Research Articles

Reactions at Interfaces As a Source of Sulfate Formation in Sea-Salt Particles

See allHide authors and affiliations

Science  18 Jul 2003:
Vol. 301, Issue 5631, pp. 340-344
DOI: 10.1126/science.1085374


Understanding the formation of sulfate particles in the troposphere is critical because of their health effects and their direct and indirect effects on radiative forcing, and hence on climate. Laboratory studies of the chemical and physical changes in sodium chloride, the major component of sea-salt particles, show that sodium hydroxide is generated upon reaction of deliquesced sodium chloride particles with gas-phase hydroxide. The increase in alkalinity will lead to an increase in the uptake and oxidation of sulfur dioxide to sulfate in sea-salt particles. This chemistry is missing from current models but is consistent with a number of previously unexplained field study observations.

Sulfate particles are a well known cause of adverse health effects (1) and play a major role in global climate (24). These particles scatter incoming solar radiation (direct effect) and act as cloud-condensation nuclei, thus altering cloud properties and the associated impacts on radiation (indirect effect) (5). Sulfate aerosols are formed from the oxidation of SO2 by OH in the gas phase and the subsequent uptake of H2SO4 by particles, or by O3 and H2O2 in the liquid phase of aerosol particles, clouds, and fogs (5). Sulfur dioxide is generated during the combustion of fossil fuels and by the atmospheric oxidation of biogenic organic sulfur compounds, particularly dimethyl sulfide (6, 7).

A number of models have been applied to predict the formation of sulfate aerosols on a global scale. Surface SO2 concentrations are typically overestimated by as much as a factor of 2; sulfate tends to be underestimated (8, 9). Although there are meteorological uncertainties in the models, there is also substantial uncertainty associated with the aerosol chemistry. For example, most models of sulfate formation and their radiative impacts do not include the uptake and oxidation of sulfur compounds inside the aqueous phase of sea-salt particles. Recently, Gong and Barrie (10) predicted that the inclusion of sea salt will increase the mean particle size for sulfate and decrease the number concentration of sulfate aerosols as a result of the uptake of gaseous H2SO4 by sea salt before it nucleates to form new (and smaller) particles. However, oxidation of sulfur dioxide in the sea-salt particles themselves was not included in that model.

Previous modeling studies of the marine boundary layer (1113) predict that SO2 will be taken up into alkaline sea-salt particles and oxidized rapidly by dissolved O3. The pH is predicted to remain in the range of ∼8 until the alkalinity is titrated and then to fall as sulfuric acid continues to be generated by the reaction between dissolved S(IV) and O3. The decreasing pH shifts the following equilibria to the left, decreasing the concentration of dissolved SO2 as well as slowing the oxidation by O3 (14) Embedded Image Embedded Image(1)

The net result is “self-quenching,” and below a pH of about 5, other oxidants, primarily H2O2, become more important, but with a rate that is still considerably slower than that by O3 in alkaline solutions.

We show here the results of laboratory studies of deliquesced NaCl particles in which involatile products of the reaction with gas-phase OH are identified by means of a combination of scanning electron microscopy with energy-dispersed analysis of X-rays (SEM-EDX) and time-of-flight secondary ion mass spectrometry (TOF-SIMS). In the marine boundary layer, the NaOH generated in this reaction will provide a previously unrecognized buffering mechanism that will prevent the pH from decreasing and thus prevent the sulfate formation rate from falling as rapidly as previously predicted. The uptake and oxidation of SO2 in solution to form sulfate should therefore be enhanced substantially. This chemistry is also shown to be consistent with a number of reported atmospheric observations of sea-salt and sulfur oxidation.

Figure 1 shows the morphology of a typical set of NaCl particles before (Fig. 1A) and after (Fig. 1B) exposure to OH; an enlarged view of one of the particles is shown in the inset in Fig. 1B. There is clear evidence of a reaction product that has segregated on the surface of the original NaCl particle. Furthermore, the particle appears to be coated with a substance that creates a halo-like appearance around the exterior. Figure 1C shows the X-ray fluorescence spectrum for a typical 0.7 μm NaCl particle before and after the reaction with OH; chlorine is depleted and oxygen is considerably enhanced after the reaction.

