Measured and modeled equilibrium sizes of NaCl and (NH 4 ) 2 SO 4 particles at relative humidities up to 99.1%

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 110,, doi: /2004jd005507, 2005 Measured and modeled equilibrium sizes of NaCl and (NH 4 ) 2 SO 4 particles at relative humidities up to 99.1% H. Wex, A. Kiselev, F. Stratmann, and J. Zoboki Institute for Tropospheric Research, Leipzig, Germany F. Brechtel Brechtel Manufacturing, Inc., Hayward, California, USA Received 12 October 2004; revised 19 July 2005; accepted 12 August 2005; published 12 November [1] The Leipzig Aerosol Cloud Interaction Simulator (LACIS) was used to measure equilibrium sizes of particles at relative humidities (RHs) up to 99.1%. Particles of substances with well-known hygroscopic behavior (NaCl and (NH 4 ) 2 SO 4 ) were used for the measurements. The intention was to check the proper functionality of LACIS for this measurement mode. Equilibrium sizes were also simulated on the basis of Köhler theory, by (1) assuming ideal solutions and (2) accounting for the nonideality of the droplets by using osmotic coefficients from the open literature. Measured and simulated equilibrium diameters were compared to measured values published in the open literature. For sodium chloride, model approaches 1 and 2 and all measured values were in good agreement. For ammonium sulfate, the results of model approach 2 (nonideal solution) were in good agreement with the measured diameters, whereas assuming an ideal solution yields equilibrium diameters that are too large by up to 20%. Measured hygroscopic growth factors were used to derive scattering coefficients and visibilities for two exemplary atmospheric dry number size distributions at different RHs. For both salts, at 90% RH the scattering coefficient increased by about a factor of 5 compared to that of the dry aerosol; at 99% RH the increase was about 21-fold. If instead the hygroscopic growth was assumed to follow the growth factors that were simulated for an ideal ammonium sulfate solution, the increase of the scattering coefficient is overestimated by 40%. This highlights the need to account for the nonideal solution behavior of droplets when calculating equilibrium diameters, even at high RHs. Citation: Wex, H., A. Kiselev, F. Stratmann, J. Zoboki, and F. Brechtel (2005), Measured and modeled equilibrium sizes of NaCl and (NH 4 ) 2 SO 4 particles at relative humidities up to 99.1%, J. Geophys. Res., 110,, doi: /2004jd Introduction [2] Atmospheric aerosol particles play an important role in the Earth s climate. By scattering and absorbing incoming solar radiation, they influence the overall radiation budget (direct effect). They also act as nuclei for cloud droplets, and determine their concentrations and sizes, and therewith cloud albedo, lifetime, and precipitation processes (indirect effect). Through these effects, anthropogenic changes of the atmospheric aerosol have an influence on the Earth climate. [3] To quantify the influence of aerosol particles on the Earth climate, several properties of the atmospheric aerosol have been examined in the past. It is known that large portions of the atmospheric aerosol consist of inorganic salts such as ammonium sulfate, ammonium nitrate, and sodium chloride [Quinn et al., 1998; Neusüß et al., 2000; Putaud et al., 2000; Quinn et al., 2000; Neusüß et al., 2002]. These Copyright 2005 by the American Geophysical Union /05/2004JD salts are hygroscopic. Their effect on climate depends on atmospheric conditions, in particular the ambient relative humidity (RH). This is because the amount of scattered light is related to the size of the scattering particles, and this changes with RH, depending on the hygroscopic properties of the particles. Model calculations that include the influence of RH on scattering [e.g., Sloane, 1986; Chuang et al., 1997; Haywood et al., 1997] require data on the hygroscopic growth of aerosol particles. Examinations of hygroscopic properties of inorganic salts have been done in the past [e.g., Rader and McMurry, 1986; Cohen et al., 1987; Tang and Munkelwitz, 1994; Tang, 1996]. [4] A new hygroscopic twin differential mobility analyzer (HTDMA) that can measure hygroscopic growth of aerosol particles up to 98% RH was introduced recently [Hennig et al., 2005]. The new instrument was tested by measuring the growth factors of ammonium sulfate particles. The results were found to agree within measurement uncertainty with growth factors given by Tang and Munkelwitz [1994] and Brechtel and Kreidenweis [2000a]. 1of9

2 [5] However, to date, there are no measurements of hygroscopic growth of atmospheric aerosol particles for RHs above 95%. [6] During the past years, several studies have been done in which the number of atmospheric cloud droplet nuclei (CCN) was derived from measurements of the hygroscopic growth of atmospheric aerosol particles [Covert et al., 1998; Brechtel and Kreidenweis, 2000b; Zhou et al., 2001; Dusek et al., 2003; Rissler et al., 2004]. In these studies, the number of ions that were in solution for the aerosol particles at 90% RH were determined from growth factors measured with a hygroscopic twin differential mobility analyzer (HTDMA). This number of ions then was used to determine the number of CCN for different supersaturations, together with Köhler theory. The derived number of CCN usually overestimated the number of CCN measured with CCN counters. This discrepancy is thought to be due to the fact that the HTDMA measurements