foam droplets generated from natural and artificial seawaters

2 Foam droplets generated from natural and artificial

3 seawaters

4 Corey A. Tyree,1 Virginie M. Hellion,1,2 Olga A. Alexandrova,3 and Jonathan O. Allen1,3

5 Received 30 June 2006; revised 2 February 2007; accepted 2 March 2007; published XX Month 2007.

6 [1] Submicrometer sea salt aerosol (SSA) particles are routinely observed in the remote7 marine boundary layer (MBL); these aerosols include cloud condensation nuclei and8 so affect the earth’s radiative balance. Here foams designed to mimic oceanic whitecaps9 were generated in the laboratory using a range of bubbling flow rates and aqueous media:10 unfiltered seawater, filtered seawater, artificial seawater, and mixtures of filtered and11 artificial seawater. The number and sizes of dried foam droplets in the particle diameter,12 Dp, range 15–673 nm were measured. Particle size distributions for natural and artificial13 seawaters were unimodal with a dN/d logDp mode at Dp � 100 nm (�200 nm at14 80% RH). The foam droplet mode falls within the range of reported mean diameters15 (Dp = 40–200 nm) for submicrometer SSA particles observed in the remote MBL. The16 present laboratory results were scaled up to estimate submicrome ter SSA particle17 fluxes; this extrapolation supports the hypothesis that foam droplets are the most important18 source of SSA particles by number. The foam droplet flux from the oceans was estimated19 to be 980 cm�2 s�1 for a fractional white cap coverage, W of 0.2%. These results20 compared well with foam droplet fluxes reported elsewhere. The origins of variability in21 foam droplet fluxes were also evaluated. Natural organic matter affected foam droplet22 flux by a factor of 1.5; this was less than (1) the effect of bubbling flow rate on foam23 droplet flux (factor of 5) and (2) the uncertainty in W (factor of 3–7).

24 Citation: Tyree, C. A., V. M. Hellion, O. A. Alexandrova, and J. O. Allen (2007), Foam droplets generated from natural and artificial

25 seawaters, J. Geophys. Res., 112, XXXXXX, doi:10.1029/2006JD007729.

27 1. Introduction

28 [2] In the remote marine boundary layer (MBL), aerosol29 particles with diameters at ambient relative humidity (RH)30 less than 1 mm are an abundant and climatologically impor-31 tant class of particles. Submicrometer particles affect the32 radiative balance in the remote marine atmosphere directly33 by scattering light [Schwartz, 1996; Murphy et al., 1998;34 Quinn et al., 1998], and indirectly by acting as cloud35 condensation nuclei (CCN) [O’Dowd et al., 1997; Murphy36 et al., 1998;Mason, 2001; Pierce and Adams, 2006]. Greater37 than 90% of particles in the MBL are submicrometer in38 diameter [Fitzgerald, 1991; Jaenicke, 1993] and their chem-39 ical composition indicate two main sources, sea salt40 [O’Dowd and Smith, 1993; Bates et al., 1998] and non-41 seasalt (nss) sulfate produced from the oxidation of gas-42 phase sulfur species [Nguyen et al., 1983; Charlson et al.,43 1987; Sievering et al., 1992; Capaldo et al., 1999]. Wind44 stress on the ocean surface produces whitecap foams, which

45are the main source of sea salt aerosol (SSA) particles46[Paterson and Spillane, 1969; Blanchard, 1983; Lewis and47Schwartz, 2004]. Here we describe laboratory experiments48examining the effect of seawater composition on the diam-49eter and flux of SSA particles generated from foams like50those at sea. These experiments are designed to supplement51particle measurements in the MBL, from which the effects of52seawater composition on particle generation cannot be53determined directly.54[3] SSA particles in the MBL usually exist as liquid drops55whose diameters depend on the water content, and so on56ambient RH. The diameter of a particle at ambient RH is57Dp,1. RH is typically 80% in the MBL. Since the mole58fraction of water in salinity 33 seawater is 0.98, the relative59humidity in equilibrium with a freshly formed SSA particle60is 98% [Lewis and Schwartz, 2004, p. 53]. The diameters of61SSA particles at RH = 80 and 98%, Dp80 and Dpo, respec-62tively, are related to their dry particle diameter by Dp � 1

263Dp80 � 1

4Dpo [Fitzgerald, 1975; Lewis and Schwartz, 2004,

64p. 53]. Submicrometer marine particles are usually grouped65by dry diameter in the Aitken (Dp = 25–85 nm), accumu-66lation (Dp = 85–250 nm), and coarse (Dp � 250 nm) modes67[Bates et al., 1998; Heintzenberg et al., 2004]. An ultrafine68mode (Dp = 5–25 nm), consisting of freshly nucleated69particles, is also sometimes observed [Bates et al., 1998;70O’Dowd et al., 2001; Heintzenberg et al., 2004]. Number71concentrations are often dominated by the ultrafine and72Aitken size modes [Bates et al., 1998; Heintzenberg et al.,

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, XXXXXX, doi:10.1029/2006JD007729, 2007ClickHere

for

FullArticle

1Chemical Engineering Department, Arizona State University, Tempe,Arizona, USA.

2Now at Ecole Nationale Sup’erieure d’Electrochimie et d’Electro-metallurgie, Institut National Polytechnique de Grenoble, Grenoble, France.

3Civil and Environmental Engineering Department, Arizona StateUniversity, Tempe, Arizona, USA.

Copyright 2007 by the American Geophysical Union.0148-0227/07/2006JD007729$09.00

XXXXXX 1 of 17

http://dx.doi.org/10.1029/2006JD007729

73 2004]. MBL aerosol particles with a dry diameter, Dp 574 80 nm are generally considered to be CCN [O’Dowd et al.,75 1997; Murphy et al., 1998; Chuang et al., 2000; Pierce and76 Adams, 2006]. Because ultrafine and Aitken mode particles77 are often too small to be CCN, accumulation mode particles78 constitute a significant fraction of marine CCN [O’Dowd and79 Smith, 1993; Chuang et al., 2000; Yoon and Brimblecombe,80 2002].81 [4] Submicrometer sea salt has been observed in the82 Aitken [Clarke et al., 2003], accumulation [O’Dowd and83 Smith, 1993; Quinn et al., 1998; Bates et al., 1998], and84 coarse modes [Bates et al., 1998; Huebert et al., 1998;85 Campuzano-Jost et al., 2003]. New nss-sulfate particles can86 form when the gas phase oxidation products of dimethyl87 sulfide (DMS) or SO2 nucleate to form ultrafine particles88 [Clarke et al., 1998]. Initially, in the ultrafine mode, these89 nss-sulfate particles can grow via heterogeneous processes90 to form Aitken [Fitzgerald, 1991; Clarke et al., 1998] and91 accumulation mode particles [Ayers et al., 1991; Hegg and92 Hobbs, 1992]. Mixtures of sea salt and nss-sulfate can form93 when gas phase sulfur-containing species condense on94 existing sea salt particles and cloud droplets [Hoppel et al.,95 1989; Fitzgerald, 1991; Sievering et al., 1992; Clarke and96 Porter, 1993; O’Dowd et al., 1997]. Because SSA particles97 are originally alkaline, nss-sulfate uptake and formation are98 enhanced on sea salt particles [O’Dowd and Smith, 1993;99 McInnes et al., 1994; Katoshevski et al., 1999; Laskin et al.,100 2003]. SSA 65 particles therefore play a dual role in the101 marine atmosphere; they are a large direct source of sub-102 micrometer particles [O’Dowd and Smith, 1993] and also103 affect the formation of nss-sulfate [Sievering et al., 1992;104 Clarke and Porter, 1993].105 [5] SSA particles can be identified by their Na content106 since Na is conserved in atmospheric particles [McInnes107 et al., 1996; Clark et al., 2001; Campuzano-Jost et al.,108 2003]. Na+ comprised approximately 30% of the ionic109 species mass of submicrometer particles near Cape Grim;110 since Na+ is 31% of sea salt mass, in this case sea salt made111 up nearly all of the submicrometer particle ionic mass112 [Huebert et al., 1998]. Sea salt comprised approximately113 80% of the ionic mass in submicrometer particles in the114 remote Southern Ocean MBL [Quinn et al., 1998]. Measure-115 ments of marine particle ionic composition are consistent116 with thermal volatility measurements that show a SSA117 particle mode at Dp � 200 nm [O’Dowd and Smith, 1993];118 for example, sea salt comprised 58% of the accumulation119 mode ionic mass (number geometric mean dry diameter,120 �Dp;g = 110 nm) and greater than 99% of coarse mode ionic121 mass (�Dp;g = 540 nm) collected over the Southern Ocean122 region [Bates et al., 1998].123 [6] Particles produced from single bubble bursting have124 been extensively studied in the laboratory [Blanchard,125 1963; Day, 1964; Resch and Afeti, 1991; Spiel, 1998].126 Particles are formed when single bubbles rise to the surface;127 the bubble film drains until the film thickness is on the order128 of 1 mm [Spiel, 1998] and then the bubble film shatters. A129 toroid composed of film material breaks apart to form130 primary ‘‘film droplets’’ [Spiel, 1998]. Secondary film131 droplets appear to be produced from impaction of the larger,132 downward moving primary film droplets. Milliseconds after133 the bubble bursts, an unstable jet can form from the collapse

