Effects of film-forming compounds on the growth of giant cloud …nenes.eas.gatech.edu/Reprints/GCCN_JGR.pdf · 2004-10-25 · Effects of film-forming compounds on the growth of giant

Effects of film-forming compounds on the growth of giant

cloud condensation nuclei: Implications for cloud

microphysics and the aerosol indirect effect

Jeessy Medina and Athanasios Nenes1

School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA

Received 20 February 2004; revised 16 June 2004; accepted 22 July 2004; published 19 October 2004.

[1] The presence of giant cloud condensation nuclei (GCCN) within stratocumulus cloudscan help the formation of drizzle by acting as collector drops. We propose that thepresence of film-forming compounds (FFCs) on GCCN may decrease their growth enoughto cease this drizzle formation mechanism. We systematically explore the accommodationproperties and amount of FFCs necessary to have a significant impact on GCCN sizeunder realistic conditions of growth inside typical stratocumulus clouds. It is foundthat even low mass fractions (as low as 0.2%) of FFCs with a modest effect on water vaporaccommodation can significantly reduce GCCN size and their potential to act as collectordrops. Our conclusions apply to both pristine and polluted aerosol conditions, whichsuggest that in the presence of FFCs, GCCN may be influencing the microphysicalevolution of clouds to a lesser extent than previously thought. INDEX TERMS: 0305

Atmospheric Composition and Structure: Aerosols and particles (0345, 4801); 0320 Atmospheric Composition

and Structure: Cloud physics and chemistry; 0345 Atmospheric Composition and Structure: Pollution—urban

and regional (0305); 1610 Global Change: Atmosphere (0315, 0325); KEYWORDS: giant cloud condensation

nuclei, film-forming compounds, cloud microphysics

Citation: Medina, J., and A. Nenes (2004), Effects of film-forming compounds on the growth of giant cloud condensation nuclei:

Implications for cloud microphysics and the aerosol indirect effect, J. Geophys. Res., 109, D20207, doi:10.1029/2004JD004666.

1. Introduction

[2] Understanding aerosol-cloud interactions is a prereq-uisite for understanding the hydrological cycle and climate.Because cloud droplets form on preexisting aerosols,also known as Cloud Condensation Nuclei (CCN), anthro-pogenic activities that increase aerosol concentrations maylead to more reflective clouds by increasing the amount ofCCN. This phenomenon is known as the ‘‘first’’ aerosolindirect effect [Twomey, 1977]. High concentrations of CCNalso may delay the formation of drizzle (the precursor ofprecipitation), which would increase cloud lifetime andcloud height; this is the so-called ‘‘second’’ aerosol indirecteffect [Albrecht, 1989]. Any process that affects the forma-tion of drizzle in clouds can be an important component ofthe ‘‘second’’ indirect effect. Marine environments, becauseof low CCN concentrations, are particularly susceptible toboth indirect effects; in particular, marine stratocumulusclouds which contribute about a third of global cloudcoverage [Albrecht, 1989].[3] Giant Cloud Condensation Nuclei (GCCN), CCN

with dry particle diameters greater than 5 mm, can influencedrizzle formation because they grow enough in cloud to

become efficient collector drops [Johnson, 1982; Tzivionet al., 1994; Cooper et al., 1997; Feingold et al., 1999].GCCN may be present in marine aerosol; number con-centrations for particles in the Northeast Atlantic in the5–150 mm range are between 0.1 and 0.5 cm"3 [Exton etal., 1986]. They are usually generated by breaking ofsurface waves, as well as other dynamically influencedmechanical processes [Fitzgerald, 1991, and referencestherein]. GCCN may also be of continental origin such asplant debris, or large dust particles [Rudich et al., 2002,and references therein]. Feingold et al. [1999] showedthat low concentrations (as low as 10"4 cm"3) of GCCNare sufficient to transition marine stratocumulus cloudsfrom a nondrizzling to a drizzling state. The same studyshowed that GCCN become more efficient in initiatingdrizzle formation as CCN concentrations increase. Thishypothesis is supported by observations from remotesensing data; Rosenfeld et al. [2002] showed that pollutedclouds developing over the Indian Ocean tended toprecipitate in contrast to polluted clouds that developedover the South Asian continent. They concluded thatprecipitation was enhanced in the clouds over the IndianOcean by the presence of GCCN generated from seaspray. Using the same remote sensing technique, Rudichet al. [2002] provided evidence that large salt-containingdust particles promoted precipitation in clouds downwindof the Aral Sea. The presence of GCCN may be asignificant component of the ‘‘second’’ indirect effect, butis currently not included in climate models.

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, D20207, doi:10.1029/2004JD004666, 2004

1Also at School of Earth and Atmospheric Sciences, Georgia Institute ofTechnology, Atlanta, Georgia, USA.

