Article pubs.acs.org/est

Secondary Organic Aerosol from Aqueous Reactions of Green Leaf Volatiles with Organic Triplet Excited States and Singlet Molecular Oxygen Nicole K. Richards-Henderson,† Andrew T. Pham,‡ Benjamin B. Kirk,§ and Cort Anastasio*,† †

Department of Land, Air and Water Resources, University of California - Davis, 1 Shields Avenue, Davis, California 95616, United States ‡ Cain Department of Chemical Engineering, Louisiana State University, South Stadium Road, Baton Rouge, Louisiana 70803, United States § Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States S Supporting Information *

ABSTRACT: Vegetation emits a class of oxygenated hydrocarbons - the green leaf volatiles (GLVs) - under stress or damage. Under foggy conditions GLVs might be a source of secondary organic aerosol (SOA) via aqueous reactions with hydroxyl radical (OH), singlet oxygen (1O2*), and excited triplet states (3C*). To examine this, we determined the aqueous kinetics and SOA mass yields for reactions of 3C* and 1O2* with five GLVs: methyl jasmonate (MeJa), methyl salicylate (MeSa), cis-3-hexenyl acetate (HxAc), cis-3-hexen-1-ol (HxO), and 2-methyl-3-butene-2-ol (MBO). Second-order rate constants with 3C* and 1O2* range from (0.13−22) × 108 M−1 s−1 and (8.2−60) × 105 M−1 s−1 at 298 K, respectively. Rate constants with 3C* are independent of temperature, while values with 1O2* show significant temperature dependence (Ea = 20−96 kJ mol−1). Aqueous SOA mass yields for oxidation by 3C* are (84 ± 7)%, (80 ± 9)%, and (38 ± 18)%, for MeJa, MeSa, and HxAc, respectively; we did not measure yields for other conditions because of slow kinetics. The aqueous production of SOA from GLVs is dominated by 3C* and OH reactions, which form low volatility products at a rate that is approximately half that from the parallel gas-phase reactions of GLVs.

■

INTRODUCTION

formed in a cascade of reactions that begin with sunlight absorption by organic chromophores (C)

Secondary organic aerosol (SOA) is a major component of atmospheric fine particulate matter (PM2.5), which affects visibility, cloud processes, human health,1,2 and global/regional climate.3,4 The traditional view of SOA formation is gas-phase oxidation of volatile organic compounds (VOCs) to form semivolatile products that can nucleate to form new particles or condense onto pre-existing particles.5−9 While gas-phase production of SOA is clearly important, it fails to completely explain atmospheric observations; i.e., models cannot accurately reproduce the magnitude, distribution, and dynamics of organic aerosol concentrations.10,11 Several recent insights might help address these gaps in our understanding of SOA formation.12,13 Among these is that SOA can form in atmospheric aqueous phases such as fog and cloud droplets: gases partition into the aqueous phase, react, and form low volatility products that remain in the particle phase after the droplet evaporates.14−28 Fog and cloud droplets contain a variety of oxidants, including hydroxyl radical, ozone, peroxides, peroxyl radicals, singlet molecular oxygen (1O2*), and triplet excited states of organic compounds (3C*).29−33 In this manuscript we focus on 1 O2* and 3C* since these oxidants have been less extensively studied for aqueous-phase SOA formation. These oxidants are © 2014 American Chemical Society

ISC

C + hν → 1C* ⎯⎯→ 3C* 3

C* + O2 → 1O2 * + C

(1) (2)

where 1C* is the short-lived singlet excited state of the organic chromophore, and ISC represents intersystem crossing. As illustrated in reaction 2, 3C* and 1O2* are intricately tied since ground state oxygen is typically the largest sink for triplet states, and this interaction forms singlet oxygen.34−36 In this study, we examine the possible roles of 3C* and 1O2* in the oxidation of green leaf volatiles (GLVs), which are a class of oxygenated organic compounds released by vegetation during a range of stresses, including severe weather, mechanical stress, insects, and pathogens.37−40 Atmospheric mixing ratios of GLVs are typically 100−800 pptv.41−43 GLVs are modestly soluble in water and have vapor pressures typical for semivolatile compounds, indicating they will partition into fog Received: Revised: Accepted: Published: 268

