Characterization of Organic Compounds Collected during

Characterization of Organic Compounds Collected during Southeastern Aerosol and Visibility Study: Water-Soluble Organic Species ...
0 downloads 0 Views 233KB Size
Environ. Sci. Technol. 2005, 39, 707-715

Characterization of Organic Compounds Collected during Southeastern Aerosol and Visibility Study: Water-Soluble Organic Species LIYA E. YU,* MICHELLE L. SHULMAN,† ROYAL KOPPERUD, AND LYNN M. HILDEMANN Environmental & Water Studies Program, Department of Civil and Environmental Engineering, Stanford University, Stanford, California 94305-4020

As part of the Southeastern Aerosol and Visibility Study (SEAVS), water-soluble organic species (WSOS) in fine aerosols collected from July 15 to August 25, 1995, at the Great Smoky Mountain National Park, Tennessee (USA), were chemically classified into seven groups, with concentrations ranging from around 1 to >200 ng/m3. Dicarboxylic acids represented the dominant identified compound class, and succinic acid was the most abundant dicarboxylic acid. The trends in data suggest that most WSOS collected in the SEAVS samples were mainly generated from secondary photochemical reactions, especially during the first (cleaner) half of the sampling campaign. High relative humidity at the sampling site resulted in substantial water uptake by the aerosols, which may have enhanced the levels of succinic acid by reducing its rate of photooxidation. Concurrent trends in malic and malonic acid concentrations suggest these were generated from the oxidation of succinic acid. Consistent with the conversion of 3-hydroxypropanoic acid to malonic acid, it appears that 4-hydroxybutanoic acid served as a major precursor contributing to high levels of succinic acid in the daytime. Nocturnal WSOS generally followed the trend of diurnal WSOS, but they exhibited different chemical compositions and lower concentrations, unlike what has been reported for an urban site. A nocturnal-to-diurnal ratio of succinic acid larger than 0.25 may indicate an atmosphere dominated by photochemical reactions, rather than by primary emissions.

Introduction Water-soluble organic species (WSOS) have been receiving increasing attention because they have been postulated to be partially responsible for the water uptake of airborne particulate matter (1). This water uptake is viewed as important because it can substantially affect the physical and chemical properties of atmospheric aerosols. Laboratory studies examining the hygroscopic properties of a limited number of oxygenated organic compounds have * Corresponding author phone: +65 6874 6474; fax: +65 6779 1936; e-mail: [email protected]. Present address: Department of Chemical and Biomolecular Engineering, Singapore. † Present address: Department of Chemistry, St. Mary’s College, CA. 10.1021/es0489700 CCC: $30.25 Published on Web 12/21/2004

 2005 American Chemical Society

shown that the water uptake of aerosols containing WSOS depends on the mixing status (external vs internal mixing) of the aerosol components, the nature of the oxygenated organic components, and the inorganic components (2-4). Limited measurements of ambient aerosols have shown that aerosol particles are hygroscopic to differing extents (5-7). Some more recent work has studied ambient WSOS generated from specific emission sources (8-10), or focused on the bulk characteristics of oxygenated substituents and the carbon content in the water-soluble fractions of collected aerosols (11-13). The concentration distribution of WSOS measured in urban areas seems to be location specific (14, 15). Thus, it may not be applicable to other cities, let alone to more remote environments. The water uptake of organic aerosols may be especially important in rural locations where polar compounds generated from biogenic emissions and photochemical reactions can predominate over anthropogenic emissions. Thus, it is of interest to characterize individual WSOS at rural locations, to extend our understanding of the potential importance of relevant chemical reactions in the atmosphere. As part of the Southeastern Aerosol and Visibility Study (SEAVS), this study chemically characterized fine organic aerosols collected from July 15 to August 25, 1995, at the Great Smoky Mountain National Park in rural Tennessee in the eastern United States. The data obtained during the SEAVS campaign are unique because comprehensive measurements of organic components, inorganic components, and condensation nuclei (CN) in parallel with meteorology and visibility data can be utilized to provide additional insights regarding atmospheric conditions and chemical reactions. This paper focuses on examining the trends in the identified WSOS to assess the relative importance of potential processes for atmospheric chemical transformations at rural locations. Correlations between the concentration distributions of diurnal and nocturnal WSOS are also examined and discussed.

