The ability of biogenic secondary organic aerosol (SOA) to contribute to the concentration of cloud condensation nuclei (CCN) in the atmosphere is examined. Aerosol is generated by the ozonolysis reaction of monoterpenes (α-pinene, β-pinene, 3-carene, and limonene) and sesquiterpenes (β-caryophyllene, α-humulene, and α-cedrene) in a 10 m3 temperature-controlled Teflon smog chamber. In some cases, a self-seeding technique is used, which enables high particle concentrations with the desired diameters without compromising particle composition and purity. The monoterpene SOA is excellent CCN material, and it activates similarly (average activation diameter equals 48 ± 8 nm at 1% supersaturation for the species used in this work) to highly water-soluble organic species. Its effective solubility in water was estimated to be in the range of 0.07–0.40 g solute/g H2O. CCN measurements for sesquiterpene SOA (average activation diameter equals 120 ± 20 nm at 1% supersaturation for the species used in this work) show that it is less CCN active than monoterpene SOA. The initial terpene mixing ratio (between 3 and 100 ppb) does not affect the CCN activation for freshly generated SOA.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Aerosols (solid or liquid particulate matter suspended in gas) in the atmosphere play an important role in climate. The impact of atmospheric aerosols on the energy balance of the planet is usually defined as the direct and indirect effects [Intergovernmental Panel on Climate Change, 2001]. Aerosols scatter and absorb solar radiation in the atmosphere (the direct effect) and alter the formation, precipitation efficiency, and the optical properties of clouds (the indirect effect).
 Secondary organic aerosol (SOA) is produced from the gas phase oxidation of volatile organic compounds (VOCs). Some of the oxidation products have lower vapor pressures than the parent VOC, and they partition to the condensed phase. SOA has been identified as a potentially important contributor to the atmospheric particulate mass [Seinfeld and Pankow, 2003]. The global biogenic emission of VOCs [Guenther et al., 1995] exceeds the anthropogenic emission of VOCs [Singh and Zimmerman, 1992]. The biogenic portion of VOCs includes monoterpenes (C10H16) and sesquiterpenes (C15H24).
 In this work, we measure the activation of laboratory-created monoterpene and sesquiterpene SOA. Through the development of a novel particle generation technique, we investigate the CCN ability of atmospherically relevant species. The CCN activity of these multicomponent particles is parameterized by using an effective solubility.
2. Experimental Methods
 The experiments were carried out using the Carnegie Mellon University Air Quality Laboratory smog chamber for SOA generation, which was connected to a Scanning Mobility Particle Sizer (SMPS) for particle sizing and a cloud condensation nuclei counter (CCNC) for measurement of CCN activity (Figure 1).
2.1. Smog Chamber
 The oxidation of the terpenes took place in a smog chamber equipped with a 10 m3 Teflon bag (Welch Fluorocarbon), though for a few initial experiments a 5 m3 Teflon bag was used. This chamber has been described in detail by C. O. Stanier and S. N. Pandis (Measurements of the volatility of aerosols from α-pinene ozonolysis, submitted to Environmental Science and Technology, 2005). Before each experiment the bag was cleaned by flowing filtered, dry air (dried using silica gel) until an aerosol concentration below one particle per cubic centimeter was reached. The bag was cleaned weekly by heating to a temperature of 40°C, irradiation with UV lights, and exposure to >500 ppb ozone for 8–12 hours. During the experiments, a Scanning Mobility Particle Sizer (Model 3936, TSI) was used to monitor the size distribution and particle number in the smog chamber. The temperature of the smog chamber was monitored using five Omega K-type thermocouples distributed throughout the smog chamber, and for all experiments the average temperature was 22° ± 2°C. The relative humidity of the bag was measured using a Vaisala HMP-233 sensor, and it was 5–8% for all experiments.
 The ozone was generated by corona discharge of oxygen gas using an Azcozon ozone generator. Ozone concentrations between 100 and 900 ppb were used with typical concentrations approximately 300 ppb. The ozone concentration in the smog chamber was monitored continuously using a Dasibi 1008 PC ozone monitor during the experiments. In some experiments, 2-butanol, a radical scavenger, was injected into the smog chamber by evaporating the liquid in a gas dispersion tube. The typical mixing ratio of 2-butanol was 6–13 ppm, which is sufficient to scavenge >95% of the OH radical produced during terpene ozonolysis [Chew and Atkinson, 1996].