July 27, 2014 October 26, 2014 November 26, 2014 November 26, 2014 dx.doi.org/10.1021/es503656m | Environ. Sci. Technol. 2015, 49, 268−276

Environmental Science & Technology

Article

μM bead, BetaBasic-18 column (Thermo Hypersil-Keystone). HPLC conditions are listed in Supporting Information Table S2. Apparent first-order rate constants were determined for the GLV and reference (R) compounds as the negative of the slopes of plots of ln(([GLV]t)/([GLV]0)) and ln(([R]t)/ ([R]0)) versus illumination time, where subscripts represent concentrations at times t and zero. Figure 1 illustrates the first-

droplets to various degrees (Table S1 in the Supporting Information (SI)). In the gas phase, GLVs are oxidized by O3 and OH to form low volatility products that produce SOA with mass yields from 0.7 to 20%.39,44−46 Our previous work shows that aqueous-phase oxidation of GLVs by OH is a minor source of SOA,47 but other oxidants − such as 3C* and 1O2* − may enhance this. For example, recent work has shown that 3C* can rapidly oxidize aqueous phenols to form low-volatility products with aqueous-phase SOA mass yields close to 100%.48 To address the potential significance of these oxidants for GLVs, we examine the bulk aqueous chemistry of five GLVs: methyl jasmonate (MeJa), methyl salicylate (MeSa), cis-3-hexenyl acetate (HxAc), cis-3-hexen-1-ol (HxO), and 2-methyl-3-butene-2-ol (MBO). For each species we quantify its aqueous-phase second-order rate constants with 3 C* and 1O2* and determine aqueous SOA mass yields for the most significant of these reactions. Finally, we use the laboratory results to assess the potential contributions of 3C* and 1O2* as sources of aqueous SOA in cloudy or foggy atmospheres.

■

EXPERIMENTAL SECTION Chemicals. All chemicals were used as received: MeJa (≥95%), MeSa (≥99%), HxAc (≥98%, FG), HxO (≥98%), MBO (98%) 3,4-dimethoxybenzaldehyde (DMB; 99%), 3′methoxyacetophenone (MAP; 99%), syringol (SYR; 99%), phenol (PhOH, 99%), furfuryl alcohol (FFA; 99%), tyrosine (Tyr; ≥ 98%), and Rose Bengal (RB; ≥ 85%) are all from Sigma-Aldrich. Sulfuric acid (Trace metal grade, 94−98%) was from Fisher. Solutions were made using purified water from a Milli-Q Plus system (18.2 MΩ-cm) with an upstream Barnstead B-Pure cartridge to remove organics. Kinetic Experiments. Bimolecular rate constants of the GLVs with 1O2* or 3C* in the aqueous-phase were determined using a relative rate approach with a reference compound. Airsaturated solutions contained the following: (1) 15−50 μM of a precursor for 3C* (DMB or MAP) or 1O2* (RB), (2) 25−100 μM of a reference compound (PhOH or SYR for 3C* experiments; Tyr or FFA for 1O2*), and (3) one or more GLVs (25−50 μM). Solutions were typically pH 5.1 ± 0.2 (unadjusted), but we also measured rate constants at pH 2.1 ± 0.2 (adjusted with sulfuric acid). Over the course of each experiment the change in pH was less than ±0.3. Each experiment was conducted in a 2 cm, far-UV, quartz cuvette (Spectrocell), placed in a temperature-controlled illumination chamber with continuous stirring. Solutions were irradiated using simulated sunlight from a 1000 W Xe lamp filtered with an AM 1.0 air mass filter (AM1D-3L, Sciencetech) and 295 nm long-pass filter (20CGA-295, Thorlabs); the spectral output of this system and the molar absorptivities of DMB and MAP are provided in Figure S1. Molar absorptivities of the GLVs were reported previously.47 For each illumination experiment two control experiments were also conducted: (1) a dark control wrapped in aluminum foil treated identically (sample composition and temperature) to the illuminated sample and (2) a direct photodegradation control with identical conditions except no oxidant precursor was added. Aliquots of sample were removed from the quartz cells at measured time intervals during illumination and concentrations of the reference compound and GLVs were measured using high performance liquid chromatography (HPLC). The HPLC system consisted of a Shimadzu SPD10A UV−vis detector, LC-10AT pump, and a 250 × 33 mm, 5