134of the bubble cavity, which breaks apart to form ‘‘jet drops’’135[Blanchard, 1989].136[7] Jet drops are a likely source of SSA particles with137Dp > 1 mm. The number of jet drops produced from single138bubble bursting decreases with increased bubble diameter.139Bubbles larger than 3 mm produced only one jet drop while140300 mm bubbles produced five to six jet drops [Cipriano141and Blanchard, 1981]. Jet drop diameters at formation were142approximately one-tenth the diameter of their parent bubble143diameter [Blanchard and Woodcock, 1957]. Submicrometer144jet drops would have to be produced from bubbles smaller145than 20 mm. However, bubbles of this diameter will not146surface in large numbers since they are expected to dissolve147before reaching the surface. For example, bubbles smaller148than 100 mm rising in 0.3 m of seawater dissolve completely149prior to reaching the surface [Blanchard and Woodcock,1501980]. Though measurements of jet drop diameter generally151cover only the supermicrometer size range [Lewis and152Schwartz, 2004, p. 193], jet drops are an unlikely source153of submicrometer SSA particles.154[8] Film droplets can also be a source of SSA particles155with Dp > 1 mm [Lewis and Schwartz, 2004, p. 204]. The156number of film droplets formed per bubble is dependent on157bubble diameter [Blanchard and Woodcock, 1957; Resch158and Afeti, 1991; Spiel, 1998]. Bubbles smaller than � 2 mm159produced no film droplets [Resch and Afeti, 1991; Spiel,1601998] while bubbles larger than �2 mm produced 100 to161200 film droplets [Spiel, 1998]. Although many authors162have suggested that film droplets can be submicrometer in163diameter [Mason, 1957; Cipriano et al., 1983; Blanchard164and Syzdek, 1988], film droplets in the range Dp � 1 mm are165largely uncharacterized.166[9] Even if single bubble bursting produced SSA particles,167the contribution of single bubble bursting to SSA particles is168expected to scale with the frequency of these events. An169alternative source of submicrometer SSA particles is foam170bubble bursting. At sea, bubble coalescence following air171entrainment leads to whitecap foam formation. Foam forma-172tion is always observed following wave breaking according173to Blanchard [1983], who estimated foam bubbles out-174number single bubbles by a factor of 10. In addition to175outnumbering single bubbles, foam bubble bursting also176produces many more droplets than does an equivalent num-177ber of single bubbles [Paterson and Spillane, 1969]. On the178basis of the relatively high SSA particle production from179foam bubbles and their ubiquity at sea, Paterson and Spillane180[1969] concluded that ‘‘the overwhelming contributor to the181sea salt aerosol population is believed to be the rafts of foam182produced by breaking waves.’’183[10] A few investigators have studied the effect of foam184formation on SSA particle production [Garrett, 1968;185Cipriano and Blanchard, 1981; Monahan et al., 1982;186Woolf et al., 1987; Martensson et al., 2003; Sellegri et al.,1872007]. The particles produced from foam bubble bursting188are referred to here as ‘‘foam droplets.’’ Woolf et al. [1987]189simulated whitecap formation by colliding two parcels of190water and measured foam droplet size distributions over the191range Dp = 0.25–10 mm. The smallest particle size bin, Dp =192250–750 nm, contained the largest number of foam drop-193lets. Hoppel et al. [1989] generated foam droplets by194bubbling through a frit in seawater and measured foam195droplet size distributions over the range Dp = 12–800 nm.

XXXXXX TYREE ET AL.: SEAWATER FOAM DROPLETS

2 of 17

XXXXXX

196 Foam droplets had a dN/dDp mode at Dp � 70–110 nm.197 Martensson et al. [2003] generated foam droplets by bub-198 bling through a frit in salinity 33% artificial seawater and199 measured foam droplet size distributions over the range200 Dp = 0.02–5 mm. At a seawater temperature of 25�C, foam201 droplets had a bimodal number size distribution with dN/d202 logDp modes in the ranges Dp = 80–120 nm and Dp = 1.5–203 2.5 mm. At 15�C, foam droplets had a bimodal number size204 distribution with dN/d logDp modes in the ranges Dp = 50–205 60 nm and Dp = 1.5–2.5 mm. These submicrometer foam206 droplet dN/d logDp modes fall within the range of mean dry207 diameters reported for submicrometer SSA particle modes,208 Dp = 40–200 nm [O’Dowd and Smith, 1993; Murphy et al.,209 1998; Nilsson et al., 2001; Clarke et al., 2003].210 [11] It has been hypothesized that variability in submi-211 crometer SSA particle flux estimates could be explained in212 part by spatial variation in the amount and nature of213 seawater organic matter [Reid et al., 2001; Geever et al.,214 2005]. Seawater organic matter consists of a complex215 mixture of dissolved and particulate species, including216 lipids, amino acids, amino sugars, cells, and cell fragments.217 Dissolved organic matter (DOM) and particulate organic218 matter (POM) concentrations vary greatly with depth, time,219 and location [Millero and Sohn, 1992]. Dissolved organic220 carbon (DOC) concentrations in the surface mixed layer,221 which extends to a depth of about 100 m, range from about222 1 mg Cl�1 [Benner et al., 1992] in the open ocean to 10 mg223 Cl�1 in coastal areas [Sharp et al., 1993]. POM concen-224 trations are about one-fifth that of DOC [Millero and Sohn,

2251992]. Under conditions that whitecap foams form, the226bubble upwelling region causes an outflow of water and227removal of the surface microlayer above the rising bubble228plume [Blanchard, 1983]. Thus it is the seawater organic229matter in the bubble entrainment zone that is most likely to230affect foam droplet production.231[12] Laboratory studies of the effect of organic matter on232droplet production have been inconclusive. In single bubble233experiments, organic matter has been shown to increase234droplet production [Blanchard, 1963], decrease droplet pro-235duction [Day, 1964;Paterson and Spillane, 1969], or have no236effect at all [Morelli et al., 1974]. Single bubble experiments237have shown that organic matter in natural seawater can238suppress film droplet production [Day, 1964; Paterson and239Spillane, 1969]. Garrett [1968] showed that bubbling240through fresh seawater resulted in a 25% decrease in foam241droplet production relative to a 30%NaCl solution. Attempts242to characterize the effect of organic matter on SSA particle243production also have been made by adding surface-active244organic species to artificial seawater. Oleic acid is often245selected as a model surface-active compound in laboratory246SSA particle experiments because it is a component of both247DOM and POM [Blanchard, 1963; Garrett, 1968; Morelli248et al., 1974; Monahan et al., 1980]. Garrett [1968] showed249that the addition of oleic acid suppressed foam formation and250increased particle production.251[13] In this work, we measure the number and size of252foam droplets produced in a controlled laboratory setting253in order to study the effect of seawater composition and254bubbling conditions on foam droplet production. The effect255of salinity was studied by generating foam droplets from256artificial seawaters with salinities in the range 0–70%. The257effect of organic content was studied by generating foam258droplets from seawater containing natural DOM and POM259(unfiltered seawater), natural DOM (filtered seawater), and260artificial seawater containing surrogate DOM (oleic acid).261The laboratory results are compared with remote, marine262particle size distributions and used to estimate the production263of submicrometer SSA particles from oceanic foams.