 The reproducibility of particle formation by terpene ozonolysis in our smog chamber is directly related to the mixing rate of the reactants. Because the alkene can be mixed more rapidly than the ozone, ozone was injected into the smog chamber first. The terpenes (Sigma Aldrich) were injected into the smog chamber using a microliter gas-tight syringe through a septum incorporated into a Swagelok T-piece. A stream of dry, filtered air (10–20 LPM) was used to vaporize and transfer the terpene into the smog chamber. Bypass air was added to the transfer line with an adjustable valve to dilute the terpene, reducing condensation and terpene losses. The injection of the terpene was performed in less than 5 min. In separate experiments, gas chromatography with gas preconcentration and flame ionization detection was used to show that this transfer method is quantitative for α-pinene and limonene (standards for the other terpenes are not available).
 Thirty minutes after terpene injection the external air circulation (air circulating inside the smog chamber but outside the Teflon reactor) was shut off in order to minimize the particle losses to the walls of the bag. Measurements of CCN activity were made between 30 min and 5 hours after terpene oxidation. The aerosol concentrations in the smog chamber ranged between 0.3 and 2 × 104 cm−3 and, after size selection, between 0.1 and 2 × 103 cm−3. The experiments took place in the dark (the UV lights of the smog chamber were not used). The experimental conditions for each terpene are shown in Table 1.
Table 1. Experimental Conditions for Terpene SOA Generation
 From the smog chamber the particles were directed to a differential mobility analyzer (3071A, TSI) for selecting an almost monodisperse aerosol for CCN analysis. The monodisperse population was directed to a system consisting of a condensation particle counter (CPC, Model 3010, TSI) and a cloud condensation nucleus counter (DH Associates) (Figure 1). The CPC samples continuously (every 3 s) at a flow rate of 1 L/min, while the CCNC samples intermittently (for 5 s out of every 30 s cycle) at a flow rate of 2.5 L/min. A 3 L/min critical orifice was added to a vacuum source at the end of the flow line and a bypass was added between the aerosol source from the DMA and the flow system end. Using this configuration, 3 L/min was pulled directly through the bypass from the aerosol source when no air was flowing through the CCNC and 0.5 L/min was pulled through the bypass when the flow in the CCNC was on. Thus the total flow from the chamber through the aerosol path of the classifier was essentially constant at 4 L/min.
 The CCNC measures the concentration of particles that act as CCN. The ratio of CCN measured by the CCNC to the total number of particles measured by the CPC is the fraction activated. The signal-to-noise ratio of the CCNC and CPC are improved by averaging measurements for 7.5 min for each data point, and the uncertainty in the data points are propagated from the standard deviation of these measurements. The experimental activation diameter (D50) was estimated from a sigmoidal curve fit,
where ds is the dry particle diameter and b is a parameter that relates to the width of the monodisperse aerosol. (NH4)2SO4 aerosol was used to check the accuracy of the experimental setup, and the activation diameter of this species was in good agreement with Köhler theory. The experimental activation diameter for (NH4)2SO4 was 29 ± 5 nm, and the theoretical activation diameter is 27 ± 4 nm at 1% supersaturation.
 The generation of secondary organic aerosol of a sufficient concentration and with desired size range for CCN experiments is challenging. A seed, typically an inorganic salt, is often used to control the number and size of aerosol particles. However, Raymond and Pandis  showed that the CCN behavior of particles made by coating an inorganic salt like sodium chloride or ammonium sulfate with an organic compound is quite sensitive to the amount of the inorganic salt. For this reason, a method of self-seeding was developed and used for these experiments to be able to gain some degree of control of the size distribution of the resulting aerosol.
 Self-seeding involves oxidizing the desired terpene in two steps. First, a small amount (between 1 and 5% of the total terpene concentration) of the investigated terpene is injected into the chamber filled with ozone in stoichiometric excess. The terpene is allowed to react for a 30–60 min, creating a few (<100 cm−3) small particles (<100 nm). Then, a second injection of terpene is performed. When the terpene is oxidized by ozone and the reaction products form, an additional nucleation process occurs and a fraction of the condensable material partitions to the already existing particle phase. With this method, we avoid contamination of SOA with chemically different seed aerosol. Also, the resultant particle size distribution is controlled by adjusting the amount of terpene used in the injections and the time between injections. This enables the formation of a convenient size distribution in the chamber for CCNC data collection.
