Generation of Internally Mixed Insoluble and Soluble Aerosol Particles to Investigate the Impact of Atmospheric Aging and Heterogeneous Processing on the CCN Activity of Mineral Dust Aerosol

Abstract

Heterogeneous reactions of trace gases with mineral dust aerosol not only impact the chemical balance of the atmosphere but also the physicochemical properties of the dust particle and the ability of the particle to act as a cloud condensation nuclei (CCN). Recent field studies have shown that carbonate minerals are preferentially associated with nitrates whereas aluminum silicates (i.e., clay minerals) are preferentially associated with sulfates. To better understand how this association can impact the climate effects of mineral dust particles, we have measured the CCN activity of a number of pure and internal mixtures of aerosols relevant to these recent field studies. The CCN activity of CaCO ₃ -Ca(NO ₃ ) ₂ aerosol, simulating the activity of mineral dust aerosol that has been partially processed by nitrogen oxides in the atmosphere, is significantly enhanced relative to CaCO₃ aerosol of the same diameter. Similar results are obtained for a clay mineral, kaolinite, internally mixed with (NH ₄ ) ₂ SO ₄ . For example, at 0.3% supersaturation, a 200 nm particle containing a soluble nitrate or sulfate component is 2 to 4 times more active than an unreacted particle. The results presented here show that when determining the contribution of mineral dust aerosol to the overall impact of the aerosol indirect effect on radiative forcing, changes in chemical composition due to atmospheric processing cannot be ignored.

INTRODUCTION

As aerosols are transported in the atmosphere, they can undergo heterogeneous reactions that lead to changes in the chemical composition of the particle surface (Bauer et al. 2004; Bian and Zender 2003; de Reus et al. 2005; Liao et al. 2004; Martin et al. 2003; Song and Carmichael 2001; Tang et al. 2004; Underwood et al. 2001). Often, tropospheric aerosols become internal mixtures of both water-soluble and insoluble components as a result of processing or coagulation (Korhonen et al. 2003). For mineral dust aerosol, it has been observed in both field and modeling studies that dust particles are frequently coated with soluble compounds such as sulfates or nitrates. Specifically, nitrate is commonly associated with calcium-rich dust, while sulfate is usually observed with aluminosilicate- and iron-rich dusts (Ooki and Uematsu 2005; Song and Carmichael 2001; Sullivan et al. 2006).

As a result of the heterogeneous chemistry that can occur on the particle surface, the physicochemical properties of the transported aerosol are expected to be different than those of the freshly emitted aerosol. These potential changes will significantly impact the role an aerosol has on climate forcing. Atmospheric aerosols affect the Earth's climate either through the scattering and absorption of solar radiation, direct radiative forcing, or by acting as cloud condensation nuclei (CCN), indirect radiative forcing. Currently, the aerosol indirect effect has the largest amount of uncertainty when estimating the impact of aerosols on climate and as such has become a focus of a number of field and laboratory studies (Abbatt et al. 2005; Dinar et al. 2006; Huff Hartz et al. 2006; Mircea et al. 2005; Petzold et al. 2005; Ramaswamy et al. 2001). There has been much discussion as to what characteristics of an aerosol govern its CCN activity (size, chemical composition, etc.) (Dusek et al. 2006a; McFiggans et al. 2006). While recent reports suggest that in most cases composition plays a secondary role to size in establishing the ability of aerosol particles to act as CCN (Dusek et al. 2006a; Kaufman and Koren 2006), it is important to establish for insoluble particles such as mineral dust how chemical changes of the dust, through heterogeneous reactions, influence CCN activity (Dusek et al. 2006b; Fan et al. 2004; Levin et al. 2005).

The carbonate component of mineral dust (calcite, CaCO₃, and dolomite, CaMg(CO₃)₂) has been shown to be particularly reactive with trace atmospheric gases such as HNO₃ and N₂O₅ (Krueger et al. 2003; Mogili et al. 2006). Single particle analysis measurements have shown that carbonates can be completely converted to nitrates in the atmosphere (Laskin et al. 2005; Matsuki et al. 2005). Consequently, it is expected that in the atmosphere freshly emitted, partially processed, and fully reacted CaCO₃ will be present. Recent laboratory studies have shown that the physicochemical properties of Ca(NO₃)₂ are markedly different than those of CaCO₃ particles (Gibson et al. 2006a; Gibson et al. 2006b). The cloud condensation nuclei activity of pure 100 nm Ca(NO₃)₂ (representative of fully processed mineral dust) is nearly two orders of magnitude greater than that of pure 100 nm CaCO₃ (freshly emitted mineral dust aerosol) at supersaturations less than 0.225% (Gibson et al. 2006a). In the atmosphere it is expected that the carbonate component of mineral dust will initially be nearly 100% carbonate and then, as it undergoes aging in the atmosphere through heterogeneous chemistry, it will be partially processed containing a carbonate core and a soluble nitrate coating. Although the association of sulfates with clay minerals will occur through a different mechanism, including particle aggregation with an already existing ammonium sulfate particle or through SO₂ oxidation on the surface of the clay, the “aged” dust particle will have a similar, albeit different composition, soluble coating. In order to gain a better understanding of the role these processed particles play in climate forcing, quantitative, composition dependent measurements of the aerosol physicochemical properties are required.

