Formation of secondary organic aerosol from irradiated α-pinene/toluene/NOx mixtures and the effect of isoprene and sulfur dioxide



[1] Secondary organic aerosol (SOA) was generated by irradiating a series of α-pinene/toluene/NOx mixtures in the absence and presence of isoprene or sulfur dioxide. The purpose of the experiment was to evaluate the extent to which chemical perturbations to this base-case (α-pinene/toluene) mixture led to changes in the gas-phase chemistry which strongly influences mass and composition of SOA and secondary organic carbon (SOC) formed. The chemical composition was examined by gas chromatography-mass spectrometry (GC-MS) and laser desorption ionization-mass spectrometry (LDI-MS). The results showed that the addition of isoprene to the base-case mixture significantly lowered the amount of toluene reacted, and thereby lowered the amount SOC produced. Simultaneous measurement of the organic NOy showed that reactions of isoprene effectively sequester NO2 by producing gas-phase organic nitrates. The addition of SO2 to the base-case mixture, while having little effect on the gas-phase chemistry, formed sulfuric acid which led to a modest enhancement of the SOC through acid-catalyzed or sulfur-incorporating reactions of α-pinene. The contribution of each hydrocarbon to the composition of the SOA was estimated using an organic tracer method. SOC from the tracer technique tended to underpredict the measured SOC, although the underprediction was especially pronounced with SO2 present. A comparison of the chromatographic results from samples of the irradiation of the α-pinene/toluene/isoprene/NOx/SO2 mixture and ambient PM2.5 showed the presence of two unique peaks that were associated with reactions of isoprene and SO2.

1. Introduction

[2] Oxygenated organic compounds constitute a significant fraction of total ambient PM2.5. These compounds can be emitted directly into the atmosphere or produced by atmospheric oxidation of volatile organic compounds (VOCs). Semivolatile and nonvolatile oxidation products partition into the aerosol phase to produce secondary organic aerosol (SOA). Both anthropogenic and biogenic VOCs can serve as precursors to SOA. Owing to the potential role of PM2.5 in visibility reduction, climate forcing, and adverse health effects [Charlson et al., 1992; Schwartz et al., 1996; Andreae and Crutzen, 1997], identification of the chemical composition and major sources of PM2.5 will help to implement strategies for the control of PM2.5. Contributions of SOA to particulate organic carbon (OC) constitute a substantial portion of PM2.5.

[3] The representation of SOA formation in atmospheric models, such as the Community Multiscale Air Quality (CMAQ) model, is based on laboratory smog chamber data on SOA yields of individual hydrocarbons. In the models, a chemical mechanism represents the oxidation of each parent VOC to yield semivolatile products that partition between the gas and aerosol phases. The parameters that govern the partitioning are derived from the laboratory data on photooxidation of single VOC/NOx mixtures. The atmosphere contains a variety of hydrocarbons that are SOA precursors, and the question arises of the extent to which the presence of multiple hydrocarbons, each of which is being oxidized to form SOA, alters SOA formation from that measured in single-hydrocarbon laboratory experiments. In addition, SO2 is ubiquitous in the atmosphere, and the same question can be posed in terms of the presence of SO2.

[4] To address this question it is helpful to consider the process of VOC oxidation as it affects SOA formation. For most atmospheric VOCs, reaction with the hydroxyl (OH) radical is the principal step initiating the mechanism to SOA formation. Initial reaction products may, themselves, react further with OH, leading eventually to the suite of semivolatile and nonvolatile products that constitute SOA. SOA formation is measured as the mass of SOA formed when a certain mass of VOC is consumed. Hydroxyl radical levels are determined by the manner in which OH is generated. In classical photooxidation experiments, OH is generated by the NOx chemistry and by the VOC degradation itself. Alternatively, OH can be generated by external means (such as, the photolysis of nitrous acid, methyl nitrite, or hydrogen peroxide). Either way, with more than one VOC, the OH level and reactions in the HOx system are perturbed from those when only a single VOC is present. In general, it is expected that each VOC oxidation mechanism will proceed as if that VOC is the sole one present, although perhaps at a different rate owing to the perturbed OH level. The mechanism leading to semivolatile products for each VOC could be different if peroxy radicals, for example, generated after the initial oxidation step inter-react; this is relevant only under low NOx conditions and, even then, is not expected to be dominant reaction path for peroxy radicals. The absorbing organic aerosol will be a mixture of products from all the VOCs present, although it is expected that the main factor governing gas-aerosol partitioning is the total mass of organic aerosol (OA), not the origin of the OA components.