Fig. 1.

Particle morphology for a set of NaCl particles (A) before and (B) after reaction with OH generated by photolysis of O3 to form O(1D), followed by its reaction with water vapor. SEM imaging was carried out with a FEI XL30 FEG digital microscope. The NaCl particles were nebulized from a 0.5 M NaCl solution prepared with Nanopure water (Barnsteal International, Boston) and then were deposited on grid-supported TEM thin-film substrates. Carbon Type-B (Ted Pella, Inc., Redding, CA) films supported by gold 200-mesh grids were used in this work. The grids were inserted into a 130 cm3 cylindrical cell with quartz windows, and the cell was evacuated. Mixtures of O2 and O3, generated with a commercial ozonizer (Model T-816, Polymetrics, U.S. Filter, Warrendale, PA), and humidified air were added to 1 atm pressure in the cell. The cell was then exposed to UV radiation [wavelength (λ) = 254 nm] for 1 hour, and its concentration was measured periodically during the irradiation by UV/visible spectroscopy; more than 90% of the initial O3 reacted. The gases were then pumped out and the cell was refilled with O3 and humid air, followed by another hour of UV radiation exposure. This procedure was repeated three times during each experiment to ensure that sufficient OH was available for the reaction. The initial ozone concentration was 3800 parts per million (ppm), and the relative humidity was 81%. The average concentration of OH was ∼1010 cm3 during the photolysis. Particles exposed to similar concentrations of O3 and water vapor without UV radiation showed no observable change in particle morphology and composition. (C) The x-ray fluorescence spectra from a typical 0.7-μm particle, both before and after reaction with OH.

Figure 2 shows the particle-by-particle elemental concentration of chlorine and oxygen (expressed as an atomic ratio to sodium) as a function of particle size for NaCl particles reacted with OH. The corresponding composition of the same particles before their reaction with OH is also shown for comparison. The smaller reacted particles show substantial amounts of chlorine ion loss and increases in oxygen. The same absolute amount of uptake and reaction of OH with smaller particles results in a greater extent of reaction of a single particle, consistent with the larger extent of conversion of the smaller particles.

Fig. 2.

Elemental ratios of (A) Cl/Na and (B) O/Na for NaCl particles before (blue open circles) and after (red solid circles) reaction with OH as described in Fig. 1. Elemental composition of individual particles was determined with a computer-controlled operation of SEM/EDX (45, 46). The dashed line in (A) corresponds to the nominal ratio of Cl/Na = 1 for NaCl. Deviations from the nominal line at larger sizes is a result of ZAF effects [atomic number (z)-dependent electron scattering, absorption (A), and fluorescence (F)] (47) pertinent for particles >1 μm, which were entirely omitted by the applied x-ray quantification method (48). Substantial replacement of chlorine by oxygen was clearly observed in submicron NaCl particles after their exposure to OH. No nitrogen or sulfur that might have resulted from reactions of trace amounts of acid in air during handling was observed. The solid lines indicate data obtained with moving averages of 20 clustered measurements from the analysis of particles prepared from solutions of NaOH and NaOCl as indicated. A moving average smooths the fluctuations in the data and thus shows the trends more clearly. The asterisk indicates that particles prepared from the NaOCl solution are a mixture of NaCl/NaOH/NaClOx where x = 2 to 4.

Previous studies (15, 16) of the OH reaction with deliquesced NaCl aerosols showed that Cl2 was a major gas-phase product, and it was postulated that it was formed in the following reaction at the air-solution interface Embedded Image Embedded Image(2)

If reaction 2 represents the chemistry, particles should become increasingly alkaline as the reaction proceeds. Molecular chlorine is rapidly hydrolyzed in basic solution, so that some of the gaseous Cl2 will be taken back up in the particles to form hypochlorite Embedded Image Embedded Image(3)

Thus, if OH is formed by the interface reaction in the particles, they will be converted from NaCl to a mixture of NaCl, NaOH, and NaOCl. The same mixture is present in commercially available solutions of NaOCl that are commonly generated with reaction 3.