Figure 1. Typical plots of the aqueous-phase decays of a GLV (MeSa) and reference compound (SYR) in the presence of the DMB triplet during illumination (298 K and pH 5.1 ± 0.2).

order losses of MeSa and reference compound SYR in the presence of the DMB triplet state (3DMB*). The measured apparent first-order rate constant of MeSa in Figure 1 is the sum of the first-order loss pathways k′MeSa = k′3C*,MeSa + k′1O2*,MeSa + k′OH,MeSa + jMeSa

(3)

where k′Ox, MeSa is the pseudo-first-order rate constant for loss of MeSa due to oxidant “Ox” (i.e., 3C*, 1O2*, or OH), and jMeSa is the rate constant for direct photodegradation. MeSa has a slow direct photodegradation (jMeSa = (3 ± 1) × 10−6 s−1 in our experiments), while there was no significant direct photodegradation loss (p > 0.05) for the other GLVs or any of the reference compounds. For each of our GLV experiments either 3 C* or 1O2* was the dominant oxidant, but we corrected each apparent first-order rate constant for the minor contributions from the other loss pathways (Section S1). The bimolecular rate constant for GLV oxidation (kGLV+Ox, where Ox = 3C* or 1O2*) was determined by eq 449 k GLV + Ox =

k′GLV kR + Ox k′R

(4)

where k′R and k′GLV are the measured apparent first-order rate constants (corrected for the other loss pathways), and kR+Ox is the second-order rate constant for the reference compound. Error bars in the GLV bimolecular rate constants are standard errors (SE) propagated from uncertainties in kR+Ox and the measured pseudo-first-order rate constants. As a test of our use of the method, we conducted relative rate experiments for two known rate constants. First we examined the 3DMB*-mediated oxidation of SYR and PhOH (in triplicate) at 298 K at pH 5.1. Our measured ratio ((k′SYR)/ (kPhOH ′ )) = 9.3 ± 3.1 is statistically smaller (p = 0.05) than the ratio calculated from literature values of the second-order rate 269

dx.doi.org/10.1021/es503656m | Environ. Sci. Technol. 2015, 49, 268−276

Environmental Science & Technology

Article

constants,48,50 which is (45 ± 21). We have more confidence in the ratio of 9.3 because the relative rate method simultaneously monitors the disappearance of both phenols in the same solution, while in our past work we determined rate constants individually. Furthermore, the SYR + 3DMB* rate constant that we recently determined,50 (3.5 ± 1.1) × 109 M−1 s−1, appears to be more accurate than the PhOH value based on a comparison with rate constants of 3′-MAP, which has a similar reactivity to DMB. The 3′-MAP triplet rate constant with SYR is (3.8 ± 0.3) × 109 M−1 s−1 (R. Kaur, unpublished data), which is very close to our SYR + 3DMB* value, while the 3′-MAP triplet rate constant with PhOH51 is 6.5 times higher than our recently determined value for 3DMB* with PhOH.50 Based on this, we suspect that our reported rate constant for PhOH + 3 DMB* is approximately a factor of 5 too low, and so we used the SYR + 3DMB* rate constant as the reference to determine the second-order rate constants with 3DMB* in this work. For experiments with 3′-MAP as the triplet precursor, we used syringol as the reference. For 1O2* experiments, relative rate experiments were performed in triplicate using Tyr as the probe to determine the FFA rate constant at 293 K and pH 5.4. Our measured value, (1.2 ± 0.1) × 108 M−1 s−1 (±1 σ), is very close to the average (±σ) of three previously reported literature values, (1.3 ± 0.1) × 108 M−1 s−1.35,52,53 Aqueous SOA Mass Yields. Aqueous-phase SOA mass yields (YSOA(aq)) were measured gravimetrically from illuminated solutions containing 100 μM GLV and 5−20 μM DMB that were blown down with N2 in a method described by Smith et al.48 Solutions were illuminated, in the same system as kinetic experiments, at 298 K until 30−50% of the GLV was degraded. Once illumination was halted, 8.0 mL of the illuminated sample and 8.0 mL dark control were removed and each transferred to a preweighed aluminum foil cup. Prior to YSOA(aq) measurements, aluminum cups were combusted for 12 h at 600 °C to remove organic compounds. Each cup was weighed three consecutive times using an electro balance (Cahn, ± 1 μg) before (no solution added) and after solution blowdowns. Once dry, the samples were weighed and YSOA(aq) was determined: YSOA(aq) =