[5] The motivation for performing photooxidation experiments with VOC mixtures is to assess the extent to which we understand the SOA formation observed and any interactions that occur. In the present work, SOA formed from the photooxidation of α-pinene/toluene/NOx mixture is considered as a base case which is to be perturbed. Experiments were conducted in a continuous flow, steady state chamber that has been used in a number of previous SOA formation studies [Kleindienst et al., 2004; Edney et al., 2005; Jaoui et al., 2006]. The experiments were carried out in four stages with a single change in reactant conditions in going from one stage to the next. The chamber operates as a continuous stirred-tank reactor and thus all concentrations and parameters (e.g., temperature, relative humidity) can be held constant for a single set of conditions. In addition, the effect of acidity on SOA formation was investigated by adding SO2 to the chamber where it is oxidized to sulfuric acid aerosol, which serves as the substrate for the condensation of semivolatile organics. In section 2 the experimental approach is described.

2. Experimental Methods

2.1. Apparatus and Instrumentation

[6] Experiments were carried out in a 14.5 m3 parallelepiped, Teflon-coated, stainless-steel chamber operated as a continuous flow reactor with a total flow rate of 40 L min−1. The general operation and procedures have been described elsewhere [Kleindienst et al., 2004; Edney et al., 2005]. Toluene and α-pinene were added to the chamber through a mixing manifold by passing air through thermostated neat liquids. Isoprene, NO, and SO2 were added throughflow controllers from high-pressure cylinders. Ammonium sulfate seed aerosol at a concentration less than 1 μg m−3 was also added to the chamber through atomization.

[7] Isoprene, toluene, and α-pinene concentrations in the inlet and within the chamber were measured by gas chromatography with flame ionization detection. Temperature, relative humidity, UV light intensity, ozone, and oxides of nitrogen were measured continuously with a random uncertainty of 5%. Prior to irradiation, the oxides of nitrogen analyzer measured both NO + NO2. After the irradiation began, other organic NOy compounds formed photochemically were detected on the total-nitrogen-oxides channel of the analyzer. (In addition, an in-line nylon filter prevented nitric acid from entering the analyzer.) For the particulate matter formed, OC was measured using a semi-continuous, transmission-type, thermal optical analyzer [Sunset Labs, Tigard, OR; Birch and Cary, 1996]. For measurements of individual SOA organic products, 47-mm glass fiber (GF) filters were taken at a flow rate of 16.7 L min−1 using an in-line carbon strip denuder to remove gas-phase organic compounds. Sulfate ions were determined from Teflon filters collected at 10 L min−1 [Lewandowski et al., 2007]. SOA for LDI-MS analysis was collected on impactor plates taken at a flow rate of 1.05 L min−1. For the semicontinuous measurements (GC, carbonyl, OC) the random uncertainty was approximately 10%.

2.2. Organic Compound Analyses

[8] SOA formed in each stage of the overall experiment was analyzed for individual organic compounds by extracting the GF filters in a 1:1 dichloromethane/methanol mixture. Ketopinic acid (KPA; 21.4 μg) and d50-tetracosane (20.8 μg) were used as internal standards. Filter extracts were then derivatized with bis(trimethylsilyl) trifluoroacetamide (BSTFA) containing 1% trimethylchorosilane as catalyst and analyzed by GC-MS in the positive CI mode. Jaoui et al. [2004] determined the efficiency of BSTFA derivatization to be greater than 95%. Additional experimental details and the GC conditions follow those utilized by Jaoui et al. [2005]. Concentrations of individual compounds were measured using the KPA calibration factor as described by Jaoui et al. [2006].

[9] SOA collected on the impactor plates, as well as that collected on the GF filters, was analyzed on an Applied Biosystems Voyager-DE Pro MALDI time-of-flight mass spectrometer operated in the positive ionization mode. Impactor plates were analyzed without sample pretreatment by placing them onto a 100-spot steel target plate (LDI-MS). GF filters were extracted by sonication with HPLC-grade methanol and 4 μL was applied and allowed to dry on a 100-spot steel target plate, which was gently brushed with graphite particles to serve as the MALDI matrix. The extraction and operating conditions of the MALDI-MS instrument are described by Surratt et al. [2006]. The LDI-MS and MALDI-MS data produced similar results, and thus only LDI-MS data are presented.

2.3. Procedures

[10] The experiment was conducted in four stages. In the first stage, the base-case mixture of α-pinene and toluene (αP/Tol) was irradiated in the presence of NOx. Reactant gases were premixed in an inlet manifold at concentrations given in Table 1 and introduced into the chamber at a constant flow. The irradiated reactants were given three chamber residence times (ca.18 h) to achieve steady state conditions. Gas and aerosol samples from the steady state chamber mixture were then collected over a 24-h period. Following this stage, isoprene (Isop) was added to the inlet stream at a concentration of 4.1 ppmC. After the mixture equilibrated to the new steady state conditions, additional samples were taken. In subsequent stages, SO2 was added to the αP/Tol mixture in one case, and isoprene and SO2 were added to the αP/Tol mixture in another. Inlet concentrations for the four stages are given in Table 1.

Table 1. Initial Concentrations for Each Stage of the Experiment
Mixtureaα-Pinene (ppmC)Toluene (ppmC)Isoprene (ppmC)NOxb (ppb)SO2 (ppb)
  • a

    Average Temperature: 24°C, RH: 30%.

  • a

    αP: α-pinene, Tol: toluene, Isop: isoprene.

