Sources and sinks of hydroxyl radical in sea-salt particles



[1] We have examined the photochemical formation of hydroxyl radical (OH) in aqueous extracts of supermicron sea-salt particulate matter (SS PM) collected from the coast of northern California. All extracts formed OH during illumination, indicating that this process is widespread in sea-salt particles. Scaling extract results to SS PM conditions reveals that in situ rates of OH photoformation are typically 100–1000 μM hr−1 in midlatitude sea-salt particles (summer, midday, 88% relative humidity). These rates are comparable to calculated rates of partitioning of gas phase OH to the particles and are 3–4 orders of magnitude greater than OH photoformation rates in surface seawater. Photolysis of nitrate was a dominant source of OH in the particle extracts, accounting for an average of 59 ± 25% of its formation. The other sources of OH have not been identified, but photolysis of organic compounds derived from seawater is likely important. The lifetimes of OH in the sea-salt particles are of the order of 10−9–10−8 s and are primarily controlled by reactions with unidentified, but probably organic, compounds. Bromide and chloride are also significant sinks of OH, typically accounting for approximately 25% of its loss. The rapid formation and destruction of OH in sea-salt particles likely significantly affects the chemistry of halides, organic compounds, and other reduced species in SS PM. In turn, these particle reactions probably alter the budgets of gases such as ozone and volatile organic compounds in the marine boundary layer.

1. Introduction

[2] Sea-salt particles are one of the most abundant aerosol types, with a tropospheric burden of approximately 27 Tg on a dry mass basis [Keene et al., 1999]. These particles have numerous effects on the radiative balance of the atmosphere as well as on the chemistry of the marine boundary layer. For example, sea-salt particles dominate the aerosol scattering by natural particles in the atmosphere [Haywood et al., 1999] and are efficient cloud condensation nuclei [Murphy et al., 1998]. In addition, sea-salt particulate matter (SS PM) is an important site for the oxidation of sulfur dioxide to form sulfuric acid, a process which reduces the formation of new sulfate particles [Keene et al., 1998; Sievering et al., 1992]. Sea-salt aerosols are also a source of gaseous reactive halogens, such as Br2 and BrCl and their corresponding halogen radicals, which increase the oxidation potential of the troposphere and react with ozone, hydrocarbons, and mercury [von Glasow and Crutzen, 2003].

[3] While these aspects of sea-salt particle chemistry are generally understood, much of the chemistry in airborne sea-salt is still probably unknown. One area in particular that is relatively unexplored is the budget and chemistry of hydroxyl radical (OH) in SS PM. Because of its reactivity, OH is likely an important oxidant for a number of reduced species in sea-salt particles (e.g., organic compounds, halides, and reduced sulfur species), as it is for many gases in the atmosphere [Finlayson-Pitts and Pitts, 2000]. One source of hydroxyl radical to sea-salt particles is the partitioning of gaseous OH. In addition, formation of OH via in situ, direct photolysis of light-absorbing species (chromophores)—such as nitrate, nitrite, hydrogen peroxide, and ozone—are also included in some models of SS PM chemistry [e.g., Pechtl et al., 2006]. While it has been proposed that photochemical reactions such as these might be important sources of OH in SS PM [Zafiriou, 1974], and might contribute to the release of reactive halogens [Keene et al., 1993], this chemistry has not been explored in actual sea-salt particles. Similarly, while numerical models of SS PM contain a variety of OH sinks [e.g., Herrmann et al., 2003; Pechtl et al., 2006], there have been no measurements of the loss of OH in actual particles and the major OH sinks have not been identified experimentally.

[4] The closest surrogate for understanding the chemistry of hydroxyl radical in sea-salt particles is probably seawater, since SS PM is formed from the injection of drops of seawater into the marine boundary layer [Lewis and Schwartz, 2004]. While photolysis of nitrate and nitrite can be important sources of OH in some seawater (e.g., in upwelled waters), the major source appears to be photolysis of dissolved organic matter (DOM), especially fluorescent components [Mopper and Zhou, 1990; Qian et al., 2001; Takeda et al., 2004]. However, the identities and relative amounts of the different DOM components are generally unknown [Joint Group of Experts on the Scientific Aspects of Marine Environmental Protection, 1995], as are the mechanisms responsible for OH photoformation by DOM. As for the sinks of OH in seawater, calculations by Zafiriou [1974] and Zafiriou et al. [1987] indicate that reaction with bromide (to form Br2) accounts for ∼90% of the OH sink, and this has been confirmed by measurements at a number of different locations [Mopper and Zhou, 1990; Qian et al., 2001; Zhou and Mopper, 1990].

[5] While these attributes of seawater are probably a useful first approximation of the photochemistry of sea-salt particles, the composition (and therefore the chemistry) of SS PM is significantly different from seawater in a number of ways. For example, relative to seawater, sea-salt particles are much more acidic, are highly enriched in organic compounds, and contain much higher concentrations of soluble atmospheric gases such as HNO3 and HOOH [Keene et al., 1998; Newberg et al., 2005; Pszenny et al., 2004]. It is unclear, however, how these differences affect the chemistry of hydroxyl radical or other reactive species in SS PM. The goal of this current work is to examine the sources and sinks of OH in sea-salt particles, with a focus on the in situ, photochemical generation of OH. Here we report the formation rates, lifetimes, and steady state concentrations of OH in supermicron SS PM (diameters (Dp) = 1.6–15 μm) illuminated with simulated sunlight. In addition, we examine the major sinks and photochemical sources of OH in sea-salt particles and the relative importance of in situ photochemical formation versus gas-to-particle partitioning. Finally, we discuss the potential significance of OH photoformation on the chemistry and composition of SS PM.

2. Experimental Methods

2.1. Aerosol-Derived Extracts and Materials

[6] The details of sea-salt particle collection, extraction, and ion analyses have been described previously [Newberg et al., 2005]. Briefly, aerosol samples were collected at a coastal site on the northeastern Pacific Ocean (Bodega Bay, California) and on two ship cruises along the California coast. Samples were collected onto Teflon substrates using a four-stage Lundgren cascade impactor with particle diameters (50% cutoff) in stages 1–4 corresponding to 15–50, 4.8–15, 1.6–4.8, and 0.48–1.6 μm, respectively. Although our collection times were quite long (typically 2–3 weeks), there are no apparent major differences in the OH kinetics (e.g., formation rates or lifetimes) between samples with short collection times (40–53 hours) and those with long collection times (385–535 hours). This suggests that the particles were not significantly modified while on the Teflon substrates in the field before sample retrieval. The air masses associated with our samples were dominantly from oceanic regions, especially from the northwest, which was associated with the highest wind speeds [Newberg et al., 2005]. However, there were also some periods of offshore flow (generally during the evenings with mild wind conditions) that likely brought continental-influenced air masses to our samplers.