Figure 2. Bimolecular rate constants for GLV reactions with 3DMB* at pH 2.1 ± 0.2 (light blue), 5.1 ± 0.3 (dark blue), and with 3MAP* at pH 5.0 ± 0.3 (green). Error bars are ±1 SE from replicate experiments at temperatures ranging from 278 to 298 K.

108 M−1 s−1 for 3DMB* and (0.52−8) × 108 M−1 s−1 for MAP*. For comparison, aqueous rate constants for GLV oxidation by OH are (5.4−8.6) × 109 M−1 s−1,47 a factor of 5 to 250 times larger than the 3C* rate constants. There is a less than a 5-fold difference in reactivity between 3MAP* and 3 DMB* with the five GLVs, with 3DMB* being more reactive with MeSa and HxAc and 3MAP* being more reactive with the other GLVs. A study by Canonica et al.51 showed that different triplets react differently with substituted phenols with rate constants varying over 1 to 2 orders of magnitude.51 While our rate constants for a given GLV with 3DMB* and 3MAP* are within a factor of 5, these results suggest that different triplets in fog waters have different reactivities with organics. The GLV-3C* bimolecular rate constants fall into two groups: HxAc, MeJa, and MeSa react readily with 3C* (kGLV+3C* > 1.2 × 108 M−1 s−1), while HxO and MBO react more slowly (kGLV+3C* ≤ 1 × 108 M−1 s−1). While most data sets for 3C* reactions are in organic solvents,54 there are some bimolecular rate constants in aqueous solutions for comparison. For example, values range from (0.13−60) × 108 M−1 s−1 for compounds of biological relevance,55−59 substituted phenols,48,51 and a few other organic compounds.60,61 Rate constants for oxidation of alcohols by 3C* range from 1 × 105−2.0 × 107 M−1 s−1;54 our values for the two GLVs with alcohol functional groups (HxO and MBO) are at the upper end of this range (Figure 2). In contrast, values of bimolecular rate constants with phenols by 3C* are generally high, with values in the range (1−60) × 108 M−1 s−1.32,48 Our values for MeSa, a phenolic GLV, are within this range (kMeSa+3C* = (8− 12) × 108 M−1 s−1). The fast triplet rate constants for MeJa and HxAc in Figure 2 are surprising; these GLVs are not phenols, but they have rate constants in the range of phenol values. To examine this, we calculated the bond dissociation energies (BDE) of different C−H bonds in the two molecules (Section S2 and Figure S3 in the SI). The allylic hydrogens on both MeJa and HxAc have low BDEs (with minimum values of 81.8 and 81.6 kcal/mol, respectively), suggesting these are the sites of hydrogen abstraction. In comparison, calculated BDEs for the phenolic O−H bond in PhOH and SYR are 83.9 and 75.4 kcal/mol, respectively.62 Thus, we expect that the reactivities of the allylic H atoms in MeJa and HxAc are intermediate between the phenolic H atoms in PhOH and SYR, which is broadly 3

(mass of illuminated sample − mass of the dark sample) mass of GLV reacted (5)

Milli-Q water blanks were also periodically blown down with nitrogen, with an average (±1σ) mass concentration of 0.95 ± 0.19 μg mL−1 (n = 6). The average mass concentrations for the illuminated GLV + 3DMB* solutions were 10 ± 0.1 μg mL−1 for MeJa, 6.3 ± 0.2 μg mL−1 for MeSa, and 4.6 ± 0.5 μg mL−1 for HxAc; corresponding values for the dark blanks were 1.9 ± 0.3, 1.1 ± 0.4, and 0.94 ± 0.3 μg mL−1, respectively.