  • b

    The initial NOx during the irradiations was greater than 98% NO.

αP/Tol1.111.2 378 
αP/Tol/SO21.211.2 378275

3. Results

3.1. Gas and Aerosol Products

[11] Steady state concentrations of NOy, O3, HNO3, and SO2 for the four stages of the experiment are given in Table 2. As seen in Tables 1 and 2, once the αP/Tol mixture reached steady state, NO and α-pinene had completely reacted while only 30% of the initial toluene (1900 μg m−3; 3.5 ppmC) was consumed. Under these conditions, the steady state concentrations of NOy, O3, and HNO3 were 111, 127, and 79 ppb, respectively. Carbonyl products from the photooxidation of the αP/Tol mixture included formaldehyde, glyoxal, methylglyoxal, benzaldehyde, and pinonaldehyde at concentrations given in Table 3. Steady state mass concentrations of SOA and SOC were measured as 211 μg m−3 and 128 μgC m−3, respectively, as given in Table 4.

Table 2. Steady State Gas Phase and Reacted Hydrocarbon Concentrations During the Irradiations
MixtureNOy (ppb)O3 (ppb)HNO3 (ppb)SO2 (ppb)Δ (αP) (μg m−3)Δ(Tol) (μg m−3)Δ(Isop) (μg m−3)
αP/Tol11112779.1 5901900 
αP/Tol/Isop19515662.1 58010001800
Table 3. Steady State Carbonyl Concentrations (ppb) During the Irradiations (G: Glyoxal; MG: Methylglyoxal; MA: Methacrolein; MV: Methyl Vinyl Ketone; BZ: Benzaldehyde; PIN: Pinonaldehyde)
αP/Tol639383  1414
αP/Tol/SO2597076  1315
Table 4. Aerosol Parameters Measured During the Irradiations for the Four Stages
MixtureSOA (μg m−3)Volume (μm3 cm−3)SOC (μgC m−3)SO42− (μg m−3)NO3 (μg m−3)NH4+ (μg m−3)

[12] The addition of 4.1 ppmC isoprene to the αP/Tol mixture substantially changed the gas- and aerosol-phase compositions. In the gas phase, the steady state NOx concentration increased to 195 ppb while at the same time the nitric acid concentration decreased to 62 ppb and O3 increased to 156 ppb. As shown in Table 2, the reacted toluene decreased from 1900 to 1000 μg m−3 with its products, benzaldehyde and glyoxal, decreasing by approximately 25%. The reaction of isoprene led to an increase in formaldehyde by a factor of seven and methylglyoxal by 70% over the values from the αP/Tol mixture. The addition of isoprene to the base-case mixture substantially decreased the SOA produced from 211 to 104 μg m−3 and SOC from 128 to 71 μgC m−3.

[13] The addition of 0.28 ppm SO2 to the αP/Tol mixture resulted in modest increases in the NOy and nitric acid concentrations and a modest decrease in the O3 concentration. Initiated by its reaction with OH radicals, SO2 was oxidized to form sulfuric acid. The carbonyl and dicarbonyl compounds changed by minor amounts, except for glyoxal, which decreased by 25%. The addition of SO2 to the αP/Tol mixture increased the mass of the aerosol in two ways: (1) from the formation of sulfuric acid condensing onto preexisting particles and (2) from increased SOA. The SOA mass was determined by subtracting the inorganic masses (SO42−, NO3, NH4+) found in Table 4 from the measured gravimetric mass. The net result was that the SOA mass increased by approximately 16% in going from the αP/Tol to the αP/Tol/SO2 mixture.

[14] Finally, the addition of both isoprene and SO2 to the αP/Tol mixture led to an increase in the steady state NOy concentration by nearly a factor of two, an increase in O3 by 14%, and a decrease in HNO3 by more than 20%. Similarly, for the carbonyl and dicarbonyl compounds, the addition of both isoprene and SO2 resulted in concentrations that were very close to those when isoprene alone was added. Adding both compounds also left the total SOA and SOC masses largely unchanged. The sulfate concentration was found to increase to 39.8 μg m−3 from that of the base-case mixture, a concentration 63% higher than that measured in the αP/Tol/SO2 mixture.

3.2. Bulk Aerosol Properties

[15] Figure 1 shows the distribution of molecular fragment ions from the composite aerosol from the (+)LDI-MS spectra in the range of 200 < m/z < 1000 for the four mixtures. The mass spectra can be directly compared because the chamber air sampled and the amount of extract analyzed for each stage were the same. A detailed examination of the (+)LDI-MS mass spectra for ions up to 800 u shows characteristic ion differences of m/z 14, 16, and 18 Da, which suggests the aerosols have an oligomeric character that is consistent with earlier work [Kalberer et al., 2004; Gao et al., 2004a, 2004b; Tolocka et al., 2004]. In mixtures containing SO2, an enhancement in the ion intensities and a shift to higher masses suggest that reactions in the condensed phase involving sulfuric acid could be occurring, which is consistent with previous detection of organosulfates in these samples [Surratt et al., 2007a]. This finding is also consistent with increased OC measured in these stages, as well as increases in OC measured in other studies where SO2 is present in irradiated HC/NOx mixtures [Edney et al., 2005; Kleindienst et al., 2006]. The addition of isoprene to the αP/Tol mixture led to a decrease in the intensities of the ions both in the presence and absence of SO2.