The elemental composition of particles prepared from solutions of NaOH and NaOCl is also shown in Fig. 2. In the case of NaOH, the O/Na ratio is greater than unity. This indicates that even under the vacuum conditions during the analysis, there is water associated with the highly hygroscopic NaOH, consistent with the known formation of hydrates (17, 18). In the case of NaOCl, the Cl/Na ratio is less than 1, and there are substantial amounts of oxygen in the particles (19). The composition of the NaCl particles reacted with OH is qualitatively similar to that of the particles from the NaOCl solution.

The molecular form of the oxygen-containing product in the NaCl reaction with OH cannot be determined with SEM/EDX. Furthermore, SEM/EDX provides elemental chemical analysis averaged over the entire particle volume, rather than the material segregated on the salt surface (Fig. 1B). To probe the products on the particle surface and to provide some insight into their molecular composition, we carried out TOF-SIMS on samples that had been exposed simultaneously with the samples analyzed by SEM/EDX. Figure 3A shows the spatial distribution of the total ion count—as well as signals for Cl, O and OH—for a portion of the sample containing particles reacted with OH (20). The oxygen-containing groups are collocated with chlorine in the particles. Also shown in Fig. 3B are similar TOF-SIMS data for a sample that was exposed to O3 and water vapor without ultraviolet (UV) radiation; in this case, only Cl is associated with the particles, indicating undetectable amounts of reaction in the absence of OH. Particles prepared from a NaOCl solution also show Cl, O and OH (Fig. 3C).

Fig. 3.

TOF-SIMS spatial images of NaCl reacted with O3 and H2O (A) with UV radiation to generate OH and (B) without UV radiation. A primary ion beam of Ga+ ions is directed at the sample surface, where the interaction of the primary ions leads to the ejection of neutral and ionized atoms, molecules, and clusters. The negative ions are collected and mass-separated in a TOF analyzer to obtain a mass spectrum. The primary ion beam is pulsed, and an entire mass spectrum is collected for each incident pulse. In conjunction with electrodynamic bunching, a primary pulse width of <1 ns is used to achieve a mass resolution of about m/Δm > 9000, with a focused primary beam of 2 μm to 3 μm in diameter. Mass-specific images are obtained by defining a mass window to monitor for the entire acquisition (images) or as a function of erosion time (depth profiles), as shown in Fig. 4. In these experiments, we carried out TOF-SIMS measurements using a Physical Electronics TRIFT II TOF-SIMS with a 69Ga+ source in a high mass-resolution mode. In this mode, a 15-kV, 600-pA electrodynamically bunched primary ion beam with <1 ns pulse width is used. Also shown (C) are similar data for NaCl/NaClO3/NaOH mixed particles obtained from a solution of NaOCl. The frame size of the spatial images is 40 μm by 40 μm for (A) and (B) and 100 μm by 100 μm for (C).

Figure 4 shows the O and OH TOF-SIMS signals as a function of depth from the surface for NaCl reacted with O3 and H2O, both with and without UV radiation. As anticipated from morphology changes (Fig. 1), the O and OH ion signals decrease with depth, consistent with a mixture of NaOH and/or NaClOx (19) products on the surface.

Fig. 4.

TOF-SIMS peak intensities for O and OH ions relative to Cl signal as a function of number of sputter cycles for NaCl reacted with O3/H2O in the absence (blue symbols) and presence of UV radiation, i.e. with OH (red symbols). The interrogated depth was estimated from SEM images to correspond to about 4 μm.

In summary, the data presented here show that OH is generated in the reaction of OH radicals with chloride ions at the interface. The subsequent uptake of some of the Cl2 into the alkaline particles generates NaOCl through reaction 2. The segregated material and the halo around the particles shown in Fig. 1B are consistent with the formation of NaOH/NaClOx products.