■

RESULTS Kinetics of GLV Oxidation by 3C* and 1O2*. Understanding aqueous SOA formation from the oxidation of GLVs by 3C* and 1O2* first requires that we know the bimolecular rate constant of these reactions. Using eq 4 with the raw kinetic data (e.g., Figure 1), we determined the second-order rate constant for the oxidation of each GLV by 3C* (at pH 5 and 2) and 1O2* (at pH 5) over a range of temperatures. We first examine the results with the two triplet states (3DMB* and 3MAP*). As summarized in Figure 2, aqueous bimolecular rate constants at pH 5 and 298 K are (0.13−15) × 270

dx.doi.org/10.1021/es503656m | Environ. Sci. Technol. 2015, 49, 268−276

Environmental Science & Technology

Article

Table 1. Rate Constants for the Aqueous-Phase Reaction of 1O2* with GLVs at Different Temperatures at a pH 5.1 ± 0.3a kGLV+1O2* (106 M−1 s−1)

a

temp (K)

HxAc

HxO

MBO

MeJa

MeSa

Tyr

298 293 288 283 278 A Ea, kJ mol−1

3.9 ± 0.8 3.1 ± 0.4 2.7 ± 0.8 1.3 ± 0.1 1.0 ± 0.2 2.2 × 1015 50 ± 7.2

2.5 ± 0.3 2.1 ± 0.2 0.9 ± 0.1 0.5 ± 0.1 0.3 ± 0.1 6.1 × 1020 82 ± 7.4

0.8 ± 0.1 0.7 ± 0.1 0.6 ± 0.2 0.5 ± 0.1 0.4 ± 0.1 6.7 × 109 22 ± 1.7

6.0 ± 0.7 2.9 ± 0.4 1.2 ± 0.2 0.7 ± 0.1 0.4 ± 0.1 3.6 × 1023 96 ± 4.8

≤0.1 ≤0.1 ≤0.1 ≤0.1 ≤0.1 N/A N/A

11 ± 0.2 8 ± 0.1 6 ± 0.1 4.5 ± 0.2 3.3 ± 0.1 1.2 × 1014 40 ± 5.1

Error bars represent ±1 SE, propagated from the standard errors of k′GLV, jGLV, and kRef+1O2*.

Figure 3. Arrhenius plots for the reactions of (a) 3DMB* and (b) 1O2* with the GLVs at pH 5.1 ± 0.3. Each dashed line in panel (a) is the average rate constant for a given GLV. In panel (b) the temperature dependence for MeSa reaction with 1O2* was not explored since the rate constant at 298 K is low (≤1 × 105 M−1 s−1). Error bars represent ±1 SE propagated from the standard errors of k′GLV, jGLV, and kRef+Ox.

Table 2. Rate Constants for the Aqueous-Phase Reaction of 3DMB* and 3MAP* with GLVs at Different Temperatures at pH 5.1 ± 0.3c kGLV+3DMB* (108 M−1 s−1) pH 2 GLVs MeJa MeSa HxAc HxO MBO

