3.1. CCN Activity
 The CCN activation curves for both samples and pure (NH4)2SO4 are shown in Figure 1. For a CCN composed of soluble non-surfactant compounds, Köhler theory suggests that the “critical supersaturation”, sc, scales with dp to the −3/2 power [Köhler, 1936]. However, for aerosol containing surfactants, this scaling will change, tending to be lower at smaller d and approaching −3/2 at larger d. This is because at smaller particle diameters, the high concentration of WSOC depresses σ, which lowers sc more than would be expected from the solute effect alone; at larger d, the WSOC concentration is insufficient to induce this effect [Padró et al., 2007].
 The activation curve of the Gulfstream sample aerosol (blue circles) is almost identical to (NH4)2SO4 (yellow triangles), despite the presence of ∼40 wt% surface-active DOM. It is likely that the surface tension depression from the organic fraction compensates for the decreased soluble mole fraction. The estuarine sample aerosol contains ∼85 wt% organic matter and is much less CCN active than (NH4)2SO4 (red circles in Figure 1). Additionally, the activation curve deviates from the −3/2 exponential power law fit (as indicated by the dotted line) at low dry particle diameters and high sc, where the concentration of organic matter is high enough to significantly affect the droplet surface tension (C > 1000 mg l−1, from Figure 2).
Figure 2. Fractional surface tension depression with respect to pure water. Shown are direct measurements (open symbols) and values inferred from KTA (solid symbols).
Download figure to PowerPoint
3.2. Köhler Theory Analysis
 The organic molar mass and surface tension depression were inferred from the CCN activity measurements using KTA, method b2 [Padró et al., 2007; Asa-Awuku et al., 2008, 2007]. For each sc/dp measurement, the fitted CCN activity (FCA) parameter, ω, is calculated from
KTA entails expressing FCA in terms of its constituents; assuming the aerosol is composed of N components [Padró et al., 2007]:
where Mw, ρw are the molar mass and density of water, respectively, R is the universal gas constant, and Mi, ρi, ɛi, vi are the molar mass, density, volume fraction, and effective van't Hoff factor of component i, respectively. Denoting the organic fraction as component “j” and rearranging Equation 3 to solve explicitly for Mj and σ yields
σ corresponds to the value at activation for a dissolved carbon concentration, of Cact, of [Padró et al., 2007],
where A = and xc,j = 0.29 is the mass fraction of carbon in the DOM, estimated from the Redfield ratio (C:N:P = 106:16:1) [Schulz and Zabel, 2006; Redfield et al., 1963]. In applying equations (4)–(6), we assume an organic density, ρj, of 1400 kg m−3 [Schulz and Zabel, 2006], an effective van't Hoff factor of 1 for the organics [Dinar et al., 2007], and 2.5 for (NH4)2SO4 and Na2SO4 [Padró et al., 2007].
 To compute Mj and σ from KTA, the following procedure is used: First, the average organic molar mass, Mj*, is estimated for each sc/dp pair (equation (4)), initially assuming the surface tension of pure water. Mj is introduced in equations (5) and (6) to estimate σ for the sc/dp data of the sample plus 60% (NH4)2SO4. The updated σ values are used to reevaluate Mj* and this process is iterated until the σ and Mj* values converge. This procedure allows the concurrent inference of Mj and σ (as a function of WSOC concentration) from the CCN activity data alone.
 Mj is estimated to be 4370 ± 26% kg kmol−1 and 4340 ± 6% kg kmol−1 for the Gulfstream and estuarine samples, respectively, (the reported uncertainty is one standard deviation from the mean Mj over all sc/dp pairs). The sensitivities of Mj to each of the independent parameters is computed with the method of Padró et al.  (Table 1). Using this method, the total estimated uncertainties of Mj for the estuarine and Gulfstream samples are 18% and 24%, respectively. As expected, the Gulfstream sample uncertainty exceeds the estuarine sample uncertainty, since the lower organic mass increases the Mj sensitivity to most of the independent parameters [Padró et al., 2007].
Table 1. Molar Mass Uncertainty, Δ Mj, From Parameter Uncertainty, Δ x, for the Estuarine (Gulfstream) Samples
| ||ω m3/2||ρj, kg m−3||vj||xc,j, %|
|Base x Value|| ||1400||1||0.29|
|Δx||1.47(2.16) × 10−15a||400b||0.2c||0.044b|
|ΔMj, %||3.7 (6.9)||10 (17)||0.2 (0.1)||15 (16)|
|Total molar mass uncertainty, %|| || || ||18 (24)|
 Both molar masses are consistent with each other and fall in the mid-range of the high molar-mass fraction that comprises 30-35% of the marine surface DOM [Ogawa and Tanoue, 2003]. The inferred Mj is also consistent with that of a theoretical Redfield-based molecule, (CH2O)106(NH3)16H3PO4 [Schulz and Zabel, 2006]. While most of the DOM is recovered, it is expected that the small amount of DOM not recovered during the ED/RO process would be low-molar-mass species that can permeate the ED membranes easiest; therefore, higher molar-mass DOM in the sample may be enriched in the process and the Mj of the sample (and hence, that inferred using KTA) may be greater than the Mj of in-situ marine DOM.
 The inferred σ is in excellent agreement with the SL fit to direct measurements (Figure 2)]. The agreement is still excellent if other functions for fitting the data are used; the variability in inferred Mj remains within the reported uncertainty (not shown). Since the inferred values are derived independently of the direct measurements, their agreement shows conclusively that diffusion of surface-active molecules to the droplet surface is sufficiently rapid to achieve equilibrium surface tension depression. Taraniuk et al.  and Asa-Awuku and Nenes  showed that humic-like organic species (∼500 kg kmol−1) in growing droplets are in equilibrium; we find this to apply for marine DOM with a tenfold higher molar mass (hence, ∼ times lower diffusivity). The latter finding is consistent with the analysis of Asa-Awuku and Nenes , as a diffusivity of 2 × 10−10 (= , with 6 × 10−10 being the diffusivity of HULIS) [Taraniuk et al., 2007] implies that the organic concentration at the droplet surface is more than 90% of its equilibrium value for the supersaturation range considered here.
 While the inferred and measured surface tensions agree, the surfactants require tenfold higher concentrations to give the same effect as organics isolated from marine aerosol [Cavalli et al., 2004]. This difference reflects the enrichment of marine aerosol in organic surfactants from the process of bubble bursting. Hence, the samples investigated here are representative of natural marine DOM, but should be interpreted as the lower limit of CCN activity of primary marine organic aerosol.
3.3. Droplet Growth Kinetics
 The mean droplet sizes for each sample measured by the DMT-STGC OPC at varying values of sc are shown in Figure 3. All samples studied exhibit growth similar to that of pure (NH4)2SO4 for most supersaturations. For the three intermediate supersaturations (0.6%, 0.8%, and 1.0%), both seawater samples appear to grow to larger droplet sizes than for pure (NH4)2SO4. Since it is not expected that organics would enhance droplet growth, the observed discrepancy may be caused by slight shifts in the laser scattering during the sizing measurement, resulting from the presence of a compressed film. (NH4)2SO4 growth could also be depressed because of water vapor depletion in the column during the calibrations; however, this is unlikely since the total CCN concentrations were relatively low (∼500 cm−3) and constant for all supersaturations. Laser shifts over long time periods are unlikely, as the results are reproducible. The data suggest that the dissolved organics, compared to pure (NH4)2SO4, do not significantly alter the droplet growth kinetics (i.e., the water vapor mass transfer coefficient).