Copyright 2004 by the American Geophysical Union.0148-0227/04/2004JD004666$09.00

D20207 1 of 14

[4] In addition to dust and salt, GCCN may containsignificant amounts of organics and black carbon (BC)[Lelieveld et al., 2001]. Under certain conditions, BCinclusions may absorb enough radiation to heat the GCCNand decrease its size to prevent it from acting as acollector drop [Nenes et al., 2002b]. Furthermore, organicspecies may form hydrophobic films on the surface ofGCCN. These films, often composed of fatty acids acquiredfrom the air/ocean surface interface [Tervahattu et al.,2002a, 2002b], may influence the growth of GCCN bydecreasing the condensation rate of water onto them. It isquite likely that organics from anthropogenic emissionsmay also have the same effect. Chuang [2003] observedparticles in Mexico City that exhibited significant growthdelay. The delay was attributed to organic films on thesurface of aerosols with an estimated mass accommodationcoefficient ranging from 1 $ 10"5 to 4 $ 10"5, more thantwo orders of magnitude less than that for pure water drops.The effect of organic films on the activation of CCN hasbeen the focus of numerous studies [e.g., Gill et al., 1983;Shulman et al., 1997; Cruz and Pandis, 1998; Feingold andChuang, 2002]; all agree that additional information isneeded to describe the effect of organics on the wateruptake of CCN.[5] As stated, the presence of GCCN can enhance the

formation of drizzle. However, it is possible that thepresence of film-forming compounds (FFCs) on GCCNmay delay their growth such that the latter became toosmall to act as efficient collector drops; hence, thismechanism of drizzle formation may cease. The potentialeffect of FFCs on GCCN growth and its implications forcloud precipitation processes are addressed in this study.Through simulations of GCCN growth within stratocumu-lus clouds, we define a range of accommodation propertiesand organic mass fractions necessary for FFCs to impartimportant reductions in GCCN size. Slow growth kineticsattributable to the dissolution of partially soluble substan-ces [Shulman et al., 1996; Shantz et al., 2003] may alsoaffect GCCN growth, but is beyond the scope of thisstudy.

2. Model Formulation

[6] The growth of GCCN within a stratocumulus cloud issimulated using the trajectory ensemble model (TEM)approach of Stevens et al. [1996]. This methodologyemploys a large eddy simulation (LES) of a cloud field thatgenerates a set of Lagrangian trajectories that describe theevolution of the cloud field. Each trajectory within the setforces a nonadiabatic parcel model that calculates thegrowth of a GCCN. A horizontal ensemble average GCCNsize throughout the boundary layer is calculated.

2.1. Model Equations: Trajectory Properties andSupersaturation Profiles

[7] Each trajectory contains variables that characterize thethermodynamic state of a material point, as it is advectedthroughout the flow field. The variables contained in thetrajectories are time t, position x, y, and z, pressure p,potential temperature in moist air ql, and the total (e.g.,liquid and vapor) water mass mixing ratio, wt. The time stepbetween two consecutive trajectory points is 2 s. All

material points are initially taken below cloud level toensure that their initial liquid water content (LWC) isapproximately zero. The tendencies of x, y, z, p, ql, and wt

are calculated by the finite difference between two consec-utive time steps.[8] Calculation of the growth of GCCN within a trajec-

tory requires the knowledge of the parcel p, temperature T,and the parcel water vapor supersaturation, S. This is notdirectly available from the trajectories, and is calculated asfollows. T is computed from ql, defined as:

ql ¼ Tpo

p

$ % RCpþDHv

Cp

po

p

$ % RCp

wl ; ð1Þ

where DHv is the latent heat of vaporization of water, Cp isthe molar heat capacity of air, wl is the liquid water massmixing ratio, po is the reference pressure (1000 mb), and Ris the universal gas constant. Equation (1) is used to solvefor T and its rate of change, dT/dt:

T ¼ qlp

po

$ % RCp"DHv

Cp

wl ð2Þ

dT

dt¼ p

po

$ % RCpdqldtþ ql

1

po

$ % RCp R

Cp

$ %p

RCp"1 dp

dt

"DHv

Cp

dwl

dt:

ð3Þ

dql/dt and dp/dt from equation (3) are approximated usingDql/Dt and Dp/Dt from the trajectory output.[9] S is calculated from the water vapor mass mixing

ratio, wv [Seinfeld and Pandis, 1998],

S ¼ wv

wv*¼ pMa

p#Mw

wv; ð4Þ

where w*v is the saturation water vapor mixing ratio, Mw

and Ma are the molar masses of water and air, respectively.wv is calculated from the conservation of water in theparcel:

wt ¼ wv þ wl: ð5Þ

Solving for the rate of change of wv and using thetrajectory output Dwt/Dt for dwt/dt, we obtain:

dwv

dt¼ dwt

dt" dwl

dt' Dwt

Dt" dwl

dt: ð6Þ

The liquid condensation rate is calculated as [Seinfeld andPandis, 1998],

dwl

dt¼ p

2

Xi

NiD2pirw

dDpi

dt; ð7Þ

where Ni is the number of droplets (in each size class)per unit mass of air, rw is the density of water, Dpi

is thedroplet diameter of each size class. The growth/evapora-

D20207 MEDINA AND NENES: CLOUD CONDENSATION NUCLEI, FILMS, AND CLOUDS

2 of 14

D20207

tion rate of each droplet is given by [Seinfeld and Pandis,1998],

dDpi

dt¼ rwRT

4p#D0vMw

þ DHvrw4k 0aT

DHvrwTR

" 1

$ %& '"11

Dpi

S " Seq" #

;