Here, we discuss the capability of the University of Iowa's Multi-Analysis Aerosol Reactor System (MAARS) to generate and characterize the physicochemical properties of internally mixed particles representative of mineral dust aerosol that has been transported and aged in the troposphere. To simulate internally mixed, aggregated, or coated aerosols consisting of soluble and insoluble components in a laboratory setting, particles can be generated through the use of several different techniques. These methods include vaporization-condensation (Abbatt et al. 2005; Garland et al. 2005; Han and Martin 2001; Katrib et al. 2004; Liao and Roberts 2006; Perry et al. 2004) or direct atomization from a mixed solution of the aerosol components (Buzorius et al. 2002; Dusek et al. 2006b; Hameri et al. 2002). In the vaporization-condensation method, an aerosol is atomized, dried, and size-selected to produce a monodisperse aerosol stream. The core aerosol is then coated via the condensation of another material onto the aerosol surface. For example, Katrib et al. (2004) have used this method to generate polystyrene latex spheres (PSLs) coated with oleic acid to better understand product formation in ozone reactions with organic aerosols. To coat aerosol particles in this manner typically requires the use of a heated flow reactor and/or furnace to obtain an air flow saturated with vapor of the coating material. Alternatively, internally mixed or coated aerosols can be atomized directly from aqueous mixtures of the aerosol components. This technique is commonly used to generate organic aerosols internally mixed with a soluble salt like (NH₄)₂SO₄ (Hameri et al. 2002). One caveat that must be considered when using this aerosol generation technique is the potential for forming an externally mixed aerosol as particles comprised solely of the coating material that may be generated (Buzorius et al. 2002).

In this study, internally mixed aerosol particles are generated from suspensions of an insoluble compound with an added amount of a soluble material. Monodisperse PSL spheres mixed with (NH₄)₂SO₄ are used to verify the formation of internally mixed particles from the atomizer. By using PSLs of a known diameter as the core of the mixed aerosol particles, the ammonium sulfate mass fraction in each particle can be quantified. Through careful control of the concentration of the aqueous salt in solution, mineral dust aerosol containing only trace amounts of soluble material can also be generated. Additionally, we show that the CCN activity of insoluble mineral dust components is enhanced dramatically when internally mixed with a small amount of an aqueous salt. Specifically, the CCN potential of CaCO₃ and kaolinite with and without, respectively, Ca(NO₃)₂ and (NH₄)₂SO₄ coatings are compared. The impact of a soluble component on the CCN activity of mineral dust aerosol and the resulting implications for indirect climate forcing are discussed.

EXPERIMENTAL METHODS

CCN measurements were carried out using MAARS, which has been discussed in great detail elsewhere (Gibson et al. 2006b). Briefly, aerosols are generated using a commercially available constant output atomizer (TSI, Inc. Model 3076) that has been modified to decrease the solution volume by using a Pyrex test tube (Fisher Scientific, 18 × 150 mm) in place of the one liter container supplied by the manufacturer. The Pyrex test tube, containing ca. 20 ml of a suspension of approximately 2 g of an insoluble powder such as CaCO₃ in water (Optima, Fisher Scientific), is positioned in the bottle so that the liquid feed tube can access any insoluble powder that has settled out on the bottom of the test tube. A drain tube (1/4″ OD Teflon tubing) has been added to the atomizer lid to ensure continuous recirculation of the sample solution.

The output of the atomizer passes through a series of diffusion dryers typically reaching a relative humidity (RH) of ca. 10%. The resulting dried, polydisperse aerosol is then fed into a flow reactor/hydration chamber where it is mixed with a dry air stream. The particles are size-selected to produce a monodisperse aerosol flow with a differential mobility analyzer (DMA-1; TSI, Inc. electrostatic classifier Model 3080), before mixing with the dry air stream. Upon exiting the flow reactor, the aerosol size distribution is measured with a scanning mobility particle sizer (SMPS; TSI, Inc. Model 3936), which consists of a second DMA (DMA-2), and a condensation particle counter (CPC; TSI, Inc. Model 3025A). SMPS size distributions are measured using a 210 s scan time and flow rates of 3.0 (sheath) and 0.3 (sample) lpm. MAARS is equipped to measure the CCN activity of aerosol particles with a continuous flow streamwise thermal-gradient cloud condensation nuclei counter (Droplet Measurement Technologies, Model CCN-2) (Roberts and Nenes 2005). In CCN measurements, the number of particles that act as cloud condensation nuclei (#CCN) relative to the total number of cloud nuclei (#CN) counted by the CPC is determined. The supersaturation generated in the CCN counter is calibrated to a temperature gradient across the diffusion chamber based on the ammonium sulfate activity data presented in Tang and Munkelwitz (1994). As previously discussed in the literature (Pradeep Kumar et al. 2003), due to counting inefficiencies, the maximum fraction of activated particles that is measured is in the range of 0.7 to 0.9 and not unity. In our system, the maximum fraction is typically near 0.9, but has been observed to range from 0.7 to 0.9. Therefore, measured #CCN/#CN ratios have been scaled by the reciprocal of the maximum ratio for each data set so that the vertical scale for these measurements goes from zero to one.