3. Theoretical Calculations
 The activation diameter of a cloud droplet is determined by the thermodynamic balance of the Kelvin and Raoult effects for the vapor pressure of water over an aqueous droplet. The CCN activity of SOA particles will be modeled using these principles. The Köhler equation [Köhler, 1936; Seinfeld and Pandis, 1998],
predicts the behavior of soluble species, such as many inorganic salts, in a supersaturated environment. For a given droplet diameter, Dp, S is the water vapor saturation relative to a flat surface, Mw is the molecular weight of water, σ is the solution surface tension, R is the gas constant, T is the temperature, ds is the dry particle diameter, γ is the activity coefficient of the solute in aqueous solution, Ms is the solute molecular weight, ρw, ρs, and ρsol are the densities of water, the solute, and the solution, respectively, ν is the average number of ions into which the solute dissolves, and ɛ refers to the dissolved mass fraction of solute particle.
 Complications may arise in Köhler theory when the nucleus comprises a mixture of compounds, potentially with quite different solubility and activity coefficients. This is a particular concern for complex organic mixtures such as those considered in this study, especially when the specific identity and properties of the individual compounds are unknown. In this study, we assume that the activity coefficient of the mixture is approximately unity, as the droplets are quite dilute at the point of activation. Consequently, any difference of behavior from ideal salts is represented by the dissolved mass fraction, ɛ. We also assume that the surface tension of the droplet is that of water.
Raymond and Pandis  introduced modifications to the Köhler equation which allow for substances with limited solubility, where ɛ < 1 for a given Dp and ds. For these cases, ɛ is defined by
where the Csat is the bulk solubility in g solute/g H2O. When the ɛ is less than unity, the density of the solution is defined by
 On the basis of the uncertainty analysis by K. E. Huff Hartz et al. (Cloud condensation nuclei activation of limited-solubility organic aerosol, submitted to Atmospheric Environment, 2005, hereinafter referred to as Huff Hartz et al., submitted manuscript, 2005), the expected error in the predicted activation diameter, D*, is 15%. The uncertainty in the measured activation diameter is 17% [Raymond and Pandis, 2002], which is based on the relative standard deviation of activation diameters of repeated CCN measurements of ammonium sulfate aerosol.
4.1. Monoterpene SOA
 The CCN activation of α-pinene ozonolysis products at 0.6% supersaturation is shown in Figure 2. The fraction of CCN particles increases with increasing particle diameter, and the data are fit to a sigmoidal equation. The data shown in Figure 2 consist of measurements from two different α-pinene SOA experiments and the activation diameter is 105 ± 18 nm (0.6% supersaturation). Figure 3 shows a summary of the results obtained for CCN activation behavior of α-pinene, β-pinene, limonene, and 3-carene SOA for 0.3% and 1.0% supersaturation. These results show that monoterpene SOA is quite CCN active.
 The effect of the initial terpene concentration on the activation diameter of the resulting SOA was examined. For limonene SOA at 0.45% supersaturation, where the initial mixing ratio varied from 3 to 50 ppb, the relative standard deviation of the activation diameter is 6% for three experiments (Table 1). In addition, the activation diameters from the α-pinene SOA generated in two separate experiments at 0.6% supersaturation, where the initial mixing ratios were 15 ppb and 100 ppb, differed by less than 6%. We find no dependence of the initial mixing ratio (for a constant initial ozone level) on the activation diameter for the conditions used in this study (terpene concentrations less than 100 ppb). For this reason, the measured activation diameters for monoterpene SOA at the same supersaturation were averaged for the results shown in Figure 3.