ð8Þ

where Seq is the water vapor saturation ratio of the droplet.Kohler theory describes Seq [Seinfeld and Pandis, 1998],

ln Seq" # ¼ A

Dp

" B

D3p

ð9Þ

A ¼ 4MwsRTrw

ð10Þ

B ¼ 6nsMw

prw; ð11Þ

where s is the droplet surface tension, and ns is the molesof solute dissolved in the droplet. The water vapordiffusivity, D0

v , and the thermal conductivity of air, k0a,(both modified to account for noncontinuum effects) aregiven by

D0v ¼

Dv

1þ 2Dv

aDp

2pMw

RT

$ %12

ð12Þ

k 0a ¼ka

1þ 2ka

aTDpraCp

2pMa

RT

$ %12

; ð13Þ

where a and aT are the mass and thermal accommodationcoefficients, respectively, and ra is the density of air. Weassume that aT is equal to unity, and the presence of FFCsbears no effect on the parameter. (Although not exploredhere, if there were an effect, aT would decrease and furtherdelay droplet growth.) The nonmodified water vapordiffusivity, Dv , and the thermal conductivity of air, ka,are given by

Dv ¼ 0:211$ 105

p

T

273

$ %1:94

ð14Þ

ka ¼ 10"3 4:39þ 0:071 * Tð Þ: ð15Þ

[10] To summarize, T, p, and S within the air parcel arecalculated from the LES trajectories in the following man-ner. Initial conditions for ql, p, and wt are used to calculatethe initial value of T, assuming that wl is negligible. Theinitial S is calculated using equation (4). The initial Dpi

arecalculated assuming the aerosol is in equilibrium with S.The derivatives of ql, p, wt with respect to time are obtainedfrom the trajectory output, and the parcel T, p, and S areobtained by integrating equations (3), (6), (7) and (8) usingthe implicit ODE solver LSODE [Hindmarsh, 1983].

2.2. CCN Populations

[11] Two supersaturation histories, one characteristic of apristine and one of an urban environment, were derived foreach LES trajectory set. The marine and urban aerosol sizedistributions of Whitby [1978] are used to represent theaerosol populations for the pristine and polluted environ-ments, respectively (Table 1). Polluted clouds tend to havelower supersaturations relative to their pristine counterparts,because of the increased competition in the latter for watervapor [e.g., Nenes et al., 2001]. As a result, the drivingforce for GCCN growth in polluted clouds is smallercompared to that of pristine clouds thus GCCN may haveless probability of becoming collector drops.[12] When calculating the supersaturation levels, a simple

chemical composition (ammonium sulfate) is assumed forthe aerosol. In reality, the presence of FFC-containingGCCN should coincide with the presence of FFCs through-out the aerosol size distribution. However, since we con-sider two extreme CCN conditions (pristine versus urban), itis rather unlikely to imagine a condition of FFC-coatedaerosol that would yield supersaturation levels outside ofthe two cases. Thus the simplified aerosol chemical com-position is sufficient for our study.

2.3. GCCN Size Calculations

[13] Figure 1 illustrates the procedure used in calculat-ing the growth of GCCN within a stratocumulus cloud.The Lagrangian trajectories were obtained from the LESsimulation and used in conjunction with either pristine orpolluted aerosol populations as inputs into the cloudparcel model described in section 2.1. The parcel modelcomputes a supersaturation history for each of the trajec-tories. Along each trajectory, droplets are subject to auniform ‘‘macroscopic supersaturation’’; we neglect con-sidering that individual droplets might be subject to localfluctuations that persist down to the millimeter scale, aswe already consider two drastically different supersatura-tion regimes (marine versus urban CCN conditions); wepresume that smaller-scale fluctuations lie within thisrange. GCCN with a prescribed chemical compositionare grown (according to equation (8)) using the supersat-uration profiles calculated for each trajectory. The tem-poral (1 hour) and horizontally averaged GCCN size isthen calculated to represent the average vertical profilesof GCCN size.

3. Simulations

3.1. Stratocumulus Clouds

[14] Lagrangian trajectories used in this study werederived from two marine stratocumulus cloud simulations

Table 1. Aerosol Distribution Parametersa

Aerosol Type

Nuclei ModeAccumulation

Mode Coarse Mode

Dg,1 s1 N1 Dg,2 s2 N2 Dg,3 s3 N3

Marine 0.010 1.6 340 0.070 2.0 60 0.62 2.7 3.1Urban 0.014 1.8 106000 0.054 2.16 32000 0.86 2.21 5.4

aDg,i, si, Ni [Whitby, 1978]. Dg,i represents the average diameter (mm), Ni

is the number concentration (cm"3), and si is the geometric standarddeviation for each mode.


3 of 14

D20207

(‘‘ASTEX-1’’ and ‘‘ASTEX-2’’) for conditions observedduring the Atlantic Stratocumulus Transition Experiment[Albrecht et al., 1995]. 500 trajectories covering 1 hour ofsimulation time were derived for each cloud. Figures 2aand 2b display important characteristics for each cloud.ASTEX-1 and ASTEX-2 have average updraft velocities ofabout 0.2 to 0.4 m s"1, respectively; both clouds areenergetic enough to maintain droplets of at least 80 mm indiameter. ASTEX-2 is a heavily drizzling cloud with ahigher LWC (0.6 g m"3) and lower cloud base (200 m)than ASTEX-1 (0.4 g m"3 and 400 m, respectively).