2642. Experimental Methods

265[14] Foam droplets were generated by bubbling air through266aqueous media (see Figure 1). Compressed air was filtered267through a high efficiency particulate air (HEPA) filter and268metered using a calibrated rotameter with a precision of2690.1 l min�1. Air then was passed through a stainless steel270tube to a fine-pore diffuser (Aquatic Eco-Systems, Model271ALR80SS, 3 � 1.5 inches, 80 mm mean pore size,272Apopka, FL). In experiments focused on the effects of273subsurface bubble diameter, the fine-pore diffuser was274exchanged for a medium-pore diffuser (Aquatic Eco-Systems,275Model ALR8, 3 � 1.5 inches, 140 mm mean pore size,276Apopka, FL). The height and diameter of the glass column277were 60 and 15 cm, respectively. The glass columnwas capped278by a cone-shaped piece of aluminum foil. The cap was formed279to the glass surface of the column and the aerosol particle-280sampling inlet. The cap was not sealed. Excess air was vented281from the headspace so that background aerosol was not282sampled. The separation between the water level and the apex283of the cone-shaped cap was 25–30 cm over the range of284bubbling flow rates studied; note that jet drops are ejected to a

Figure 1. Schematic diagram of foam droplet apparatus.Compressed air was filtered through a high efficiencyparticulate air (HEPA) filter and metered using a rotameter(F), and then passed through a diffuser to generate bubbles.Aerosol was sampled through a diffusion drier. Particle sizedistributions were measured using a Scanning MobilityParticle Sizer (SMPS).


3 of 17

XXXXXX

285 maximum height of 20 cm [Spiel, 1997]. The displacement286 height, and so the void fraction in the column, was measured287 for each bubbling flow rate.288 [15] The diffuser was placed approximately 1 cm above the289 bottom of the glass column so that bubble flow was unaf-290 fected by the column bottom. The column was filled with291 aqueous media to a height of 40.6 cm (±0.2 cm), which292 corresponds to a volume of 7.2 l. The bubble rise distance in293 the column was then 32–39.5 cm, which approximates the294 natural circulation depth of oceanic bubbles [Lamarre and295 Melville, 1992;Woolf, 1997]. Entrained air bubbles can reach296 depths of several meters [Thorpe, 1986], but the majority of297 entrained air is located within �50 cm of the sea surface298 [Lamarre andMelville, 1992]. Note that the circulation depth299 utilized here is an order of magnitude greater than that used in300 prior experiments [Garrett, 1968; Martensson et al., 2003;301 Sellegri et al., 2007].302 [16] Any effect of organic matter on SSA particle pro-303 duction is likely due to organic matter which partitions to304 the air-water interface of the bubble during ascent. Some305 time is necessary for this interface to reach steady state. The306 bubble residence time for this apparatus was �5 s. Similar307 to the characteristic bubble residence times of oceanic308 bubble plumes, which evolve rapidly beneath breaking309 waves, is on the order of 1–5 s [Graham et al., 2004].310 This residence time is significantly shorter than that esti-311 mated using a Stokes terminal velocity; a plume of bubbles312 causes upwelling flow which in turn causes bubbles in a313 plume to rise faster than single bubbles in still water.314 [17] The aerosol sampling inlet was a 0.64 cm i.d. copper315 tube placed approximately 15 cm above the top of the foam316 layer. The aerosol was sampled at 300 cm3 min�1, corre-317 sponding to laminar flow in the sampling tube. Foam318 droplets were dried using a diffusion dryer, which had a319 length of 30 cm, an annular diameter of 2.5 cm, and a320 residence time of approximately 30 s. In the dryer, aerosol321 flow was surrounded by CaSO4 dessicant separated by a322 polypropylene mesh tube. The desiccant was replaced after323 approximately 2 hours of use. The total residence time of the324 foam droplets in the transport line was approximately 35 s.325 The calculated RH at the exit was 20% [Tyree and Allen,326 2004], which is below the crystallization point, 45%, for327 NaCl [Biskos et al., 2006].328 [18] Sizes of the dried particles were measured using a329 TSI (St. Paul, MN) Model 3080 Scanning Mobility Particle330 Sizer (SMPS) equipped with a Model 3081 Long Differen-331 tial Mobility Analyzer and a Model 3025A Condensation332 Particle Counter. The SMPS system recorded the concen-333 tration of particles with mobility diameters in the range334 14.9–673 nm. Four 2.5-min measurements (one scan per335 measurement) were recorded for most experiments; only336 two 2.5-min measurements were recorded for artificial337 seawater experiments using salinities of 1, 10, and 20%.338 Following a change in airflow or aqueous media, the339 apparatus was run for at least 5 min in order to purge the340 system before acquiring data.341 [19] The apparatus was thoroughly cleaned in order to342 minimize contaminants including naturally occurring organic343 species and detergents. The glass column was first rinsed344 with water in order to remove visible salt residues. The345 column was cleaned with two sequential washes of 0.5 l of346 Alkanox detergent solution, followed by two sequential

347washes with deionized (DI) water. The column was then348rinsed with three sequential washes of 50 mL of isopro-349panol (Burdick and Jackson, High Purity Solvent, 99.9+%)350and three sequential washes of 50 mL of dichloromethane351(DCM) (Burdick and Jackson, GC/GC-MS grade,35299.9+%); the column was allowed to dry in a fume hood.353The diffuser stone was rinsed with DI water, 50 mL of354isopropanol, and DI water again. The diffuser was then355sonicated in a precleaned beaker containing DI water for35610 min and rinsed with DI water. For the final cleaning357step, the diffuser was placed in a precleaned beaker358containing DI water, where air was bubbled through the359diffuser for at least 10 min. The DI water was changed and360the process repeated twice. Stainless steel tubing and tube361fittings were cleaned with 5 mL of detergent solution, rinsed362with DI water, and rinsed with three sequential 5 mL washes363of isopropanol and then three sequential 5-mL washes of364DCM. The interior surface of the aluminum cap was rinsed365with three sequential 5-mL washes of isopropanol and then366three sequential 5-mL washes of DCM.367[20] The particle size distributions of foam droplets were368measured for a range of aqueous media and bubbling369superficial velocities (see Table 1). Artificial seawater was370prepared from analytical grade salts and DI water. The salt371mixture contained, by mass, 87.9% NaCl (Fluka, ACS372reagent grade, 99.5+%), 11.7% MgSO4 � 7H2O (Sigma-373Aldrich, ACS reagent grade, 98+%), and 0.4% NaHCO3

374(Sigma-Aldrich, ACS reagent grade, 99.7+%). This salt375mixture includes Na+, Cl�, Mg2+, and SO4