Experimental Section A total of 42 daytime and 10 nighttime samples were collected during the SEAVS study. The sampling procedure and chemical analyses have been described in detail elsewhere (16). In brief, each 10-12 h sample was collected onto a set of Teflon-coated glass fiber filters downstream of cyclones that achieved a 2.1 µm size cut. Filter samples from every other day (along with all nighttime samples) underwent water extraction followed by solvent extraction. The water extracts were derivatized using a silylation reagent before GC-MS analysis. A butylation method was also utilized to specifically measure the concentration of oxalic acid. Solvent blanks and field blanks were analyzed to assess for background contaminants. An internal standard (perdeuterated pimelic acid or perdeuterated tetracosane, Cambridge Isotope, USA) in conjunction with a co-injection standard (1-phenyldodecane, Aldrich, USA) allowed consistent quantification of the identified compounds. The internal standard automatically adjusts for variations in sample recovery and derivatization efficiency, while the coinjection standard corrects for any variations in sample injection efficiency and instrument response. The percentage match against the library standard mass spectra is included in the species quantification. Because it was not possible to collect duplicate samples in the field for an experimental determination of precision, the uncertainties must instead be estimated theoretically. The sources of uncertainty we VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

707

TABLE 1. Concentration (ng/m3) and Occurrence of Identified WSOS in 21 Daytime and 10 Nighttime SEAVS Samples daytime concentration monocarboxylic acid glyoxylic acid 3-oxobutanoic acid 4-oxopentanoic acid decanoic acid hydroxycarboxylic acid 2-hydroxy-2-propenoic acid 3-hydroxypropanoic acid 4-hydroxybutanoic acid 3-hydroxybutanoic acid 3-hydroxy-3-methylvaleric acid 2-hydroxypentanoic acid 3-methyl-2-hydroxy-2-pentenoic acid 2,3,4-trimethoxymandelic acid dihydroxycarboxylic acid 2-methyl-2,3-dihydroxypropanoic acid 2,3-dihydroxypropanoic acid dicarboxylic acid methylpropanedioic (methylmalonic) acid propanedioic (malonic) acid butanedioic (succinic) acid bromobutanedioic acid hexanedioic (adipic) acid heptanedioic (pimelic) acid 1,2-benzenedicarboxylic (phthalic) acid 4-oxo-heptanedioic acid 1,4-benzenedicarboxylic (terephthalic) acid 2-hexenedioic acid 2-methoxyiminohexanedioic acid 1,2-cyclohexanedicarboxylic acid hydroxydicarboxylic acid hydroxybutanedioic (malic) acid 2-hydroxypentanedioic acid 2,3-dimethyl-3-hydroxyglutaric acid alcohol and polyols glycerol 2-methyl-1,4-dihydroxybutane 2-methyl-2-hydroxyl-3-butene 3-methylene-1,4-dihydroxybutane 2-hydroxy-3-penten γ-lactone pentonic acid 4,5-dimethyl-2,6-dihydroxypyrimidine galactose oxime levoglucosan others 4-hydroxycyclohexanone 1,3-butanediamine 2,3,5-trimethyl-1H-indole 3-pyridinecarboxaldehyde

nighttime occurrence

3.3b 5.7-21.1a 4-4.1b 3.2-11.6c

1 6 2 9

NDd 1.2-64.5a 5.4-178.4a 16.1a 4.2b 9.7-20.1c 8.5-12.3c 52.3c

10 17 1 1 3 2 1

2.2-11.4a 1.7-6.9b

8 6

concentration NDd 2.8-4.9a NDd NDd 2.8b 1.5b 21.2-364.7a NDd NDd NDd NDd NDd

occurrence

2

1 1 10

0.8-3.9b NDd

4 2 1 6

2 3 1 1 1 1

13.7-19.1b 2.3a 12.6-27.6a NDd NDd 3.6b NDd NDd NDd NDd NDd NDd

2.5-38.5a 2.6-11.8b 1.5-4.0b

9 2 2

4.7-14a 1.5b NDd

2 1

1.4-9.6c 2.1-19.7c 3.1-7.7c 3.5b 2.8-21.1b 13.4a 4.1a 4.0b 4.1-11.9a

3 6 6 1 7 1 1 1 4

19.9c 1.8b NDd 1.1b NDd NDd NDd NDd 3.5b

1 1

4.9b 4.6b 6.2b 6.6b

1 1 1 1

NDd NDd 4.3-4.6b NDd

52.1a 4.4-28.6a 20.3-209.5a 22.2c 2.1-18.7b NDd 6.5-10.7a 10.7-28.1b 9a 5b 28.6a 12.6b

1 8 18 1 9

1

1

1

2

a Probable: match with library mass fragmentation g70% and the compounds confirmed with standards. b Possible: 50% < match with library mass fragmentation < 70%. c Tentative: occurrence > 2 and 30% < match with library mass fragmentation < 50%, or concentration > one-fourth of total concentration of identified WSOS or SSOS. d ND, not detected.

considered are the measured sample flow rates and the physical measurement of the internal standard and the coinjection standard. Propagating these sources of uncertainty, we theoretically estimate that each quantification of an identified species is precise to within a confidence interval of (5%.