[7] Sampled substrates were placed in individual containers and then stored in the refrigerator at 4°C, typically for less than two months but up to 6–12 months for some samples. Particles from a given sample were extracted using small volumes of Milli-Q water (Millipore; ≥ 18.2 MΩ cm), then filtered (0.22 μm Teflon) and stored at 4°C in 2-mL PTFE bottles prior to analysis and illumination. Samples were generally illuminated (see below) within a few weeks of being extracted.

[8] The UV–visible absorption spectra of the filtered, aerosol-derived aqueous extracts (see below) were measured in quartz microcells using a Shimadzu UV-2501PC spectrophotometer with a 3 mm mask on the cell holder and Milli-Q water as the reference. Base 10 light absorption coefficients at wavelength λ in the extracts (αλ,EXT) were determined using

equation image

where the path length was 1 cm. The pH in each extract was measured using a microelectrode (Microelectronics, MI-414-6cm) with ≈ 20 μL of extract.

[9] Acetonitrile (Optima), perchloric acid (Optima), sodium borate (certified ACS), and concentrated sulfuric acid (Optima) were all purchased from Fisher Scientific, while benzoic acid (99%), metahydroxybenzoic acid (99%) and sodium benzoate (99%) were from Aldrich. All chemicals were used as received.

2.2. OH Kinetics

2.2.1. Method

[10] The formation of hydroxyl radical in the aerosol extracts was monitored using a chemical probe technique [Anastasio and McGregor, 2001; Zhou and Mopper, 1990] that uses a mixture of benzoic acid (HBA) and benzoate (BA). Reactions of these species with OH (kHBA = 4.3 × 109 M−1 s−1, kBA− = 5.9 × 109 M−1 s−1) [Ross et al., 1998] forms the orthohydroxy, metahydroxy and parahydroxy isomers. We followed the formation of metahydroxybenzoic acid (m-OHBA), which has a yield of 19.1% from the reaction of OH with HBA/BA [Anastasio and McGregor, 2001]. Concentrations of m-OHBA were measured using a high-pressure liquid chromatography (HPLC) method with UV-Vis detection that has been described previously [Anastasio and McGregor, 2001]. Probe solutions (a mixture of HBA and BA) and calibration solutions of m-OHBA in HBA/BA were prepared from stock solutions on each day of an illumination experiment. The pH of the probe solution was set at the pH of the aerosol extract (typically ≈ 5.5; Table S1) by using different proportions of HBA and BA (pKa = 4.19) [Lide, 1999].

[11] Illumination solutions (≈ 200 μL) were prepared by combining known volumes of particle extract (VEXT), probe solution (VP), and Milli-Q (VW). The ratio of particle extract volume to the total volume of illuminated sample solution is defined as the dilution factor (DF):

equation image

Dilution factors in our experiments were typically ∼ 0.50. 200 μl of each illumination solution was put into a GE 021 quartz tube (4 mm I.D. × 58 mm L) with a PTFE stopper and uniformly illuminated at room temperature with simulated sunlight from a 1000 W Xe lamp with optical filters [Faust and Allen, 1993]. During illumination, aliquots of sample solution (≈40 μL) were removed from the quartz tube at known exposure times, adjusted to pH ≈ 2 with 5 μL of 100 mM H2SO4 to protonate m-OHBA (pKa = 4.06) [Lide, 1999], and injected into the HPLC. Spikes of m-OHBA into aliquots of sample solution showed that there were no matrix effects.

2.2.2. Rates of OH Photoformation

[12] The initial rate of m-OHBA formation during illumination, RP*, was determined from a linear regression of [m-OHBA] versus illumination time or, in some cases, from an exponential-rise-to-maximum fit to the data [Anastasio and McGregor, 2001]. For most samples we measured RP* at three different probe concentrations in order to fully determine OH kinetics using competition kinetics. For these samples a plot of 1/RP* versus 1/(probe concentration) yields a straight line where the slope and y intercept are used to determine ROH,EXP, the rate of OH photoformation in the illumination solution, and kOH,EXP, the pseudo first-order rate constant for OH destruction due to natural scavengers in the sample [Anastasio and McGregor, 2001; Zhou and Mopper, 1990]. In samples with limited volumes, we measured RP* at only one, high, probe concentration (5.0 mM) in order to determine just ROH,EXP [Zhou and Mopper, 1990].

[13] We converted each value of ROH,EXP to a value of ROH,EXT, the rate of OH formation expected in the particle extract illuminated with midday, summer solstice, sunlight at 38°N latitude, using

equation image

where DF is the extract dilution factor (equation (2)); j2NB,SUM and j2NB,EXP are the rate constants for photolysis of the chemical actinometer 2-nitrobenzaldehyde (2NB) on midday of the summer solstice at 38°N (0.013 s−1) [McGregor, 2000] and in the solar simulator on the day of a given experiment, respectively [Anastasio et al., 1994], and ROH,BLANK (0.076 μM hr−1) is the average OH formation rate observed in blank experiments (illuminated solutions containing only probe or both probe and the extract from a field blank filter).

[14] The rate of OH photoformation in a given particle extract was converted to ROH,SS, the rate expected in ambient sea-salt particles (midday, summer solstice, 38°N) at a relative humidity (RH) of 88%, using

equation image

where [Na+]SS is the concentration of Na+ in ambient sea-salt particles at our reference relative humidity of 88% (3.11 M; calculated on the basis of data for the water activity [Hamer and Wu, 1972] and density of aqueous NaCl [Lide, 1999]), and [Na+]EXT is the concentration of Na+ in the sample extract (Table S1). As described in section 3.2, our experiments indicate that ROH,SS scales linearly with particle dilution factor, consistent with equations (3) and (4). Similarly, the concentration of any chemical species X in our particle extracts can be linearly scaled to conditions of SS PM at 88% RH using

equation image

Relative to the reference sodium concentration of 3.11 M, our SS PM extracts were diluted by factors of 11–38 for the Bodega Bay particles and 40–320 for the ship particles.

[15] In a given sea-salt particle sample the fraction of OH formed from nitrate photolysis is

equation image

where ROH,NO3 is the rate of OH formation due to nitrate photolysis in the sea-salt particles

equation image

In our solar simulator the value for jOH,NO3, the rate constant for OH formation from nitrate photolysis, is 1.4 × 10−7 s−1 under conditions where the rate constant for 2NB photolysis (j2NB,NO3) is 0.0070 s−1 [Anastasio and McGregor, 2001].

2.2.3. Rate Constant for Loss of OH

[16] As described above, we also use our competition kinetics results to determine kOH,EXP, the pseudo first-order rate constant for loss of OH due to natural scavengers in the illuminated samples. Each value of kOH,EXP was converted to the result expected in the extract (kOH,EXT) by dividing by the dilution factor of the illuminated sample; as described below, our experimental results indicate that kOH,EXP is proportional to 1/DF for our samples.

