Alkene ozonolysis SOA : inferences of composition and droplet growth kinetics from Köhler theory analysis

Alkene ozonolysis SOA: inferences of composition and droplet growth kinetics from Köhler theory analysis A. Asa-Awuku, A. Nenes, S. Gao, R. C. Flagan, and J. H. Seinfeld School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA, USA School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA, USA Department of Environmental Science and Engineering, California Institute of Technology, Pasadena, CA, USA Department of Chemical Engineering, California Institute of Technology, Pasadena, CA, USA now at: the Department of Atmospheric Science, University of Arizona, Tuscon, AZ, USA Received: 7 June 2007 – Accepted: 14 June 2007 – Published: 26 June 2007 Correspondence to: A. Nenes (nenes@eas.gatech.edu)


Introduction
Aerosols, by acting as cloud condensation nuclei (CCN), have a profound impact on the hydrological cycle and climate.Carbonaceous material (organic carbon, OC) can comprise up to 90% of aerosol mass (Andreae and Crutzen, 1997;Cachier et al., 1995;Yamasoe et al., 2000), 10-70% of which may be water-soluble (WSOC).Studies have shown that WSOC can influence aerosol hygroscopicity and surface tension (Decesari et al., 2003;Saxena and Hildemann, 1996;Shulman et al., 1996) and must be characterized to quantify the impact of these aerosols on cloud droplet formation.WSOC
In this study we report the experimental investigation of the CCN activity of the water soluble fraction of SOA generated in laboratory chamber ozonolysis of alkenes; these measurements are then used to obtain thermodynamic properties (e.g., molar mass and surface tension depression), which are inferred using K öhler Theory Analysis (KTA) (Asa-Awuku et al., 2007;Padr ó et al., 2007).Furthermore, we characterize WSOC SOA droplet growth kinetics, relative to pure (NH 4 ) 2 SO 4 .Finally, we evaluate

EGU
(by comparing inferred properties to direct measurements) the applications of KTA for complex organic aerosol systems.

Filter extraction and chemical composition
Secondary organic aerosol is generated from the dark ozonolysis of three parent alkenes (cycloheptene, 1-methyl cycloheptene and terpinolene) and collected upon Teflon filters.The ozonolysis experiments were performed in the Caltech dual 28 m 3 teflon chambers, a detailed description of which can be found in Keywood et al. (2004).SOA chemical speciation information measured by liquid chromatography/massspectrometry and ion trap mass spectrometry are available for the cycloheptene and 1-methylcycloheptene precursors from Gao et al. (2004a) (Table 1).No chemical speciation data are available for SOA generated from terpinolene.The presented analysis is the first study to characterize the CCN-relevant properties of WSOC from cycloheptene and 1-methyl cycloheptene ozonolysis.Table 1 presents the estimated average (mole fraction weighted) molar mass and carbon to organic carbon mass ratio for the speciated organics.Following the protocols outlined in Sullivan and Weber (2006), the WSOC in the filter samples was extracted in water during a 1.25 h sonication process with heat (water bath temperature ∼60 • C).WSOC concentration was then measured with a Total Organic Carbon (TOC) Turbo Siever analyzer (Sullivan and Weber, 2006) (Lance et al., 2006;Roberts and Nenes, 2005).The total concentration (CN) of sized particles is also counted, so that the ratio of CCN to CN can be determined.The process is repeated for different particle sizes.
For each supersaturation, s, the cut-off diameter, d , (defined as the point at which CCN/CN=0.5)provides a quantitative characterization of the SOA CCN activity (i.e., for a given s, a larger d corresponds to a lower CCN activity).

Addition of inorganic salts
The impact of adding electrolytes to the CCN activity of the WSOC is explored by mixing a pre-calculated amount of (NH 4 ) 2 SO 4 to the dissolved SOA sample (so that the salt mass fraction in the atomized aerosol is known).The mass of organic carbon, m organic , in the extracted sample is determined by multiplying the measured WSOC carbon concentration by an organic carbon-to-carbon ratio of 2, as suggested by the speciated information provided by Gao et al. (2004a) (Table 1).The inorganic mass to be added, m inorganic , to obtain the resulting inorganic mass fraction, α, is then computed where m organic = 2 × [WSOC] V sample , [WSOC] is the the WSOC concentration (mg C L −1 ), and V sample is the sample volume (ml).The 18 Mohms of filtered water used to extract the water during sonication contributes negligible ions to the particulate matter (Table 2).