In this work, our goal is to produce internally mixed aerosol particles consisting of an insoluble (CaCO₃, clay, or PSLs) and soluble (nitrate or sulfate) component that can then be used to further study the impact of heterogeneous processing of mineral dust aerosol on climate. The individual mineral dust components and soluble salts under investigation in this work are commercially available and were used as received. Ammonium sulfate (Fisher Scientific, ACS) and calcium nitrate (ACS, 99.0–103.0%) aerosols are generated from an aqueous solution of the salt, ranging in concentration from 0.01–1 weight percent by volume (wt%). CaCO₃ (OMYA) and kaolinite (Al₂Si₂O₅(OH)₄; Source Clays Repository, KGa-1, low-defect) were atomized from a suspension of the insoluble powder in water. To generate internally mixed aerosol particles, the insoluble mineral dust component was atomized from a suspension of the powder in a solution of the soluble material. The fraction of soluble material included in the internally mixed particle is dependent on the concentration of the solution from which the aerosol is generated. Parameters for the lognormal size distributions of pure calcium nitrate and ammonium sulfate generated from solutions ranging in concentration from 0.01–0.1 wt%, equivalent to a molality range of 0.6–8 × 10^{− 3} m, are provided in Table 1.

TABLE 1 Parameters for lognormal size distributions ^a of (NH₄)₂SO₄ and Ca(NO₃)₂ aerosols generated from solutions of varying concentration.

Concentration (wt %)	Mean Diameter, x0 (nm)	Width
(NH4)2SO4
0.01	49	0.58
0.03	55	0.64
0.05	59	0.66
0.06	62	0.67
0.10	66	0.68
Ca(NO3)2
0.01	54	0.63
0.05	62	0.70
0.10	68	0.70

The methodology for generating internally mixed aerosols and quantifying the amount of soluble material in each particle is developed using PSLs of varying diameters with manufacturer's stated sizes of 202 ± 10, 356 ± 14, and 535 ± 10 nm (Polysciences, Inc., Cat. # 07304, 07306, and 07307, respectively), aerosolized from a suspension of the PSLs in (NH₄)₂SO₄. As the PSLs are dried, (NH₄)₂SO₄ crystallizes on the particle surface. The resultant aerosol stream from the atomizer does include a certain amount of pure ammonium sulfate particles in addition to the coated PSLs, but at these ammonium sulfate concentrations there is little overlap between the ammonium sulfate and PSLs size distributions. Visual evidence of (NH₄)₂SO₄ aggregated with PSL spheres was obtained using transmission electron microscopy (TEM; JEOL JEM-1230). TEM samples of 535 nm PSL in 0.1 wt% (NH₄)₂SO₄ were prepared by exposing mesh grids to the aerosol stream from a DMA that has been set to size-select particles from 505–583 nm, outside of the range where the pure ammonium sulfate distribution would fall. Size-selection was done to ensure minimal contribution from pure ammonium sulfate to the TEM samples.

RESULTS AND DISCUSSION

Determination of the Soluble Mass Fraction of Internally Mixed Particles

Figure 1 displays TEM images of 535 nm PSL mixed with 0.1 wt% (NH₄)₂SO₄. The images ((a)–(c)) show that, when aerosolized from a solution containing ammonium sulfate, the surface of each PSL sphere has distinct (NH₄)₂SO₄ crystallites (indicated by arrows in Figure 1). These are unlike the images of pure PSLs ((d)–(e)) which do show signs of some contamination either from water impurities or remaining surfactant from the original PSL suspension. These impurities, however, do not appear crystalline in nature. Additionally, it should be noted that some of the differences in the appearance of the (NH₄)₂SO₄ crystallites from image to image in Figure 1 can be attributed to electron beam damage from the TEM (Huang and Turpin 1996), which occurs within a few minutes upon exposure to the electron beam.

FIG. 1 TEM images of 535 nm PSL internally mixed with (NH₄)₂SO₄ from 0.1 wt% solutions are shown in (a)–(c). Crystallites of (NH₄)₂SO₄, designated by the arrows, are visible as islands on the surface of the PSLs. TEM images of 535 nm PSLs without (NH₄)₂SO₄ coatings are shown in (d) and (e). The PSLs without (NH₄)₂SO₄ are more spherical compared to the particles with (NH₄)₂SO₄ although the uncoated particle surfaces do show some small “bumps” on the surface that may be due to impurities in the water.