[17] Scaling kOH,EXT to sea-salt particle conditions at 88% RH requires first determining the importance of Br and Cl as sinks for OH in the extract solution. For example, the pseudo first-order rate constant in the extract for the loss of OH due to reaction with Br is

equation image

where k1 is the rate constant for the OH + Br reaction (reaction (R1), Table 1), [Br]EXT is the bromide concentration in the extract, and ϕBr,EXT is the fraction of BrOH formed in reaction (R1) that does not go on to form products (i.e., that decomposes to regenerate OH and Br):

equation image

Rate constants here are for the reactions in Table 1, while the concentrations refer to extract values. Using reactions (R7)–(R11) (Table 1), we can derive an analogous equation to determine the pseudo first-order rate constant for the loss of OH due to reaction with Cl in the extract. We also examined reactions with nitrate and sulfate as sinks for OH. However, the rate constants with OH for these compounds are slow enough (Table 1) that they are insignificant OH sinks in our SS PM samples (each accounting for < 0.1% of the total OH sink in any sample) and so are not considered further. The rate constant for loss of OH due to “other” sinks (i.e., not Br or Cl) in an extract was then determined using

equation image
Table 1. Reactions of OH with Br, Cl, NO3, and HSO4/SO42−
ReactionRate ConstantaReference
  • a

    Units are M−1 s−1 or s−1.

  • b

    As cited by Arakaki and Faust [1998].

  • c

    This is the pseudo first-order rate constant (units of s−1) that includes the water concentration.

  • d

    Value is for solutions at pH ∼ 1, i.e., a mixture of bisulfate and sulfate.

  • e

    As cited by Pechtl et al. [2006].

Bromide Reactions
(R1)OH + BrBrOH1.1 × 1010Ross et al. [1998]
(R2)BrOH → Br + OH3.3 × 107Zehavi and Rabani [1972]
(R3)BrOH → Br + OH4.2 × 106Zehavi and Rabani [1972]
(R4)BrOH + H+ → Br + H2O4.4 × 1010Zehavi and Rabani [1972]
(R5)BrOH + BrBr2 + OH1.9 × 108Zehavi and Rabani [1972]
(R6)BrOH + ClBrCl + OH1.9 × 108Anastasio and Matthew [2006]
Chloride Reactions
(R7)Cl + OH → ClOH4.3 × 109Ross et al. [1998]
(R8)ClOH → Cl + OH6.1 × 109Jayson et al. [1973]
(R9)ClOH + H+ → Cl + H2O2.1 × 1010Jayson et al. [1973]
(R10)ClOH + ClCl2 + OH1 × 105Grigor'ev et al. [1987]
(R11)ClOH + BrBrCl + OH1 × 109Anastasio and Matthew [2006]
Nitrate Reactions
(R12)NO3 + OH → NO3 + OH≤ 5 × 105Farhataziz and Ross [1977]b
(R13)NO3 + H2O → HNO3 + OH50cHerrmann [2003]
(R14)NO3 + Br → NO3 + Br3.8 × 109Herrmann et al. [2000]
(R15)NO3 + Cl → NO3 + Cl3.5 × 108Herrmann [2003]
Sulfate Reactions
(R16)HSO4/SO42− + OH → SO4 + OH3.5 × 105dRoss et al. [1998]
(R17)SO4 + H2O → SO42− + OH690Buxton et al. [1999]
(R18)SO4 + Br → SO42− + Br2.1 × 109Jacobi [1996]e
(R19)SO4 + Cl → SO42− + Cl6.1 × 108Buxton et al. [1999]

[18] This value in the extract solution was converted to that expected in ambient sea-salt particles at 88% RH (kOH,OTHER,SS) by scaling with [Na+]:

equation image

The implicit assumption in this scaling is that the rate of loss of OH due to reactions with “other” sinks is first order in the concentration of “other.” As described in section 3.2, this is consistent with our experimental results. We next calculated the contributions of Br and Cl toward OH scavenging in the sea-salt particles by calculating values of ϕi,SS and kOH,i,SS using equations (8) and (9) (and analogous versions for Cl) with sea-salt particle concentrations of each halide. Values of [H+] in a given sea-salt particle sample were assumed to be the same as in the corresponding extract (Table S1). The total rate constant for loss of OH in the sea-salt particles (kOH,SS) was calculated by summing the contributions from Br, Cl,and “other” sinks:

equation image

The fraction of OH loss in SS PM due to reaction with any individual sink i was calculated as

equation image

2.2.4. Lifetimes and Steady State Concentrations of OH and Errors

[19] The lifetime of OH (τOH) in a solution (e.g., particle extract or sea-salt particles) is equal to the inverse of the corresponding value of kOH (e.g., for the sea-salt particles, τOH,SS = 1/kOH,SS). Steady state concentrations of hydroxyl radical in ambient sea-salt particles, [OH]SS, were determined using

equation image

[20] Errors in ROH, kOH, and [OH] were determined by propagating uncertainties from the slopes and intercepts of the inverse plots of 1/RP* versus 1/(probe concentration) and from the j2NB,EXP actinometry values. The resulting average relative standard errors are 15% for OH photoformation rates, 20% for pseudo first-order rate constants for the destruction of OH, and 25% for OH steady state concentrations. Error bars in figures represent ±1 standard error.

3. Results and Discussion

3.1. Sample Overview

[21] Size-segregated marine particles were collected at two sites: Bodega Bay, California, between June and October with sampling periods of 18–22 days, and on ship cruises off the California coast in January–February and May–June with sampling periods of approximately 3 days [Newberg et al., 2005] (Table S1). As described previously [Newberg et al., 2005], we refer to the supermicron particles with diameters (Dp) of 1.6–15 μm as “sea-salt particles” or SS PM since these particles contain the bulk of the sea-salt mass and, on the basis of cation ratios, they contain insignificant amounts of soil dust or other particle types. These sizes of SS PM are collected on stages 2 (Dp = 4.8–15 μm) and 3 (Dp = 1.6–4.8 μm) of our collector. The average enrichment factor for bromide in the samples is 0.54 (i.e., on average, the samples contain 54% of the bromide expected in fresh sea-salt particles; see Table 2), while there are no significant depletions of chloride (Table S1 and Newberg et al. [2005]).

Table 2. Budget of OH in Sea Salt Particlesa
Sample and Stage[Na+]SS/[Na+]EXTα300,SS,b cm−1EF(Br)c[NO3]SS, mMOH FormationdOH Destructione[OH]SS, 10−16 M
ROH,SS, μM hr−1fOH,NO3, %kOH,EXT, s−1τOH,SS, nsfSINK,Cl, %fSINK,Br, %fSINK,OTHER, %
  • a

    Sea-salt particle quantities are expressed for a relative humidity of 88% (see text).

  • b

    Light absorption coefficient of sea salt particles at 300 nm.

  • c

    Enrichment factor for bromide in the SS PM, defined as ([Br]SS/[Na+])SS/([Br]SW/[Na+]SW), where SW indicates seawater values [Wilson, 1975].

  • d

    ROH,SS is rate of OH photoformation in SS PM (equation (4)); fOH,NO3 is fraction of OH due to nitrate photolysis (equation (6)).