3.2. Composition and Size of GCCN

[15] We consider GCCN composed of soluble and insol-uble material with an average density of 1760 kg m"3.The soluble material is assumed to be (NH4)2SO4 and theinsoluble fraction is treated as a mixture of FFCs andcore material. We consider initial dry CCN diameters,Dp,dry, of 1 mm, 2.5 mm, 5 mm, 7.5 mm, 10 mm, 12.5 mm,15 mm, and 25 mm. For each particle size, simulationswere performed for different soluble mass fractions, es, of25%, 50%, and 75%. Finally, for each es, the FFC massfraction, eo, was varied from 0.2% to 20%. A total of2016 simulations were done for the ASTEX-1 andASTEX-2 trajectory sets.

3.3. FFCs and Their Effect on Droplet Growth Rate

[16] We adopt a ‘‘film-breaking’’ model [Feingold andChuang, 2002] to describe the effect of FFCs on dropletgrowth rate (Figure 3). When present, FFCs are initiallyassumed to form a film on the CCN surface. This makes theparticle experience slow growth, expressed by a low valueof the accommodation coefficient, aslow. If the particlegrows enough to break its film, the FFCs are incorporatedwithin the insoluble core material. The droplet surfacebecomes an aqueous solution, and the droplet is assumedto enter a rapid growth regime, expressed by a ‘‘pure’’ waterarapid = 0.042 [Pruppacher and Klett, 1997]. Publishedvalues of arapid vary considerably, ranging between 0.04to 1. Fung et al. [1987] were able to fit arapid with a valueclose to unity from condensational growth measurements ona pure NaCl droplet using Mie resonance spectroscopy. Thevalue of arapid is closer to 0.01 for aged atmosphericdroplets [Pruppacher and Klett, 1997] and maybe as lowas 0.04 for pure water [Shaw and Lamb, 1999; Li et al.,2001]. For this study, we used 0.042 as it is widely acceptedfor atmospheric droplets in the atmospheric community[Pruppacher and Klett, 1997].[17] As the chemical composition of FFCs and their

accommodation properties are not known, we considervalues for aslow ranging from 10"5 to 10"3. The lower

Figure 1. Overview of the methodology used in calculating the growth of GCCN using the LES-derived Lagrangian trajectories.


4 of 14

D20207

end value of aslow is consistent with values estimated byChuang [2003]. The ‘‘critical’’ film thickness (film thick-ness required for the film to break) is assumed to equal amonolayer coverage of cetyl alcohol molecules (&10"10 m)[Feingold and Chuang, 2002; Nenes et al., 2002a] andrepresents the maximum effect a film can exert on the

growth of a droplet. The film thickness is related to thevolume of FFCs in a particle, which in turn is directlyrelated to eo and Dp,dry. Figure 4 presents the thresholddiameter required for the film on a droplet to break as afunction of its Dp,dry and eo. Although it may be possible fora film to break before reaching its ‘‘critical’’ film thickness,

Figure 2. Vertical profiles of updraft velocity characteristics for stratocumulus cloud (a) ASTEX-1 and(b) ASTEX-2.


5 of 14

D20207

this is not explored; the current simulations however can berelated to films that break at any ‘‘critical’’ film thickness.

4. Simulation Results

4.1. GCCN Maximum Size Reductions

[18] To systematically explore the influence of FFCs onGCCN growth, we examine rmax, the ratio of the maximumin-cloud diameter of GCCN if they contain FFCs, Dp

max (eo),over the maximum diameter they attain in the absence ofFFCs, Dp

max (eo = 0):

rmax ¼Dmax

p eoð ÞDmax

p eo ¼ 0ð Þ : ð16Þ

The lower the value of rmax, the more effective FFCs are ininhibiting the condensational growth of GCCN. Figure 5presents rmax as a function of Dp,dry for supersaturationtrajectories derived from ASTEX-1 for pristine aerosolconditions. rmax is primarily affected by aslow; when aslow isequal to 10"3, rmax ranges between 0.7 and 1.0 but whenaslow is equal to 10"5, rmax is between 0.1 and 0.4. Similar

Figure 4. The diameter required for a growing droplet to break its organic film as a function of eo andDp,dry .

Figure 3. Illustration of the ‘‘breaking film’’ model adapted in this study. Film-covered dropletsexperience slow condensational growth. When the threshold size required to break the film is reached, thedroplet enters a rapid growth phase.


6 of 14

D20207

behavior is seen in both ASTEX-1 and ASTEX-2trajectories for pristine and polluted conditions (not shown).In addition to aslow , rmax also depends on Dp, dry , eo, and es.Figure 6 (which is the same as Figure 5, but for aslow =10"5) is used to explain the effect of each parameter on rmaxthrough three examples. ‘‘Case 1’’ represents rmax as afunction of Dp, dry, ‘‘case 2’’ presents rmax as a function ofeo, and ‘‘case 3’’ shows rmax as a function of es.[19] ‘‘Case 1’’ corresponds to eo = 5% and es = 50%. For

small Dp,dry , rmax decreases with increasing Dp, dry until itreaches a minimum value. For large values of Dp,dry , rmaxcurves converge to a common curve. Typically, GCCNwith Dp,dry % 15 mm approach this limit in all the cloudconditions considered in this study. This asymptotedepends on the value of aslow (Figure 5). The behaviorof rmax can be rationalized if it is related to the GCCN drysize:

rmax ¼Dmax

p eoð ÞDmax

p eo ¼ 0ð Þ 'Dp;dry þ DDmax

p;cg

Dp;dry þ DDmaxp;cg eo ¼ 0ð Þ : ð17Þ

DDp,cgmax and DDp,cg

max (eo = 0) represent the maximum averagecondensational growth when FFCs are present and absent,respectively. GCCN with small Dp,dry grow enough in mosttrajectories to break their films and experience significant

growth with arapid. Under these conditions, DDp,cgmax '