Gaussian fits to the measured SMPS size distributions for the 202, 356, and 535 nm PSLs only and mixed with varying concentrations of ammonium sulfate are shown in Figure 2. The Gaussian fit is used to determine the peak diameter of the size distribution. First, it can be seen in Figure 2 that the measured mobility diameters are all within the range of the manufacturer's specifications for the PSLs with the exception for the smallest PSL, which shows the particles are actually approximately 10% larger than the specified diameter, 228 nm compared to 202 nm. Second, it can be clearly seen that as the PSLs are mixed with (NH₄)₂SO₄ solutions, the particle diameter increases. Furthermore, as the (NH₄)₂SO₄ concentration in the solution increases, the peak diameter of the distribution shifts to increasingly larger diameters. In general, the change in the measured peak diameter ranges from less than one to approximately 18 nm, and is dependent on the concentration of the (NH₄)₂SO₄ solution. The smallest changes in peak diameter are observed for the larger PSLs, a result that is not unexpected. For example, if a 202 nm and a 535 nm PSL were coated with the same amount of soluble material, there would be a greater change in diameter for the smaller particle. Also, at larger diameters there is lower resolution in the SMPS, as the spacing between size bins increases, making it more difficult to detect smaller changes in particle size. Additionally, as the concentration of the (NH₄)₂SO₄ solution is increased, broadening of the aerosol size distribution is observed for each of the three PSL core sizes, most noticeably for the smallest PSLs. This may be due to the fact that while the core PSL is a constant diameter, the amount of (NH₄)₂SO₄ that can crystallize on the surface is not. As will be discussed in further detail below, the atomizer produces a distribution of droplets that vary in diameter. This results in the same size core PSL particles being aggregated with different quantities of ammonium sulfate which then fall into different SMPS size bins. Thus, a more polydisperse distribution of internally mixed PSL-(NH₄)₂SO₄ particles is measured as the (NH₄)₂SO₄ solution concentration increases. Furthermore, for a given concentration, broadening of the measured size distribution is smaller as particle diameter increases due to the loss of resolution in the SMPS.

FIG. 2 Size distributions of 202, 356, and 535 nm PSLs shown mixed with various concentrations of aqueous (NH₄)₂SO₄. Each size distribution displayed is a Gaussian fit to the measured SMPS distribution. The peak diameter of each distribution has been normalized to one. The distributions have been offset for clarity.

The mass fraction of the ammonium sulfate aggregated to the PSL spheres can be estimated from the change in the peak diameter of the aerosol size distributions. The particle size and composition used to calculate the contribution of (NH₄)₂SO₄ to the particles is shown pictorially in Figure 3. As the TEMs in Figure 1 show, there is deviation from a spherical shape when the (NH₄)₂SO₄ aggregates on the PSL surface. The mobility diameter measured for the internally mixed PSLs will be of a sphere equivalent in volume to the PSL-(NH₄)₂SO₄ aggregate. This diameter is represented by the dashed line in Figure 3a. Because the total amount of (NH₄)₂SO₄ that crystallizes on the PSL surface will vary from particle to particle, when calculating the mass fraction of (NH₄)₂SO₄ it is assumed that the ammonium sulfate forms a uniform crystalline layer on the PSL surface (Figure 3b). The thickness of this layer is the difference between the measured mobility diameter and the known diameter of the PSL core. The mass of the PSL core (m_PSL) is calculated from the peak diameter that is determined from a Gaussian fit to the size distribution of the pure PSL and its known density of 1.05 g/cm³. Assuming sphericity of the internally mixed PSL-(NH₄)₂SO₄ particles as well as volume additivity, the mass of the ammonium sulfate (m_solute) is equivalent to:

where ρ_solute is the density of (NH₄)₂SO₄ (1.769 g/cm³) and D_p and D_PSL are the diameters of the dry, internally mixed particle and the PSL core, respectively. Knowing the mass of (NH₄)₂SO₄ in the internally mixed aerosol particle allows the mass fraction of ammonium sulfate (f_solute) to be determined as:

FIG. 3 A representative schematic of the particle shape and diameter used to determine the mass fraction of (NH₄)₂SO₄ (f_solute) associated with each PSL sphere. A representation of the internally mixed particles based on the TEM images from Figure 1 is shown in (a), where solid crystallites of (NH₄)₂SO₄ are formed on the PSL surface. When the particle is sized by the SMPS, the measured diameter is that of a sphere equivalent in volume to the internally mixed PSL particle. This diameter is represented by the dashed black line. In (b), the particle is assumed to have a uniform, crystalline layer of (NH₄)₂SO₄ coating the PSL surface, where the thickness is the difference between the measured particle diameter, DP, and known PSL diameter, DPSL.

The value of f_solute determined, reported as percent of the total particle mass, will vary as a function of the initial solution concentration. The ammonium sulfate mass fraction calculated using the density ranges from 1–30%, 5–21%, and 5–10% for the 202, 356, and 535 nm PSLs, respectively, from the lowest (0.01 wt%) to the highest (0.1 wt%) ammonium sulfate solution concentrations. (NH₄)₂SO₄ does not crystallize in a uniform coating on the PSL surface under the vacuum conditions of the TEM. The assumed formation of a uniform crystalline layer on the PSL surface overestimates f_solute and therefore, the mass determined using the density of (NH₄)₂SO₄ represents an upper limit to the amount of soluble material internally mixed with each PSL sphere.

Alternatively, the concentration of the solution the particles are atomized from in g/cm³ (c_solute) can be used in place of ρ_solute in Equation (1) to determine the mass of soluble material in an internally mixed particle. The use of solution concentration in place of density would be appropriate when the soluble material does not crystallize on the insoluble PSL core but instead remains in the aqueous phase. This is the case with nitrate salts such as Ca(NO₃)₂.