were performed at the relatively low RH of 90%, where the number of ions in solution might differ from that at larger RHs because of the presence of organic compounds in the aerosol. Thus it is desirable to perform measurements of hygroscopic growth factors at larger RHs. Also, in these studies, the number of CCN were derived under the assumption that droplet solutions behaved like ideal solutions at the critical supersaturation. The influence of the nonideal behavior of the solutions also has to be explored through measurements at saturations as close as possible to the critical supersaturations. [7] In this study we explore, for the first time, the suitability of the Leipzig Aerosol Cloud Interaction Simulator (LACIS) [Stratmann et al., 2004] to measure hygroscopic growth factors at high relative humidities up to 99.1%. LACIS is a laminar flow tube that can operate at a stable RH, which can be adjusted from almost 0% up to 99.1%. Alternatively, it can be used at supersaturation with maximum values up to several percent. For this study, the equilibrium diameters and growth factors of sodium chloride and ammonium sulfate particles were measured in a range from 85.8% RH up to 99.1% RH. Measured growth factors were compared to values derived from different model calculations and also to experimental data from the literature [Tang and Munkelwitz, 1994; Tang, 1996]. [8] To show the influence of the hygroscopic growth factor on calculated scattering coefficients, scattering coefficients of two different model aerosols were calculated for different relative humidities. Dry number size distributions were taken from the literature [O Dowd et al., 1999; Tuch et al., 2003]. The growth factors obtained in this work were used to determine the number size distributions at different RHs from 90% to 99%, for which then scattering coefficients were calculated. The increase of the scattering coefficient at the different RHs compared to the value for the dry aerosol was derived. It is shown for ammonium sulfate that the use of humidity growth factors calculated under the assumption of an ideal droplet solution can result in a significant overestimation of the increase of the scattering coefficient. [9] Hygroscopic growth factors increase with increasing relative humidity, and situations with relative humidities above 90% occur frequently in the atmosphere. Therefore the presented influence of the hygroscopic growth on modeled scattering coefficients stresses the importance of the use of proper hygroscopic growth factors for the derivation of the aerosol direct effect, and it also stresses the importance of measurements of growth factors of atmospheric aerosol at high relative humidities, well above 90%. 2. Experimental Setup [10] Figure 1 shows the experimental setup. Particles were generated by atomizing an aqueous solution of 1 g salt per liter double deionized water (atomizer: TSI 3075, TSI Inc., St. Paul, Minnesota, USA). The resulting aerosol particles were dried in a diffusion dryer. A differential mobility analyzer (DMA [Knutson and Whitby, 1975], type Vienna medium ) was used to select a dry particle size. The mobility diameter of the selected dry particles was 200 nm for both salts unless otherwise noted. (NH 4 ) 2 SO 4 was assumed to form spherical particles, whereas the cubic shape of NaCl particles had to be accounted for by using a shape factor of 1.08 [Kelly and McMurry, 1992]. Therewith the mass equivalent diameter of the respective selected NaCl particles was 185 nm. [11] Number concentrations of the quasi monodisperse aerosol after the DMA were determined with a CPC (TSI 3010, TSI Inc., St. Paul, Minnesota, USA), and were kept at cm 1 with a dilution system up stream of the DMA. All flows were controlled with mass flow controllers (MKS 1179, MKS Instruments Deutschland GmbH, Munich, Germany) and were checked daily with a bubble flowmeter. [12] Before entering LACIS, the aerosol passed through a saturator (Perma Pure MH S-4, Perma Pure, Toms River, New Jersey, USA). The saturator consisted of a tube made from nafion, which was surrounded by temperaturecontrolled water. The temperature of the water jacket was kept at a defined value by circulating the water through a thermostat (HAAKE C25P, HAAKE GmbH, Karlsruhe, Germany). The temperature of the thermostat was regulated by the signal of a Pt-100 resistance thermometer that measured the temperature of the water at the aerosol outlet of the saturator. Therewith, the saturator was used in a counter flow principle and it was assured that the temperature at the saturator outlet was at the desired value. Downstream of the saturator, the dew point temperature of the aerosol was equal to the temperature at the saturator outlet. This was verified with a dew point mirror (Dew Prime I-S2, Edge Tech, Milford, Massachusetts, USA), which measures with an accuracy of 0.1 K. A similar setup was used to humidify particle free sheath air. Humidified aerosol and sheath air, both with the same dew point temperatures, were combined in the LACIS head. The aerosol was confined by the sheath air in a narrow beam (2 mm in diameter) at the center axis of LACIS. [13] LACIS itself is a laminar flow tube with a diameter of 15 mm and a length of 1 m, surrounded by a water jacket with thermostat (thermostat: HAAKE C40P, HAAKE GmbH, Karlsruhe, Germany). The difference between the dew point temperature of the aerosol and sheath airflow and the temperature of the LACIS water jacket determines the RH inside LACIS. For this study, LACIS was kept at a constant temperature of 22.5 C. The saturator temperatures 2of9