Results and Discussion The identified compounds in the SEAVS samples are categorized into three groups: (1) organics that are water-soluble regardless of pH values (i.e., very water soluble), such as malonic acid; (2) organics that can be water-soluble, depending on pH values (e.g., 1,2-benzenedicarboxylic acid); and (3) organics that are non-water-soluble regardless of pH values (solvent-soluble). Considering the variation in temperatures and pH values expected in a typical atmospheric 708

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005

environment, groups 1 and 2 are classified as water-soluble organic species (WSOS), while group 3 represents solventsoluble organic species (SSOS) in this study. Such a classification scheme is necessary for consistent assessment of “WSOS” versus “SSOS” (or non-water-soluble species) in different aerosol samples. While the SSOS are given in detail elsewhere (16), the discussion here focuses on the identified WSOS in the daytime and nighttime samples during the SEAVS study. These WSOS are listed in Table 1 along with the ranges of concentrations measured and their frequencies of detection. Most of the WSOS were further classified into six categories based on the number of carboxylic groups and hydroxyl substituents: (1) monocarboxylic acids, (2) hydroxyl-carboxylic acids, (3) dihydroxy-caboxylic acids, (4) dicarboxylic acids, (5) hydroxyl-dicarboxylic acids, and (6) alcohol and

FIGURE 1. Concentration distribution of daytime WSOS classes during the SEAVS. polyols. WSOS containing elements or structures that did not belong to any of these six categories appear under a seventh category, “others”, in Table 1. Overall, most of the identified WSOS contained multiple oxygenated substituents, demonstrating the complicated composition of atmospheric organics in the Great Smoky Mountain National Park. The WSOS concentration dominated the total identified organic compounds at the Great Smoky Mountains during the summer of 1995, typically ranging between 51% and 98% of the total for the daytime samples; the relationship between WSOS and SSOS identified during the SEAVS study is discussed in more detail elsewhere (16). Trends in WSOS with Local Meteorological Conditions. Figure 1 shows the mass concentration of total identified WSOS and the individual compound classes throughout the sampling period between Julian Days (JDs) 196-237. In general, during the first two-thirds of the study period, the trends in total WSOS appeared to be similar to the volume concentration of dry accumulation mode aerosols (defined as particulate matter with diameters between 0.1 and 0.5 µm, or PM0.1-0.5) reported by Ames et al. (17) and the mass concentration of PM2.5 aerosols reported by Andrews et al. (18) as well as Malm et al. (19); all three groups reported especially low PM concentrations during JDs 203-211 and 214-220. A forest fire such as the one that occurred during JDs 196-204 (18, 20) would generate a significant amount of oxygenated compounds, which could explain the unusually high level of WSOS on JD 196 (Figure 1); this represented 79% of the total identified organic compounds on that day. The meteorological conditions reported by Sherman et al. (21) included a dust event (characterized by high soil dust concentrations along with higher-than-average wind speeds) during JDs 204-207. During this time, the total WSOS concentration remained low (Figure 1), suggesting that this long-range transport event was an insignificant source of WSOS. While WSOS concentrations decreased during the episode of Hurricane Erin (JDs 215-217), a substantial increase in WSOS occurred during the subsequent transition period (JDs 221-225), as shown in Figure 1. This is consistent with the trends observed in aerosol number and volume for PM0.1-0.5 (17), as well as for the PM2.5 mass (18, 19); however, the PM2.5 organic carbon mass concentration stayed relatively constant (21). Stagnant weather conditions during the transition period (JDs 221-225) may have enhanced WSOS by allowing increased concentrations of gaseous precursors to accumulate.