Figure 1.

LDI-MS positive ion mode spectra generated from impactor plates collected in each stage of the experiment. The same volume of air was sampled on each impactor.

[16] Photographs of the filters (F1–F4) collected from each stage, shown in Figure 2, indicate distinct differences in the products formed for the different mixtures. The filter from the αP/Tol mixture (F1) shows a yellow hue that is close in appearance to filters previously collected in this laboratory from toluene/NOx experiments. Consistent with a decrease in reacted toluene in the αP/Tol/Isop system, filter F2 has lost most of the color seen in F1. The addition of SO2 to the αP/Tol mixture produced material (filter F3) of a deep brown color. Finally, the aerosol from the αP/Tol/Isop/SO2 mixture resulted in filter F4 having a pronounced brown color, although considerably lighter than that from the αP/Tol/SO2 mixture.

Figure 2.

Effect of mixture changes in filter appearance in going from the αP/Tol/NOx (base-case) filter (F1) to subsequent mixtures. The same volume of air was sampled on each filter.

3.3. Nitrogeneous Organic Compounds

[17] A number of individual organic compounds have been detected in the SOA extracts of the mixtures (Figure 3). Of these, a group of five compounds (N-1 to N-5; Table 5) exhibit odd molecular weights for their BSTFA derivatives, suggesting the presence of a nitrogenous moiety. Compounds N-1 and N-2 were detected only when isoprene, NOx, and SO2 were present in the irradiated mixture, while N-3, N-4, and N-5 were detected only when toluene was present. The common BSTFA fragments for the derivatization of acidic or alcoholic OH groups were detected for these compounds [Jaoui et al., 2004]. Figure 4 shows mass spectra recorded for compounds N-1, N-2, and N-5. The mass spectrum of N-1 shows fragments and adducts consistent with a derivative compound with MW 253 Da., while N-2 fragments and adducts are consistent with a compound with MW of 313 Da [Jaoui et al., 2007]. Compounds N-3, N-4, and N-5 have similar fragmentation patterns, with the ion at m/z M+. + 1 (314) showing the highest intensity. While N-2 shows similar ions as N-3, N-4, and N-5, it also exhibits a strong M+. - 15 ion at m/z 298. Peaks associated with N-3, N-4, and N-5 are also consistent with the results from a combined carbonyl-OH derivatization that suggests the absence of carbonyl (C = O) groups. N-5 was identified as 2-hydroxy-5-nitrobenzyl alcohol (MW of 169 Da) by comparison to a standard of the authentic compound. N-3 and N-4 were tentatively identified as structural isomers of N-5 from the similarity of their mass spectra with 2-hydroxy-5-nitrobenzyl alcohol. Several isomers of 2-nitrophenol and 2-methyl-2-nitrophenol were also detected in the mixtures. Table 6 gives concentrations for the nitrogenated compounds, based on standards or as ketopinic acid where standards are not available.

Figure 3.

GC-MS extracted ion chromatograms in CI mode of BSTFA derivatized particle phase collected in each of the four stages of the experiment. Ions extracted: 165, 238, 298, 321, 405, 409, 314, 349, 377, and 495.

Figure 4.

Mass spectra of nitrogen containing compounds N-1, N-2, and N-5.

Table 5. Organic Compounds in the Mixtures Identified From the GC-MS Analysis
Chemical NameIDMWMWaPrecursor VOC
  • a

    MW as BSTFA derivative.

  • b

    Detected when SO2 and isoprene present.

  • c

    Detected when toluene present.

Tracer Compounds
2-Methylglyceric acidI-1120336Isop
3-Hydroxyglutaric acidP-1148364αP
2-Hydroxy-4,4-dimethyl glutaric acidP-2176392αP
Pinic acidP-3186330αP
3-Methyl-1,2,3-butanetricarboxylic acidP-4204420αP
2,3-Dihydroxy-4-oxo pentanoic acidT-1148364Tol
Other Compounds
2-Methylglyceric acid dimerbI-4222510Isop
Nitrogen containing compoundbN-1181253Isop
Nitrogen containing compoundbN-2169313Isop
2-Hydroxy-5-nitrobenzyl alcohol and isomerscN-5, N-3, N-4169313Tol
Table 6. Estimated Concentrations (μg m−3) of Nitrogen Containing Compounds Using the Response Factor for the Ketopinic Acid BSTFA Derivative
αP/Tol  1.570.301.80
αP/Tol/SO2  1.810.361.39

3.4. Organic Tracer Compounds and SOC Contributions

[18] The oxidation of individual biogenic and aromatic hydrocarbons also leads to a variety of oxygenated compounds which have been identified as tracer compounds for their parent VOC; many of these have also been detected in ambient PM2.5 [Edney et al., 2005; Kleindienst et al., 2004; Jaoui et al., 2005; Surratt et al., 2007a, 2007b; Claeys et al., 2007; Gómez-González et al., 2008]. The organic SOA tracers for the parent VOCs were found in all stages. Table 5 gives compound names, their MWs, and the derivative MWs for the identified tracer compounds and the associated parent VOC [Kleindienst et al., 2007], who also discuss the uncertainties associated with the method.