2�, which together376account for 97% of the inorganic salt mass in seawater377[Horne, 1969]. NaHCO3 served to buffer the pH of the378artificial seawater [Pilson, 1998]. The solution pH was 7.8,379which is comparable to the typical surface seawater pH of3808.2. The artificial sea salt mixture was prepared so that the381ionic mass ratios of Na+, Cl�, Mg2+, and SO4

2� were382comparable to that in seawater. The mass ratios (and their383corresponding mass ratios in seawater) were Cl� = Na+ =3841.54 (1.80), SO4

2� =Mg2+ = 3.95 (2.11), SO42�/Cl� = 0.09

385(0.14), and Mg2+=Na+ = 0.033 (0.12). Artificial seawaters386were prepared with the following salinities (±0.05%):38770, 33, 20, 10, 1, and 0%; these media were designated388AS70, 245 AS33, AS20, AS10, AS1, and AS0 (pure DI389water), respectively.390[21] Natural seawater was collected off the Scripps Insti-391tute of Oceanography (SIO) Pier (32 ± 520N, 1170160W) on3929 June 2004 and 20 February 2006. Water was siphoned393through a tube suspended approximately 20 cm beneath a394surface float to avoid sampling of the surface microlayer.395Seawater was stored in precleaned brown glass bottles396maintained at 0�C. A portion of each seawater sample397was filtered using 0.7 mm pore Whatman GF/F glass fiber398filters in a separatory funnel to remove POM. Prior to399filtration, the filters were cleaned using sequential isopro-400panol and DCM washings aided by sonication for 30 min.401Because foam droplet production may be dependent on402water temperature [Martensson et al., 2003], seawater was403warmed to at least 20�C before experimentation. The404seawater temperature was maintained at 20�–25�C for405each experiment. All natural seawater experiments were406completed within 48 hours of seawater collection.407[22] For the June 2004 seawater sample, concurrent in408situ temperature and salinity measurements were provided


4 of 17

XXXXXX

409 by the Network for Environmental Observations of the410 Coastal Ocean (NEOCO). The NEOCO instrumentation411 includes a Sea-Bird Seacat CTD instrument located 2 m412 below the mean low water level. The temperature and413 salinity of the near surface were 20.3�C and 33.47%.414 Concurrent chlorophyll fluorometric measurements at the415 SIO Pier were provided by the University of California416 Coastal Environmental Quality Initiative. The seawater417 had a chlorophyll concentration of 1.8 mgm�3. For the418 February 2006 seawater sample, concurrent measurements419 were provided by the Southern California Coastal Ocean420 Observing System (SCCOOS). The instrumentation is part421 of the SCCOOS automated shore station at the SIO pier,422 which includes a suite of automated sensors located at the423 water surface. The temperature, salinity, and chlorophyll424 concentration were 14.5�C, 33.31%, and 0.10 mg m�3.425 [23] Filtered seawater and an artificial seawater blank426 were analyzed for DOC. These samples were acidified with427 quartz-distilled hydrochloric acid to pH 2 and the dissolved428 inorganic carbon (DIC) removed by sparging with nitrogen429 (ultra-high purity grade) for 5 min [Sharp, 2002]. Samples430 were stored at 0�C and the DOC content was determined431 within 48 hour of seawater collection using a Shimadzu432 TOC-V Total Organic Carbon Analyzer. 271 The DOC433 concentrations were 3.07 and 2.28 mg Cl�1 for the June434 2004 and February 2006 samples, respectively. The DOC435 concentration in AS33 was 0.05 mg Cl�1.436 [24] In order to study the effect of natural organic matter437 on foam droplet production, foam droplets were generated438 from aqueous media with a range of organic content. These439 aqueous media included 100% filtered seawater (FS100)440 and 100% unfiltered seawater (UFS100). Solutions with441 lower concentrations of natural DOM were made by dilut-442 ing filtered seawater with salinity 33% artificial seawater;443 these were composed of 10% filtered seawater and 90%444 artificial seawater by volume (FS10) and 1% filtered sea-445 water and 99% artificial seawater (FS1). Foam droplet446 experiments using FS100 were conducted with June 2004

447and February 2006 seawater samples. All other natural448seawater media (i.e., FS1, FS10, and UFS100) used only449the June 2004 seawater sample.450[25] In order to isolate the effect of a surrogate DOM451compound on foam droplet production, foam droplets were452generated from aqueous media with a range of oleic acid453concentrations (Fisher Chemical, NF/FCC grade). Solu-454tions were made from salinity 33% artificial seawater with455oleic acid concentrations of 0.1mg l�1 (0.08mgCl�1, OA0.1),4561 mg l�1 (0.8 mg Cl�1, OA1), and 10 mg l�1 (8 mg Cl�1,457OA10). The oleic acid was added to AS33 in the column and458mixed using the bubbling airflow for at least 5 min prior to459measurement.460[26] In order to compare our results with those from other461experiments, it is convenient to express the bubbling flow462rate as a superficial gas velocity, vg, which is equal to the463bubbling flow rate ( _V air) divided by the column area (A). A464standard bubbling superficial velocity, vg = 1.2 cm s�1, was465used for all aqueous media and diffusers. Foam droplets were466also generated from the AS33, AS70, FS100, and UFS100467using vg = 0.090, 0.33, and 1.6 cm s�1. These velocities were468selected because observations showed foam morphologies469were different for each of these vg.470[27] In order to isolate the effect of subsurface bubble471diameter on foam droplet diameter and production, foam472droplets were generated from AS33 using a fine-pore and473medium-pore diffuser stone. Photographs of the bubble474plume produced by the fine-pore and medium-pore diffusers475at a water depth of 5–15 cm were analyzed to determine the476diameter of bubbles produced. The diameters of 200 bubbles477were determined with a measurement precision of ±50 mm by478comparing the bubble width to the known width of a479submerged stainless steel tube.

4803. Results

4813.1. Subsurface Bubbles

482[28] Subsurface bubbles rise through the water column483and coalesce into a foam at the surface. The connection is

t1.1 Table 1. Summary of Foam Droplet Size Distribution Results for a Range of Superficial Bubbling Velocities (vg) and Aqueous Mediaa

vg = 0.090 cm s�1 vg = 0.33 cm s�1 vg = 1.2 cm s�1 vg = 1.6 cm s�1t1.2

Aqueous MediaNtotal,cm�3

�Dp;g,nm sg

Ntotal,cm�3

�Dp;g,nm sg

Ntotal,cm�3

�Dp;g,nm sg

Ntotal,cm�3

�Dp;g,nm sgt1.3

AS70: 70% artificial seawater – – – – – – 24,000 98.8 1.77 – – –t1.4AS33: 33% artificial seawater (Oct 2004) 7,070 86.0 1.61 7,100 91.0 1.60 17,000 97.8 1.70 23,300 99.7 1.72t1.5AS33: 33% artificial seawater (Jan 2006) 6,830 89.9 1.57 6,450 97.3 1.62 16,800 105 1.80 22,200 114 1.82t1.6AS33-MP: 33% artificial seawater

medium-pore diffuser3,390 83.4 1.70 4,000 89.2 1.75 5,610 89.4 1.77 6,190 89.4 1.80t1.7

AS20: 20% artificial seawater – – – – – – 12,100 90.1 1.67 – – –t1.8AS10: 10% artificial seawater – – – – – – 9,980 86.1 1.73 – – –t1.9AS1: 1% artificial seawater – – – – – – 2,650 62.5 1.75 – – –t1.10AS0: DI water 40.3 58.5 2.04 – – – 0 NA NA – –t1.11OA10: 10 mg l�1 oleic acid in AS33 12,500 94.5 1.66 – – – 31,100 84.0 1.82 – – –t1.12OA1: 1 mg l�1 oleic acid in AS33 12,100 94.1 1.66 – – – 31,700 85.0 1.73 – – –t1.13OA0.1: 0.1 mg l�1 oleic acid in AS33 10,500 92.3 1.70 – – – 22,400 89.4 1.71 – – –t1.14FS100: 100% filtered seawater collected