DDp,cgmax (eo = 0) allowing rmax to approach unity. As Dp,dry

increases, the amount of organic material also increases andfewer GCCN experience film rupture. Thus there is lesscondensational growth and DDp,cg

max is smaller than DDp,cgmax

(eo = 0), forcing rmax to decrease. At a characteristic drydiameter, D*p, dry , the films in all trajectories do not rupture;thus, the condensational growth (and rmax) reaches aminimum value. For Dp, dry > D*p, dry , DDp,cg

max does notchange much and rmax varies monotonically with Dp, dry . Itis important to examine the values of D*p,dry to assess whichrange of GCCN sizes display the strongest sensitivity to thepresence of FFCs. From Figures 5 and 6, we consistentlysee that D*p,dry ranges between 5 and 12 mm, which is thesize range where most GCCN are likely to exist [Exton etal., 1986]. This means that GCCN are more likely to exhibitthe maximum sensitivity to the presence of FFCs.[20] Since all trajectories in the cloud do not exhibit the

same supersaturation history, not all GCCN will concur-rently experience bursting of their films. The relativeproportion of ‘‘rapidly’’ to ‘‘slowly’’ growing GCCN willthus depend, in addition to the parcel supersaturation, on eo.‘‘Case 2,’’ represented by three simulation points on Figure 6,illustrates this effect. When eo is less than 0.2%, rmax

approaches unity. As eo increases to 0.5%, rmax decreasesto approximately 0.6. When eo is about 1%, rmax is 0.25;

Figure 5. Value of rmax as a function of Dp,dry , for a variety of aslow (10"3, 10"4, 10"5), es (25%, 50%,75%) and eo (0.2%, 0.5%, 1%, 5%). The simulations are for the ASTEX-1 cloud with pristine aerosolconditions.


7 of 14

D20207

additionally increasing eo does not further decrease rmax.Equation (17) can be used to explain this dependence. Forsmall values of eo, many droplets experience rapid conden-sational growth as their films rupture. This translates toDDp,cg

max ' DDp,cgmax (eo = 0), or rmax approximately unity. As

eo increases, less and less of the GCCN can grow enough tobreak their films; at a characteristic e*o, none of the GCCNcan break their films and DDp,cg

max becomes minimum for theGCCN in all trajectories. Thus, for eo > e*o (in our case e*o &1%), rmax remains constant (see also Figure 7).[21] The hygroscopicity of GCCN is determined by the

amount of soluble material present. The more hygroscopicGCCN are, the larger the driving force for condensationalgrowth (equation (8)); furthermore, their equilibrium sizewith ambient RH is larger below cloud [Seinfeld andPandis, 1998]. Both factors contribute to a larger wet sizeof the GCCN in cloud when compared to less hygroscopicCCN with the same dry diameter; thus increasing thehygroscopicity would facilitate film rupture. ‘‘Case 3’’examines the effect of es on rmax. When the es equals25%, rmax is 0.35. When es equals 50% and 75%, rmax is0.7 and 0.85, respectively. The soluble fraction effect ismore pronounced at small Dp,dry , as less growth (comparedto larger Dp,dry) is necessary to rupture the films.

4.2. GCCN Sizes

[22] The analysis in section 4.1 was an attempt torationalize and parameterize the effect of FFCs on GCCNgrowth. In terms of the microphysical evolution of a cloud,what is ultimately important is the absolute size of the

GCCN in the cloud. As proposed by Feingold et al. [1999],we consider GCCN as effective collector drops if their sizein cloud is 40 mm or greater.[23] Figure 8 presents the average growth of a 5 mm

GCCN under pristine conditions for different values of aslow

and with eo equal to 0.2% (Figure 8a), and 0.5% (Figure 8b).In Figure 8, FFC-free GCCN can grow to 50 mm (e.g., it canact as a collector drop). In Figure 8a, for aslow equal to 10"3

and 10"4, the reduction in size is minimal. This is expected;under these conditions of eo and aslow , rmax approaches 1(Figure 5). For aslow equal to 10"5, the GCCN growth isinhibited and the maximum size reached is below thethreshold of 40 mm. Increasing eo (Figure 8b) results in amore pronounced reduction in GCCN size; in contrast toFigure 8a, a reduction of about 10 mm (which is significant,given that the GCCN is now about the 40 mm size threshold)is seen throughout the cloud for aslow equal to 10"4. Almostcomplete inhibition in growth is seen for aslow = 10"5; theGCCN only grows to a size of 10 mm, and is effectivelyindistinguishable from any other cloud droplet. Increasingeo to 1% further decreases the GCCN size when aslow is10"4, and as expected, almost complete inhibition in growthis seen when aslow is equal to 10"5 (not shown).[24] Figure 9 presents simulations for a 5 mm GCCN

under polluted conditions. Under these conditions, theFFC-free GCCN can still exceed the 40 mm size thresholdand act as an effective collector drop. If a small amount ofFFC is present (eo = 0.2%), negligible reductions in sizeare seen for aslow between 10"3 and 10"4 (Figure 9a).Strong reductions in size, however, are seen if aslow = 10"5.