 The recent works of Ziemann  and Keywood et al.  have shown that the presence and type of radical scavenger affects the products of alkene oxidation. We may expect a different CCN activation diameter of SOA products with different levels of radical scavenger. We observe a small difference in the CCN activation of β-pinene SOA when 2-butanol scavenger is used. For β-pinene SOA (1% supersaturation), the measured activation diameter in the presence of 2-butanol was 50 ± 9 nm, while in the absence of 2-butanol, the measured activation diameter was 65 ± 11 nm. This difference is significant at the 95% confidence level, but insignificant at the 99% confidence level. A change in the product distribution due to the presence or absence of an OH scavenger may be responsible for the small difference in activation diameters that we observe. However, different ozone levels were used in these experiments, which may also cause different reaction products. Thus it is difficult to conclusively determine the effect of radical scavenger on CCN activity.
 The ozonation aerosol products of α-pinene include pinic acid, norpinic acid, and cis-pinonic acid [Jang and Kamens, 1999; Yu et al., 1999; Glasius et al., 2000; Koch et al., 2000]. These species represent between 10% and 50% (by mole) of the α-pinene SOA. The activation diameter of these three α-pinene ozonation products has been measured [Raymond and Pandis, 2002; Huff Hartz et al., submitted manuscript, 2005] as single-component particles. The experimental activation diameters ranged from 38 ± 6 to 72 ± 12 nm at 1% supersaturation and from 88 ± 18 nm to 114 ± 23 nm at 0.3% supersaturation for pinic acid, norpinic acid, and cis-pinonic acid. The experimental activation diameter for α-pinene SOA is 60 ± 10 nm at 1% supersaturation and 121 ± 21 nm at 0.3% supersaturation. These values agree within the error with the single-component aerosol activation diameter, tending toward the higher-diameter (less active) end of the range.
 The oxidation products of β-pinene have also been characterized and they include pinic and pinonic acid [Yu et al., 1999; Jaoui and Kamens, 2003a]. The activation diameter of single-component pinic acid aerosol is 38 ± 6 nm at 1% supersaturation and 92 ± 18 nm at 0.3% supersaturation [Raymond and Pandis, 2002]. The activation diameter of single-component pinonic acid aerosol ranges between 50 ± 9 and 72 ± 12 nm at 1% supersaturation and is 114 ± 23 nm at 0.3% supersaturation [Raymond and Pandis, 2002; Huff Hartz et al., submitted manuscript, 2005]. The measured activation diameters of β-pinene SOA are 50 ± 9 nm at 1% supersaturation and 165 ± 28 nm at 0.3% supersaturation. The activation diameter of β-pinene SOA is in good agreement with the activation diameters of two of the components of β-pinene SOA at 1% supersaturation, but is higher than the components at 0.3% supersaturation. This could be due to the CCN properties of the remaining species in the SOA.
 Although several products of 3-carene and limonene have been characterized [Yu et al., 1999; Koch et al., 2000; Glasius et al., 2000], their activation diameters have not been measured. For this reason, a comparison to the 3-carene and limonene SOA experimental activation diameter is not possible.
4.2. Sesquiterpene SOA
Table 1 summarizes the experiments completed for the CCN activation of sesquiterpene SOA and Figure 3 shows the average activation diameters of the ozonation products of β-caryophyllene, α-cedrene, and α-humulene. CCN data at only 1% supersaturation were obtained for sesquiterpene SOA. The products of sesquiterpene ozonolysis are less CCN-active than the products of monoterpene ozonolysis, and the CCN activity at lower supersaturations could not be measured because of experimental limitations in the dry particle size selection.
 Although some products of β-caryophyllene, α-cedrene, and α-humulene have been identified [Calogirou et al., 1997; Jaoui and Kamens, 2003b; Jaoui et al., 2003, 2004], the activation diameters of the individual species have not been measured. For this reason, a comparison to the β-caryophyllene, α-cedrene, and α-humulene SOA experimental activation diameter is not possible at this stage.
Figure 3 shows the comparison of activation diameters of terpene SOA at 1% supersaturation. The monoterpene ozonolysis products (average activation diameter for four monoterpenes at 1% is 48 ± 8 nm) tend to be 2.5 times more CCN active on a diameter basis and approximately an order of magnitude more CCN active than the sesquiterpene products on a volume and mass basis (average activation diameter for three sesquiterpenes at 1% is 120 ± 27 nm). Both types of terpenes produce carboxylic acids, aldehydes, ketones, and other species with mixed functionalities upon reaction with ozone. However, monoterpenes (C10H16) produce species with smaller carbon numbers than sesquiterpenes (C15H24). For this reason, sesquiterpene SOA is expected to be less water soluble (on a molar basis), and as result, to have a higher activation diameter than monoterpene SOA. Direct relationships between organic solubility and CCN activation have been observed in most studies [Raymond and Pandis, 2002; Huff Hartz et al., submitted manuscript, 2005].