Jun 2004, 3.07 mg Cl�14,770 86.9 1.64 7,510 96.3 1.54 17,500 102 1.66 23,200 105 1.69t1.15

FS100: 100% filtered seawater collectedFeb 2006, 2.28 mg Cl�1

6,730 89.4 1.61 9,060 96.3 1.66 17,700 102 1.75 24,000 102 1.77t1.16

FS10: 10% (vol) filtered seawater; 90% AS33 – – – – – – 16,800 96.8 1.69 – – –t1.17FS1: 1% (vol) filtered seawater; 99% AS33 – – – – – – 16,900 96.8 1.68 – – –t1.18UFS100: 100% unfiltered seawater

collected Jun 20046,390 89.0 1.67 10,400 93.8 1.58 18,700 99.3 1.65 26,700 103 1.71t1.19

aNtotal is total number concentration, �Dp;g is dried number geometric mean diameter, and sg is geometric standard deviation.t1.20


5 of 17

XXXXXX

484 therefore indirect between the diameter, Db, and concentra-485 tion, Nb, of subsurface bubbles and the nature of the foam486 from which particles are produced. Nevertheless, the appa-487 ratus used here was designed to generate subsurface bubbles488 comparable in Db and Nb to those observed under breaking489 waves. Db was measured optically using two different490 diffusers at two bubbling superficial velocities, vg, in491 AS33 (salinity 33% artificial seawater). The fine-pore dif-492 fuser produced bubbles with mean Db, �Db = 230 mm493 (standard deviation, s = 80 mm) for vg = 0.33 cm s-1 and494 �Db = 250 mm (s = 95 mm) for vg = 1.6 cm s�1 (see Figure 2).495 The medium-pore diffuser produced bubbles with �Db =496 1000 mm (s = 310 mm) for vg = 0.33 cm s�1 and �Db =497 950 mm (s = 330 mm) for vg = 1.6 cm s�1. Thus bubble498 diameters were approximately independent of vg.499 [29] Nb was determined for vg = 0.33 cm s�1 using Db

500 and the void fraction measurements; these bubble size501 distributions were compared to oceanic bubble size distri-502 butions summarized by Lewis and Schwartz [2004, p.239]503 [Kolovayev, 1976; Walsh and Mulhearn, 1987; Bowyer,504 2001; de Leeuw and Cohen, 2002] (see Figure 2). These505 measurements have shown that the oceanic bubble size506 distribution spans the range Db = 50–800 mm. But as noted507 previously, the bubbles in the range Db = 50–100 mm are508 expected to dissolve as they circulate beneath a breaking509 wave [Blanchard and Woodcock, 1980]. Therefore bubbles510 in the range Db = 100–800 mm are the most relevant to SSA511 particle production. The fine-pore diffuser produced bubbles512 in the range Db = 60–800 mm. The medium-pore diffuser513 produced bubbles in the rangeDb = 550–2000 mm. Since the514 Db range produced by the fine-pore diffuser more closely515 matches that found at sea, the fine-pore diffuser was used for516 all other experiments.517 [30] Note that the oceanic Nb shown in Figure 2 generally518 included periods with no wave breaking. In order to

519compare the subsurface Nb measured here with those in520the ocean, the laboratory Nb should be multiplied by521fractional whitecap coverage, W. Since the average W is522on the order of 0.01 [Monahan, 1986], the number of523bubbles generated from the fine-pore diffuser for vg =5240.33 cm s�1 is comparable to the high end (top of shaded525band in figure) of the estimated range for oceanic bubble526concentrations. Also note that since the bubble diameter is527independent of vg, Nb for vg = 1.6 cm s�1 is approximately5285 times higher than for vg = 0.33 cm s�1.529[31] TheNb andDb generated from the fine-pore diffuser in530media containing natural organic matter were similar to those531measured for AS33. In contrast, the addition of oleic acid to532artificial seawater caused a dramatic increase in Db which533was dependent on oleic acid concentration. Many of the534bubbles in OA0.1 (salinity 33% artificial seawater with5350.1 mg l�1 oleic acid added) had Db � 250 mm, compa-536rable to the bubbles present in AS33; however, some of537the bubbles coalesced near the diffuser to form bubbles538with Db > 0.2 cm. In OA10 nearly all bubbles coalesced to539form bubbles with Db � 0.5 cm. Though the effect of540salinity on Db was not quantified, observations indicated541that Db decreased slightly with increased artificial seawa-542ter salinity from 1 to 70%. Bubbles in AS0 were much543larger, Db � 0.5 cm.

5443.2. Foam Characterization

545[32] Bubbling through liquids with varying salinity, organic546content, and vg produced varying 341 foam morphologies,547from surfaces with no foam to those with continuous foam548greater than 1 cm thick. Foamwas formed in every experiment549except the AS0 (DI water) and OA10 experiments. Since pure550liquids do not foam [Bikerman, 1973], the absence of foam551when bubbling through AS0 suggested that the apparatus was552free of contaminants. Foam formation was observed upon the

Figure 2. Bubble size distributions in salinity 33% artificial seawater (AS33) produced by fine-poreand medium-pore diffusers using a bubbling superficial velocity (vg) of 0.33 cm s�1. The shaded bandencompasses the majority of the oceanic bubble size distributions reported for depths �1.5 m and windspeeds 10 m above the sea surface (U10) greater than 10 ms�1 [Lewis and Schwartz, 2004, p. 241].


6 of 17

XXXXXX

553 addition of small amounts of sea salt (AS1). Increased salinity554 from 1 to 70% caused only small increases in foam thickness555 and foam bubble diameter.556 [33] Foam thickness in AS33was dependent on vg. For vg =557 0.090 cm s�1, the surface was approximately 90% covered by558 rafts of foam bubbles. In most cases, the foam bubble559 diameters were approximately 0.5 cm, indicating that bubble560 coalescence occurred at the surface. The coverage and561 thickness of the foam was enhanced as vg was increased to562 0.33 cm s�1. For vg = 0.33 cm s�1, the surface was covered by563 aged foam and an active upwelling region; this upwelling564 region appeared intermittently where a stream of bubbles565 breached the surface. For vg = 1.2 and 1.6 cm s�1, thicker566 foam layers formed and the active upwelling region appeared567 less frequently.568 [34] The presence of organic matter in the aqueous media569 produced noticeable changes in foam morphology. The570 appearance of foams formed from artificial seawater can571 be described as ‘‘soda pop like,’’ since these contained572 bubbles with relatively short surface lifetimes and nearly573 uniform size. In contrast, the appearance of natural seawater574 foams can be described as ‘‘beer-like,’’ since these575 contained bubbles with relatively long surface lifetimes576 and heterogeneous sizes. The presence of organic matter577 in the aqueous media also produced changes in foam578 thickness. For vg = 0.090 and 0.33 cm s�1, FS100 (100%579 filtered seawater) and UFS100 (100% unfiltered seawater)580 foams were similar in thickness to those of AS33. However,581 FS100 and UFS100 produced coarser foams with the largest582 bubbles having diameters on the order of 1 cm. For vg = 1.2583 and 1.6 cm s�1, FS100 and UFS100 foams were noticeably584 thicker than AS33 foams. A foam layer was maintained across585 the entire liquid surface during experiments with media con-586 taining natural organicmaterial and vg� 0.33 cm s�1. UFS100587 and FS100 foams were similar in morphology. June 2004588 FS100 and February 2006 FS100 foams were also similar in589 morphology. The dilution of dissolved organic matter content590 had no noticeable effect on foammorphology and thickness, as591 foams produced from FS100, FS10, and FS1 were similar.592 [35] The effect of oleic acid on foam formation was593 concentration dependent. Foams formed on OA0.1 and594 OA1 contained a mixture of bubble sizes and were slightly595 thinner than those on AS33. Foams did not form on OA10;596 instead the large bubbles burst independently at the water597 surface.