Unlike PSLs, CaCO₃, and kaolinite are not monodisperse when aerosolized and as a result it is more difficult to determine how much soluble material is associated with each particle. CaCO₃, for example, has a somewhat broad size distribution that is lognormal in nature (192 nm mean diameter, width of 0.96). Therefore, when a 200 nm particle is size-selected from the polydisperse mixed aerosol stream of CaCO₃ and Ca(NO₃)₂, the CaCO₃ core will vary in diameter. Additionally, because the distribution of droplets produced from the atomizer is polydisperse as well (0.35 μ m mean diameter, width of 1.9), insoluble CaCO₃ particles will be enclosed in solution droplets of varying volume. Thus, two insoluble CaCO₃ particles that are equivalent in diameter may have different fractions of soluble nitrate associated with them when internally mixed.

The mass of the CaCO₃ core (m_core) and the nitrate component at a given dry particle diameter can be calculated using Equations (3) and (4) (Dusek et al. 2006b):

where D_d is the diameter of a solution droplet containing an insoluble CaCO₃ core, as generated from the atomizer and prior to the diffusion dryer, and ρ _core is the density of the CaCO₃ core. The value of D_d used in Equations (3) and (4) ranges from 200–1700 nm. The lower limit of this size range is the diameter where the particles are size-selected, while the upper limit corresponds to the full width at half maximum of the atomizer droplet distribution. For a size-selected 200 nm dry particle generated from an aqueous Ca(NO₃)₂ solution at a given concentration, each droplet diameter (D_d) corresponds to a specific mass fraction of soluble material. Calculated CaCO₃ core diameters and f_solute are shown in Figures 4a and 4b, respectively, as a function of increasing droplet diameter for different concentrations of Ca(NO₃)₂. As D_d increases, the diameter of the insoluble CaCO₃ component of each 200 nm internally mixed particle can decrease to as small as 182 nm when generated from a 0.1 wt% Ca(NO₃)₂ solution. Because D _p is fixed at 200 nm, the size of the core CaCO₃ particle must be reduced in order to accommodate the increasingly greater volume of solution in each droplet.

FIG. 4 (a) Calculated core CaCO₃ diameter of 200 nm internally mixed CaCO₃-Ca(NO₃)₂ particles generated from 0.01 (filled diamonds), 0.05 (open circles), and 0.1 (filled triangles) wt% Ca(NO₃)₂ solutions, shown as a function of atomizer droplet diameter (D_d). (b) Calculated mass fraction (f_solute) of Ca(NO₃)₂ in 200 nm internally mixed CaCO₃-Ca(NO₃)₂ particles as a function of D_d. (c) Estimate of the frequency of occurrence of internally mixed 200 nm CaCO₃-Ca(NO₃)₂ particles with a given f_solute size-selected from carbonate-nitrate mixtures of varying Ca(NO₃)₂ concentration.

As expected, the fraction of Ca(NO₃)₂ in each aerosol particle at a given droplet diameter increases with solution concentration. The maximum percentage by mass of nitrate aggregated with CaCO₃ is 2% and 23% for 0.01 wt% and 0.1 wt% Ca(NO₃)₂ solutions, respectively. Figure 4c displays the frequency of occurrence of different values of f_solute that are expected based on the distribution of droplets produced from the constant output atomizer. For the CCN results presented in the following section, the soluble mass fraction is determined to range from < 1 to 11% and < 1 to 5% for CaCO₃-Ca(NO₃)₂ generated from 0.01 and 0.05 wt% Ca(NO₃)₂ solutions and kaolinite-(NH₄)₂SO₄ generated from 0.01 and 0.02 wt% ammonium sulfate solutions, respectively. Although the fraction of soluble material in each internally mixed particle has a wide range, according to Figure 4c the most likely value of f_solute is less than one percent of the total particle mass. Equations (3) and (4) can also be used to estimate the fraction of ammonium sulfate from a 0.1 wt% solution aggregated to the PSL spheres. Doing so returns maximum calculated values of the (NH₄)₂SO₄ mass as 30, 8, and 3% of the total particle mass for the 202, 356, and 535 nm PSL particles, respectively. These results are in good agreement with those calculated using Equation (1).

Once the composition of the internally mixed aerosol particles is better understood, the implications of trace amounts of soluble material associated with an insoluble mineral dust particle, in terms of indirect climate forcing, can be explored. The measurements of the CCN activity of internally mixed CaCO₃-Ca(NO₃)₂ and kaolinite-(NH₄)₂SO₄ particles discussed here demonstrate that even minimal quantities of a soluble component (< 1%) can dramatically alter the indirect climate forcing of mineral dust aerosol.

CCN Activity of Single Component and Internally Mixed Aerosols

While it is important to understand the behavior of single components of atmospherically aged mineral dust aerosol, this may not accurately represent typical atmospheric conditions. It is possible that the carbonate or aluminosilicate component of mineral dust aerosol will only be partially processed, thus developing a thin coating or inclusion of soluble nitrate or sulfate. Therefore, we have examined the CCN activity of individual mineral dust components and internal mixtures of these components with soluble inclusions, both as a function of percent supersaturation (% SS) and mobility diameter. Internally mixed aerosols are generated from a mixture of CaCO₃ in an aqueous solution of either 0.01 or 0.05 wt% Ca(NO₃)₂ or a mixture of kaolinite with 0.01 and 0.02 wt% (NH₄)₂SO₄. The use of dilute solutions minimizes the contribution of pure Ca(NO₃)₂ and (NH₄)₂SO₄ aerosol to the measured #CCN/#CN ratio of the internally mixed particles at diameters greater than 150 nm. Below 150 nm, there can be significant contribution of pure Ca(NO₃)₂ or (NH₄)₂SO₄ to the measured CCN activity.