[19] The tracers present in the GC-MS extracted ion chromatograms of the four stages are shown in Figure 3. Compounds associated with the photooxidation products of isoprene are 2-methylglyceric acid (I-1), 2-methylthreitol (I-2), and 2-methylerythritol (I-3). Peak I-4 has recently been found to be associated with the NOx photooxidation of isoprene and identified as a 2-methylglyceric acid dimer (2-MGAD) [Szmigielski et al., 2007a]. This compound was detected in the αP/Tol/Isop and αP/Tol/Isop/SO2 mixtures. The concentration of 2-MGAD was found to increase significantly with increasing SO2, as did N-1 and N-2 [Jaoui et al., 2007]. Compounds associated with α-pinene SOA included 3-hydroxyglutaric acid (P-1), 2-hydroxy-4,4-dimethylglutaric acid (P-2), and 3-methyl-1,2,3-butanetricarboxylic (P-4) which are among the largest peaks detected in the GC-MS total ion chromatogram (TIC) and consistent with previous laboratory and field experiments [Jaoui et al., 2005; Szmigielski et al., 2007b]. For toluene, a single compound detected in SOA (2,3-dihydroxy-4-oxopentanoic acid: T-1) is used as the tracer compound. For stages with SO2 present, a strong BSTFA-derivative peak for particle-phase sulfuric acid is also detected at an early retention time. Finally, several compounds, measured in these experiments using ESI-MS but previously reported [Surratt et al., 2007a], were characterized as organosulfates of isoprene and α-pinene [Surratt et al., 2007b; Gómez-González et al., 2008].

[20] The contribution to the SOC mass from the individual precursors can be determined using an SOC tracer method developed in this laboratory [Kleindienst et al., 2007] and evaluated with binary hydrocarbon mixtures using 14C analysis [Offenberg et al., 2007]. The method consists of measuring, for each hydrocarbon precursor, the concentrations of the organic tracer compounds (Table 5) and SOC from laboratory hydrocarbon/NOx irradiations to obtain an SOC mass fraction, fsoa,hc, for each precursor. To further test this tracer technique for mixtures, the identified organic tracers from Table 5 are used with the previously derived mass fractions (0.23, 0.0079, and 0.16 for α-pinene, toluene, and isoprene SOC, respectively; [Kleindienst et al., 2007]) to obtain estimated SOC masses which can then be compared to the SOC measured in this experiment. Table 7 gives the concentrations of SOC attributed to α-pinene, toluene, and isoprene for each mixture using this tracer method. For the αP/Tol mixture, the estimated and measured SOC concentrations are 123 and 128 μg m−3, respectively. For the αP/Tol/Isop mixture, the α-pinene and toluene contributions decrease significantly and while there is a small SOC contribution from isoprene, the net result is a substantial decrease in the estimated SOC from 123 to 84 μg m−3. With SO2 added to the irradiated αP/Tol mixture, the estimated SOC contribution increased slightly from 123 to 125 μgC m−3. Finally, for the addition of SO2 to the αP/Tol/Isop mixture, the estimated SOC increased from 84 to 95 μgC m−3, which is in the same direction but a lower difference than that measured (71 to 122 μgC m−3).

Table 7. Comparison of the Sum of the Contribution of Individual Hydrocarbon and the Measured OC Concentrations μgC m−3
Mixtureα-Pinene SOCToluene SOCIsoprene SOCEstimated SOCMeasured SOC
αP/Tol4875 123128
αP/Tol/SO27055 125146

[21] From the SOC contributions in Table 7 and the reacted hydrocarbon concentrations (ΔHC) in Table 2, an SOC yield, Ysoc (i.e., OCHC/ΔHC), can be determined. Table 8 gives the SOC yields for α-pinene, isoprene, and toluene for each mixture. For α-pinene, the largest change in yield occurred (35%) when SO2 was added to the αP/Tol mixture which is a similar increase as that seen by Kleindienst et al. [2006] when SO2 was added to α-pinene/NOx mixtures. For the four mixtures, the mean yields for α-pinene, toluene and isoprene were 0.107, 0.044, and 0.005, respectively. Recent laboratory studies have shown that SOA yields from aromatic hydrocarbons and monoterpenes can be a function of the initial NOx concentration [Ng et al., 2007a, 2007b].

Table 8. Precursor Yields Based on the SOC Masses From Tracer Method Given in Table 7a
Mixtureα-Pinene SOCIsoprene SOCToluene SOC
  • a

    Reacted hydrocarbon masses are found in Table 2.