  • e

    kOH,EXT is rate constant for loss of OH in the particle extract; τOH,SS is lifetime of OH in sea-salt particles; and fSINK,i is fraction of OH loss in SS PM that is due to reaction with compound i (i = Cl, Br, and “other”).

Bodega Bay Samples
Stage 4 (Dp = 0.48–1.6 μm)
B99046424< 0.24400818056.1 × 1062.51.4< 3.5> 9557
Stage 3 (Dp = 1.6–4.8 μm)
B9903162.40.5283130582.5 × 1072.31.97.2910.84
B9904265.80.40290920291.0 × 1073.51.98.1908.9
Stage 2 (Dp = 4.8–15 μm)
B9903163.40.731861279.7 × 1065.46.123710.91
B9904113.50.7184220354.9 × 106131353347.8
B9905223.40.6982120641.2 × 1073.13.413841.0
Stage 2 + 3 aggregates
B9801385.30.5799100919.2 × 105161553324.4
B9802335.70.59130340366.2 × 1064.34.115814.0
B9804176.10.65120560192.0 × 1072.82.711874.3
B9805134.70.41160370411.6 × 1074.52.811864.7
B9902152.60.644069541.8 × 1073.33.412840.63
Stage 1 (Dp = 15 – 50 μm)
Ship Samples
Stage 3 (Dp = 1.6 – 4.8 μm)
Stage 2 (Dp = 4.8 – 15 μm)
Stage 2 + 3 aggregates
 88190.3210301020947.6 × 1061.40.62.7974.1
Statistical Summary
Stage 3 (Dp = 1.6 – 4.8 μm)
Med475.10.40270400551.8 × 1072.91.97.7904.9
Avg604.90.46260500551.8 × 1072.91.97.7904.9
1 σ431.80.25120330171.1 × 1070.
Stage 2 (Dp = 4.8 – 15 μm)
Med413.40.69100120649.7 × 1065.46.123711.0
Avg803.20.6196140669.0 × 1067.17.630633.3
1 σ1100.90.134861293.7 × 1065.05.221263.9
Stage 2 + 3 aggregates
Med255.50.58130360481.2 × 1073.83.112854.2
Avg347.20.53260410561.2 × 1075.34.817783.7
1 σ285.90.13380350307.7 × 1065.25.218231.5
All samples from stages 2 + 3
Med384.40.55120220581.0 × 1073.53.412844.1
Avg595.00.54200340591.2 × 1075.45.019763.8
1 σ723.70.17230300257.3 × 1064.64.818222.8

[22] All of the sea-salt particle extracts absorb light between 290 nm (approximately the lowest wavelength of sunlight in the troposphere) and 550 nm, indicating significant potential for in situ photochemical transformations. In contrast, field blanks had no significant absorption in this range. Base 10 light absorption coefficients at 300 nm for our sample extracts, α300,EXT, range from 0.015 to 0.38 cm−1. Multiplying each extract value by the ratio of [Na+]SS/[Na+]EXT in the sample produces the absorption coefficient at 300 nm expected in ambient SS PM at 88% RH; these α300,SS values range from 1.7 to 19 cm−1, with a median value of 4.4 (Table 2). On average, nitrate accounts for approximately 25% of this absorption at 300 nm, but is less important at longer wavelengths.

[23] The absorption coefficients at 300 nm in the sea-salt particles are ∼100 times greater than values in surface seawater [e.g., Blough et al., 1993; D'Sa et al., 1999]. This enhancement in absorption is likely due to two factors: the enrichment of dissolved organic matter (DOM) and other chromophores in SS PM, relative to bulk seawater, during particle formation [Lewis and Schwartz, 2004], and secondary sources of chromophores, such as gaseous HNO3, to the particles during their lifetimes in the atmosphere [Finlayson-Pitts, 2003]. As seen in Table 2, values of α300,SS are generally greater in the smaller (stage 3) sea-salt particles than in the larger (stage 2) particles. This is consistent with observations that the relative amount of organic compounds in SS PM increases with decreasing particle size [Oppo et al., 1999] but could also be due to the fact that the smaller particles are more efficient scavengers (on a volume-normalized basis) of soluble and reactive gases. Compared to other atmospheric hydrometeors, values of α300 here are roughly a factor of 10–1000 times larger than values in midlatitude, continental cloud and fog waters [Anastasio and McGregor, 2001; Faust and Allen, 1992] but are approximately 10 times smaller than values in sulfate particles from the Canadian Arctic [Anastasio and Jordan, 2004].

3.2. Scaling OH Results From Particle Extracts to Ambient SS PM

[24] Because of the very limited volumes in our sea-salt particle samples, we need to extract (and therefore dilute) the samples into Milli-Q water before we can measure hydroxyl radical kinetics. Thus, in order to calculate the rates of OH photoformation and destruction in ambient SS PM we scale our measurements in illuminated extracts to ambient particle conditions using a conservative tracer (Na+). To determine the appropriate scaling, we examined how the rates of formation (ROH), and rate constants for destruction (kOH), depend upon sample dilution in two samples. Figure 1 shows one set of results: both the rate of OH formation (ROH,EXP) and the rate constant for OH destruction (kOH,EXP) are approximately linearly dependent upon dilution factor (R2 = 0.999 and 0.950, respectively). Results from the second sample (B9804A01) also show good linear correlations for ROH,EXP and kOH,EXP with DF (R2 = 0.961 and 0.923, respectively). These results are consistent with previous work where rates of OH photoformation were directly proportional to dilution factor in extracts of bulk aerosol particles from Alert, Canada [Anastasio and Jordan, 2004].

Figure 1.

Effect of dilution on the rate of OH formation (ROH,EXP, normalized to summer solstice sunlight, circles and solid line) and the rate constant for OH destruction (kOH,EXP, squares and dotted line) in illumination solutions of sample B9805A01. DF = 0 is defined as infinite dilution where ROH,EXP and kOH,EXP are assumed to be zero, while DF = 1 is the concentration of the initial particle extract (equation (2)). Error bars represent ±1 standard error.

3.3. OH Photoformation Rates and Sources

3.3.1. Overview

[25] Hydroxyl radicals are formed at appreciable rates in all of our sea-salt particle samples during illumination with simulated sunlight. Rates of OH photoformation in the particle extracts under summer solstice sunlight (ROH,EXT) range from 0.30–35 μM hr−1 and are generally correlated with extract light absorbance at 300 nm (Figure 2). On the basis of the rate constant for OH formation from nitrate photolysis in our system (section 2.2.2.), and the molar absorptivity of NO3 at 300 nm (7.05 M−1 cm−1 [Chu and Anastasio, 2003]), illumination of a nitrate solution would yield a slope of 72 μM hr−1 cm in Figure 2. This is similar to the slopes of the extract data in Figure 2, both for the actual sample extracts (solid line: 82 ± 15 μM hr−1 cm) as well as for the case where the nitrate contribution to OH formation and light absorption has been removed (dashed line: 65 ± 16 μM hr−1 cm). While there is significant scatter in the data, these results suggest that sea-salt particle chromophores other than nitrate have values of ROH/α300 that are broadly similar to that of nitrate.