This interface reaction to generate OH should also occur with Br in sea-salt particles, because molecular dynamics (MD) simulations (21, 22) show that the surface concentration of bromide relative to the bulk is much higher than that for chloride.

The generation of alkalinity in sea-salt particles by the interface reactions in air will neutralize H2SO4 as it is formed inside the particles either by oxidation of S(IV) by O3 or through uptake from the gas phase. Assuming a sea-salt particle concentration of 10 cm3 of air and a typical particle diameter of 2 μm, the rate of generation of OH in the particles can be calculated from the rate of collisions of OH with the particle surfaces and the reaction probability for the OH-Cl reaction. MD simulations show that 85% of gas-phase OH radicals striking a salt surface are taken up (23), consistent with experiments on the uptake of OH into the quasiliquid layer of ice containing HCl where the uptake coefficient is greater than 0.2 (their upper limit for quantitative measurements) (24). A net probability for uptake and reaction that approaches unity therefore is reasonable for the OH-Cl interface reaction. Taking an average daytime OH concentration of 5 × 106 radicals cm3, the rate of formation of OH from the interface reaction for 10 particles cm3 is 9.6 × 104 s1 cm3 of air, or 3.8 × 106 M s1 in the particle liquid phase.

Sea-salt particles are initially alkaline, which enhances uptake of SO2 into the aqueous phase through the equilibria (reaction 1). Oxidation by O3 in alkaline solution is relatively fast, generating sulfuric acid in the particles. This acid is neutralized by carbonate in seawater until the initial alkalinity is consumed, at which point the pH falls (11). The increasing acidity of the particles decreases the equilibrium amounts of dissolved SO2 [S(IV) = SO2(aq) + HSO3 + SO32–] and as a result decreases kinetics of the O3 oxidation (14), which is the sum of contributions from the reactions of SO2(aq) (k0 = 2.4 × 104 L mol1 s1), HSO3 (k1 = 3.7 × 105 L mol1 s1), and SO32– (k2 = 1.5 × 109 L mol1 s1) Embedded Image Embedded Image(4) where α0, α1, and α2 are the fractions of total dissolved S(IV) in the form of SO2(aq), HSO3 and SO32–, respectively, and can be calculated from the known equilibrium constants (5) for dissolved SO2. With O3 at a concentration in the gas phase of 30 parts per billion (ppb) (Henry's Law constant of 1.0 × 102) and an SO2 concentration of 60 parts per trillion (ppt) in the gas phase (25), applying the equilibria and rate constants shows that the rate of acid generation from the reaction of S(IV) with O3 will be equal to the rate of OH formation from the OH-Cl interface reaction at a pH of 6.9.

Thus, freshly generated 2-μm alkaline sea-salt particles will take up SO2, and sulfuric acid will be formed, causing the pH to fall. At a pH of ∼7, the hydroxide generated by the interfacial oxidation of OH neutralizes the acid during the day as it is formed, resulting in a higher pH than predicted without the interface chemistry. This will lead to a higher equilibrium concentration of dissolved S(IV), more rapid oxidation by O3, and a larger rate of generation of sulfate in sea-salt particles than predicted without the interface reaction.

A second source of acidity (26) in sea-salt particles is the uptake of gaseous H2SO4 formed from the gas-phase reaction of SO2 and OH. Gas-phase H2SO4 concentrations have been measured at coastal sites (25, 27, 28), with average daytime concentrations of ∼5 × 106 cm3. Assuming an uptake coefficient of unity (29) for H2SO4 on sea-salt particles, the rate of formation of H+ in the particles during the day from the uptake of H2SO4 is 3.2 × 106 M s1, again similar to the rate of generation of OH in the particles from the interface reaction. Thus, the generation of hydroxide ions by the interfacial reaction of OH with halogens can also modulate the rate of acidification of the particles by the uptake of sulfuric acid from the gas phase.