αP/Tol0.102 0.051
αP/Tol/SO20.136 0.046

4. Discussion

[22] Tables 9 and 10 present measured chemical and physical parameters and the percentage changes in going between the αP/Tol mixture and the other mixtures. For example, in adding isoprene to the αP/Tol mixture, the NOx concentration increases by 76%. Four comparisons are presented: (1) αP/Tol – αP/Tol/Isop; (2) αP/Tol – αP/Tol/SO2; (3) αP/Tol – αP/Tol/Isop/SO2; and (4) αP/Tol/Isop – αP/Tol/Isop/SO2. The systematic errors in this work are not considered to be a major source of experimental uncertainty since the same procedure was used in each stage of the experiment and percent changes from the base case are reported in Tables 9 and 10. However, in many cases, the percent changes in the tables are less than the uncertainties of the measurements and are therefore not discussed.

Table 9. Percentage Change in Concentrations for Values Going From the Base-Case to Those for Different Mixturesa
ParameterUnitsαP/Tol (Base Case)+Isop (%)+SO2 (%)+Isop/SO2 (%)
  • a

    Values in comparison to the mixture containing αP/Tol/Isop.

  • b

    Absolute values.

ΔTolμg m−31900−47−15−53
ΔαPμg m−3586−2+9+12
Methyl glyoxalppb83+72−8+61
SOAμg m−3211−51+16−5
VolumenL m−3152−29+17+9
OCμgC m−3128−46+14−5
N-1bμg m−3bdlbdlbdl1.6
N-2bμg m−3bdlbdlbdl3.1
N-3μg m−31.6−84+15−60
N-4μg m−30.3−90+20−60
N-5μg m−31.8−92−23−66
Tracer SOCμgC m−3123−32+2−23
Tracer Tol SOCμgC m−375−56−27−52
Tracer αP SOCμgC m−348−10+46+8
Table 10. Percentage Change in Concentrations for Adding SO2 to the αP/Tol/Isop Mixturea
ParameterUnitsαP/Tol/Isop+SO2* (%)
  • a

    Values in comparison to the mixture containing αP/Tol/Isop.

  • b

    Absolute values.

ΔTolμg m−31000−10
ΔαPμg m−3580+14
Methyl glyoxalppb143−6
Methyl vinyl ketoneppb577
SOAμg m−3104+113
VolumenL m−3108+54
OCμgC m−371+72
N-1bμg m−3bdl1.6
N-2bμg m−3bdl3.1
N-3μg m−30.25+152
N-4μg m−30.03+300
N-5μg m−30.14+336
Tracer SOCμgC m−384+13
Tracer tol SOCμgC m−333+9
Tracer isop SOCμgC m−38−13
Tracer αP SOCμgC m−343+21

4.1. α-Pinene/Toluene Versus α-Pinene/Toluene/Isoprene Systems

[23] The first comparison examines the influence of adding a hydrocarbon, isoprene, expected to strongly perturb the radical concentrations in the system. Irradiation of the αP/Tol mixture alone resulted in complete reaction of α-pinene and 30% of toluene giving an SOA concentration of 211 μg m−3 (128 μgC m−3 for SOC; Table 4). When isoprene is added to the αP/Tol mixture, because of the large OH-isoprene rate constant of 1.0 × 10−10 cm3 molec−1 s−1 [Atkinson, 1989] the OH production and consumption are perturbed. As noted in Table 2, the steady state NOy level changed from 111 to 195 ppb, and the nitric acid level decreased from 79 to 62 ppb. Since HNO3 is produced exclusively from the OH – NO2 reaction, a decrease of HNO3 must occur from a decrease of OH, a decrease of NO2, or a decrease of both. While the NO2 concentration could not be measured directly, the increased NOy concentration is likely a result of the formation of peroxyacyl nitrates and other organic nitrates known to form in the isoprene system [Carter and Atkinson, 1996] rather than from an increase in NO2 itself. Since the reactions producing organic nitrates serve as sinks for NO2, it is likely that the steady state NO2 concentration is suppressed with isoprene present compared to the αP/Tol system. This suppression has an indirect feedback effect by decreasing the production of NO through the photolysis of NO2, thereby suppressing the NO + HO2 reaction, the chief means of producing OH. Thus it is anticipated that the reduced HNO3 concentration results from decreases in both the OH and NO2 concentrations. The reduced OH concentration is consistent with a decrease in reacted toluene as well as decreases in benzaldehyde and glyoxal, major toluene reaction products. The significant role of isoprene in the system is also consistent with substantial increases in the carbonyl and dicarbonyl products from isoprene oxidation, formaldehyde, methyl vinyl ketone, methacrolein, and methylglyoxal. The lower production of methylglyoxal from toluene is more than compensated by an increased of methylglyoxal from isoprene resulting in an overall increase of 72%. Changes in the carbonyl and dicarbonyl concentrations also strongly affect the formation of new radicals by carbonyl and dicarbonyl photolysis.