3 Figure 1. Setup of the particle generation and the LACIS flow tube. Solid lines with arrows indicate the flow of aerosol; dashed lines indicate water flow from thermostats to the saturators and the LACIS flow tube. (MFC, mass flow controller; DMA, differential particle analyzer; CPC, condensation particle counter.) were varied from 20 C to C, resulting in a relative humidity from 85.8% to 99.1% inside the LACIS flow tube. Inside the flow tube, the RH reaches a constant value at approximately 20 cm downstream of the head. The residence time of the aerosol inside LACIS is 2 s, which is sufficient for the salt particles to reach equilibrium at subsaturation conditions. [14] At the outlet of the flow tube the size of the grown particles/droplets is measured with an optical particle sizer that was designed and built especially for LACIS. For a detailed description, see Kiselev et al. [2005]. The optical particle sizer is a white light, wide angle scattering instrument. A Xenon arc lamp provides white light. The light beam is shaped and illuminates the particle beam through a small windowless slit in the LACIS flow tube, with the light beam being perpendicular to the particle beam. The light scattered by the particles is collected with two elliptical mirrors for solid angles from 10 to 50 in the forward direction and is passed on to two photomultipliers. The duration and shape of the signals of the two photomultipliers are compared, so that only the droplets that pass through the center of the measuring volume are included in the size measurement. [15] The optical particle sizer was calibrated daily with PSL particles with diameters of 300 nm, 500 nm and 1532 nm to determine its response function. The day to day variability of the PSL signals was less than 7%. For the determination of the size of grown aerosol particles/droplets, the refractive index of the solution was estimated using a volume mixing rule approach accounting for the dissolved solute mass and water [e.g., Seinfeld and Pandis, 1998]. [16] For the measurements of equilibrium diameters of salt particles at different RHs, the size of a grown particle at the LACIS outlet corresponds to a growth factor g, which is defined as the ratio of the size of the grown particle (d droplet ) to the dry mass equivalent diameter of the particle (d dry ): g ¼ d droplet =d dry 3. Modeling the Hygroscopic Growth [17] For modeling equilibrium diameters in this study we used FLUENT 6 [Fluent, Inc., 2003] together with the Fine Particle Model (FPM) [Wilck et al., 2002; Particle Dynamics, Inc., 2003] in its moving monodisperse mode of operation. FLUENT 6 together with the FPM has been used earlier to describe the coupled fluid flow, heat/mass transfer and particle/droplet dynamical processes taking place in LACIS [Stratmann et al., 2004]. With this numerical model, particle/droplet sizes as a function of position in LACIS can be calculated. This is done dynamically, for both supersaturated [Stratmann et al., 2004] and subsaturated conditions (this study) using the growth law according to Barrett and Clement ¼ ð 2pd p S S p Þ M R vt 1 w vmv D vp v;ef Mass þ L2 v Sp R vt 2 k gf Heat p /@t is the mass growth rate of the droplets, d p is the particle/droplet diameter, S and S p are the water vapor saturation ratios in the gas phase and at the droplet surface, respectively. R v is the vapor specific gas constant, T is the droplet temperature, w v is the vapor mass fraction, M and M v are the molecular weights of the vapor-gas mixture and ð1þ ð2þ 3of9

4 Figure 2. Osmotic coefficient for NaCl and (NH 4 ) 2 SO 4 as taken from Pruppacher and Klett [1997] (thin lines) and from Pitzer and Mayorga [1973] (thick lines). the vapor, respectively. D v is the vapor diffusion coefficient (following Hall and Pruppacher [1976]), p v,e is the equilibrium vapor pressure, f Mass and f Heat are the mass and heat transfer transition functions, respectively, and L v is the latent heat of vaporization. k g is the carrier gas heat conductivity, and for the air water vapor mixture considered here k g was determined with the Mason-Saxena equation [Pruppacher and Klett, 1997]. S p is defined as follows: S p ¼ exp 4M ws sol M wnfm ; ð3þ RTr w d p 1000 where M w is the molecular weight of water, s sol is the droplet surface tension, R is the universal gas constant, r w is the density of water, d p is the droplet diameter, n is the total number of ions the soluble substance dissociates into, f is the osmotic coefficient and m is the molality of the solution. In this study, we used the surface tension of water for s sol. The density of the droplets was calculated using the mass fraction weighted bulk densities of the single droplet components. [18] The particles reach equilibrium in the flow tube, p /@t = 0 and S = S p. Consequently, the numerical model computes equilibrium diameters as given in equation (3), i.e., according to Köhler theory. [19] In highly diluted solutions, ion-ion interactions are limited and the solution can be treated as ideal. In many real droplets, however, ions do interact with each other, which causes deviations from ideal behavior. One way to correct for these nonideal interactions is to introduce the van t Hoff factor. Another possibility is the use of the osmotic coefficient f [see, e.g., Pruppacher and Klett, 1997], which is preferable because, as mentioned by Pruppacher and Klett [1997], tabulated data exists to a greater extent and also because f represents the variability of solution nonideal behavior with changing molality. The FPM was used to calculate the equilibrium sizes of NaCl and (NH 4 ) 2 SO 4 particles at the LACIS outlet for different RHs and for two different assumptions concerning ideality: (1) assuming ideal behavior (i.e., f = 1) and (2) taking values of f from Pruppacher and Klett [1997, p. 112, Table 4.2] and interpolating between the values with a polynomial fit. Figure 2 shows f derived from Pruppacher and Klett [1997] and also from Pitzer and Mayorga [1973] for comparison. Using a van t Hoff factor instead of f would mean to approximate f with a constant value over the whole range of molalities. [20] To verify the model results, the equilibrium diameters were also calculated using Köhler equation (3) with a parameterization of the osmotic coefficient according to Pitzer and Mayorga [1973], as described by Brechtel and Kreidenweis [2000a]. In these calculations, an additional correction for the surface curvature (Kelvin term) was included, although the influence of the curvature is strongly diminishing for particle/droplet sizes above 100 nm. [21] Figure 3 shows the comparison of g obtained from the FPM using f according to Pruppacher and Klett [1997] and from Köhler equation (3) with f according to Pitzer and Mayorga [1973]. For both, NaCl and (NH 4 ) 2 SO 4, values from the two calculations differ by less than 1.5% at the different RHs. This confirms that the FPM properly computes equilibrium particle sizes applying equations (2) and (3). Consequently, g derived with the FPM is used in the comparison with the measured values in the following. [22] Additionally, the equilibrium diameters were derived from parameterizations based on measurements by Tang [1996] for NaCl particles and by Tang and Munkelwitz [1994] for (NH 4 ) 2 SO 4 particles. Tang [1996] and Tang and Munkelwitz [1994] determined the hygroscopic growth of particles with dry sizes in the range of 6 8 mm for RHs < 92% using the single particle levitation technique. From these measurements, a parameterization of the hygroscopic growth was derived. The growth factors that can be obtained from this parameterization do not account for the curvature of the particles, since it has no influence for particles in the size range used by Tang [1996] and Tang and Munkelwitz [1994]. As already stated, also for the particle size we use for this study, the influence of the curvature on the equilibrium diameter is small. Nevertheless, a Kelvin correction was included in the calculations when we derived the equilibrium diameters Figure 3. Comparison of growth factors modeled with the FPM and modeled according to Brechtel and Kreidenweis [2000a]. 4of9

5 [23] Measurements of the equilibrium diameter (or the growth factor g, respectively) made with LACIS show an uncertainty of 3% to 4%, derived from the day to day variability of the measurements, of which a part is ascribed to the repeatability of the RH adjusted in the LACIS flow tube. Figures 4 and 5 show g measured with LACIS at different RHs for NaCl particles with a dry size of 185 nm and for (NH 4 ) 2 SO 4 particles of 200 nm, respectively. Shown are also the results from the two FPM calculations using the different assumptions on ideality, and values of g according to Tang [1996], Cohen et al. [1987], and Tang and Munkelwitz [1994]. In Figure 5, the larger squares indicate LACIS measurements for which the geometry of the optical particle sizer was different from the one described above, using solid scattering angles from 5 to 15 (instead of 10 to 50 ). At 97% to 98.5% RH, where measurements were performed for both geometries, measured values were similar for the two different geometries. Figures 4 and 5 also show the deviation between the LACIS measurements and the different simulations. They are given (in percent) as: DEVIATION ¼ 100 * g calc 1 g LACIS where g LACIS are the growth factors measured with LACIS and g calc are the growth factors calculated with the FPM or with the parameterization from Tang [1996] and Tang and Munkelwitz [1994]. [24] At RHs of 98% and 99%, measurements were also performed for mass equivalent diameters of the dry particles of 139 nm and 278 nm for NaCl and 150 nm and 300 nm for (NH 4 ) 2 SO 4, respectively. The values differ for the two different salts, because the same mobility diameters were selected for both salts, and, as already stated above, a shape factor according to Kelly and McMurry [1992] has to be applied for NaCl. Measurements for each size and each RH were performed on four different days over the course of three weeks. Again, day to day variability of LACIS gave ð4þ Figure 4. (top) Measured and modeled growth factors and equilibrium diameters for NaCl (dry diameter 185 nm) at different RHs, including data derived from parameterizations by Tang [1996] and by Cohen et al. [1987]. (bottom) Deviation between the LACIS measurements and the different calculated values, following equation (4). on the basis of the parameterizations of Tang [1996] and Tang and Munkelwitz [1994]. 4. Comparison of Measurements and Model Figure 5. (top) Measured and modeled growth factors and equilibrium diameters for (NH 4 ) 2 SO 4 (dry diameter 200 nm) at different RHs, including data from Tang and Munkelwitz [1994]. (bottom) Deviation between the LACIS measurements and the different calculated values, following equation (4). 5of9