Measuring and inferring surface tension of the CCN
A CAM 200 pendant drop method goniometer is used to directly measure surface tension.A description of the method and procedure can be found in Asa-Awuku et al. (2007).Since the surface tension depression strongly depends on [WSOC] (Decesari et al., 2003;Henning et al., 2005;Kiss et al., 2005), surface tension, σ, is measured at numerous concentrations.The measurements are then fit to the Szyskowski-Langmuir isotherm (Asa-Awuku et al., 2007;Langmuir, 1917), where σ w is the surface tension of pure water at temperature, T , (obtained by infinitely diluting our sample with deionized ultra-filtered water), and α, β are empirical constants obtained from the fit.Unfortunately, direct measurement of σ of WSOC solutions at concentrations relevant for CCN activation (10 3 ppm and above) requires significant amount of mass (10 3 µg and above) or usage of dilute WSOC sample.If α, β are based on using dilute samples, extrapolation of Eq. ( 2) to higher concentrations is often subject to substantial uncertainty because i ) the uncertainty α and β can translate to large uncertainty in σ and ii) may not be applicable at concentrations close to or above the initial micelle concentration; we propose the following alternate method of inferring σ from CCN measurements.
As one approaches the critical micelle concentration for a solution containing organic surfactants and electrolytes, the surface tension of droplets would tend to vary little with Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion EGU carbon concentration.Adding electrolytes can enhance surfactant partitioning to the surface layer (otherwise known as "salting-out" effect) (Asa-Awuku et al., 2007;Kiss et al., 2005).A ubiquitous bivalent ion, such as SO 2− 4 can be a very effective "saltingout" agent, so that CCN containing surfactant and sulfate may have a constant surface tension (but lower than that of water).Salting-out and its effect on surface tension and CCN activity, has been seen in (NH 4 ) 2 SO 4 -HULIS mixtures (Kiss et al., 2005) and hydrophobic water-soluble organics isolated from freshly collected biomass-burning aerosol (Asa-Awuku et al., 2007;Kiss et al., 2005).
Furthermore, if the salt mass fraction exceeds 50%, the majority of dissolved solute, n s , is usually from the inorganic salt, and one could then infer the droplet surface tension, σ, at the point of activation using a combination of CCN activation experiments and K öhler theory, as follows.For particles composed of soluble and insoluble fractions, the critical supersaturation, s c , is (K öhler, 1936;Seinfeld and Pandis, 1998), where , R is the universal gas constant, T is droplet temperature, n s are the moles of dissolved solute, with an effective Van't Hoff factor ν. M w and ρ w are the molecular weight and density of water, respectively, and σ is the surface tension of the droplet at the point of activation.The assumption that the inorganic salt contributes the dominant solute implies it is the only component that contributes to B (otherwise known as the "Raoult term"), where d is the CCN dry diameter, M i is the molecular weight of the inorganic constituent, ε i is the volume fraction of the inorganic which relates to mass fraction, m, where "i " and "o" subscripts refer to inorganic and organic components, respectively.If the organic is not a strong surfactant, then s c for the dry diameter d and volume fractionε i , should be given by where σ w corresponds to the surface tension of pure water.However, if the organic depresses surface tension to σ (less than σ w ), then the critical supersaturation is given by If s c and d are known from the CCN activity measurements, Eqs. ( 6) and ( 7) can be combined to give σ: where s c * is given by Eq. ( 6); Eq. ( 8) represents the extension of K öhler Theory Analysis to infer surface tension from activation experiments.If the organic contribution to the Raoult term (Eq.4) is not negligible, then it must be accounted for in Eqs.(6-8).Thus a molar volume must be calculated to estimate the organic contribution and the inferred surface tension is determined in conjunction with KTA (Sect.2.5).
KTA has been shown to constrain molecular weight estimates of known inorganic and organic mixtures to within 20% (Padr ó et al., 2007) and has also been applied to complex biomass burning WSOC with an estimated 40% uncertainty (Asa-Awuku et al., 2007).
The measured variables employed in the KTA analysis are summarized in Table 3.In applying KTA, we assume that the effective organic Van't Hoff factor, v organic =1.
Molecular weights are presented assuming an average organic density of 1.4 g cm −3 (Turpin and Lim, 2001).The uncertainty in inferred molar volume can be computed as ∆ , where ∆xis the uncertainty in of each of the measured parameters x,(i.e., any of σ, ω, and υ) and is the sensitivity of molar volume to x, , derived from Eq. ( 9).Table 4 provides a list of Φ x .