298 Ka 3.6 29 14 0.33 0.28

± ± ± ± ±

0.5 2 2 0.04 0.1

kGLV+3MAP* (108 M−1 s−1)

pH 5 298 K 4.1 12 15 0.24 0.12

± ± ± ± ±

1 4 4 0.1 0.05

288 K 4.1 12 15 0.27 0.15

± ± ± ± ±

2 4 4 0.06 0.05

pH 5 avb

278 K 4.5 13 12 0.21 0.12

± ± ± ± ±

2 4 4 0.04 0.04

4.2 12 14 0.24 0.13

± ± ± ± ±

298 K 3 4 7 0.1 0.07

1.2 8.0 7.9 1.1 0.56

± ± ± ± ±

0.3 0.4 2 0.2 0.1

288 K 0.98 8.4 6.9 0.8 0.54

± ± ± ± ±

0.2 0.4 0.7 0.5 0.1

278 K 1.3 7.6 7.1 1.7 0.56

± ± ± ± ±

0.3 0.5 0.8 0.4 0.1

av 1.2 8.0 7.3 1.2 0.55

± ± ± ± ±

0.5 0.8 2 0.8 0.2

Experiments conducted at a pH 2.1 ± 0.2. bBimolecular rate constants measured at pH 5.1 ± 0.3. cError bars represent ±1 SE, propagated from the standard errors of k′GLV, jGLV, and kRef+3C*. a

s−1 (Table 1). MeSa reacted much more slowly with 1O2*, showing less than a 7% loss over a 48-h period, which corresponds to a bimolecular rate constant of ≤1 × 105 M−1 s−1. Our measured values are within the range of bimolecular rate constants for small aromatics and structurally similar ketones and alcohols.36 As shown in Figure 3, we also determined the temperature dependence for the rate constants of GLV oxidation by 3C* and 1O2*. In the case of the triplet reactions, there is no apparent temperature dependence within the tested range (278−298 K; Table 2), consistent with results reported for 3 DMB* with phenol at two temperatures.48 In contrast, 1O2* reactions are strongly dependent on temperature, with values at 298 K higher by factors of 1.7 (for MBO) to 16 (for MeJa) compared to those at 278 K. All of these rate constants are fit well (R2 = 0.94−0.98) by the Arrhenius equation (Figure 3b)

consistent with our measured rate constants. This is interesting because it suggests that unsaturated aliphatic hydrocarbons as a group are reactive with triplet excited states in atmospheric waters, which is a new finding. We also examined the effect of pH on the rate constants for GLV reactions with 3DMB*, which has a pKa of 3.3.31,48 As shown in Figure 2, for MeJa and HxAc there is little difference in the rate constants with the neutral (pH 5) and protonated (pH 2) forms of the DMB triplet. In contrast, for the phenol MeSa, the rate constant at pH 2 is 2.5 times faster. This is consistent with past observations that phenols react more quickly with a protonated carbonyl triplet state.31,48 It is unclear, however, why phenols exhibit this behavior, while the other four GLVs we examined show no difference in reactivity with the neutral and protonated forms of the DMB triplet. In the case of 1O2* reactions, at 298 K rate constants with HxAc, HxO, MBO, and MeJa ranged from (8.2−60) × 105 M−1 271

dx.doi.org/10.1021/es503656m | Environ. Sci. Technol. 2015, 49, 268−276

Environmental Science & Technology ln(k GLV + Ox ) = ln(A) − Ea /RT

Article

reactions. There is some evidence that triplet excited states and OH give similar types of products during the aqueous oxidation of phenols as well: SOA yields from OH and 3DMB* oxidation are similar, with average values of (90 ± 8)% and (107 ± 12%), respectively.48,65 Our aqueous SOA yields from GLVs are generally higher than values reported for the gas phase: values of YSOA(g) for oxidation by OH are 0.93% (HxAc), 3.1% (HxO), 0.7% (MBO), and 20% (MeSa),39,44,45,66 while yields during oxidation by O3 are 1.2−8.6% (HxAc), 3.3−9.5% (HxO), and 1% (MBO)44,46 (Table S5 in the SI). To our knowledge there are no reported gas-phase SOA mass yields for oxidation of MeJa by OH or O3 or for MeSa oxidation by O3. The higher SOA yields in the aqueous-phase relative to gas-phase yields could be that aqueous-phase reactions more efficiently produce oligomers and/or functionalized/oxygenated products relative to the gas-phase. Significance of Aqueous-Phase Formation of SOA by GLVs. To compare the relative importance of each oxidant in generating aqueous SOA from the GLVs we start by comparing oxidant reactivities in fog drops. Typical fog oxidant steadystate concentrations are 7.7 × 10−15 M for OH67 and 2.3 × 10−13 M for 1O2*;29 furthermore, as described in section S5, we estimate [3C*] = 9 × 10−14 M based on past work from Davis fogs68 and assume 30 ppbv O3, corresponding to an aqueous concentration of 3.9 × 10−10 M. Based on these oxidant concentrations ([Ox]ss) and measured bimolecular rate constants at 278 K (Table 1, Table 2, Table S4, and Richards-Henderson et al.47) the pseudo-first-order rate constant for loss of GLV due to a given oxidant is