6 Table 1. Growth Factor g Measured With LACIS and Derived From (NaCl) and Tang and Munkelwitz [1994] ((NH 4 ) 2 SO 4 ) and the Deviation Between Measured and Calculated Values, Following Equation (4) NaCl (NH 4 ) 2 SO nm 185 nm 278 nm 150 nm 200 nm 300 nm g Measured With LACIS RH 98% RH 99% g Derived From the Literature RH 98% RH 99% Deviation a RH 98% RH 99% a Deviations are given in percent. an uncertainty in the measured values of less than 4%. The measured growth factors are shown in Table 1. Also shown are the values of g as derived from Tang [1996] for NaCl and Tang and Munkelwitz [1994] for (NH 4 ) 2 SO 4 for the respective sizes and RHs, and the deviations between the measured and calculated g according to equation (4). [25] Values of g for NaCl derived with the FPM are shown in Figure 4. For NaCl, g derived with the FPM simulations assuming an ideal solution differs less than 3% from the results of the simulation that included nonideal behavior. The deviation between the two model approaches and the measurements from either Tang [1996] or LACIS is equal to or less than 4%. Figure 4 also shows values of g taken from a parameterization of hygroscopic particle growth by Cohen et al. [1987]. Both the Tang [1996] and Cohen et al. [1987] parameterizations were based on measurements of equilibrium diameters for RHs up to 92% and 90%, respectively. Results from our measurements and both parameterizations agree for RHs up to 97%, but for larger RHs [Cohen et al., 1987] underestimate g compared to our measurements and to those of Tang [1996], with an underestimation of 10% at 99% RH. Nevertheless, it is worth noting that the difference of our measurements and the FPM results and of the parameterization by Tang [1996] does not exceed 4%, even though Tang [1996] used very large particles (6 8 mm) at the relatively low maximum RH of 92%. [26] For (NH 4 ) 2 SO 4, the most striking feature in Figure 5 is the large deviation between results from growth calculations with f = 1 (i.e., ideal solution) and the other data. The FPM calculation using f = 1 leads to values of g that exceed those from calculations with realistic values of f and those from measurements by 5% to 20%, with increasing deviations for an increasing RH. All the other data, as there are the FPM simulation with the more realistic f, the data from Tang and Munkelwitz [1994], and the LACIS measurements, differ from each other by 5% at the most. Again, it can be seen that the parameterization given by Tang and Munkelwitz [1994] agrees very well with our measurements and the results of the FPM. As for NaCl, this is surprising, since again, the range of our measurements is well outside the range used by Tang and Munkelwitz [1994]. Nevertheless, this agreement is valuable information. It ensures that the parameterization of Tang and Munkelwitz [1994] and also that of Tang [1996] can be used to derive g for different dry particle sizes and RHs up to 99% when calibrating instrumentation that measures g, e.g., LACIS or HTDMAs operating at larger RHs. [27] As already stated, there is an obvious influence on the assumption of solution ideality on the hygroscopic growth derived with the FPM for (NH 4 ) 2 SO 4. It can be derived from Figure 2 that the molality in the examined droplets is lower than 4.5 mol/kg. The good agreement between measured and modeled growth for NaCl assuming f = 1 is due to the fact that f does not vary significantly from unity for this salt at the molalities in this study, whereas f is between 0.6 and 0.7 for (NH 4 ) 2 SO 4 for these molalities. This explains the deviation of the FPM simulation assuming ideal behavior for (NH 4 ) 2 SO 4. While the NaCl solution can be treated as an ideal solution for the range of molalities examined here, nonideal behavior has to be considered in modeling the growth of (NH 4 ) 2 SO 4 particles. It is important to emphasize the fact that even at 99% RH there is still a significant deviation between g derived for ideal and for nonideal behavior for the (NH 4 ) 2 SO 4 particles, although there is an increasing thinning of the solution with increasing particle/droplet size. The dilution of a droplet grown on (NH 4 ) 2 SO 4 at 99% RH is not large enough to allow for the neglect of the nonideal behavior of the solution. [28] Overall, the above comparisons establish the capability of the FPM to derive equilibrium diameters, as well as the ability of LACIS to measure equilibrium diameters for RHs up to 99.1%. We have shown that results of Tang [1996] and Tang and Munkelwitz [1994] are consistent with results from the FPM and with LACIS measurements, even at high RH. The importance of accounting for nonideal behavior of droplets has been shown, where Pruppacher and Klett [1997] and Pitzer and Mayorga [1973], the two parameterizations for f used in this work, produced similar results (see Figure 3). In the following, we use the growth factors measured with LACIS and, for comparison, g derived from Tang [1996] and Tang and Munkelwitz [1994], to examine the influence of hygroscopic growth and the assumption of nonideal behavior on the optical properties of atmospheric aerosol. 5. Influence of Hygroscopic Growth on Aerosol Light Scattering [29] Mie calculations using the Mie code provided by Bohren and Huffman [1983] were performed to derive the scattering coefficients of two exemplary particle number size distributions at different RHs. This was done to emphasize the importance of the use of proper growth factors for the atmospheric aerosol, especially at high RHs, when modeling the aerosol direct effect. Calculations were done at a wavelength of 550 nm. Exemplary dry number size distributions were taken from measurements of marine aerosol [O Dowd et al., 1999] and from measurements performed in an urban environment (Tuch et al. [2003], urban aerosol), both shown in Figure 6. The seasalt number size distribution given by O Dowd et al. [1999] was derived from measured sea-salt number size distribu- 6of9