Droplet growth kinetics
When exposed to the same s profile, an activated CCN will grow to cloud droplets of similar diameter, D p , provided that the mass transfer coefficient of water vapor to the 8991 Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion

EGU
growing droplet and the critical supersaturation is the same.The DMT CCN counter measures droplet sizes by an optical particle counter and therefore can be used to explore the impact of organics on the droplet growth kinetics.By comparing the droplet sizes of activated SOA particles against (NH 4 ) 2 SO 4 particles at identical s c , we directly assess the impact of organics on CCN growth kinetics.This is done by observing the wet diameter, D p , that corresponds to particles with s c equal to the instrument saturation, s, (i.e., CCN with a dry diameter equal to the cutoff diameter, d ) and subsequently evaluating D p versus s.

CCN activity
The dependence.This suggests that the SOA are soluble hydrophilic relatively low molecular weight compounds that are not strong surfactants.For all three parent hydrocarbons, the original SOA samples activate at diameters larger than that of (NH 4 ) 2 SO 4 ; this is expected as organics are, in general, less CCN active than (NH 4 ) 2 SO 4 .The activation curves are well represented with a power law consistent with a d −3 / 2 dependence; this implies that the water-soluble SOA do not exhibit limited solubility (Padr ó et al., 2007).

Surface tension
Figure 4 shows the direct measurements of surface tension for all SOA samples and the Szyskowski-Langmuir fits to the data (α and β parameters of the fits are given in

EGU
Table 2).None of the samples demonstrate significant surface tension depression at measured concentrations, even when extrapolated to concentrations relevant for CCN activation (100 mg C L −1 and above) (Fig. 4).If surfactants do exist in the SOA, it is likely they are not concentrated enough in the extracted samples to have a notable impact on surface tension; even for strong surfactants extracted from a biomass burning sample (Asa-Awuku et al., 2007), the depression for concentrations up to 100 mg C L −1 is within the measurement uncertainty (Fig. 4).Thus, direct surface tension measurements for dilute samples would not conclusively reveal the presence of surfactants.Acquiring sufficient sample for σ measurement is challenging, so we infer surface tension using the method described in Sect.2.4.For large mass fractions of salt (>90%), the inferred surface tension approaches that of water used to extract the WSOC from the SOA filter samples (∼71 mN m −1 ) (Table 5).However, for the 33% mixture of sulfate with cycloheptene and terpinolene, the inferred σ is ∼60 mN m −1 (Table 5), ∼15% depression from pure water, suggesting that surface active components do exist in the WSOC.The extent of surface tension depression suggests that the surfactants are appreciably strong, which is expected given the amphiphilic nature of the oxidation products; the presence of humic-like polymers (unless if in very small quantities) is unlikely, however given that expected depression is much higher at the point of activation (Asa-Awuku et al., 2007;Kiss et al., 2005;Salma et al., 2006).

Molecular weight estimates and uncertainty
Using the inferred values of surface tension (Table 5) and assuming an aerosol density of 1.4 g mol −1 (Turpin and Lim, 2001), KTA gives effective organic molecular weights of 162±28 , 101±20, 207±54 g mol −1 for terpinolene, 1-methylcylcoheptene, and cycloheptene SOA, respectively (Tables 3 and 6), which are close to (that is within uncertainty of) the estimates from the Gao et al. (2004a) speciation (Table 1).If water surface tension was used to infer the molecular weight of the organics, large deviations from the Gao et al. (2004) speciations would be found (Tables 1 and 3
In terms of molar volume uncertainty, the assumption that ν organic = 1, does not account for the partial dissociation of the organic species.The greatest source of uncertainty in the calculations arise from ν organic (Table 6); ν organic larger than unity suggests larger molar volumes.The potential dissociation of organics (up to 20% as measured in HULIS titration experiments; Dinar et al., 2006), contributes roughly 23% uncertainty to the molar volume estimates.As in previous KTA studies (Asa-Awuku et al., 2007;Padr ó et al., 2007), the contributions of σ and ω variability to the inferred molar volume uncertainty are around 10% each.Uncertainty in molar mass (not molar volume) also arises from the value of density; varying from 1.4 g cm −3 to 1.6 g cm −3 (Turpin and Lim, 2001) increases molar masses by 14% (though relatively small compared to the uncertainty from ν).The total estimated uncertainty in molar mass is approximately 25% for all SOA samples (Table 6).