(6)

where A is the pre-exponential factor, Ea is the activation energy, and R is the gas constant. The activation energies for the GLV + 1O2* reactions vary between 20−96 kJ mol−1 (Table 1). Limited temperature dependence data is available for 1 O2* reactions, but our results encompass Ea values reported for furfuryl alcohol (FFA), imidazole, and tyrosine (Tyr) of 22.7, 30.2, and 40.2 kJ mol−1, respectively.52 SOA(aq) Mass Yields. Determining how aqueous oxidation of GLVs by 3C* and 1O2* contributes to SOA(aq) formation also requires knowing the yields of low volatility products. SOA(aq) mass yields (YSOA(aq)) were determined from the mass of low-volatility material formed from GLV oxidation after correction for the corresponding mass in the dark sample (eq 5), which was (30−50)% of the illuminated value. SOA(aq) mass yields were not determined for 1O2* or for HxO and MBO with 3DMB* due to their slow rate constants. Half-lives of the GLVs oxidized by 3DMB* for our YSOA(aq) experiments were 8−18 h; based on our bimolecular rate constants (Table 2) this indicates that aqueous 3DMB* steady-state concentrations ranged from (1−10) × 10−14 M. Reconstitution of blown-down material showed, on average, ≤5% of the initial GLV remained after blowdown, indicating that low-volatility products are secondary. As shown in Figure 4, aqueous SOA yields for 3DMB* reactions are, on average (±1 SE), (38 ± 18)%, (84 ± 7)%, and

k′GLV,Ox = k GLV + Ox[Ox]

(7)

Ranges for k′GLV for each oxidant with the five GLVs are (0.6− 9) × 10−5 s−1 for 3C*, (3−5.1) × 10−5 s−1 for OH, (0.0003− 1.6) × 10−5 s−1 for O3, and (0.0023−0.023) × 10−5 s−1 for 1 O2*. Even though O3 steady-state concentrations in fog drops are approximately 430 and 5.1 × 104 times higher than 3C* and OH, respectively, this is offset by the ≤4 × 104 and ≤430 times faster bimolecular GLV-OH and GLV-3C* rate constants, respectively. Therefore, the oxidant reactivity order in fog drops is approximately 3C* ≥ OH > O3 ≥ 1O2* for the five GLVs we are examining. While it appears that triplet excited states dominate the aqueous processing of our GLVs, the steady-state concentration of triplets requires several assumptions and is quite uncertain (section S5). To more specifically assess the contributions of GLV reactions to aqueous SOA, we next compare the rates of SOA formation from each GLV in the aqueous and gas phase. For these calculations, we use typical Davis, CA fog conditions (278 K, liquid water content = 1.0 × 10−7 L-aq/L-gas), and the oxidant concentrations, and estimated Henry’s Law constants69 to determine the gas-aqueous partitioning of the GLVs, as described in detail in section S4 of the SI. Figure 5a shows that the order of contributions of the oxidants to the rate of aqueous SOA formation from GLVs is 3 C* ≥ OH > O3 ≥ 1O2*. The two most important oxidants for SOA formation, 3C* and OH, have total rates of aqueous SOA formation of 65 and 20 ng m−3 h−1, respectively. Singlet molecular oxygen and O3 were minor contributors (