of O2, NO, or RO2, the alkoxy radical would then decompose to dialdehyde, which oxidizes to 3-acetyl pentanedioic acid. However, the oxidation of the aldehyde groups could occur at other steps in the mechanism. Pathways depicting the formation of 3-acetyl pentanedioic acid formation via H-atom abstraction of pinalic-4-acid aldehydic hydrogen are shown in the lower half of Scheme 4. The formation of hydroxylated compounds (U-5 and U-6) from R-pinene, β-pinene, and d-limonene photooxidation are difficult to explain. However, since none of these compounds were found in the gas phase, it is also possible that they are due to particle-phase reactions. Heterogeneous reactions in the particle phase or gas-surface reactions might be occurring. Therefore, experimental and possibly theoretical studies are needed to explain the origin of these compounds. Quantification of Tracer Compounds in Field Samples. Quantitative analysis of reaction products in Table 4 were performed using a single derivatization with BSTFA. As a check of the method, known compounds (PA, NPA, PNA, and PAHD) were quantified using their authentic standards. An initial analysis of the tracer compounds, however, uses 5672
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 15, 2005
the pinic acid response factor as surrogate for the true response factors. For each compound, the TIC or extracted ion chromatogram of the most intensive ions was used for quantification, depending on the degree of coelution, especially for the field samples. The most significant source of error was in the surrogate calibration and ion extraction uncertainties. The systematic errors are unknown but estimated to be within a factor of 2-3 of the true value. For example, when the BSTFA derivative of norpinic acid is used as surrogate to quantify the BSTFA derivative of pinic acid, the concentration of pinic acid is overestimated by onethird. If the extracted ions are inappropriately selected the concentration can be overestimated or underestimated by as many as 3 orders of magnitude depending on the specific ions used. The concentrations of products were analyzed relative to the two internal standards KPA and PMD. When pinic acid and norpinic acid were quantified using their authentic standards, but with KPA and PMD as internal/recovery standards, the systematic error was about 60% for both compounds. A detailed investigation showed that KPA gives more reliable results when used as the internal/recovery standard than PMD. Thus, KPA was used as the internal/ recovery standard for all quantitative values reported in this study. Using our best estimate of the calibration factors, concentrations for each of the target compounds for FS03-176 were determined and are provided in Figure 4. For this particular field sample, concentrations for U-2 through U-5 ranged from 30 to 50 ng m-3. Other unknown target compounds were in the range of 5-10 ng m-3. These values are typical of the range of values often seen for individual nonpolar organic compounds detected in ambient samples. Efforts are currently underway to try to improve the reliability of the calibration factors and other analytical uncertainties inherent in quantifying these types of compounds in ambient PM2.5. Authentic standards for the tracer compounds identified in this study (U-1 through U-7) are not available commercially. To accurately quantify the link between biogenic and ambient PM2.5, several of these compounds are in the process of being synthesized. Therefore, their yield from chamber experiments are not reported in this study and will be reported in a subsequent study.
(7)
(8)
(9)
(10)
FIGURE 4. Concentration of unknown compounds in addition to PA identified in the 2003 summer field sample (FS03-176).
(11)
Acknowledgments The U.S. Environmental Protection Agency through its Office of Research and Development funded and collaborated in the research described here under Contract 68-D-00-206 to Alion Science and Technology. It has been subject to Agency review and approved for publication. Mention of trade names or commercial products does not constitute an endorsement or recommendation for use. The authors wish to thank Eric W. Corse for operating the smog chamber.
(12)
(13)
Supporting Information Available Results and Discussion section about the identification of unknown compounds. This material is available free of charge via the Internet at http://pubs.acs.org.