598 3.3. Foam Droplet Experiments

599 [36] Preliminary experiments were conducted to investi-600 gate the presence of extraneous particles in the apparatus.601 These particles may have had three sources: (1) back-602 ground particles from laboratory air, (2) particles in the603 bubbling air, and (3) particles from aqueous media impu-604 rities. Particle number concentrations with no bubbling605 flow averaged 551 cm�3; this is the same as the concentration606 in ambient laboratory air and is due to flow of laboratory air607 into the head space of the apparatus. Number concentrations608 were also measured while bubbling air through AS0 in order609 to detect extraneous particles from either the bubbling air or610 aqueous media impurities.611 [37] Total number concentrations, Ntotal, were 40.3 cm�3

612 for vg = 0.090 cm s�1 and 0 cm�3 for vg = 1.2 cm s�1.613 Particle counts recorded for vg = 1.2 cm s�1 were equal to

614those for HEPA384 filtered air. These results suggest that615for vg = 1.2 cm s�1 sufficient excess airflow excluded616ambient aerosol from the headspace while some background617aerosol was sampled for vg = 0.090 cm s�1. Note that the618concentration of background aerosol particles was at least619two orders of magnitude less than those observed for media620other than AS0.621[38] Particle size distribution measurements were made for622many combinations of aqueous media and vg (see Table 1). In623most cases, four particle size distributions were measured624during a 10-min sampling period. The two replicate experi-625ments for AS33 and vg = 1.2 cm s�1 were representative.626Within each experiment, the size distributions for each of four627measurements were similar; each distribution had one dis-628tinct mode atDp� 100 nm andNtotal of each distribution were629within 1% (see Figure 3). This indicates that the particle size630distributions were stable over the 10-min sampling period.631Because particle size distributions were stable and unimodal,632the dried number geometric mean diameter, �Dp;g, geometric633standard deviation, sg, and Ntotal were determined from the634mean of the four measurements. The particle size distribu-635tions were represented by dN/d logDp; references to a peak or636mode in these distributions are based on this representation.637[39] We determined whether experimental results were638equivalent using a two-sample t test on Ntotal. Sample639standard deviations, si, for each experiment (usually four640runs) were comparable and in the range 67–1300 cm�3,641with the exception of OA10 and vg = 1.6 cm s�1 which had642an si of 3700 cm�3. Excluding this experiment, values of si643were not correlated with Ntotal or any experimental param-644eter; therefore the underlying distributions were assumed to645have the same standard deviation. We then calculated646DNtotal as the difference between Ntotal for every run and647the mean for the runs in an experiment excluding the OA10648and vg = 1.6 cm s�1 experiment. The ensemble sample649standard deviation of DNtotal for all runs across all experi-650ments, s0, was 440 cm�3. Below, we refer to experimental651results as ‘‘indistinguishable’’ if the Ntotal were indistin-652guishable at 95% confidence based on a two-sample t test653using s0 = 440 cm�3. For �Dp;g, the s and relative standard654deviation among the runs were 1.1 nm and 1%. Comparing655the two AS33 experiments, �Dp;g values differ by as much as65615 nm for the same vg. Results were not quantitatively657compared based on �Dp;g since lognormal curve fitting658appears to introduce artificial variability.

6593.4. Effect of Salinity

660[40] The effect of salinity on foam droplet production was661examined using artificial seawaters with a range of salinities662and vg = 1.2 cm s�1 (see Table 1). Ntotal of foam droplets663increased from 2650 to 24,000 cm�3 for salinities of 1 to66470%, respectively (see Figure 4). �Dp;g of the particles also665increased from 62.5 to 98.8 nm over the same range of666salinity. The geometric standard deviations were in the range6671.67–1.80 and were comparable for all artificial seawater668solutions.669[41] The effect of vg on Ntotal generated from AS33 was670examined by generating foam droplets using vg = 0.090, 0.33,6711.2, and 1.6 cm s�1. In both AS33 experiments, Ntotal values672were indistinguishable for vg = 0.090 and 0.33 cm s�1. Ntotal

673linearly increased with vg over the range vg = 0.33–1.6 cm s�1

674(see Figure 5).


7 of 17

XXXXXX

675 [42] The effect of subsurface bubble diameter on foam676 droplet production was examined by generating foam677 droplets using a fine- and medium-pore diffuser with678 AS33. Ntotal generated using the medium-pore diffuser679 was less than that generated using the fine-pore diffuser680 for each vg (see Figures 5 and 6). Ntotal generated from the681 medium-pore diffuser increased linearly with vg. Larger682 bubbles from the medium-pore diffuser tended to produce683 foam droplets slightly smaller than those from the fine-pore684 diffuser.

685[43] We investigated the effect of salinity on foam droplet686size at formation (i.e., wet droplet diameter), which can be

calculated from �Dp;g as

�Dpo;g ¼ �Dp;grssmo

rswmd

� �1=3

ð1Þ

689where �Dp;g is the number geometric mean diameter at690formation (98%RH), rss is the density of sea salt (2.2 g cm

�3)691[Lewis and Schwartz, 2004, p. 49], rsw is the density of

Figure 4. Particle size distributions of dried foam droplets generated from artificial seawater withsalinities in the range 0–70%. Bubbles were generated using a fine-pore diffuser and a bubblingsuperficial velocity (vg) of 1.2 cm s�1.

Figure 3. Particle size distributions of dried foam droplets generated using salinity 33% artificialseawater (AS33). Bubbles were generated using a fine-pore diffuser and a bubbling superficial velocity(vg) of 1.2 cm s�1. Lognormal distributions from two experiments shown as lines; four runs were madefor each experiment, these data are shown as symbols.


8 of 17

XXXXXX

692 seawater (1.03 g cm�3 for salinity 33% seawater), andmo/md

693 is the ratio of seawater to solute mass at formation (for694 example, mo/md = 1000/33 for salinity 33%) [Lewis and695 Schwartz, 2004, p. 54]. The geometric mean diameter of696 foam droplets at 80% RH can also be calculated from �Dp;g as

�Dp80;g ¼ �Dp;grss m80

rswmd

� �1=3

ð2Þ

698 where �Dp80;g is the number geometric mean diameter at699 80% RH and m80/md is the ratio of seawater mass to solute

700mass for 80% RH. The diameters at formation and 80%701RH conditions were calculated assuming volume additivity702and that the salinity of the wet droplet was that of the703aqueous media from which the droplets originated. Foam704droplets formed from low salinity media were significantly705larger at formation than those from high salinity media (see706Figure 7).707[44] The number and sizes of foam droplets were mea-708sured over the salinity range 1–70%. The salinity of surface709seawater typically varies over the much narrower range 33–71037% [Millero and Sohn, 1992]. Using a linear interpolation

Figure 5. Total particle number concentration (Ntotal) as a function of bubbling superficial velocity (vg)for different diffusers and aqueous media: salinity 33% artificial seawater (AS33), 100% filteredseawater (FS100), and 100% unfiltered seawater (UFS100). Unless specified a fine-pore diffuser wasused.

Figure 6. Particle size distributions of dried foam droplets generated using fine-pore and medium-porediffusers and a bubbling superficial velocity (vg) of 1.2 cm s�1.


9 of 17

XXXXXX

711 of the present laboratory results for salinities 10–70%, it is712 estimated that Ntotal would vary by less than 5% over the713 oceanic salinity range. AS1 was excluded from the interpo-714 lation since the size distributions of bubbles in such low715 salinity waters appears to differ qualitatively from that with716 salinities greater than �10% [Lewis and Schwartz, 2004,717 p. 249]. Foam droplet size would also not differ significantly718 over the range 33–37%. In some instances, surface seawater719 salinity can be much lower (for example, the Black Sea has a720 salinity of 17%) or slightly higher (for example, the Red Sea721 has a salinity of 40%) than the typical range described. Ntotal

722 of foam droplets produced from salinity 17% seawater723 would be 30% lower than that produced from salinity724 33% seawater.