Figure 2.

Photoformation rates of OH in sea salt particle extracts versus light absorption coefficients at 300 nm. Solid circles represent the rates of OH photoformation and light absorption in the particle extracts, while the open circles represent the same quantities after removing the contributions from nitrate. The slopes (μM hr−1 cm) and y intercepts (μM hr−1) for linear regressions to these data are 82 ± 15 and −2 ± 7 for the extracts (solid line, R2 = 0.64) and 65 ± 16 and –2 ± 6 for the extracts with nitrate contributions removed (dashed line, R2 = 0.49), where uncertainties represent ±1 standard error.

[26] Scaling from these rates of OH photoformation in the sample extracts to the values expected in sea-salt particles at 88% RH (equation (4)) produces values of ROH,SS that range between 61–1020 μM hr−1, with an average of 340 μM hr−1, for particles on stages 2 and 3 (Table 2). Although the sodium concentration in sea-salt particles at 88% RH is only a factor of 6.5 greater than the seawater value (0.48 M [Wilson, 1975]), the rates of OH photoformation in our SS PM are approximately 103–104 times greater than rates in midlatitude, surface seawater [Mopper and Zhou, 1990; Takeda et al., 2004]. As described below, this enormous enhancement is due to secondary nitrate in the particles and, probably, the enrichment of primary organic compounds derived from the sea surface during particle formation. Compared to other atmospheric samples, our rates of OH formation in SS PM are approximately 100 times greater than typical values observed in midlatitude cloud and fog waters [Anastasio and McGregor, 2001; Arakaki and Faust, 1998; Faust and Allen, 1993], but are comparable to the ∼1000 μM hr−1 value reported for Arctic aerosol particles at Alert, Canada in spring [Anastasio and Jordan, 2004].

3.3.2. Size Dependence of OH Photoformation Rates

[27] The average (± 1σ) rate of OH photoformation in the smaller (stage 3) sea-salt particles is 500 ± 330 μM hr−1, 3.6 times greater than the corresponding value for the larger, stage 2 particles (140 ± 61 μM hr−1; see Table 2). This difference is consistent in the six samples where OH photoformation was measured in both stages 2 and 3: the rates in the stage 3 particles (Dp = 1.6–4.8 μm) are 2–4 times greater than in the corresponding stage 2 particles (Dp = 4.8–15 μm). To examine this effect on a larger range of particle sizes, we extracted stages 1–4 for one sample (B9904) and measured the rates of OH formation during illumination. As shown in Table 2, rates of OH photoformation in the sea-salt particles increase dramatically with decreasing particle size. Relative to the rate in the largest (stage 1) particles (36 μM hr−1), the sodium concentration–normalized rates of OH formation are 1:6:26:230 for particles on stages 1–4, respectively. While these values represent the rates within the particles of each size, we can also examine the rates of OH formation on an air parcel volume basis to account for differences in the sea-salt mass at each size. Relative to the rate in the stage 1 particles (43 molecules cm3 air s−1), these air volume-normalized rates of OH formation are 1:38:69:250 stages 1–4, respectively (Figure 3). Particles from stages 2 and 3 are relatively more important from this point of view because these particle sizes contribute the most to the overall aerosol mass [Newberg et al., 2005]. Although the stage 4 sample has the highest rate of OH photoformation, in this small size range (0.48–1.6 μm) there are probably significant contributions from particle types other than sea-salt, such as sulfate aerosols [Newberg et al., 2005]. Thus, for this stage, it is unclear whether the observed OH formation is representative of sea-salt particles of this size or whether mixing of different particle types on this sample stage significantly altered the chemistry.

Figure 3.

Rates of hydroxyl radical photoformation in the particles (expressed relative to the air volume of the aerosol, circles), and the fraction of OH formation due to nitrate photolysis (fOH,NO3, squares), as a function of particle size for sample B9904. Rates of OH photoformation were scaled from an extract volume basis (μmol L−1 extract hr−1) to an aerosol air volume basis (molecules cm−3 air s−1) on the basis of the volume of air sampled (2377 m3), the fraction of Teflon substrate extracted (0.50 for each stage), and the extract volume (0.600 mL for each stage). Particle diameters are plotted at the midpoint of each stage.

3.3.3. Sources of OH

[28] Nitrate (NO3) is a major chromophore responsible for the formation of hydroxyl radical in our sea-salt particles via the sequence [Chu and Anastasio, 2003; Zafiriou and True, 1979]

equation image
equation image

On average, nitrate photolysis accounts for 59 ± 25% of photoformed OH in the stage 2 and 3 particles, but this contribution varies significantly, ranging from 19–100% (Table 2). With one exception (B9903), nitrate contributed more to OH formation in the larger, stage 2 particles compared to the smaller, stage 3 particles from the same sample. The same trend is seen by comparing the average stage 2 and 3 results, but the difference is not statistically significant (fOH,NO3 = 66 ± 29% and 55 ± 17%, respectively; see Table 2). In sample B9904 we determined the contribution of nitrate photolysis as an OH source for all four impactor stages. As shown by the squares in Figure 3, nitrate accounts for most of OH formation in the largest particles, but this contribution decreases with decreasing particle size. On the basis of our results (Figure 3 and Table 2), chromophores other than nitrate are significant sources of OH in the bulk of the sea-salt particle mass size range (i.e., Dp = 1.6–15 μm) and are dominant in the smaller particles.

[29] Past reports have shown that the majority of OH photoformation in surface seawater is due to reactions of unknown chromophores, while photolysis of nitrate, nitrite, and hydrogen peroxide are minor or insignificant. On the basis of good correlations between OH photoformation rates and measures of colored dissolved organic material (CDOM), such as sample fluorescence and/or light absorbance, it appears that unidentified, organic chromophores are responsible for most of OH photoformation in seawater [Mopper and Zhou, 1990; Takeda et al., 2004]. Other reports have shown that, relative to conservative tracers such as sodium, sea-salt particles are enriched in organic compounds by factors of 102–103 compared to the levels in seawater and that this concentration effect generally increases with decreasing particle size [Gershey, 1983; Hoffman and Duce, 1977; Oppo et al., 1999; Turekian et al., 2003]. Combined with our observations in Figure 3, this suggests that CDOM photoreactions could be a significant source of OH in sea-salt particles and could be the dominant source in the smaller, submicron particles. Indeed, multiplying previously measured rates of OH formation from unknown chromophores in coastal surface seawater (∼ 0.1 μM hr−1) [Mopper and Zhou, 1990] by the concentration factor of DOM in sea-salt particles relative to seawater (102–103), provides estimated rates of OH formation from CDOM photolysis in sea-salt particles (10–100 μM hr−1) that are roughly comparable to the rates due to unknown chromophores in our samples (Table 2). While photolysis of CDOM is broadly consistent with the production of OH from unknown chromophores in our samples, other mechanisms might also contribute significantly. These pathways include the photo-Fenton reaction (i.e., reaction of photoproduced Fe(II) with HOOH), photolysis of metal-carboxylate complexes, or nitrite photolysis [Arakaki and Faust, 1998; Arakaki et al., 2006; Faust and Zepp, 1993; Zepp et al., 1992]. Nitrite is always below the detection limit in our samples (Table S1), but even at these low levels it could make a significant contribution to OH formation in some of the samples.