The increased uptake of SO2 and the higher rate of sulfate formation in sea-salt particles will result in more rapid formation of sulfate in the marine boundary layer. Sea-salt particles are much larger than the sulfate aerosol formed from sulfuric acid nucleation. This size difference results in a larger mean diameter for sulfate-containing particles, which are more rapidly removed by deposition. As a result, sea-salt particles will help to remove SO2 from the atmosphere, but the amount of suspended sulfate particles will not increase proportionally (30, 31). This is consistent with the tendency of models of sulfate formation in the atmosphere to overpredict the SO2 concentration substantially and to underpredict sulfate, but by smaller amounts (8, 9, 32, 33). For example, in a comparison of 11 large-scale models, SO2 in the boundary layer was overpredicted by a factor of 2 or more, whereas sulfate predictions were usually within ∼20% of observations (9).

The higher alkalinity of sea-salt particles generated by the interface reaction may also affect sea salt–cloud relationships. Sea-salt particles act as cloud-condensation nuclei (3436), as do acid sulfate particles. As discussed by O'Dowd et al. (34), cloud droplets formed from sea-salt particles will have a higher pH, and hence take up and oxidize SO2 more rapidly than droplets formed from acidic sulfate particles. The results of our experiments suggest that during the day, even those sea-salt particles that already contain substantial amounts of sulfate will be less acidic than previously anticipated, enhancing formation of sulfate in cloud droplets.

A strong correlation between the formation of non–sea-salt sulfate (nss-SO42–)— sulfate in excess of that found in sea salt— and the loss of bromide and, to a lesser extent, chloride from sea-salt particles has been reported (37). These observations were interpreted as support for the proposed mechanism of oxidation of S(IV) to sulfate by HOCl and HOBr (38). However, the formation of OH in the particles considered here and its modulation of oxidation of S(IV) in sea-salt particles would lead to the same observations, that is, depletion of chloride and bromide from the particles and increased formation of sulfate. Higher bromide ion depletions relative to chloride ion loss is anticipated because of the greater propensity of bromide ions to reside at the air-water interface and hence to react with OH radicals (21, 22), as well as because bromide ions are more readily oxidized than chloride.

Keene et al. (39) proposed a surface reaction of ozone Embedded Image Embedded Image(5) to explain a measured loss of chloride from sea-salt particles that was larger than the increase in nitrate and non–sea-salt sulfate, an observation that has been replicated by a number of researchers [e.g. (40, 41)]. They proposed that HCO3 is the “missing anion” that is held in the aqueous phase by the increased alkalinity. However, the contribution of reaction 5 cannot be substantial, because Cl2 is not generated in the dark when deliquesced NaCl or sea-salt particles are exposed to O3 (15, 16); this is consistent with the known slow kinetics of oxidation of chloride by ozone in solution (42). Reaction 5, however, is equivalent to the photolysis of ozone in the presence of water vapor to generate OH, followed by reaction 2, the mechanism we propose here.

Kerminen et al. (43) report a much larger depletion of chloride from sea-salt particles than is consistent with the uptake of H2SO4 from the gas-phase or the aqueous-phase oxidation of SO2 inside sea-salt particles. They propose that the processing of these particles by nonprecipitating clouds will enhance the conversion of SO2 to sulfate and lead to enhanced chloride loss by acid displacement as the cloud droplet subsequently evaporates. An alternative explanation is the mechanism proposed here, in which chlorine is displaced from the interface as Cl2 and the associated generation of OH in the sea-salt particles leads to enhanced S(IV) concentrations and oxidation by O3 in the particles.

Finally, the formation of NaOH may also affect the reactivity of the particles under conditions in which they would normally crystallize into solid particles, for example, when they are carried inland or to higher altitudes (44). The SEM data reported here indicate that NaOH forms a shell around the NaCl. NaOH is very hygroscopic and forms a stable hydrate at 298 K (17, 18). As the salt particles are transported away from sources of the oxides of sulfur and nitrogen and their associated acids, the OH formed in the interface reaction may not be fully neutralized. Under these conditions, the particles will form an alkaline hygroscopic coating as they dry out, similar to that seen in Fig. 1. This coating will have liquidlike properties. The uptake and reaction of a variety of gases on such an alkaline liquidlike surface will clearly be different from those on a dry NaCl surface.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article