External and Internal Cloud Condensation Nuclei ( CCN ) Mixtures : Controlled Laboratory 1 Studies of Varying Mixing States

19


Introduction
It is well accepted that the water content and the point of activation is dependent on more factors than just the 54 supersaturation and dry diameter for CCN active aerosols (Dusek et al., 2006;Petters and Kreidenweis, 2007). The 55 droplet growth and activation of slightly soluble organics can be modified when internally mixed with inorganic 56 salts that readily deliquesce (Cruz et al., 1998;Padró et al., 2002;Svenningsson et al., 2006). Although inorganic 57 salts are well characterized, the quantification of CCN activity is complicated when they are internally mixed with a 58 complex organic. Consequently, simple mixing rules may no longer be appropriate. It has been observed that mixed 59 aerosols can activate at lower supersaturations than their bulk constituents and organic compounds that may not 60 traditionally be considered as water soluble may aid in the formation of a cloud droplet by acting as a surfactant, 61 depressing surface tension, or simply by contributing mass (Cruz et al., 1998;Padró et al., 2007;Svenningsson et al., 62 2006). In addition, when there is a sufficiently large enough fraction of salt, the slightly soluble core can dissolve 63 before activation, thus lowering the required supersaturation (Sullivan et al., 2009). Thus, the mixing state and 64 extent of mixing can substantially influence CCN activity. 65 To help minimize the complexity in characterizing aerosol hygroscopic and CCN activation properties, CCN 66 data analysis has traditionally been simplified by assuming that i) the aerosols share a similar or uniform 67 hygroscopicity over a particle size distribution, ii) the CCN particle size can be described by the electrical mobility 68 diameter, iii) CCN consists of few multiply charged aerosols and iv) all CCN active aerosols readily dissolve at 69 activation. As a result, a singular sigmoidal fit is commonly applied over the entire CCN activation. However, this 70 method of analysis may not be fully representative of the heterogeneous mixing state occasionally present in the 71 aerosol sample. Thus a CCN mixture refers to the diversity of activated aerosols in the particle population (not the 72 property of an individual particle; i.e., Winkler, 1973 pressure, relative humidity) change in a flow tube, it is assumed that the external mixture may transition into an 132 internally mixed aerosol system. A flow tube mixing apparatus was constructed to test this assumption and modify 133 the extent of mixing of multiple components ( Fig. 1 & 2). the Aerosol Flow Tube. A brief description of the flow tube is provided here. The first aerosol stream is introduced 141 into the flow tube by a ¼ inch stainless steel (SS) tube. The second aerosol stream is also introduced by a ¼ inch SS 142 tube, but is expanded to an outer concentric ¾ inch SS tube using a SS Swagelok tee connection. The two aerosol 143 flows are initially mixed together at the exit of ¼ inch tube and aerosol mixes within the ¾ inch SS tube for an 144 additional 12 inches before entering the quartz tube where it continues to mix. In this study, the pressure and 145 temperature of the flow tube is maintained at ambient conditions. The extent of mixing in the flow tube mixer has 146 been modeled by Computational Fluid Dynamics simulation (CFD -Comsol) to test and improve the aerosol mixing 147 capabilities of the flow tube mixer (Fig. A1). The focus of this work is not the mixing apparatus but the CCN 148 behavior that results from changes in the extent of mixing. It is noted that particle losses likely occur within the 149 flow tube system but do not affect the intrinsic aerosol and CCN properties (activated fractions) presented here.  Roberts and Nenes, 2005). Aerosols with a S c lower than the supersaturation in the column activate and form 160 droplets. These droplets are detected and counted using an optical particle counter at the exit of the column.
! ! ! is the surface tension, ! ! is the molecular weight of water, ! the universal gas constant, ! is the temperature at 204 activation, and ! ! is the density of water. Surface tension and density of water were calculated according to 205 temperature dependent parameterizations presented by Seinfeld and Pandis (1998) and Pruppacher and Klett (1997).

9
The surface tension of the solution is assumed to be that of pure water. Traditional Köhler theory is known to work 207 reasonably well for inorganic salts and slightly-soluble and hygroscopic organics like succinic acid. 208 209