7 Figure 6. Dry number size distributions used in Mie calculations to derive scattering coefficients. tions typical for the Northeast Atlantic. The urban number size distribution taken from Tuch et al. [2003] is an average distribution derived from continuous measurements during 5 years ( ) for working days in summer in Leipzig. These distributions are representative for a sea-salt and an urban aerosol, respectively. The marine aerosol consisted mainly of sea salt, which we assumed to consist of pure NaCl, and a small fraction of non-sea-salt sulfate, which was treated as (NH 4 ) 2 SO 4. The two substances were externally mixed. The urban aerosol was assumed to consist of (NH 4 ) 2 SO 4 entirely. This is a simplification, since it is known that atmospheric aerosol contains organic material up to several tens of percent. Some of this organic material is hydrophilic and participates in the hygroscopic growth. This is an area of ongoing research and of many unresolved issues [Kanakidou et al., 2004]. Among the open questions is the degree to which the organic compounds dissociate in solution, i.e., the osmotic coefficient that has to be used when modeling the behavior of the organic compounds. We decided to use a simple chemical composition for the aerosol in our calculations. Because of this, our results have to be looked at as upper boundaries. [30] To account for the hygroscopic growth of the particles of the dry number size distributions, g was taken from LACIS measurements for one set of Mie calculations, and from Tang [1996] and Tang and Munkelwitz [1994] for a second set of Mie calculations. Calculations were performed for RHs of 90%, 95%, 98%, and 99% for the two number size distributions. The measured values of g were interpolated linearly to get the respective values of g at 90%, 95%, 98%, and 99% RH. For particle sizes up to 175 nm, g measured for dry particle sizes of 150 nm was used, from 175 nm to 250 nm g measured for 200 nm was used, and for larger particles the measurements of g made for the dry particle size of 300 nm was taken. The parameterizations of Tang [1996] and Tang and Munkelwitz [1994] were used to derive g at 90%, 95%, 98%, and 99% RH. At these four RHs, g was derived separately for particle sizes of 20 nm, 60 nm, 200 nm, 600 nm, and 2000 nm, to account for the change in g with particle size. [31] The refractive index of the grown particles was derived using the volume mixing rule [e.g., Seinfeld and Pandis, 1998], with a real part of the refractive index of 1.55 for NaCl, 1.53 for (NH 4 ) 2 SO 4 (both from Toon et al. [1976]), and 1.33 for water [Seinfeld and Pandis, 1998]. Results from the calculations are presented in Table 2 as the ratio F of wet-to-dry light scattering coefficients at the different RHs. The non-sea-salt fraction in the marine aerosol added less than 4% to the scattering coefficient and values of F for the marine case were dominated by the sea-salt fraction; therefore Table 2 shows values of F only for the sea-salt fraction. [32] Results for the increase in the scattering coefficient due to hygroscopic particle growth are similar for the LACIS measurements and the growth derived from Tang [1996] and Tang and Munkelwitz [1994]. The largest difference in F between the two sets of calculations exists for sea salt at 99% RH, where an F of 20.6 was derived from LACIS measurements, whereas F derived from Tang [1996] was Still, considering the large overall increase, results from the two set of calculations are remarkably similar. This is based on the similarity of g derived from LACIS measurements with those derived from Tang [1996] and Tang and Munkelwitz [1994]. [33] Although the marine and urban number size distributions (Figure 6) are significantly different, the values of F given in Table 2 for the two distributions are similar. This is due to the high sensitivity of the light scattering coefficient to particle size. At 90% RH, the scattering coefficient increases by a factor of five, compared to the dry conditions. F gets larger as the humidity increases, getting larger than 20 at 99% RH. In terms of visibility (at the 550 nm wavelength used to derive the scattering coefficients) this is equal to a decrease from 50 km to 10 km for the sea-salt aerosol and from 63 km to 18 km for the urban aerosol for an increase in RH from dry conditions to 90% RH. At 99% RH the visibility decreases to 3 km for both aerosols. Table 2. Factors by Which the Scattering Coefficients of Two Exemplary (Dry) Number Size Distributions Increase Because of the Hygroscopic Growth of the Particles at High Relative Humidity a According to Literature LACIS Measurements 90% 95% 98% 99% 90% 95% 98% 99% Sea salt (NaCl) Urban ((NH 4 ) 2 SO 4 ) Urban ((NH 4 ) 2 SO 4 )(f = 1) a In the first two rows, the increase is given for g derived from Tang [1996] (NaCl) and Tang and Munkelwitz [1994] ((NH 4 ) 2 SO 4 ) (columns 1 4), and it is also given for g as measured with LACIS (columns 5 8). The third row gives the factors of the increase if the growth of the (NH 4 ) 2 SO 4 particles was assumed to follow that of an ideal solution. 7of9