Figure 6. Same as Figure 5, except that simulations are for one value of aslow (10"5).


8 of 14

D20207

Increasing eo to 0.5% (Figure 9b) significantly reducesthe GCCN size for aslow = 10"4; almost complete inhibitionin growth is seen for aslow = 10"5. By comparing Figures 8and 9, a striking observation arises: the ability of a GCCN togrow is strongly dependent on eo (e.g., film thickness) andaslow, but not on the cloud supersaturation characteristics.[25] It is important to assess whether the conclusions

from Figures 8 and 9 apply to GCCN of larger Dp,dry.Figure 10 presents growth curves for a 10 mm GCCN forboth pristine and polluted cloud conditions. In the absenceof FFCs (Figure 10a), the 10 mm GCCN grows to about60 mm in diameter and can act as a collector drop. When theGCCN contains 0.2% FFCs, the growth is reduced some-what when aslow ranges between 10"4 and 10"3, but notenough to prevent the GCCN from growing past 40 mm(Figure 10a). Nevertheless, for aslow equal 10"5, the GCCNis prevented from becoming a collector drop; in fact, bothpristine and polluted simulations overlap and display thesame growth behavior. Increasing the FFC mass fraction to1% exemplifies the growth inhibition for both polluted andpristine conditions (Figure 10b). In Figure 10b, significantreductions in size are seen even for aslow equal to 10"4.[26] Up to this point, we have examined the growth of

individual GCCN within the cloud trajectory ensemble. Inreality, there is a size distribution of GCCN present within acloud; it is therefore instructive to extend our analysis to apolydisperse GCCN population using Whitby [1978] distri-butions within a size range of 1 to 25 mm. Figure 11 showsthe fraction of this GCCN population whose size exceeds

40 mm as a function of eo. The results were obtained fromASTEX-1 trajectories under pristine conditions. In theabsence of FFCs (e.g., eo = 0), about 30% of the GCCNbecome larger than 40 mm, thus potentially acting ascollector drops. When FFCs are included, the fraction ofGCCN whose size reaches threshold significantly decreaseseven if a small amount of FFCs are present. For example,when eo is about 2%, less than 10% of the GCCN withinthe population exceed the 40 mm threshold.

5. Summary and Conclusions

[27] Our analysis indicates that the presence of FFCs inGCCN can influence the microphysical evolution of cloudsthrough this previously unexplored mechanism. FFCs de-crease the rate of mass transfer of water vapor to/from theGCCN expressed by a reduction in the accommodationcoefficient, aslow. This study shows that for aslow rangingfrom 10"3 to 10"5, GCCN within the trajectory setsexperienced a 30–90% reduction in size when comparedto GCCN growing with a ‘‘pure water’’ accommodationcoefficient of 0.042. For GCCN with dry diameters greaterthan 15 mm, aslow is the primary parameter affecting thedroplet size; not so with GCCN with dry diameters less than15 mm, which were found to be dependent on the initial drydiameter as well as the FFC content. Whether or not the filmruptures is a deciding factor for the droplet size. Loweringthe FFC mass fraction and increasing the hygroscopicity ofthe GCCN tend to facilitate the rupture of films.

Figure 7. Value of rmax as a function of eo with a constant es = 25%, for a variety of Dp,dry (2.5 mm,5 mm, 10 mm) and aslow (10"3, 10"4, 10"5). The simulations are for the ASTEX-1 cloud for pristineaerosol conditions.


9 of 14

D20207

Figure 8. Average growth of a 5 mm GCCN, for the ASTEX-1 cloud with pristine aerosol conditions.The es is constant at 25%. When FFCs are absent (e.g., eo = 0%), aslow = arapid = 0.042; (a) eo = 0.2%,(b) eo = 0.5%.


10 of 14

D20207

Figure 9. Same as Figure 8, except that simulations are for the ASTEX-1 cloud with polluted aerosolconditions; (a) eo = 0.2%, (b) eo = 0.5%.


11 of 14

D20207

Figure 10. Same as Figure 8, except that simulations present a 10 mm GCCN for both pristine (white)and polluted (black) conditions; (a) eo = 0.2%, (b) eo = 1%.