 Köhler theory can be used to predict the activation behavior of aerosol. To apply this theory to a complex mixture (SOA), which contains twenty or more species, average values for the parameters needed to calculate activation behavior were estimated. Table 2 shows the parameters that were used in equations (2)–(4). The air-liquid surface tension (σ) of the droplet was assumed to be essentially the surface tension of water, 0.0725 J/m2 at 295 K [Vargaftik et al., 1983]. Turpin and Lim  recommended 1.2 g/cm3 for the density of organic aerosol (ρs), which is consistent with the densities of aliphatic dicarboxylic (C2-C9) and ketocarboxylic (C2-C3) acids. The concentration of organic in the growing droplet is low, therefore the activity coefficient (γ) was assumed to be equal to one. The acid dissociation constants of dicarboxylic acids are low, Ka = 10−4–10−5 M [Morrison and Boyd, 1992]; therefore at least 99.9% of each acid is in its undissociated, protonated form and the average number of species into which the solute dissolves (ν) is practically one. The average molecular weight of SOA (Ms) is equal to the molecular weight of each species found in the aerosol averaged by the corresponding mass fraction of that species [Calogirou et al., 1997; Jang and Kamens, 1999; Glasius et al., 2000; Koch et al., 2000; Jaoui et al., 2003; Jaoui and Kamens, 2003a, 2003b; Jaoui et al., 2004]. For monoterpene SOA, 175 g/mol is used, and for sesquiterpene SOA, 250 g/mol is used. The effective solubility of SOA (C*sat) is the most uncertain parameter of these and it was selected as part of the best fitting procedure.
Table 2. Parameters for Calculation of Activation Diameter of Terpene SOA at 295 K
Surface tension σ
SOA density ρs
Activity coefficient γ
Number of dissolution ions ν
SOA molecular weight Ms
Sesquiterpene SOA, freshly generated
Effective saturation concentration C*sat
0.11 g/g H2O (range 0.07–0.4 g/g H2O)
Sesquiterpene SOA, freshly generated
0.10 g/g H2O (range 0.055–0.14 g/g H2O)
 The best fit of the experimental activation diameters for monoterpene SOA to the theoretical activation diameters, using the parameters given in Table 2, occurs when the effective solubility of SOA (C*sat) is 0.11 g/g H2O. The fit is not improved by using a different C*sat for each terpene. However, the effective solubility values that cover the range of the measured activation diameters is 0.07 to more than 0.4 g/g H2O. At 0.4 g/g H2O, the monoterpene SOA is completely soluble in the droplet. Therefore the effective solubilities greater than 0.4 g/g H2O give the same calculated activation diameter.
 The use of one average molecular weight for all monoterpene SOA might seem too simplistic. The mass-fraction-weighted average molecular weight of the SOA products depends on the type of monoterpene and the oxidation conditions. However, the calculation of activation diameter is not very sensitive to molecular weight. For example, if the molecular weight for monoterpene SOA is decreased from 175 g/mol to 160 g/mol, the calculated activation diameter only decreases from 98 nm to 94 nm at 1% supersaturation. Even if there is a wide range of molecular weights of the SOA species, only small changes in activation diameter will result.
 The average activation diameter of sesquiterpene SOA at 1% supersaturation is 120 nm. The mass yield averaged molecular weight of sesquiterpene SOA, based on the aerosol product studies of β-caryophyllene, cedrene, and α-humulene [Calogirou et al., 1997; Jaoui and Kamens, 2003b; Jaoui et al., 2003, 2004], is 250 g/mol. Using the parameters in Table 2 and equations (2)–(4), the effective solubility of sesquiterpene SOA is 0.10 g/g H2O, where the range of effective solubilities for the three sesquiterpene SOA species varies from 0.055 to 0.14 g/g H2O. The effective solubility of sesquiterpene SOA is smaller than that of monoterpene SOA by a factor of 1.5 on a per mole basis.