[30] Concentrations of nitrate in surface seawater are quite low, of the order of 10−6 M or less, which corresponds to a NO3/Na+ molar ratio of approximately the same size, i.e., ≤10−6. In contrast, values of NO3/Na+ in our sea-salt particle samples are typically 104–105 times higher, with an average (±1 σ) value of 0.065 ± 0.072 mol mol−1 and a median of 0.038 (range = 0.006 to 0.33). Thus essentially all of our sea-salt particle nitrate is secondary, derived from the partitioning of oxidized nitrogen species such as HNO3, NO3, and N2O5 to sea-salt particles [Finlayson-Pitts, 2003]. Values of NO3/Na+ in our sea-salt particles are similar to previously reported values, which range from ∼ 0.03 in the cleanest areas to ∼ 0.2 in polluted marine air masses, and which are typically 0.05–0.07 for moderately clean marine air [Cavalli et al., 2004; Huebert et al., 1996; Johansen et al., 1999; Savoie and Prospero, 1982]. The fact that these previously reported samples were collected at a variety of locations over the Atlantic, Pacific, and Indian Oceans indicates that nitrate photolysis in sea-salt particles is widespread in the marine boundary layer. If NOx emissions from ships continue to increase in the future [Eyring et al., 2005], concentrations of nitrate in sea-salt particles should likewise rise, leading to greater rates of OH photoformation and a more vigorous processing in sea-salt particles. Such vigorous nitrate photochemistry in sea-salt particles is already occurring in coastal urban regions, where much, or even all, of the chloride in sea-salt particles can be replaced by nitrate [Gard et al., 1998], corresponding to a molar NO3/Na+ ratio of up to 1.2.

3.4. OH Sinks and Lifetimes

[31] Lifetimes of OH in our SS PM samples at 88% RH (τOH,SS) range from 1.4 to 16 ns, with an average of 5.4 (± 4.6) ns (Table 2). This is approximately 100 times shorter than the lifetime of OH in seawater, where Br is the dominant sink, consuming approximately 93% of the photoproduced OH [Qian et al., 2001; Zafiriou, 1974; Zafiriou et al., 1987; Zhou and Mopper, 1990]. Bromide was a much less important (though still significant) sink for OH in our sea-salt particles, accounting for an average of 19% (± 18%) of OH loss. This pathway would have been more important if not for the fact that the particles, on average, were missing approximately half of their initial bromide (i.e., EF(Br) = 0.54 ± 0.17; see Table 2). Chloride was a minor sink for OH, accounting for 5.0 ± 4.8% of OH loss (Table 2), while nitrate and sulfate were both insignificant, together accounting for <0.1% of OH loss in any sample (data not shown). The chloride contribution is small because although the rate constant for Cl with OH is very rapid ((R7) in Table 1), less than 0.1% of the ClOH formed goes on to make products (i.e., (1 − ϕCl,SS) < 0.001) while the remaining 99.9+% breaks down to reform OH in the SS PM. In contrast, approximately 95% of BrOH goes on to make oxidized bromide products under sea-salt particle conditions.

[32] Most OH in our sea-salt particles reacts with unknown, “other” components, which account for an average of 76 ± 22% of the loss of hydroxyl radical (Table 2). It is likely that dissolved organic matter (DOM) accounts for most of this unknown sink since organic compounds are abundant in sea-salt particles [Cavalli et al., 2004; Hoffman and Duce, 1977] and they typically react very rapidly with OH [Ross et al., 1998]. In addition, although the size-segregated data for τOH,SS are limited (Table 2), they indicate that OH has a shorter lifetime in the smaller, stage 3 particles compared to the larger, stage 2 particles, which is consistent with the greater enrichment of organic compounds in smaller particles [Oppo et al., 1999; Turekian et al., 2003].

[33] The average value of the pseudo first-order rate constant for loss of OH due to “other” sinks in the sea-salt particles at 88% RH (i.e., kOH,OTHER,SS) is (2.5 ± 1.8) × 108 s−1. Using an estimated rate constant for OH with sea-salt DOM of 109 M−1 s−1, we estimate that the average concentration of dissolved organic material in our SS PM is 250 (± 180) mM with a range of 20–670 mM (determined as [DOM] = kOH,OTHER,SS/kOH+DOM). This represents an average DOM/Na+ molar ratio of 0.08 and, on the basis of a typical sea-salt particle loading of 230 nmol Na+ m−3 air at Bodega Bay, California [Newberg et al., 2005], corresponds to a sea-salt aerosol loading of 18 nmol DOM m−3 air. It should be noted that since our OH measurements were performed on filtered, aqueous extracts of sea-salt particles these results are only for the dissolved organic compounds and do not include the influence of any nonsoluble organic surface coatings or particle inclusions [Tervahattu et al., 2002]. Finally, in addition to their role as the dominant sink for OH, organic compounds are probably also significant sinks for other reactive species in sea-salt particles, including ozone and reactive halogens.

3.5. OH Concentrations

[34] Steady state concentrations of OH in ambient SS PM at midday on the summer solstice, [OH]SS, range from (0.63 to 8.9) × 10−16 M, with no apparent difference between the stage 2 and 3 particles (Table 2). It is interesting to note that the concentration in the stage 4 sample is 15 times greater than the average in the stage 2 and 3 particles; however, as described above, reactivity in the smaller, stage 4, particles could be enhanced by mixing of different particle types and might not be representative of sea-salt particles in the MBL. The average value of [OH]SS in the stage 2 and 3 particles, (3.8 ± 2.8) × 10−16 M, is roughly 100 times greater than values observed in ocean waters [Mopper and Zhou, 1990; Zhou and Mopper, 1990]. This steady state concentration represents a balance between the rate of OH formation and the rate constant for OH loss (equation (14)). The much greater value of [OH] in the SS PM shows that, relative to bulk seawater, the dissolved sources of OH in sea-salt particles (i.e., nitrate and, probably, CDOM) are much more enhanced relative to the increase in the OH sinks (i.e., DOM and bromide).