12 of 14

D20207

[28] The absolute wet diameter of GCCN in the cloudsimulations is important to assess its effectiveness as acollector drop; in this study, we used a threshold diameterof 40 mm to classify the GCCN as a collector drop. Thisstudy shows that the conditions to exceed the threshold is aweak function of the cloud supersaturation history (i.e.,whether it corresponded to pristine or polluted aerosolconditions); this opens the possibility of parameterizing thismechanism. The results also indicate that very low massfractions of organic film-forming compounds (FFCs) areneeded to render a GCCN an inefficient collector drop.Under certain conditions, FFC mass fractions as low as0.5%, delayed the growth of a 5 mm GCCN to such anextent that its final size was indistinguishable from a typicalcloud droplet (&10 mm). It is quite likely that the thresholddiameter for becoming a collector drop would vary fromcloud to cloud. This uncertainty does not have a significantimpact on our conclusions, as the effect of FFCs on growthis potentially very strong.[29] In addition to affecting the accommodation proper-

ties, FFCs, being surfactants may, together with the water-soluble organics, decrease droplet surface tension. The latterhas been shown to have an important effect on dropletnumber [e.g., Facchini et al., 1999; Nenes et al., 2002a].Such effects are neglected here, but are not expected to havea significant impact on our results; GCCN already have verylow critical supersaturations, Sc, (&0.01%), so an additionaldecrease in Sc is not expected to appreciably affect growth.FFCs however may be partially soluble, so they can affect

droplet growth by introducing another kinetic limitation,being the finite dissolution time [e.g., Shantz et al., 2003].The significance of this mechanism remains to be explored.The additional hygroscopicity from the soluble fraction,although not explicitly considered here in the model, wouldnot exceed that of (NH4)2SO4 and thus lie within the rangeexplored. It is also possible that the hydrophobic films mayundergo oxidative reactions [Eliason et al., 2004] and beconverted to water-soluble compounds. The timescale re-quired for air masses to be aged (oxidized) is much largerthan the lifetime of freshly emitted marine GCCN whichmay be on the order of hours [Gong et al., 2002]. Thus it islikely that the films may retain their hydrophobic state overthe course of the GCCN lifetime. As both biogenic andanthropogenic sources emit large amounts of hydrophobiccompounds that potentially can act as FFCs, it is likely thatlocal sources of FFC are ubiquitous throughout the atmo-sphere and responsible for the accommodation properties ofthe CCN.[30] The most striking result of this study is that small

quantities of FFCs, if present in GCCN, may have thepotential to change a cloud from a precipitating to a non-precipitating state. Together with the synergistic effect ofblack carbon [Nenes et al., 2002b], GCCN may be influ-encing the microphysical evolution of clouds to a lesserextent than previously thought. Therefore understanding thefrequency of occurrence and accommodation propertiesof FFCs is required to advance the understanding andmodeling capability of the hydrological cycle.

Figure 11. Fraction of GCCN that exceeds the 40 mm size threshold as a function of eo. The simulationscorrespond to the ASTEX-1 cloud with pristine aerosol conditions.


13 of 14

D20207

[31] Acknowledgments. This work was supported by NASA Head-quarters under the Earth System Science Fellowship grant NGT5-30506.We would also like to thank Graham Feingold and Bjorn Stevens forproviding the trajectories from the LES simulations.

ReferencesAlbrecht, B. A. (1989), Aerosols, cloud microphysics, and fractionalcloudiness, Science, 245, 1227–1230.

Albrecht, B. A., C. S. Bretherton, D. Johnson, W. H. Scubert, and A. S.Frisch (1995), The Atlantic Stratocumulus Transition Experiment:ASTEX, Bull. Am. Meteorol. Soc., 76, 889–904.

Chuang, P. Y. (2003), Measurement of the timescale of hygroscopic growthfor atmospheric aerosols, J. Geophys. Res., 108(D9), 4282, doi:10.1029/2002JD002757.

Cooper, W. A., R. T. Bruintjes, and G. K. Mather (1997), Calculationspertaining to hygroscopic seeding with flares, J. Appl. Meteorol., 36,1449–1469.

Cruz, C. N., and S. N. Pandis (1998), The effect of organic coatings on thecloud condensation nuclei activation of inorganic atmospheric aerosol,J. Geophys. Res., 103, 13,111–13,123.

Eliason, T. L., J. B. Gilman, and V. Vaida (2004), Oxidation of organicfilms relevant to atmospheric aerosols, Atmos. Environ., 38, 1367–1378.

Exton, H. J., J. Latham, P. M. Park, M. H. Smith, and R. R. Allan (1986),The Production and Dispersal of Maritime Aerosol: Oceanic Whitecapsand Their Role in Air-Sea Exchange Processes, edited by E. C. Monahanand G. Mac Niocaill, pp. 175–193, D. Reidel, Norwell, Mass.

Facchini, M. C., M. Mircea, S. Fuzzi, and R. Charlson (1999), Cloudalbedo enhancement by surface-active organic solutes in growingdroplets, Nature, 401, 257–259.

Feingold, G., and P. Chuang (2002), Analysis of the influence of film-forming compounds on droplet growth: Implications for cloud microphy-sical processes and climate, J. Atmos. Sci., 59, 2006–2018.

Feingold, G., W. R. Cotton, S. M. Kreidenweiss, and J. T. Davis (1999),The impact of giant cloud condensation nuclei on drizzle formation instratocumulus: Implication for cloud radiative properties, J. Atmos. Sci.,56, 4100–4117.

Fitzgerald, J. W. (1991), Marine aerosols—A review, Atmos. Environ.Part A, 25, 533–545.

Fung, K. H., I. N. Tang, and H. R. Munkelwitz (1987), Study of condensa-tional growth of water droplets by Mie resonance spectroscopy, Appl.Opt., 26, 1282–1287.

Gill, P. S., T. E. Graedel, and C. J. Weschler (1983), Organic films onatmospheric aerosol particles, fog droplets, cloud droplets, raindrops,and snowflakes, Rev. Geophys., 21, 903–920.