Results and Discussion
Equation (3) (Fig. 2b). At particle mobility diameters between ~35 and 45 nm there is an 242 asymptote, η ~ 0.6 ( Fig. 4a., 4b., and 4c). The activation curves were characteristic of AS and SA, and the measured 243 activation diameters agreed well with Köhler Theory and the single parameter (κ) thermodynamic predictions of 244 droplet activation (Fig. 4a). The external mixture was maintained for an hour as indicated by the separate and stable 245 activation diameters derived from multiple sigmoid analysis. One hour after initial injection into the flow tube, the active heating column was turned off. It should be 258 noted that atomized aerosol continued to be dried through the silica gel diffusion dryers, as is commonly done. The 259 relative humidity after the dryer in both cases is small (< 20%) and thus the activation diameters of very hygroscopic 260 AS calibration aerosol are not affected with or without active heating (Fig. A2). However, as soon as active heating 261 was turned off, particles in the mixing flow tube became more mixed (Figure 4). Thus, it is likely that minute 262 amounts of aerosol water promoted internal mixing and shifted aerosol mixing from external to internal in the 263 mixing flow tube system. 264 CCN activation curves for the two compounds remained distinct and separate until internal mixing 265 conditions dominated and the multiple CCN activation curves converged into a single curve (Fig. 4b and 4c). 266 Results suggest aerosol water plays a significant factor in mixing and CCN activation. This is consistent with 267 previous work that indicates that the presence of water led to lowered aerosol viscosity and increased diffusivity ( To help track the change in organic/inorganic fractions during the transition from external to internal, the 272 mixed aerosols were analyzed with a high-resolution time of flight mass spectrometer (HR-Tof-AMS) to provide 273 mass fraction information. The mass size distribution was integrated and normalized for each compound per scan 274 according to the total mass that was measured. The mass size distribution was then converted to number size 275 distribution and the diameters were converted from aerodynamic diameter to electrical mobility diameter. Then for 276 each superstation and fraction, the EMF was calculated between the two respective activation diameters and 277 correlated to the EMF that was determined from SMCA to determine the plateau height (!). we also refer to carbonaceous aerosol as black carbon mixtures (simply, BC mixtures). The CCN activated fraction 309 data from soot was fit to a singular sigmoidal curve (Fig. 7). There are no plateaus in the activation curve and the single sigmoid fit indicates that the aerosol generated is a homogenous internal mixture. Combustion aerosol 311 activated at a mobility diameter of 133 nm at 2.2% supersaturation. The apparent hygroscopicity of combustion 312 aerosol was !=0.001, and is consistent with the order of magnitude and kappa values reported for fresh combustion 313 aerosol from diesel engine sources (Fig. 7)  normalized size distribution data show that there are few BC-like particles at small sizes (<50 nm), the majority of 331 particles in this range are inorganic. Thus for both the AS and NaCl external mixtures, the activation diameters 332 derived from a singular fit were consistent with the expected dp 50 < 50 nm of the respective inorganic salts. At the 333 larger sizes (> inorganic d p50 ), the BC mixture concentration increased and the CCN/CN was depressed. The 334 combustion aerosol alone is not CCN active at this S c or size and the depressions are reflective of the non-335 hygroscopic combustion aerosol fraction in the aerosol sample. Notably, plateaus are dynamic. As the concentration 336 of inorganic salts increase, the increased activated fraction is reflected in the CCN spectra; the plateau heights 337 increase with increasing hygroscopic concentrations. In these particular externally mixed experiments, the initial CCN/CN plateau can be as large as one, subsequently decrease, and then will likely increase to one after the BC 339 critical diameters are reached. BC externally mixed with very hygroscopic material is more CCN active than the 340 soot alone. 341 342 343 Figure 8. Combustion aerosols externally mixed with inorganics a) NaCl externally mixed with concentrations 344 modified from 51% to 85% over the course of 60 min b) AS externally mixed with concentrations from 41% to 86% 345 over a course of 75 min. Cross symbols represent the initial size distribution of the combustion aerosol. 346 347 Succinic acid (SA) was mixed with combustion aerosol to investigate the external to internal mixing and transition 348 of slightly hygroscopic organic with non-hygroscopic insoluble but wettable aerosols. The laboratory system mimics 349 observed increases in SOA mass fractions on combustion aerosols during atmospheric aging. The SA was 350 introduced to the flow tube at various concentrations, followed by the combustion aerosol from the APG, under dry 351 externally mixed conditions and bimodal size distribution peaks were observed (Figure 9). 352

353
The normalized size distributions of the aerosol leaving the flow tube are presented. The initial soot size distribution 354 is similar to those presented in Figure 8. Assuming the first of the two peaks is SA, and the second is a mixture of 355 the combustion aerosol and SA, the initial point of activation agrees with that of succinic acid where SA aerosols all 356 activate. After a mobility diameter, dp > 50 nm, the concentration of combustion aerosol in the mixture increases 357 and the CCN/CN ratio is < 0.2, indicative of a lower SA concentration relative to the non-activated BC 358 concentration. 359 360 To induce internal mixing, active heating was again turned off for the atomized aerosol source. Again, internal 361 mixing was promoted and the multiple activation curves converge into a single sigmoid for the BC and SA system. 362 This is consistent with the AS/SA experiment and previous work that showed a strong influence of insoluble 363 compounds on activation when internally mixed with a more soluble compound. (9) With continued mixing, a shift 364 to larger activation diameters was observed towards the end of the experiment (scan 93) and there was a slight 365 depression in the plateau of the CCN spectra.

367
The data suggests that only a small amount of soluble inorganic and organic material is required to make the soot 368 more active than that observed alone, especially as the aerosol becomes more internally mixed.