Along with high particle concentrations and poor visibility, the hazy period (JDs 226-230) had the largest concentrations of sulfate and ozone in this study (21); however, the total WSOS concentration on JDs 228-230 was quite low, as shown in Figure 1. On the basis of this, we hypothesize that the levels of PM became elevated due mainly to emissions from fossil-fuel combustion sources, not secondary photochemistry of natural emissions. Consistent with this, the concentration of identified SSOS increased during the hazy period, particularly on JD 228, when more than 30% of the identified organics were relatively nonpolar (16). The high WSOS concentration on JD 226 (Figure 1) may reflect the residual effects of stagnant weather on accumulated WSOS during the transition period; this day had by far the lowest average wind speed of the entire study period (21). Major Classes and Species of WSOS. Dicarboxylic acids were the dominant WSOS class identified, followed by alcohol-polyols and hydroxy-carboxylic acids (Figure 1). These three most prominent classes accounted for, on average, 83% of the total identified WSOS in the daytime samples. Most of the dominant species in the three compound classes had relatively short chain lengths of 3-5 carbons along with multiple oxygenated substituents (Table 1). Our WSOS categories were based on the numbers of hydroxyl and carboxylic substituents, so aldehyde/ketonecontaining monocarboxylic acids were listed in the same category as the alkanoic acids (Table 1). Surprisingly, only one n-alkanoic acid, decanoic acid, appeared in the SEAVS samples. It was ubiquitous throughout most of the study period, except during the hurricane episode (JDs 215-217), posthurricane period (JDs 218-220), and polluted periods (JDs 226-237). This suggests that the presence of decanoic acid was mainly associated with natural rather than anthropogenic emissions. Most of the identified WSOS contained multiple oxygenated substituents. As shown in Table 1, succinic acid and 4-hydroxybutanoic acid were the most dominant species in terms of concentration and frequency of detection, in both the daytime and the nighttime samples. Many prominent WSOS in various classes exhibited concentration trends closely associated with some of the dicarboxylic acids. Thus, instead of analyzing individual WSOS classes separately, the following discussion focuses on the concentration distributions of dicarboxylic acids and incorporates relevant species from the other classes of WSOS. VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

709

Dicarboxylic Acids. The dicarboxylic acids found in this study accounted for 27-94% of the total identified WSOS, with a peak concentration of 210 ng/m3 on JD 224. In addition to the C3 to C7 R,ω-dicarboxylic acids, Table 1 shows that a methylated dicarboxylic acid (methylmalonic acid, C3), a keto dicarboxylic acid (4-oxopimelic acid, C7), an unsaturated dicarboxylic acid (2-hexenedioic acid), two aromatic dicarboxylic acids (phthalic acid and terephthalic acid), and a cyclic dicarboxylic acid (1,2-cyclohexanedicarboxylic acid) were identified. A large amount of dicarboxylic acids was observed during the stagnant weather conditions present during the transition period, peaking on JD 224; this coincided with an increase in the number of dry-accumulation-mode particles (PM0.1-0.5) (17), but there were insignificant changes in the measured fine organic mass and PM2.5 mass (18, 19). This suggests that secondary atmospheric chemical reactions, rather than additional emissions, resulted in the high concentration of dicarboxylic acids on JD 224. The total concentration of dicarboxylic acids peaked on JD 232, which was after, rather than during, the highly polluted period, giving support to the hypothesis that the dicarboxylic-acid class resulted from secondary reactions in the study area. Distributions of r,ω-Dicarboxylic Acids. Consistent with previous studies conducted in urban areas (14, 22-24), in polar regions (25-27), and on wet samples (28), the concentration of dicarboxylic acids dominated over that of other polar compounds for this study. However, the concentration distributions of various R,ω-dicarboxylic acids found in the SEAVS samples differed markedly from those reported by other studies. In other studies, oxalic acid (C2) has been generally found at the highest concentrations, followed by malonic acid (C3) or succinic acid (C4), and occasionally along with substantial amounts of adipic acid (C6) and/or phthalic acid (C8) (9, 14, 22, 24, 25, 29). However, this study shows that succinic acid was present at the highest concentrations among the identified dicarboxylic acids, followed by malic acid (hC4); concentrations of malonic acid and adipic acid were detected only intermittently (Table 1). Except for succinic acid, all of the dicarboxylic acids found in the SEAVS samples were present at substantially lower concentrations than in urban samples (14, 30). While primary emissions, such as diesel- and gasoline-powered vehicle exhausts, significantly affect the distribution of C2-C10 R,ωdicarboxylic acids in urban environments (8), the absence of oxalic acid, glutaric acid (C5), maleic acid, and azelaic acid (C9) further distinguished the WSOS distribution in the SEAVS samples from other studies (e.g., refs 22, 25, and 31). Hence, the unique distribution of dicarboxylic acids in this study implies a general and strong influence of photochemical reactions in this national park region. In fact, this is supported by the seasonal observation of various dicarboxylic acids in both the Arctic and the Antarctic regions: the domination of photochemical reactions over the influx of anthropogenic pollutants resulted in the concentration of succinic acid increasing much more than that of oxalic acid and malonic acid during polar sunrise in the Arctic (29) and during the summertime in Antarctica (27). Therefore, the dominance of succinic acid over other WSOS during the SEAVS study appears to be associated with strong photochemical reactions, with little if any input from anthropogenic emission sources. In addition to the distribution of R,ω-dicarboxylic acids observed, the concentration trends in cyclohexene and phthalic acid in the SEAVS samples further suggest that anthropogenic emissions had a negligible impact during the first half of the sampling campaign. Cyclohexene is known to be an anthropogenic pollutant and a precursor for C4-C6 dicarboxylic acids, in particular succinic acid (C4) and adipic acid (C6) (14). The scarcity of cyclohexene (it was only detected 710