Figure 5 shows the measured CCN activity of several different single aerosol components and internally mixed aerosols size-selected at 200 nm. Specifically, the CCN activity of calcium carbonate is compared to that of calcium nitrate, and the activity of kaolinite is compared to ammonium sulfate. The critical supersaturation (S_c), where 50% of the particles are activated, for CaCO₃ and kaolinite are determined from a sigmoidal fit to the activation data shown in Figures 5a and 5b. The value of S_c for Ca(NO₃)₂ and (NH₄)₂SO₄ was not measurable and lies outside the range of our instrument (< 0.1% SS). The results displayed in Figures 5a and 5b show that soluble Ca(NO₃)₂ and (NH₄)₂SO₄ are much more effective CCN than the insoluble mineral dust components CaCO₃ and kaolinite. This is most apparent in the 0–0.2% SS region. At 0.2% SS, both Ca(NO₃)₂ and (NH₄)₂SO₄ are 100% activated while CaCO₃ and kaolinite are less than 15 and 2% active, respectively. Comparing CaCO₃ to Ca(NO₃)₂ in Figure 5a shows how chemical reactions in the atmosphere will alter the CCN activity of a particle that is completely converted from a carbonate to a nitrate salt.

FIG. 5 CCN activity as a function of % supersaturation for 200 nm (a) pure CaCO₃ (filled circles) and Ca(NO₃)₂ (open circles); (b) pure kaolinite (filled diamonds) and (NH₄)₂SO₄ (open diamonds); (c) CaCO₃ coated with 0.01 wt% (open triangles) and 0.05 wt% (filled triangles) Ca(NO₃)₂; and (d) kaolinite internally mixed with 0.01 wt% (open inverted triangles) and 0.02 wt% (filled inverted triangles) (NH₄)₂SO₄. Each data point is the average of 2–10 measurements and the error bars represent the standard deviation of these multiple measurements. In (a) and (c), the solid black lines represent sigmoidal fits to the activity of pure Ca(NO₃)₂ and CaCO₃, while the dashed black lines are sigmoidal fits to the internally mixed CaCO₃ data. Similarly, the solid black lines in (b) and (d) represent sigmoidal fits to the activity of pure kaolinite and (NH₄)₂SO₄ and the dashed black lines sigmoidal fits to the internally mixed kaolinite data. The Ca(NO₃)₂ and (NH₄)₂SO₄ fits were calculated using a critical supersaturation value determined from Köhler theory. In each plot, the dashed gray line represents the 50% activation point. The data have all been normalized such that the maximum value of #CCN/#CN is one.

Enhanced CCN activity of CaCO₃ and kaolinite is observed for size-selected 200 nm particles when internally mixed with Ca(NO₃)₂ or (NH₄)₂SO₄ (Figure 5c and 5d), which are more representative of partially processed particles. At 200 nm, there is almost no contribution (< 5%) from pure Ca(NO₃)₂ or (NH₄)₂SO₄ to the internally mixed aerosols. The critical supersaturation for CaCO₃ decreases by as much as 46%, from 0.37 ± 0.02% SS for the pure CaCO₃ aerosol to 0.20 ± 0.02% SS when mixed with 0.05 wt% Ca(NO₃)₂. The value of S_c for 200 nm kaolinite is reduced from 0.44 ± 0.02 % SS to 0.21 ± 0.02 % SS when mixed with 0.02 wt% (NH₄)₂SO₄. For kaolinite especially, the reported S_c values should be considered an upper limit, as there may be significant contributions to the measured activity from multiply charged particles larger than 200 nm. The values of S_cfor all the experiments shown in Figure 5 are summarized in Table 2.

TABLE 2 Critical supersaturation values for 200 nm size-selected pure and internally mixed mineral dust components

	Sc (% SS) ^a
CaCO3	0.37 ± 0.02
CaCO3+ 0.01 wt% Ca(NO3)2	0.29 ± 0.02
CaCO3+ 0.05 wt% Ca(NO3)2	0.20 ± 0.02
Ca(NO3)2 ^b	< 0.1
Kaolinite	0.44 ± 0.02
Kaolinite + 0.01 wt% (NH4)2SO4	0.31 ± 0.03
Kaolinite + 0.02 wt% (NH4)2SO4	0.21 ± 0.02
(NH4)2SO4 ^b	< 0.1

The influence of soluble material on the CCN activity of mineral dust aerosol is further illustrated in the 0–0.2% SS range, where the efficacy of CaCO₃ as a CCN increases by a factor 4 when internally mixed with Ca(NO₃)₂, determined by comparing the integrated area of this region for each aerosol. In the same range, the presence of (NH₄)₂SO₄ increases the CCN potential of kaolinite by a factor of 6.

Figure 6 displays the CCN activity at 0.3% SS of Ca(NO₃)₂ in comparison to insoluble CaCO₃ as a function of mobility diameter. The critical diameter (D_c), where 50% of the particles are activated, is determined from a sigmoidal fit to the data for each aerosol. It can be seen that for a given supersaturation, Ca(NO₃)₂ activates at a critical diameter that is ca. 4 times smaller than the insoluble mineral dust aerosol. Although not shown, a similar difference is observed when the CCN activity of another carbonate, dolomite (CaMg(CO₃)₂), and a 1:1 molar mixture of Ca(NO₃)₂ and Mg(NO₃)₂ are compared. A comparison of the CCN activity of kaolinite and (NH₄)₂SO₄ (not shown) yields results analogous to those obtained for CaCO₃ and Ca(NO₃)₂.