8 [34] A further set of Mie calculations was performed to examine the sensitivity of the derived scattering coefficients to the assumption on the osmotic coefficient in the hygroscopic growth model. This was done for the urban aerosol, only. Values of g were taken from the FPM simulations assuming f = 1, which resulted in overestimated values of g (see Figure 5). Larger growth factors lead to larger scattering coefficients. Results from the calculations are shown in the bottom row of Table 2. F is overestimated by 30% to 50%, if the (NH 4 ) 2 SO 4 particles are assumed to grow according to Köhler theory but neglecting the nonideal behavior. [35] This shows clearly that information on particle growth factors at high RHs is required for accurate estimates of the scattering coefficients of atmospheric aerosol. Particle size at the atmospheric ambient conditions is one of the most important factors determining aerosol optical properties, and conditions above 90% RH are frequently found in the atmosphere, so measurements of growth factors of atmospheric aerosol at high RHs are needed. With this study, we introduced a new device that is able to perform measurements of hygroscopic growth up to 99% RH. 6. Conclusions [36] Equilibrium diameters (i.e., hygroscopic growth factors) of NaCl and (NH 4 ) 2 SO 4 particles were measured with LACIS at RHs up to 99.1% and were modeled using a numerical model, i.e., Fluent 6 together with the FPM, by including Köhler theory in the model. Comparison of measured and modeled values showed the necessity to account for nonideal behavior of the grown particles/ droplets in the model. It was found that it is important to account for nonideal behavior of (NH 4 ) 2 SO 4 solutions even at large RHs, at least up to 99.1% RH. [37] Deviations between the measured and the modeled growth factors were smaller than 4% if the nonideal behavior was accounted for in the simulations. In the case of (NH 4 ) 2 SO 4 particles, omitting the nonideal behavior leads to deviations between measured and modeled growth factors of 5% to 20%. [38] In addition, measured growth factors from this work compared favorably with values from parameterizations given by Tang [1996] for NaCl and Tang and Munkelwitz [1994] for (NH 4 ) 2 SO 4. Differences in growth factors were less than 4% for all RHs examined, up to 99.1% RH. [39] Scattering coefficients were derived for two model particle number size distributions (marine aerosol treated as NaCl and urban aerosol treated as (NH 4 ) 2 SO 4 ) at different RHs, using Mie theory. Growth factors measured with LACIS and those calculated from Tang [1996] and Tang and Munkelwitz [1994] were used to derive the sizes of the particles of the model aerosol at 90%, 95%, 98%, and 99% RH. The increase of the scattering coefficient with increasing RH was derived to be about fivefold from dry conditions to 90% RH and about 21-fold from dry conditions to 99% RH. This increase has to be looked at as an upper boundary, since we assumed the atmospheric aerosol to consist only of NaCl or of (NH 4 ) 2 SO 4 ), which overestimates the hygroscopic growth of atmospheric aerosol. Nevertheless, using the growth factors that were derived for (NH 4 ) 2 SO 4 assuming ideal behavior yielded an overestimation of the increase of the scattering coefficient of 30% to 50%, which clearly shows that an error will occur in calculations of the direct effect of atmospheric aerosol if the nonideal behavior of the aerosol particles/droplets is not taken into account. [40] Altogether, we showed the necessity of accounting for solution nonideal behavior, even at the high RH of 99%, and we introduced a new device, LACIS, which was used successfully for the first time to perform measurements of equilibrium diameters and hygroscopic growth factors, respectively, at these high RHs. References Barrett, J. C., and C. F. Clement (1988), Growth rates for liquid drops, J. Aerosol Sci., 19(2), Bohren, C. F., and D. R. Huffman (1983), Absorption and Scattering of Light by Small Particles, John Wiley, Hoboken, N. J. Brechtel, F. J., and S. M. Kreidenweis (2000a), Predicting particle critical supersaturation from hygroscopic growth measurements in the humidified TDMA. Part I: Theory and sensitivity studies, J. Atmos. Sci., 57, Brechtel, F. J., and S. M. Kreidenweis (2000b), Predicting particle critical supersaturation from hygroscopic growth measurements in the humidified TDMA. Part II: Laboratory and ambient studies, J. Atmos. Sci., 57, Chuang, C. C., J. E. Penner, K. E. Taylor, A. S. Grossman, and J. J. Walton (1997), An assessment of the radiative effects of anthropogenic sulfate, J. Geophys. Res., 102(D3), Cohen, M. D., R. C. Flagen, and J. H. Seinfeld (1987), Studies of concentrated electrolyte solutions using the electrodynmaic balance: 1. Water activities for single-electrolyte solutions, J. Phys. Chem., 91, Covert, D. S., J. L. Gras, A. Wiedensohler, and F. Stratmann (1998), Comparison of directly measured CCN with CCN modeled from the numbersize distribution in the marine boundary layer during ACE1 at Cape Grim, Tasmania, J. Geophys. Res., 103(D13), 16,597 16,608. Dusek, U., D. S. Covert, A. Wiedensohler, C. Neusüß, D. Weise, and W. Chantrell (2003), Cloud condensation nuclei spectra derived from size distributions and hygroscopic properties of the aerosol in coastal southwest Portugal during ACE-2, Tellus, Ser. B, 55, Fluent, Inc. (2003), FLUENT 6 user s guide, technical