[24] The changes in the gas-phase chemistry result in the SOA and SOC concentrations in the mixture with isoprene decreasing by 51 and 46%, respectively. Additional aerosol data give a consistent picture of the changes in the SOC. From Tables 7 and 9, the decrease in the measured SOC is reflected in the reduced contributions to SOC from toluene (−56%) and α-pinene (−12%) using the tracer method. The added contribution of 8 μgC m−3 from isoprene SOC makes up for only a small fraction of the SOC lost from toluene and α-pinene. Moreover, the aerosol-phase products from the toluene reaction, N-3, N-4, and N-5, also decreased sharply. Finally, the decrease in the aerosol mass is consistent with the reduced intensity of the LDI-MS spectra and the lighter yellow color from F1 to F2 (Figure 2).

4.2. α-Pinene/Toluene Versus α-Pinene/Toluene/SO2 Systems

[25] The second comparison examines the effects of the addition of SO2 to the αP/Tol mixture. In the gas phase, the influence of SO2 is found to be minor since the OH + SO2 rate constant (0.9 × 10−12 cm3 molec−1 s−1 at 298K and 1 atm [Atkinson et al., 2004]) is a factor of seven smaller than that for the slowest reacting hydrocarbon, toluene. The competition between toluene and SO2 for the available OH radicals leads to a modest decrease in the amount of toluene reacted, as seen in Table 2. The changes in the gas-phase chemistry are minor and consistent with the small changes in most of the carbonyl products (Table 3). The main effect of SO2 is that its oxidation by OH leads to the formation of sulfuric acid aerosol in the irradiated mixture. Table 9 indicates that the aerosol parameters, SOA and SOC, increased by about 15% in the presence of SO2 compared to the αP/Tol mixture. In spite of this modest increase, a substantial increase in the intensity and molecular weight of the aerosol products in the LDI-MS spectra in Figure 1 suggests that acid-catalyzed, or sulfur-incorporating, reactions in the aerosol phase may be occurring. For example, whereas carbonyl compounds were within 10% of the base-case values, glyoxal was found to decrease in the gas phase by 25%, which would be consistent with an increased uptake on the acidic aerosol [Liggio et al., 2005; Kroll et al., 2005]. The filter substrates of the αP/Tol and αP/Tol/SO2 mixtures also show pronounced color change from light yellow to dark brown, which also would be consistent with aerosol-phase reactions occurring.

[26] From the SOC estimates given in Table 7, the contributions from α-pinene and toluene show offsetting effects with a 46% increase attributed to α-pinene SOC due to the presence of acid aerosol offset and 27% decrease in the toluene contribution owing to the decreased toluene consumption. The α-pinene SOC increase is consistent with that previously found for irradiated α-pinene/NOx/SO2 mixtures [Kleindienst et al., 2006]. The net result is an estimated SOC concentration that is essentially unchanged from the αP/Tol mixture but with a considerably different composition indicated by the LDI-MS measurements and filter color observations.

4.3. α-Pinene/Toluene Versus α-Pinene/Toluene/Isoprene/SO2 Systems

[27] The addition of isoprene and SO2 to the αP/Tol mixture was found to substantially reduce the amount of reacted toluene and increase the gas-phase NOy concentration. The aerosol mass is seen to be very similar to that of the αP/Tol mixture, which suggests offsetting effects, with the reduction in toluene aerosol products being compensated by the increase in α-pinene and isoprene products, probably through acid-catalyzed, aerosol-phase reactions [Edney et al., 2005; Kleindienst et al., 2006]. From an analysis of the tracers, the reduction in the contribution of toluene SOC is over 50%, the increase of the α-pinene SOC contribution from the presence of acidic aerosol is 8%, and isoprene SOC contributes 10% to the total estimated SOC. These results are consistent with the LDI-MS data where the two spectra are qualitatively similar. By contrast, even though the two mixtures have essentially the same aerosol mass, F1 shows a yellow color while F4 is light brown, suggesting a difference in the products.

4.4. α-Pinene/Toluene/Isoprene Versus α-Pinene/Toluene/Isoprene/SO2 Systems

[28] Table 10 shows a comparison of the αP/Tol/Isop mixtures in the presence and absence of SO2. In this comparison, the concentrations of the gas-phase compounds remain largely unchanged as seen by the concentrations of the carbonyl compounds, most of which are within 10% of each other, except for methacrolein, for which the difference is 27%. By contrast, the SOA and SOC masses increase by 113 and 72%, respectively. These sizable increases are probably a result of acid-catalyzed, or sulfur-incorporating reactions in the aerosol phase. As seen in Table 6, N-1 and N-2 appear to form only in the presence of both isoprene and acidic aerosol. The LDI-MS data show an increase in the mass intensities that is consistent with the increased SOA and SOC masses. Previous work using (-)ESI-MS showed the formation of nitroxy-organosulfates from the photooxidation of isoprene/NOx/SO2 [Surratt et al., 2007a; Gómez-González et al., 2008], and higher H+ air concentrations led to increased masses of these compounds [Surratt et al., 2007b].