(14)
Literature Cited
(15)
(1) Charlson, R. J.; Schwartz, S. E.; Hales, J. M.; Cess, R. D.; Coakley, J. A.; Hansen, J. E.; Hoffmann, D. J. Climate forcing by anthropogenic aerosols. Science 1992, 255, 423-430. (2) Haywood, J.; Boucher, O. Estimates of the direct and indirect radiative forcing due to tropospheric aerosols: a review. Rev. Geophys. 2000, 38, 513-543. (3) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Aerosols, climate, and the hydrological cycle. Science 2001, 294, 21192124. (4) Samet, J. M.; Dominici, F.; Curriero, F. C.; Coursac, I.; Zeger, S. L. Fine particulate air pollution and mortality in 20 U.S. cities, 1987-1994. N. Engl. J. Med. 2000, 343, 1742-1749. (5) Gamble, J. F. PM2.5 and mortality in long-term prospective cohort studies: cause-effect or statistical association? Environ. Health Perspect. 1998, 106, 535-549. (6) Edney, E. O.; Kleindienst, T. E.; Conver, T. S.; McIver, C. D.; Weathers, W. S. Polar organic oxygenates in PM2.5 at a
(16) (17)
(18)
southeastern site in the United States. Atmos. Environ. 2003, 37, 3947-3965. Yu, J.; Flagan, R. C.; Seinfeld, J. H. Identification of products containing -COOH, -OH and CdO in atmospheric oxidation of hydrocarbons. Environ. Sci. Technol. 1998, 32, 2357-2370. Kleindienst, T. E.; Conver, T. S.; McIver, C. D.; Edney, E. O. Determination of secondary organic aerosol products from the photooxidation of toluene and their implications in ambient PM2.5. J. Atmos. Chem. 2004, 47, 79-100. Jaoui, M.; Kleindienst, T. E.; Lewandowski, M.; Edney, E. O. Identification and quantification of aerosol polar oxygenated compounds bearing carboxylic and/or hydroxyl groups. 1. Method development. Anal. Chem. 2004, 76, 4765-4778. Kleindienst, T. E.; Smith, D. F.; Edney, E. O.; Driscoll, D. J.; Speer, R. E.; Weathers, W. S. Secondary organic aerosol formation from the oxidation of aromatic hydrocarbons in the presence of dry submicron ammonium sulfate aerosol. Atmos. Environ. 1999, 33, 3669-3681. Smith, D. F.; Kleindienst, T. E.; Hudgens, E. E.; Bufalini, J. Measurement of organic atmospheric transformation products by gas chromatography. Int. J. Environ. Anal. Chem. 1994, 54, 265-281. Lewandowski, M.; Kleindienst, T. E.; Edney, E. O.; Jaoui, M. Composition of PM2.5 during the summer of 2003 in Research Triangle Park, North Carolina, USA. Presented at the 22nd Annual American Association for Aerosol Research National Meeting, Anaheim, CA, October 2004. Guenther, A.; Hewitt, C. N.; Erickson, D.; Fall, R.; Geron, C.; Graedel, T.; Harley, P.; Klinger, L.; Lerdau, M.; McKay, W. A.; Pierce, T.; Scholes, B.; Steinbrecher, R.; Tallamraju, R.; Taylor, T.; Zimmerman, P. A global model of natural volatile organic compound emissions. J. Geophys. Res.: Atmos. 1995, 100, 88738892. Kuba´tova´, A.; Vermeylen, R.; Claeys, M.; Cafmeyer, J.; Maenhaut, W.; Roberts, G.; Artaxo, P. Carbonaceous aerosol characterization in the Amazon basin, Brazil: novel dicarboxylic acids and related compounds. Atmos. Environ. 2000, 34, 5037-5051. Jaoui, M.; Kamens, R. M. Mass balance of gaseous and particulate products analysis from alpha-pinene/NOx/air in the presence of natural sunlight. J. Geophys. Res.: Atmos. 2001, 106, D12, 12541-12558. McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra; University Science Books: Sausalito, CA, 1993. Jaoui, M.; Kamens, R. M. Mass balance of gaseous and particulate products from β-pinene/O3/air in the absence of light and β-pinene/NOx/air in the presence of natural sunlight. J. Atmos. Chem. 2003, 43, 101-141. Vereecken, L.; Peeters, J. Theoretical Study of the Formation of Acetone in the OH-Initiated Atmospheric Oxidation of R-Pinene. J. Phys. Chem. A 2000, 104, 11140-11146.
Received for review November 30, 2004. Revised manuscript received May 18, 2005. Accepted May 18, 2005. ES048111B
VOL. 39, NO. 15, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5673