Droplet growth kinetics
Figure 5 presents the droplet size measurements at the instrument OPC for all supersaturations and samples considered.For all points, the flow rate within the instrument was maintained constant at 0.5 L min −1 and the sheath to aerosol ratio is 10:1; this ensures that all the particles were exposed to similar supersaturation profiles.From Fig. 5 we conclude that the droplet growth kinetic curves for all SOA samples are virtually indistinguishable for all s values examined; compared to (NH 4 ) 2 SO 4, SOA particles grow to very similar sizes.Only in some cases, does the organic CCN appear to grow slightly larger at higher supersaturations; this is attributed to water depletion effects from the Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU high concentrations of (NH 4 ) 2 SO 4 particles within the instrument.Fewer particles of SOA (∼600 cm −3 ) do not deplete water vapor after activation, while whereas the higher concentration of (NH 4 ) 2 SO 4 aerosols (∼1800 cm −3 ) after their activation deplete vapor faster than can be provided by diffusion.Although this does not affect CCN measurements, the supersaturation profile in the instrument changes slightly for the (NH 4 ) 2 SO 4 calibration experiments, and D p attained at the OPC is slightly decreased.Despite this, almost all of the growth kinetics experiments lie within the measurement uncertainty, so we conclude that the growth kinetics (or water vapor mass transfer coefficient) are uniform and equal to that of (NH 4 ) 2 SO 4.

Summary and implications
In this study, we explore the CCN activity, composition, and droplet growth kinetic characteristics of SOA generated from the ozonolysis of biogenic precursors.A novel method is presented to infer surface tension depression from CCN activation experiments, which requires a much smaller aerosol sample than direct surface tension measurements at CCN-relevant concentrations.From the inferred values of surface tension we conclude that surfactants are likely present in the water-soluble fraction of the SOA, but with a smaller effect than expected for HULIS; together with the small average molar mass inferred from KTA (100 to 200 g mol −1 ), this suggests that HULIS are not an important component of the WSOC fraction of the SOA studied here.KTA results are consistent with available composition data when using inferred surface tensions which validate the applicability of the method for complex mixtures.Finally we find that the presence of organic surfactants does not affect droplet growth kinetics; all the SOA samples exhibit growth kinetics similar to that of (NH 4 ) 2 SO 4.
cut-off diameter, d , as a function of supersaturation and (NH 4 ) 2 SO 4 mass fraction are shown for all SOA samples in Figs.1-3.WSOC from the SOA for the three parent alkenes studied (Figs. 1, 2, and 3) indicate that as the mass fraction of (NH 4 ) 2 SO 4 increases, the aerosol smoothly transitions to pure (NH 4 ) 2 SO 4 behavior with roughly a m −1/2 i
can Introduction Table 2 provides a summary of the offline WSOC chemical composition measurements and nominal anion concentrations in the extracted samples; as expected, the ion concentrations are very low and contribute Introduction Awuku et al., 2007;Padr ó et al., 2007)is used to infer average molar volume (molecular weight, M, over density ρ) of the water-soluble organic fraction of the SOA.KTA (method b 1 , Padr ó et al., 2007) employs measurements of dry diameter versus critical supersaturation, s c, which are then fit to the expression,s c =ωd

Table 1 :
Characteristics of parent hydrocarbons and water-soluble fraction of SOA a Information obtained from 1-methylcyclohexene ozonolysis due to its structural similarity.Introduction

Table 3 .
K öhler Theory Analysis Properties and Molar Volume Results.

Table 4 .
Formulae for the Sensitivity of Molar Volume to the dependant parameters σ, ω, and ν o .
Property Sensitivity, Φ x

Table 5 .
σ values inferred at the point of activation.

Table 6 .
Molar Volume Sensitivity Analysis for SOA.