 The range in values for effective solubility of monoterpene SOA is comparable, in terms of the moles of dissolved molecule, to the solubility of ammonium sulfate, 0.76 g/g H2O, and sodium chloride, 0.36 g/g H2O [Raymond and Pandis, 2002]. This effective solubility is higher than the solubility of three of the known products of SOA: pinic acid (0.085 g/g H2O), norpinic acid (0.047 g/g H2O), and pinonic acid (0.0064 g/g H2O). In single-component studies [Raymond and Pandis, 2002; Huff Hartz et al., submitted manuscript, 2004] these species were more CCN active than their respective solubilities would predict, and higher solubilities were used to model their CCN activation. This was explained in part because of the particle generation process for the single-component studies, where aerosol was prepared from atomized solutions. Although the particles were dried by silica gel in these previous studies, it is thought that residual water remains. However, in the smog chamber experiments, the aerosol is generated by ozonation, and the relative humidity of the chamber is less than 8%. There is no opportunity for the particles to go through the same process as atomization, where the relative humidity of the particle decreases from saturation to 4% during the drying process.
 One possible explanation for these relatively high effective solubilities for monoterpene SOA is that there are species other than carboxylic acids that contribute to the effective solubility of the particle. Aldehydes and hydroxyl aldehydes species have been identified in α-pinene and β-pinene SOA [Jang and Kamens, 1999; Yu et al., 1999; Glasius et al., 2000; Koch et al., 2000; Jaoui and Kamens, 2003a]. These species have the potential to be more soluble than their carboxylic acid counterparts, which would increase the overall solubility of SOA. Another potential explanation is the reduction of the surface tension of the droplets by some of the unknown compounds.
 A comparison of experimental activation diameters to the calculated activation diameters is shown in Figure 4. The activation diameters of limonene and 3-carene at 0.3% supersaturation are significantly higher than predicted by the parameters used for the remaining data. This may indicate that the product distribution is not uniform over all particle sizes, where larger particles contain less soluble, and as a result, less CCN active species.
 Aging of the aerosol by heterogeneous oxidation reactions could affect the measured activation diameters. For this reason, only aerosol that was freshly generated is used in this study, and extended ozone/aerosol contact times were avoided. This presents a potential complication with the self-seeding method is aging of the initial seed. However, as the seed particles comprise only 1–5% of the ultimate SOA mass and are presumably identical in both initial and final composition, to first-order “pre-aging” of a small fraction of the SOA should only cause a small change in the effective exposure time of the bulk aerosol.
 The CCN behavior of the ozonolysis products of biogenic secondary organic aerosol in the absence of inorganic seed aerosol has been measured. The activation diameter of the complex mixture of products of α-pinene/ozone and β-pinene reaction generated in this study is similar to activation diameters of the three of the individual components (pinic acid, norpinic acid, and cis-pinonic acid) known to exist in this type of aerosol. We find that that terpene mixing ratio has no effect on CCN activity. The presence or absence of a radical scavenger may affect the CCN activity of SOA, though the relationship is uncertain. The composition of secondary organic aerosol from terpene/ozone reaction could also vary with oxidation temperature, ozone and radical scavenger mixing ratio, the relative humidity, and light. The effect of these factors on CCN activity is not known and is the subject for future research.
 The activation diameters of the products of the monoterpene/ozone reaction have been modeled using Köhler theory, where the effective solubility of SOA is similar to the solubility of inorganic species like ammonium sulfate and sodium chloride. Low CCN activation diameters indicate that monoterpene SOA would be a good source of CCN in the atmosphere. Though monoterpene SOA shows higher activation diameters than inorganic salts, this is mainly due the higher average molecular weight of SOA, not solubility differences.
 The products of sesquiterpene ozonolysis are less CCN active then the products of monoterpene ozonolysis when the aerosol is freshly generated. To model the activation diameter of monoterpenes and sesquiterpene aerosol a solubility value of 0.1 g/g H2O can be used as a first approximation. Further research, particularly with corresponding aerosol speciation for all terpenes, is needed to predict the CCN activity of SOA quantitatively.
 The authors would like to thank the National Science Foundation (ATM-336296) for funding and Joshua Tischuk for experimental assistance. K.E.H.H. acknowledges the Dreyfus Foundation for funding. M.B. acknowledges the Danish Natural Science Research Council for funding.