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005

on JDs 200 and 224) and the lack of correlation between cyclohexene and adipic acid concentrations in this study (not shown) suggest that cyclohexene from anthropogenic emissions insignificantly contributed to C4-C6 R,ω-dicarboxylic acids in the Great Smoky Mountain National Park. While it is possible that cyclohexene contributed to succinic acid, the large difference in their concentrations and the absence of detectable adipic acid on JD 224 indicate that the transformation of other chemicals contributed more aggressively to the substantial amount of succinic acid present on that day. Phthalic acid was detectable in only two of the SEAVS samples (JDs 214 and 222), reflecting the insignificant impact of urban sources on this region. In addition, the absence of intermediates such as maleic acid (never detected) and methylmaleic acid (only detected once), which should be formed during the oxidation of vehicle-emitted hydrocarbons, further supports the finding that WSOS were not significantly connected with anthropogenic combustion sources during the study period. Oxalic Acid. In other studies measuring atmospheric polar organics, oxalic acid has been ubiquitous and was the predominant dicarboxylic acid reported (9, 14, 29, 32-34). Therefore, the absence of oxalic acid in the analyzed SEAVS samples is quite noteworthy. However, the absence of oxalic acid in these samples is supported by other measurements collected in parallel during SEAVS. Blando et al. (35) used FTIR to sequentially characterize fine particulate matter samples via a series of solvent rinsings. They observed that the carbonyl absorbance was removed in the acetone rinse step, suggesting that it was not oxalic acid (which tests showed was not removed by acetone). The lack of primary emission sources (such as vehicle emissions) would result in less airborne oxalic acid, but the absence of oxalic acid further implies that the depletion of oxalic acid outweighed its formation via photochemical oxidation of precursors at our study site. Oxalic acid and malonic acid have been hypothesized as secondary products from the degradation of dicarboxylic acids with higher carbon numbers (33). The small amount of malonic acid measured in our SEAVS samples (Table 1) supports the hypothesis that little, if any, oxalic acid was generated from this secondary chemical transformation pathway during the SEAVS study. In addition, while glyoxylic acid has been postulated as one of the precursors forming oxalic acid through atmospheric oxidation processes (25), and reported to be ubiquitous (9, 23-25), the presence of a small amount of glyoxylic acid on just one day (JD 214, not shown) lends further support to the absence of oxalic acid at the SEAVS sampling site. The strength and/or nature of photodegradation pathways could also potentially influence the concentration of succinic acid relative to other diacids. For example, a strong effect of photon-induced reactions on the distribution of dicarboxylic acids would be reflected by a higher concentration of malonic acid than succinic acid in aerosol samples (14). If the smaller R,ω-dicarboxylic acids (such as oxalic acid (C2) and malonic acid (C3)) observed were derived from the oxidation of larger ones, as hypothesized by Kawamura et al. (25), additional factors at the sampling site appeared to hinder the subsequent oxidation of succinic acid. Interestingly, succinic acid and adipic acid were most frequently observed in the SEAVS samples, but exhibit relatively low solubility and low vapor pressures among the C2-C9 R,ω-dicarboxylic acids, as shown in Table 2. This indicates that, while these two diacids tend to mainly partition into the particle phase, they will not exist in high concentrations in an aqueous solution. Unlike other rural sites where oxalic acid has been reported (25, 27, 34), the SEAVS samples were collected in the summertime under conditions of warm temperatures (21-

TABLE 2. Particle-Phase Concentration, Vapor Pressure, and Solubility of Selected r,ω-Dicarboxylic Acids

compound

solubility in watera (g solute per 100 g of water)

vapor pressureb (mmHg)

particle phasec (ng/m3)

oxalic acid (C2) malonic acid (C3) succinic acid (C4) glutaric acid (C5) adipic acid (C6) pimelic acid (C7) azelaic acid (C9)

14.3 76.3 8.3 160 3.1 5 0.2