Gong, S. L., L. A. Barrie, and M. Lazare (2002), Canadian Aerosol Module(CAM): A size-segregated simulation of atmospheric aerosol processesfor climate and air quality models: 2. Global sea-salt aerosol and itsbudgets, J. Geophys. Res., 107(D24), 4779, doi:10.1029/2001JD002004.

Hindmarsh, A. C. (1983), ODEPACK: A Systematized Collection of ODESolvers, Scientific Computing, edited by R. S. Stepleman, pp. 55–64,North-Holland, New York.

Johnson, D. B. (1982), The role of giant and ultragiant aerosol particles inwarm rain initiation, J. Atmos. Sci., 39, 448–460.

Lelieveld, J., et al. (2001), The Indian Ocean Experiment: Widespread airpollution from south and southeast Asia, Science, 291, 1031–1036.

Li, Y. Q., P. Davidovits, Q. Shi, J. T. Jayne, C. E. Kolb, and D. R. Worsnop(2001), Mass and thermal accommodation coefficients of H20(g) on

liquid water as a function of temperature, J. Phys. Chem. A, 105,10,627–10,634.

Nenes, A., S. Ghan, H. Abdul-Razzak, P. Y. Chuang, and J. H. Seinfeld(2001), Kinetic limitations on cloud droplet formation and impact oncloud albedo, Tellus, Ser. B, 53, 133–149.

Nenes, A., R. J. Charlson, M. C. Facchini, M. Kulmala, A. Laaksonen, andJ. H. Seinfeld (2002a), Can chemical effects on cloud droplet numberrival the first indirect effect?, Geophys. Res. Lett., 29(17), 1848,doi:10.1029/2002GL015295.

Nenes, A., W. C. Conant, and J. H. Seinfeld (2002b), Black carbonradiative heating effects on cloud microphysics and implications for theaerosol indirect effect: 2. Cloud microphysics, J. Geophys. Res.,107(D21), 4605, doi:10.1029/2002JD002101.

Pruppacher, H., and J. Klett (1997), Microphysics of Clouds and Precipita-tion, 2nd ed., Kluwer Acad., Norwell, Mass.

Rosenfeld, D., L. R. Khain, and M. Pinsky (2002), The role of sea spray incleansing air pollution over ocean via cloud processes, Science, 297,1667–1670.

Rudich, Y., O. Khersonsky, and D. Rosenfeld (2002), Treating clouds witha grain of salt, Geophys. Res. Lett., 29(22), 2060, doi:10.1029/2002GL016055.

Seinfeld, J. H., and S. N. Pandis (1998), Atmospheric Chemistry and Phy-sics: From Air Pollution to Climate Change, John Wiley, Hoboken, N. J.

Shantz, N. C., W. R. Leaitch, and P. F. Caffrey (2003), Effect of organics oflow solubility on the growth rate of cloud droplets, J. Geophys. Res.,108(D5), 4168, doi:10.1029/2002JD002540.

Shaw, R. A., and D. Lamb (1999), Experimental determination of thethermal and condensation coefficients of water, J. Chem. Phys., 111,10,659–10,663.

Shulman, M., M. Jacobson, R. Charlson, R. Synovec, and T. Young (1996),Dissolution behavior and surface tension effects of organic compounds innucleating cloud droplets, Geophys. Res. Lett., 23, 277–280.

Shulman, M. L., R. J. Charlson, and E. J. Davis (1997), The effects ofatmospheric organics on aqueous droplet evaporation, J. Aerosol Sci., 28,737–752.

Stevens, B., G. Feingold, W. Cotton, and R. Walco (1996), Elements of themicrophysical structure of numerically simulated nonprecipitating strato-cumulus, J. Aerosol Sci., 53, 980–1006.

Tervahattu, H., K. Hartonen, V. Kerminen, K. Kupiainen, P. Aarnio,T. Koskentalo, A. F. Tuck, and V. Vaida (2002a), New evidence of anorganic layer on marine aerosols, J. Geophys. Res., 107(D7), 4053,doi:10.1029/2000JD000282.

Tervahattu, H., J. Juhanoja, and K. Kupiainen (2002b), Identification of anorganic coating on marine aerosol particles by TOF-SIMS, J. Geophys.Res., 107(D16), 4319, doi:10.1029/2001JD001403.

Twomey, S. (1977), Minimum size of particle for nucleation in clouds,J. Atmos. Sci., 34, 1832–1835.

Tzivion, S., T. Reisin, and Z. Levin (1994), Numerical simulation of hygro-scopic seeding in a convective cloud, J. Appl. Meteorol., 33, 252–267.

Whitby, K. T. (1978), The physical characteristics of sulfur aerosols, Atmos.Environ., 12, 135–159.

"""""""""""""""""""""""J. Medina and A. Nenes, School of Chemical and Biomolecular

Engineering, Georgia Institute of Technology, 331 Ferst Drive, Atlanta,GA 30332-0340, USA. ([email protected])


14 of 14

D20207

Documents

Effects of film-forming compounds on the growth of giant cloud …nenes.eas.gatech.edu/Reprints/GCCN_JGR.pdf · 2004-10-25 · Effects of film-forming compounds on the growth of giant