725 3.5. Effect of Organic Matter

726 [45] The effects of natural DOM and POMon foam droplet727 production were examined using filtered and unfiltered728 seawater. The seawater collected from the SIO Pier in June729 2004 had a DOC concentration of 3.07 mgCl-1, characteristic730 of coastal seawater. A comparison of foam droplet production731 from 100% unfiltered seawater (UFS100) and filtered sea-732 water (FS100) collected in June 2004 demonstrated that the733 presence of POM led to an approximately 35–40% increase734 in Ntotal relative to FS100 for vg = 0.090 and 0.33 cm s�1 (see735 Table 1). The enhanced droplet production for UFS100736 relative to FS100 was smaller (�10%), but significant for737 vg = 1.2 and 1.6 cm s�1 (see Figures 5 and 8). Ntotal generated738 from UFS100 and FS100 increased linearly with vg (see739 Figures 5 and 9). The presence of POM did not appear to740 affect the particle size distribution, which was unimodal and741 centered atDp� 100 nm (dN/dDp mode was centered atDp�742 80 nm).743 [46] The effect of natural DOM on foam droplet production744 was also examined using mixtures of salinity 33% artificial745 seawater and filtered seawater collected in June 2004. Mix-746 tures were prepared containing 10% filtered seawater and 90%

747artificial seawater (FS10) by volume, and 1% filtered seawater748and 99% artificial seawater (FS1). Foam droplets were gener-749ated from FS10 and FS1 using vg = 1.2 cm s�1. Ntotal from750FS100, FS10, and FS1 were similar (see Figure 8); these were751indistinguishable from each other, except for FS100 and752460 FS10 for which Ntotal differed by less than 5%.753[47] Foam droplets were also generated from FS100754collected from the SIO Pier in February 2006. Ntotal for755February 2006 FS100 was approximately 20–40% higher than756that from June 2004 FS100 for vg = 0.090 and 0.33 cm s�1.757Ntotal increased linearly with vg for both FS100 media (see758Figure 5). Ntotal from the two FS100 media were indistin-759guishable for vg = 1.2 cm s�1 and differed only slightly (3%)760for vg = 1.6 cm s�1. The seawater had a slightly lower DOC761concentration, 2.28 mg Cl-1 than that collected in June 2004,7623.07 mg Cl�1. The summer seawater (June 2004) had a763chlorophyll concentration, 1.8 mgm�3, one order of magni-764tude greater than in the winter seawater (February 2006),7650.10 mgm�3. The seasonal differences in DOC composition766may have affected foam droplet production for low vg.767[48] Foam droplets from seawater solutions with organic768matter can be compared to those from AS33 which had no769DOM. For vg � 1.2 cm s�1, foam droplets generated from770AS33 and FS100 (June 2004 and February 2006) were771indistinguishable. For lower bubbling superficial velocities,772vg = 0.090 and 0.33 cm s�1, Ntotal from AS33 was773approximately constant, while the presence of DOM and774POM leads to Ntotal either significantly greater or less than775that from AS33.776[49] Droplets were generated from artificial seawater to777which oleic acid, which is sometimes used as a surrogate778for natural DOM, was added in concentrations from 0.1 to77910 mg l�1. For vg = 0.090 and 1.2 cm s�1, the addition of7800.1mg l�1 oleic acid (OA0.1), 1mg l�1 (OA1), and 10mg l�1

781(OA10) caused significant increases in Ntotal relative to782FS100 (see Figure 10). Ntotal was dependent on oleic acid783concentration up to 1 mg l�1; Ntotal for OA1 and OA10 were

Figure 7. Geometric mean diameters for particles generated from artificial seawater with sea salt massfractions of 0.001–0.07 (salinities 1–70%). Particles were dried before measurement; particle diametersat 80% RH and formation (wet, 98% RH) were calculated from dried diameters. Bubbles were generatedusing a fine-pore diffuser and a bubbling superficial velocity (vg) of 1.2 cm s�1.


10 of 17

XXXXXX

784 indistinguishable. Though Ntotal was affected markedly by785 oleic acid, the particle size distributions produced from each786 oleic acid media were unimodal and centered at Dp � 90 nm.787 The magnitude of the oleic acid effect on foam droplet788 concentration varied with particle diameter. For example, the789 concentrations of 30, 50, and 90 nm particles generated from790 OA10 using vg = 1.2 cm s�1 were factors of 3.1, 1.9, and 1.4,791 respectively, higher than that generated from AS33. This size792 dependent effect of oleic acid on foam droplet concentration793 was not observed for media containing natural organic matter.

794[50] The nature of organic matter appears to affect foam795droplet production, especially for low vg. June 2004 FS100,796February 2006 FS100, and OA1 have DOC concentrations797in the range 1–3 mg Cl�1. Ntotal produced from these three798media differed by up to 150%. The addition of oleic acid799suppressed foam formation and led to increase in subsurface800bubble diameters relative to FS100. The morphologies of801foams formed from June 2004 FS100 and UFS100, and802February 2006 FS100, were qualitatively identical; how-803ever, Ntotal from these samples differed by up to 40% for

Figure 9. Particle size distributions of dried foam droplets generated from June 2004 filtered seawater(FS100) using a fine-pore diffuser and bubbling superficial velocities (vg) in the range 0 to 1.6 cm s�1.

Figure 8. Particle size distributions of dried foam droplets generated from the following solutions:100% unfiltered seawater (UFS100), 100% filtered seawater (FS100), 10% filtered seawater and 90%artificial seawater (FS10), 1% filtered seawater and 99% artificial seawater (FS1), and salinity 33%artificial seawater (AS33). Bubbles were generated using a fine-pore diffuser and a bubbling superficialvelocity (vg) of 1.2 cm s�1.


11 of 17

XXXXXX

804 vg � 0.33 cm s�1. Below we use the variability in Ntotal

805 which results from differences in natural organic matter806 to bound estimates of foam droplet flux over the world’s807 oceans.

809 4. Discussion

810 4.1. Comparison With Other Measurements

811 [51] The present size distributions compare well with812 other foam droplet laboratory experiments. Laboratory813 foams generated by wave breaking simulation [Woolf et814 al., 1987], bubbling through frits [Hoppel et al., 1989;815 Martensson et al., 2003; Sellegri et al., 2007], splashing816 with a weir [Sellegri et al., 2007], and bubbling through a817 diffuser frit (this paper) all produce dried droplets with818 peak number concentrations in the submicrometer size819 range. This paper demonstrates a dN/d logDp mode at820 Dp �100 nm (dN/dDp mode at Dp �80 nm), which falls821 within the range of reported values, Dp = 50–110 nm, for822 the location of the submicrometer SSA mode [Hoppel et823 al., 1989; Martensson et al., 2003; Sellegri et al., 2007].824 The variation in foam droplet diameter with water tempera-825 ture may explain variations in size distributions observed in826 the MBL [Martensson et al., 2003; Sellegri et al., 2007].827 Martensson et al. [2003] also reported a coarse dN/d logDp

828 mode for foam droplets at Dp �2.5 mm. Supermicrometer829 particles could not be detected in the present experiments.830 [52] Laboratory experiments have shown that the pres-831 ence of oleic acid affects droplet production from single832 bubbles [Blanchard, 1963; Morelli et al., 1974] and foams833 [Garrett, 1968]. In these experiments, oleic acid was either834 added to seawater or NaCl solutions in order to approximate835 the effect of natural organic matter [Blanchard, 1963;836 Garrett, 1968; Morelli et al., 1974]. In the work of Garrett837 [1968] and the present work, foams were suppressed upon838 addition of oleic acid. The vg for the experiments of Garrett