[29] The aerosol carbon contributions obtained when SO2 is added to the αP/Tol/Isop mixture show an increase in the α-pinene contribution. Ng et al. [2007a] have shown that SOA production from aromatics is not affected by the presence of an acid aerosol. Somewhat unexpectedly, the isoprene contribution did not increase in the presence of acidic aerosol, in contrast to that previously observed in isoprene/NOx/SO2 irradiations [Edney et al., 2005; Kleindienst et al., 2006]. Given that there is a substantial deficit in the measured SOC contribution compared to that observed, a possible explanation may be an insufficient amount of the BSTFA derivatizing reagent for a condition where the sulfuric acid was present at high levels in the aerosol and where excess BSTFA was expected to be present (see sulfuric acid peak in Figure 5a). This explanation may also be relevant in terms of the SOC difference between the αP/Tol/Isop/SO2 and αP/Tol mixtures in Table 7.

Figure 5.

GC-MS extracted ion chromatograms in CI mode of BSTFA derivatized particle phase (a) sample collected in stage αP/Tol/Isop/SO2 and (b) PM2.5 sample collected in Research Triangle Park during the summer of 2003. Ions extracted: 165, 238, 298, 321, 405, 409, 314, 349, 377, and 495.

4.5. Applications to Field Samples

[30] The applicability of these results to ambient conditions can be demonstrated by comparing selected ion chromatograms of the αP/Tol/Isop/SO2 mixture with an ambient PM2.5 sample collected during a previous field study [Lewandowski et al., 2007]. The comparison is shown in Figure 5, where the organic compounds identified are the same as those given in Table 5. Moreover, in addition to the organic tracers and nitrogenated compounds N-1 and N-2, sulfate esters were detected both in the irradiated chamber mixtures containing SO2 and in ambient aerosol [Surratt et al., 2007a, 2007b]. The high levels of α-pinene and isoprene tracer compounds in ambient samples during the summer [Kleindienst et al., 2007] indicate that α-pinene and isoprene probably contribute strongly to the SOC component of ambient OC. In addition to the isoprene tracer compounds, 2-MGAD (I-4) also an isoprene product is seen in the field sample at relatively large abundances. Heretofore, this compound had not been detected in ambient aerosol, and thus, indicates that particle-phase organic esterification can occur under ambient conditions, consistent with thermodynamic calculations [Barsanti and Pankow, 2006]. The toluene tracer was also detected but is typically less intense in the ambient chromatogram than the isoprene or α-pinene tracers owing to the low sensitivity for the toluene tracer.

[31] In order to determine the role of acid-catalyzed or sulfur-incorporating reactions in PM2.5, it is likely that similar tracer compounds will have to be identified that are specific to the sulfur containing system. The large peak seen early in the chromatogram comes from a derivatization reaction of H2SO4 and provides an initial indicator of the presence of an acidic aerosol. Compounds N-1 and N-2 seen in both the laboratory and ambient samples could be candidates to fulfill the role as organic tracers [Jaoui et al., 2007]. However, while this study considers photochemical systems that provide insight to processes occurring in ambient atmospheres, only a single set of concentrations were studied. Thus an examination of mixtures such as these under a broader range of conditions will be required to more fully assess the applicability of these results to OC formation in ambient PM2.5.

5. Conclusion

[32] Laboratory data show that the formation of SOA of individual hydrocarbon precursors can be significantly affected by the presence of other hydrocarbons as well as SO2. For example, the formation of peroxyacyl and multifunctional nitrates, compounds serving as NOx sink, appear to, at least for this particular set of reactant concentrations, inhibit photochemical chain reactions, thereby decreasing OH radical concentrations which reduce SOA formation from aromatic hydrocarbons. The presence of SO2 in the system generates acidic aerosol, which can lead to sulfur-incorporating reactions occurring in the particle phase. Furthermore, two tracer compounds were identified that could serve as tracer compounds for these aerosol-phase reactions. Finally, the tracer technique is found to be a useful tool for estimating SOC yields from systems containing multiple hydrocarbons.


[33] The U.S. EPA through its Office of Research and Development also funded research described here under Contract EP-D-05-065 to Alion Science and Technology. This article also has been jointly developed and published by the U.S. EPA and the California Institute of Technology. It was produced under Cooperative Agreement CR83194001 and is subject to 40 CFR 30.36. The article has been reviewed by EPA personnel under EPA scientific and technical peer review procedures and approved for joint publication based on its scientific merit, technical accuracy, or contribution to advancing public understanding of environmental protection. However, the Agency's decision to publish the article jointly with Caltech is intended to further the public purpose supported by Cooperative Agreement No. CR83194001 and not to establish an official EPA rule, regulation, guidance, or policy through the publication of this article. Mention of trade names or commercial products does not constitute an endorsement or recommendation. Jason Surratt was supported in part by the U.S. EPA under the Science to Achieve Results (STAR) Graduate Fellowship Program.