report, Lebanon, N. H. Hall, W. D., and H. R. Pruppacher (1976), The survival of ice particles falling from cirrus clouds in supersaturation, J. Atmos. Sci., 33, Haywood, J., V. Ramaswamy, and L. Donner (1997), A limited-area-model case study of the effects of sub-grid scale variations in relative humidity and cloud upon the direct radiative forcing of sulfate aerosol, Geophys. Res. Lett., 24(2), Hennig, T., A. Massling, F. J. Brechtel, and A. Wiedensohler (2005), A Tandem DMA for highly temperature-stabilized hygroscopic particle growth measurements between 90% and 98% relative humidity, J. Aerosol Sci., 36(10), Kanakidou, M., et al. (2004), Organic aerosol and global climate modelling: A review, Atmos. Chem. Phys., 5, Kelly, W. P., and P. H. McMurry (1992), Measurement of particle density by inertial classification of differential mobility analyzer-generated monodisperse aerosols, Aerosol Sci. Technol., 17, Kiselev, A., H. Wex, F. Stratmann, A. Nadeev, and D. Karpushenko (2005), White light optical particle spectrometer for in-situ measurements of condensational growth of aerosol particles, Appl. Opt., 44(22), Knutson, E. O., and K. T. Whitby (1975), Aerosol classification by electric mobility: Apparatus, theory and applications, J. Aerosol Sci., 6, Neusüß, C., D. Weise, W. Birmili, H. Wex, A. Wiedensohler, and D. Covert (2000), Size-segregated chemical, gravimetric and number distributionderived mass closure of the aerosol in sagres, portugal during ACE-2, Tellus, Ser. B, 52, Neusüß, C., H. Wex, W. Birmili, A. Wiedensohler, C. Koziar, B. Busch, E. Brüggemann, T. Gnauk, M. Ebert, and D. S. Covert (2002), Characterization and parameterization of atmospheric particle number-, mass-, and chemical-size distributions in central Europe during LACE 98 and MINT, J. Geophys. Res., 107(D21), 8127, doi: /2001jd O Dowd, C., J. A. Lowe, M. H. Smith, and A. D. Kaye (1999), The relative importance of sea-salt and nss-sulphate aerosol to the marine CCN population: An improved multi-component aerosol-droplet parameterisation, Q. J. R. Meteorol. Soc., 125, of9

9 Particle Dynamics, Inc. (2003), FPM user s guide, technical report, St. Louis, Mo. Pitzer, K. S., and G. Mayorga (1973), Thermodynamics of electrolytes: II. Activity and osmotic coefficients for strong electrolytes with one or both ions univalent, J. Phys. Chem., 77(19), Pruppacher, H. R., and J. D. Klett (1997), Microphysics of Clouds and Precipitation, pp , Springer, New York. Putaud, J.-P., et al. (2000), Chemical mass closure and assessment of the origin of the submicron aerosol in the marine boundary layer and the free troposphere at Tenerife during ACE-2, Tellus, Ser. B, 52, Quinn, P. K., D. J. Coffman, V. N. Kapustin, T. S. Bates, and D. S. Covert (1998), Aerosol optical properties in the marine boundary layer during the first Aerosol Characterization Experiment (ACE1) and the underlying chemical and physical aerosol properties, J. Geophys. Res., 103(D13), 16,547 16,563. Quinn, P. K., et al. (2000), Surface submicron aerosol chemical composition: What fraction is not sulfate?, J. Geophys. Res., 105(D5), Rader, D. J., and P. H. McMurry (1986), Application of the tandem differential mobility analyzer to studies of droplet growth or evaporation, J. Aerosol Sci., 17(5), Rissler, J., E. Swietlicki, J. Zhou, G. Roberts, M. O. Andreae, L. V. Gatti, and P. Artaxo (2004), Physical properties of the sub-micrometer aerosol over the Amazon rain forest during the wet-to-dry season transition: Comparison of modeled and measured CCN concentrations, Atmos. Chem. Phys., 4, Seinfeld, J. H., and S. N. Pandis (1998), Atmospheric Chemistry and Physics, p. 1134, John Wiley, Hoboken, N. J. Sloane, C. (1986), Effects of composition on aerosol light scattering efficiencies, Atmos. Environ., 20, Stratmann, F., A. Kiselev, S. Wurzler, M. Wendisch, J. Heintzenberg, R. J. Charlson, K. Diehl, H. Wex, and S. Schmidt (2004), Laboratory studies and numerical simulations of cloud droplet formation under realistic super-saturation conditions, J. Atmos. Oceanic Technol., 21, Tang, I. N. (1996), Chemical and size effects of hygroscopic aerosols on light scattering coefficients, J. Geophys. Res., 101(D14), 19,245 19,250. Tang, I. N., and H. R. Munkelwitz (1994), Water activities, densities and refractive indices of aqueous sulfates and sodium nitrate droplets of atmospheric importance, J. Geophys. Res., 99(D9), 18,801 18,808. Toon, O. B., J. B. Pollack, and B. N. Khare (1976), The optical constants of several atmospheric aerosol species: Ammonium sulfate, aluminum oxide, and sodium chloride, J. Geophys. Res., 81(33), Tuch, T., B. Wehner, M. Pitz, J. Cyrys, J. Heinrich, W. G. Kreyling, H. E. Wichmann, and A. Wiedensohler (2003), Long-term measurements of size-segregated ambient aerosol in two German cities located 100 km apart, Atmos. Environ., 37, Wilck, M., F. Stratmann, and E. R. Whitby (2002), A fine particle model for FLUENT: Description and application, paper presented at 6th International Aerosol Conference, Chin. Assoc. for Aerosol Res. in Taiwan, Taipei. Zhou, J., E. Swietlicki, O. H. Berg, P. P. Aalto, K. Hämeri, E. D. Nilsson, and C. Leck (2001), Hygroscopic properties of aerosol particles over the central Arctic Ocean during summer, J. Geophys. Res., 106(D23), 32,111 32,123. F. Brechtel, Brechtel Manufacturing, Inc., 1789 Addison Way, Hayward, CA 94544, USA. A. Kiselev, F. Stratmann, H. Wex and J. Zoboki, Institute for Tropospheric Research, Permoser Strasse 15, D Leipzig, Germany. (wex@tropos.de) 9of9