FIG. 6 CCN activity at 0.3% SS for (a) Ca(NO₃)₂ (open circles) and CaCO₃ (filled circles) and (b) CaCO₃ internally mixed with 0.01 wt% (open triangles) and 0.05 wt% (filled triangles) Ca(NO₃)₂. The data are plotted as the fraction of particles that can nucleate a cloud (#CCN/#CN) as a function of mobility diameter. These particles represent partially processed mineral dust aerosol. Each data point represents the average of 2–14 measurements, and the error bars are equivalent to the standard deviation of these multiple measurements. The data have been normalized by multiplying each point by the reciprocal of the maximum activation. The dashed gray line indicates the critical diameter where 50% of the particles can activate a cloud. The solid black lines represent sigmoidal fits to the Ca(NO₃)₂ and CaCO₃ data. For CaCO₃, data points that are affected by multiple charging of the aerosol particles are not included in the sigmoidal fit. In (a), the points that have been excluded from the fit are shown in gray. The dashed black lines are sigmoidal fits to the internally mixed CaCO₃-Ca(NO₃)₂ data.

It should be noted that CaCO₃ and kaolinite, like most mineral dust components, are non-spherical in nature, unlike the fully processed nitrate or ammonium sulfate aerosol. The actual value of the critical diameter is expected to decrease when particle shape effects are taken into consideration, based on the relationship between mobility and volume equivalent diameter (Hinds 1999; Khlystov et al. 2004; Zelenyuk et al. 2006; Zelenyuk and Imre 2007). As previously mentioned, the presence of multiply-charged particles may also lead to some difficulty in determining D_c. Contributions from multiply-charged particles are more prevalent when the critical diameter and the peak diameter of the particle size distribution are large, i.e., greater than 100 nm. The measured activated fraction will be artificially high due to multiply-charged particles that are greater in diameter than the singly-charged particles that have been size-selected (Petters et al. 2006). This is especially true below 50% activation, and helps to explain the multiple peaks observed in the CCN activation curve for CaCO₃ shown in Figure 6a. Consequently, the sigmoidal fit used to determine D_c is only applied to a portion of the data for CaCO₃ (above 250 nm) to help account for these contributions. Additionally, particle shape and chemical heterogeneity may also lead to broadening of the measured CCN activation curves for CaCO₃ and kaolinite (Petters et al. 2006).

Figure 6b displays the CCN activity at 0.3% SS of CaCO₃ aerosol particles internally mixed with 0.01 and 0.05 wt% Ca(NO₃)₂ as a function of mobility diameter. As was observed in the 200 nm data, these results show that the presence of Ca(NO₃)₂ enhances the ability of CaCO₃ to act as a cloud condensation nuclei. Furthermore, as the concentration of Ca(NO₃)₂ internally mixed with the CaCO₃ aerosol increases, the CCN activity of the CaCO₃ increases as well. The value of D_c decreases to smaller diameters by as much as 42% with as dilute as 0.05 wt% Ca(NO₃)₂ relative to pure CaCO₃. The critical diameter for kaolinite at 0.3% SS (not shown) also decreases significantly when mixed with 0.01 wt% ammonium sulfate, and becomes similar to that of pure (NH₄)₂SO₄ aerosol when the sulfate solution concentration is increased to 0.02 wt%. The results obtained in these studies suggest that understanding the amount of chemical processing mineral dust aerosol undergoes as it is transported in the atmosphere will be critical in determining its CCN activity, as it is significantly altered by even minute amounts of soluble components.

Predicted CCN Activity of Externally Mixed CaCO₃ and Ca(NO₃)₂ Versus Measured CCN Activity

To further verify that the aerosol particles are indeed an internal mixture of mineral dust and soluble material, the measured CCN activity can be compared to the CCN activity that is predicted based on an external mixture. The CCN activity of CaCO₃ externally mixed with Ca(NO₃)₂ can be predicted from the CCN activity and size distributions of the pure components. The individual size distributions of CaCO₃ and Ca(NO₃)₂ can be summed to determine the size distribution of an externally mixed aerosol (not shown). The activity at a specific diameter can be calculated through the use of the following equation:

where f is the fraction of the size distribution at a given diameter that is either CaCO₃ or Ca(NO₃)₂ and A is the CCN activity (#CCN/#CN) of the individual component at that diameter. Using a mixture of 0.05 wt% Ca(NO₃)₂ and CaCO₃ as an example, at 200 nm 3% of the aerosol particles would be Ca(NO₃)₂ droplets, while 97% would be CaCO₃. The activity of Ca(NO₃)₂ and CaCO₃ at this diameter and 0.3% SS are 1 and 0.39, respectively, determined from the fixed % SS results shown in Figure 6a. Using Equation (5) a total activity of 0.41 would be expected for the external mixture. The predicted activity can be compared to the measured #CCN/#CN ratio shown in Figure 7 for CaCO₃-Ca(NO₃)₂ particles atomized from 0.01 and 0.05 wt% Ca(NO₃)₂ solutions in the 150–400 nm range. If the aerosol generated from the CaCO₃-Ca(NO₃)₂ slurry was an external mixture, the data points would display a linear trend with a slope of one. The actual measured ratio of #CCN to #CN is greater than would be predicted assuming an external mixture by a factor of nearly 2 for a 200 nm particle generated from a 0.05 wt% Ca(NO₃)₂ solution. The predicted ratio trends toward an activation ratio of one at larger diameters where pure CaCO₃ and kaolinite particles would both be completely activated. Thus, these data provide additional evidence that the generated aerosol particles are an internal mixture of Ca(NO₃)₂ and CaCO₃ as all the data points are above the 1:1 line predicted for an external mixture. Similar results are obtained for kaolinite-(NH₄)₂SO₄ mixtures.

FIG. 7 Measured #CCN/#CN versus predicted #CCN/#CN for CaCO₃ internally mixed with 0.01 wt% (open triangles) and 0.05 wt% (filled triangles) Ca(NO₃)₂ for diameters ranging from 150–400 nm at 0.3% SS. Moving across the x-axis the diameter of each data point increases by 25 nm. The predicted activities assume an external mixture and are based on the measured SMPS size distributions and CCN activity (Figure 6a) of the individual components. The 200 nm point discussed in the text is indicated in the plot.

CONCLUSIONS AND ATMOSPHERIC IMPLICATIONS

Internally mixed aerosols from aqueous mixtures of insoluble and soluble components have been generated, characterized, and used in CCN activity measurements. The amount of soluble material associated with each PSL, CaCO₃, or kaolinite particle can be calculated and is shown to increase as a function of soluble solution concentration. Measurements of the CCN activity of internally mixed CaCO₃-Ca(NO₃)₂ and kaolinite-(NH₄)₂SO₄ show that even trace quantities (calculated as typically less than 1% of the total particle mass for a mixed 200 nm particle) of a soluble component can alter the CCN potential of an insoluble aerosol, in good agreement with previous studies on internally mixed soot particles (Dusek et al. 2006b) and S_c values predicted from Köhler theory for aqueous droplets containing different sized insoluble cores. These theoretical values of S_c calculated for Ca(NO₃)₂ and (NH₄)₂SO₄ droplets containing insoluble cores that comprise various percentages of the total particle volume are shown as a function of diameter in Figure 8.

FIG. 8 Theoretical S_c, shown as a function of dry particle diameter, for Ca(NO₃)₂ (solid lines) and (NH₄)₂SO₄ (dashed lines) droplets containing varying size insoluble cores. S_c has been calculated according to Köhler theory and assumes complete dissociation of Ca(NO₃)₂ and (NH₄)₂SO₄. The series of lines shown are for a droplet containing no core and cores that represent 94.1, 97.0, and 99.7% of the total particle volume (corresponding to 98, 99, and 99.9% of the total particle diameter, respectively).

The data obtained in this work provide important insight into the influence of atmospheric chemistry on the climate impact of mineral dust aerosol through enhanced cloud nucleation activity. The ability of mineral dust aerosol to act as cloud condensation nuclei is strongly dependent on the chemical composition of the aerosol. In general, as mineral dust is transported and processed in the atmosphere it can develop a soluble coating typically composed of sulfates or nitrates, or organic species, consequently changing the physicochemical properties of the aerosol. Because the presence of soluble species on the dust surface increases the particle CCN activity, there will be an enhancement in the concentration of cloud droplets formed from mineral dust aerosol, consequently decreasing precipitation efficiency and changing the radiative properties of the cloud (Levin et al. 2005; Rosenfeld et al. 2001; Yin et al. 2002). This work shows that CaCO₃ that has undergone chemical processing by nitrogen oxides to form calcium nitrate acts as a much more efficient CCN than freshly emitted CaCO₃. The enhancement of the CCN activity will depend on the extent of chemical processing. Additionally, this enhancement is dependent on the chemical composition of both the insoluble and soluble species, as the behavior of internally mixed CaCO₃-Ca(NO₃)₂ differs from internally mixed kaolinite-(NH₄)₂SO₄. The results presented here are particularly applicable to submicron mineral dust aerosol, which has been shown to comprise 8–31% of submicron aerosol mass (Arimoto et al. (2006)). In this size range, internal mixing of mineral dust with soluble material, especially sulfate, is prevalent (Sullivan et al. 2006). The results shown here indicate that there is a large impact on CCN activity for mineral dust aerosol even when trace amounts of soluble material are associated with the particle. Thus, the importance of the chemical reactions leading to these soluble components must be taken into account when determining mineral dust aerosol indirect radiative forcing.

Acknowledgments

This material is based upon work supported by the National Science Foundation under Grant Nos. CHE0503854 and ATM0613124. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors would like to thank Jean Ross of the Central Microscopy Research Facility for her assistance in obtaining TEM images. VHG would like to thank Professor Kim Prather and Ryan Sullivan for helpful discussions.

Notes

^a Lognormal distribution: y = Aexp [].

^a Reported error based on the calculated sigmoidal fit of the data shown in Figure 5.

^b S_c is outside the range of the CCN counter.