[35] Our calculations (not shown) indicate that the steady state concentration of OH from in situ chemistry is relatively independent of changes in the ambient relative humidity (RH), despite the fact that these changes alter the amount of water in sea-salt particles, and therefore the concentrations of in situ OH sources and sinks. On the basis of the formulation of Lewis and Schwartz [2004], between a relative humidity of 76% (the deliquescence RH of sea-salt particles) and 99% the diameter of a sea-salt particle grows by a factor of ≈ 2.7, corresponding to a decrease in the aqueous concentrations of nonvolatile particle constituents by approximately a factor of 20. This increase in RH will decrease the rate of OH formation (ROH,SS) in the particles by approximately a factor of 20, but it will have a similar effect on the rate constant for OH destruction (kOH,SS). Thus these changes essentially cancel and the steady state concentration of OH from in situ chemistry is approximately independent of RH. This can be seen, for example, by the fact that the OH concentration in a given sea-salt particle sample (Table 2) is essentially the same as the concentration in the corresponding particle extract, which is significantly more dilute (Table S1). However, although OH concentrations from in situ reactions are relatively independent of RH, the rate of OH delivery from gas-to-particle partitioning, which is also a significant source to sea-salt particles, depends upon particle diameter and therefore RH (section 3.7).

3.6. Effects of pH on the Photoformation Rate and Lifetime of OH

[36] The measurements of OH formation and loss reported above were all performed at the pH values of the SS PM extracts, which ranged from pH 4.9 to 5.9 with an average (± 1 σ) of 5.5 ± 0.3 (Table S1). While we cannot determine pH values of the sea-salt particles from these extract measurements, these values are roughly comparable to median values previously reported for supermicron marine particles, i.e., pH 3 to 5.5 [Keene et al., 2004; Pszenny et al., 2004]. These pH values for aged sea-salt particles are much lower than the estimated value for newly formed SS PM (pH ∼ 8) as a result of the particles accumulating acids such as HNO3 and H2SO4 as they age [Keene et al., 1998].

[37] To examine how OH kinetics are affected by this change in pH over the lifetime of a sea-salt particle, we measured the formation rate and lifetime of OH in different portions of the same extract, each adjusted to a different pH using H2SO4 or borate. Figure 4 shows two distinct cases for the pH dependence of ROH,EXT. The first is a sample where OH formation is dominated by nitrate photolysis (solid circles; average fOH,NO3 = 0.91) and there is little change in ROH,EXT between pH 2.1 and 7.5. This behavior is expected since the quantum yield for OH from nitrate photolysis is independent of pH in this approximate range [Zellner et al., 1990]. The second case is for a sample where nitrate is not the dominant source of OH (open circles; fOH,NO3 = 0.36 at pH 5.7). In this extract the photoformation rate of OH varies by a factor of three, with a peak at pH 4.6 and the lowest value at pH 7.3 (where approximately 70% of OH formation is from nitrate photolysis). If we consider only the portion of ROH,EXT due to unknown chromophores (i.e., the total rate minus the nitrate contribution), the OH photoformation rate increases by nearly a factor of 9 between pH 7.3 (1.6 μM hr−1) and pH 4.6 (13.7 μM hr−1). This indicates that, at least for this sample, the nonnitrate mechanisms for producing OH become much more efficient as fresh sea-salt particles age and become acidified.

Figure 4.

The pH dependence of the rate of OH photoformation (ROH,EXT, circles) and lifetime of OH due to reaction with natural scavengers (τOH,EXT, squares) in sample extracts B9802 (open symbols) and B9801 (solid circles). Each extract is an aggregate of particles from stages 2 and 3 for each sample.

[38] Figure 4 also shows how the lifetime of OH in the extract of sample B9802 varies with pH. Within one standard error, τOH,EXT is essentially the same (∼ 130 ns) at pH values above 4.5, but is nearly 3 times smaller at pH 2.1 (48 ns). This difference is consistent with the strong pH dependence of the net reaction of OH with Cl (Table 2 and section 2.2.3.). From pH 5.8 to 2.1, the rate constant for OH loss due to Cl (kOH,Cl,EXT) in this sample increases by a factor of 1200. This change is because of the pH-dependent conversion of ClOH to Cl (reaction (R9) in Table 2), which enhances the term (1 − ϕCl,EXT) by a factor of 1200 (calculated using the chloride equivalent to equation (9)). This change elevates chloride from an insignificant sink in the pH 5.8 extract (where fSINK,Cl,EXT = 0.18%) to the dominant sink in the pH 2.1 extract (fSINK,Cl,EXT = 66%) and highlights the important pH dependence of OH reactions with sea-salt particle chloride. Since reaction with chloride can explain essentially all of the pH dependence for τOH,EXT here, these results suggest that unknown sinks have very little pH dependence in their reaction with OH, at least for this one sample. Finally, it should be noted that acidic conditions also enhance the efficiency of bromide as a sink for OH (reaction (R4) in Table 2), but only by a factor of two since the term (1 − ϕBr,EXT) is already relatively high (0.43) in the pH 5.8 extract.

3.7. OH Gas-to-Particle Partitioning Rates Versus In Situ Photoformation

[39] The significance of the in situ photochemical formation of OH in sea-salt particles depends, in part, on how its rate compares with the rate of partitioning of gaseous OH to the particles. To make this comparison we first calculate the rate of gas-to-particle partitioning of OH to SS PM in the MBL using the Fuchs-Sutugin transition regime formula [Seinfeld and Pandis, 1998] with a modeled, midday, gas phase OH concentration of 4 × 106 molecules cm−3 [von Glasow et al., 2002] and two values for the mass accommodation coefficient of OH (αOH = 0.01 and 1.0) to reflect the range reported by Takami et al. [1998].

[40] As shown in Figure 5, at the lower value for αOH the rates of OH formation from photolysis of aqueous chromophores are approximately 8 times greater than the rates of OH mass transport. At the upper bound value for αOH of unity, rates of OH photoformation in sea-salt particles are comparable to the rates of mass transport. However, regardless of the value of the accommodation coefficient, the spatial distributions of hydroxyl radical from these two sources will be very different because the lifetime of OH (10−9 – 10−8 s; see Table 2) is very short compared to the characteristic time for liquid phase diffusion in sea-salt particles (∼ 10−4–10−2 s) [Seinfeld and Pandis, 1998]. Thus, while in situ OH will form and react relatively uniformly throughout the bulk volume of the particles (reflecting the distribution of its precursor chromophores), OH from the gas phase will react only at the surfaces of the particles. Recent modeling suggests that the reaction of gaseous OH with chloride on the surface of SS PM is a major source of gaseous Cl2 [Thomas et al., 2006], but it is unclear how this interfacial chemistry is affected by the presence of surface organics [Tervahattu et al., 2002].

Figure 5.

Comparison of the rates of in situ OH photoformation in sea salt particles with rates of transfer of OH from the gas phase. The boxes show, for each stage of the impactor (dashed boxes) and for the stage 2 and 3 aggregates (solid box), the ranges of particle sizes collected and measured rates of aqueous phase OH photoformation (ROH,SS; see Table 2). Note that stages 1 and 4 each represents results from only a single sample and that the full particle size range for stage 4 (0.48–1.6 μm) is not shown. The dotted lines represent the rates of mass transport of gas phase OH (at a concentration of 4 × 106 molecules cm−3) as a function of particle size using mass accommodation coefficients of 1.0 (top line) and 0.01 (bottom line).

3.8. Implications of OH Photoformation in Sea-Salt Particles

[41] Because it is highly reactive, OH in sea-salt particles likely affects the chemical cycles of a number of particle constituents, such as organic compounds, halides, and reduced sulfur species (e.g., S(IV)). For dissolved organic compounds that react with near diffusion-controlled rate constants with OH (i.e., 109–1010 M−1 s−1) [Ross et al., 1998], their lifetimes with respect to OH reaction in sea-salt particles will be of the order of 102–103 hours (midday, summer solstice, 38°N), on the basis of our average value of [OH]SS (Table 2). Thus hydroxyl radical will only be a significant sink for those organic compounds with which it reacts very rapidly.

[42] While this calculation shows that the effect of OH on any particular organic compound might not be large, the aggregate impact of the OH reactions could be significant because of the large flux of OH in the particles. If we assume that the unknown, “other” sinks for OH are organic compounds (section 3.4), then approximately 75% of the OH flux is shuttled into reacting with organics, which likely increases the hygroscopicity and hydrophilicity of sea-salt organic matter [Rudich, 2003]. In addition, these reactions could affect the composition of the gas phase by release of volatile organic compounds (VOCs) that are produced by the photo-oxidation of sea-salt organics [Riemer et al., 2000; Zhou and Mopper, 1997]. On the basis of the size-segregated data in Figure 3, the in situ OH flux in the aerosol is of the order of 104 molecules cm−3 air s−1. Thus, if we consider an upper bound, where every OH reaction with sea-salt organics emits one VOC molecule into the gas phase, the total flux of VOCs could be up to 104 molecules cm−3 air s−1. On the basis of reported gas phase mixing ratios and lifetimes for a number of VOCs measured on the coast of northern California [Millet et al., 2004], this OH-mediated release of VOCs from sea-salt particles might be an important source for longer-lived oxygenated compounds such as acetone, methanol, and ethanol.

[43] While, on average, approximately 75% of OH in sea-salt particles reacts with unknown (likely organic) compounds, the remaining ∼25% reacts irreversibly with bromide and chloride (Table 2). These reactions form halogen radicals, such as Br and Br2, as well as volatile species, such as Br2 and BrCl [Anastasio and Matthew, 2006; Thomas et al., 2006]. These reactive halogens can combine with unsaturated organic compounds to abiotically form halogenated organic compounds [e.g., Anastasio and Matthew, 2006; Matthew et al., 2003], while the volatile species can also evaporate into the gas phase. Photolysis of gaseous Br2 and BrCl forms halogen radicals, which can destroy boundary layer ozone and react with gaseous organic compounds, ozone, and mercury [von Glasow and Crutzen, 2003]. Laboratory studies and preliminary modeling indicate that the release of Br2 from reactions of halides with in situ and gaseous OH in sea-salt particles can be a significant source of gaseous, reactive halogens in midlatitudes [Frinak and Abbatt, 2006; George and Anastasio, 2007; Matthew et al., 2003]. Similar OH reactions in particles, snow, sea ice, and frost flowers might also be important in the chemistry of halogens and ozone in polar regions [Anastasio and Jordan, 2004; Mozurkewich, 1995].

[44] The channel of nitrate photolysis that forms OH also produces equimolar amounts of nitrogen dioxide [Chu and Anastasio, 2003], most of which probably escapes to the gas phase. On the basis of the data for sample B9904 shown in Figure 3, the flux of NO2 to the boundary layer in this aerosol is approximately 2000 molecules cm−3 air s−1, which is quite low and will only be significant in regions of the MBL with very low levels of NOx. In addition to this nitrate chemistry, it is likely that there is a wealth of chemistry occurring through photoreactions of unidentified chromophores in sea-salt particles. The enormous optical depths in our samples (Table 2), and the fact that most of this light absorption is due to unknown species, indicates that there is a huge potential for direct (as well as indirect) photochemical reactions in sunlit SS PM. This set of reactions includes the photoformation of OH that we have characterized here and, probably, the formation of other reactive species that have been found in seawater and atmospheric drops and particles, such as peroxides and peroxyl radicals, singlet molecular oxygen, and excited triplet states of organic compounds [Faust, 1994; Zafiriou et al., 1984]. These species undoubtedly play important roles in the chemistry and composition of sea-salt particles, and the resulting reactions likely release volatile species to the gas phase, which could affect its composition and chemistry.

4. Conclusions

[45] We have found that the photoformation of OH in sea-salt particles is ubiquitous: all of our samples produce OH upon illumination, with an average rate (±1σ) of 340 (±300) μM hr−1 for midday, summer sunlight at 38°N, and a relative humidity of 88%. This rate is approximately 103 to 104 times greater than the rate of hydroxyl radical photoformation in surface seawater, reflecting the great concentration of chromophores (light absorbing species) in sea-salt particles relative to seawater. Typically half or more of the in situ OH formation in our samples is due to photolysis of nitrate, which is derived from atmospheric sources rather than from seawater. The chromophores responsible for the remaining portion of OH formation are unknown but probably include dissolved organic compounds derived from seawater. The average (± 1σ) lifetime of OH in our sea-salt particles is 5.4 ± 4.6 ns and is primarily controlled by unknown, but likely organic, compounds. Bromide and chloride together account for approximately 25% of the loss of OH in the sea-salt particles, in contrast to seawater where Br accounts for over 90% of OH loss. The average steady state concentration of OH in ambient sea-salt particles is (3.8 ± 2.8) × 10−16 M, approximately 100 times greater than the level in surface seawater.

[46] The photoformation of hydroxyl radical in sea-salt particles will affect the chemistry and composition of the particles and, probably, that of the surrounding gas phase as well. For example, the processing of sea-salt organic compounds by OH will make these compounds more hydrophilic and will probably also release volatile organics to the gas phase. In addition, reactions of OH with bromide and chloride will form reactive halogens, including volatile species such as Br2. On the basis of the very high light absorption coefficients of sea-salt particles, the formation of OH is just one pathway in a collection of in situ photochemical reactions that generates reactive oxidants and affects chemistry in the marine boundary layer.

[47] On the basis of our results, the chemistry of hydroxyl radicals in SS PM reflects both the birth of the particles (that is, the injection of drops of seawater into the atmosphere [Lewis and Schwartz, 2004]) as well as the subsequent lives of the particles (i.e., the aging of SS PM in the boundary layer as they accumulate acids and other soluble, low-volatility chemical species [Finlayson-Pitts, 2003; Keene et al., 1998]). These same factors likely shape the cycles of other reactants in sea-salt particles and complicate the chemistry of SS PM compared to the chemistry of surface seawater.


[48] This work was supported by the Atmospheric Chemistry Program of the National Science Foundation (ATM-9701995), the California Agricultural Experiment Station (project CA-D*-LAW-6403-RR) and by a Jastro Shields Fellowship from the University of California, Davis, to J.T.N. Generous logistical support for shipboard sampling was provided by NOAA and Steve Ralston from the National Marine Fisheries Service. We also thank the Bodega Marine Laboratory (UC Davis) for logistical and financial support and Dawnnica Williams for laboratory assistance.