839[1968] is not available but is estimated to be in the range8400.1–0.5 cm s�1. Here Ntotal from OA10 was 1.8 times841greater than that from UFS100 for vg = 0.090 and 1.2 cm s�1.842This agrees well with Garrett [1968], who found that the843addition of 12.5 mg l�1 oleic acid to a 30%NaCl solution led844to a factor of 1.6 increase in Ntotal relative to unfiltered845seawater. Note that in both of these experiments, oleic acid846was added in concentrations comparable to that of total847natural DOC. However, oleic acid makes up approximately84810�5 of natural DOCmass [Marty et al., 1979]. Thus it is not849surprising that droplet production from saline solutions with850oleic acid was substantially different from droplet production851from natural seawater.852[53] Thermal volatility, ionic composition, and particle853flux measurements indicate a submicrometer SSA particle854mode can be present in the range Dp = 40–200 nm [O’Dowd855and Smith, 1993; Murphy et al., 1998; Kreidenweis et al.,8561998; Nilsson et al., 2001; Clarke et al., 2003, 2006]. The857present results, that sea foams produce large numbers of SSA858particles with a mode at Dp � 100 nm, are consistent with859these measurements. Other investigators have found low860concentrations of accumulation mode particles that contain861Na [Meszaros and Vissy, 1974; Gras and Ayers, 1983;862Jennings and O’Dowd, 1990; Bigg et al., 1995]. Since the863accumulation mode is almost always observed in the MBL864[Heintzenberg et al., 2004], these measurements indicate865that a second source, such as growth of Aitken mode866particles that do not contain Na [Hoppel and Frick,8671990], can also contribute to accumulation mode aerosol868number concentrations.

8694.2. Estimation of Oceanic Foam Droplet Flux

870[54] SSA particle production is commonly estimated from871wind speed at an elevation of 10 m, U10 [Andreas et al.,8721995]. These parameterizations are motivated by the rela-873tion between wind speed and whitecap coverage. Estimates

Figure 10. Particle size distributions of dried foam droplets generated from the following solutions:salinity 33% artificial seawater containing 0 (AS33), 0.1 (OA0.1), 1.0 (OA1), and 10 (OA10) mg l�1

oleic acid. Bubbles were generated using a fine-pore diffuser and a bubbling superficial velocity (vg) of1.2 cm s�1.


12 of 17

XXXXXX

874 of supermicrometer SSA particle flux vary by more than an875 order of magnitude even for similar wind speeds [Andreas876 et al., 1995; Lewis and Schwartz, 2004]. Although labora-877 tory experiments have found that bursting bubbles produce878 submicrometer SSA particles [Mason, 1957; Cipriano et al.,879 1983; Martensson et al., 2003; Sellegri et al., 2007], few880 parameterizations of submicrometer SSA particle flux are881 available. It has been hypothesized that variations in sub-882 micrometer SSA particle concentration in the MBL could be883 explained in part by variations in the concentration and884 nature of seawater organic matter thought to affect SSA885 particle production [Reid et al., 2001; Geever et al., 2005].886 Seawater organic matter is known to affect whitecap foam887 formation and decay [Garrett, 1967] and was shown in the888 laboratory to affect SSA particle production from bubble889 bursting [Garrett, 1968; Paterson and Spillane, 1969].890 [55] Parameterizations of SSA flux can be developed891 from the number concentrations measured here by first892 estimating the size-dependent flux per whitecap foam area,893 dFw/d logDp80, by the flux per foam area in the laboratory

dFw

d logDp80

¼ vgdN

d logDp80

ð3Þ

895 dFw/d logDp80 can then be scaled to the size-dependent flux896 per ocean area, dFo/d logDp80

dFo

d logDp80

¼ WdFw

d logDp80

ð4Þ

898 whereW is the fraction of the sea surface covered bywhitecap899 foams, which has been parameterized byMonahan [1986] as

W ¼ 3:84� 10�6U3:4110 ð5Þ

902 [56] In order to estimate dFw/d logDp80 from these experi-903 ments, one must estimate the range of laboratory vg that are904 representative of oceanic whitecap foams. At sea, a dense905 subsurface bubble plume forms just after wave breaking906 [Monahan and Lu, 1990]. As the wave stops spilling, a907 residual, decaying foam patch is present. The characteristic908 time for whitecap foam decay is approximately 5 s [Monahan909 et al., 1980]. In contrast, the foams generated in the present910 experiments were approximately constant in thickness and911 constrained by glass walls. It is expected that the character-912 istic lifetime of foam bubbles in the laboratory apparatus is913 more uniform than that on the oceans. Bubbling through914 100% unfiltered seawater (UFS100) using vg = 0.090, 0.33,915 1.2, and 1.6 cm s�1 produced foam thicknesses of 0.2, 0.8, 2,916 and 3 cm, respectively; all of these are similar to oceanic917 whitecap foam thickness, �1 cm [Wilheit, 1979]. The range918 of vg tested here appears to produce foam morphologies919 which span whitecap foams at all stages. Therefore we920 estimate oceanic flux of submicrometer foam droplets and921 its variability using the laboratory measurements over the922 range vg = 0.090–1.6 cm s�1. The total flux from whitecap923 foams is expected to be within the fluxes estimated here for924 extreme values of vg because foam droplets are produced over925 all stages of whitecap evolution.

926[57] Estimates of dFw/d logDp80 were bounded by a low927and high estimate, which correspond to the dN/d logDp80

928distributions obtained from June 2004 FS100 for vg =9290.090 cm s�1 and UFS100 for vg = 1.6 cm s�1, respectively930(see Figure 11). The foam droplet fluxes from these experi-931ments cover the range of fluxes from all the experiments with932AS33, FS1, FS10, FS100, andUFS100 at all values of vg. The933variability due to vg is >5which is themaximum variability of934Ntotal over vg = 0.090–1.6 cm s�1 for any single media. TheB935symbol denotes a multiplicative variability. The variability936due to organic matter is estimated to be B1.5 which is the937maximum variability of Ntotal between media for any single938vg. We assume that these sources of variability are indepen-939dent, so that the total variability in extrapolating the labora-940tory measurements to estimate dFw/d logDp80 from sea foams941is B7.5.942[58] The variability in dFo/d logDp80 is also subject to943uncertainty in the parameterization ofW as a function of U10

944[see equation (5)]. Uncertainty in W has been estimated as945B7 for U10 = 8.5 ms�1 and B5 for U10 = 15 ms�1 [Lewis946and Schwartz, 2004, p. 266]. The total uncertainty in dFo/d947logDp80 including variability in dFw/d logDp80 is then B53948for U10 = 8.5 ms�1 and B38 for U10 = 15 ms�1.949[59] The present estimates of foam droplet flux were950compared to the supermicrometer SSA particle flux esti-951mates ofMonahan [1986], applicable forDp80 = 1.6–20 mm,952and those of Smith et al. [1993] (Dp80 = 2–50 mm) (see953Figure 11). Comparisons were made for selected values of954U10, 6 ms�1 (W = 0.002) and 15 ms�1 (W = 0.04). The955parameterization of Monahan [1986] is based on laboratory956whitecap measurements. The parameterization of Smith et al.957[1993] is based on concentration measurements at 10-m958height. For the purposes of comparison, each parameteriza-959tion was converted to dFo/d logDp80 and particle diameter at96080% RH, Dp80, using the relationship Dp80 = 2Dp. At W =9610.002, the total foam droplet flux per ocean area, Fo, for962particles in the accumulation mode size range (Dp80 =963170–500 nm, Dp = 85–250 nm), is approximately 10 and964100 times Fo for supermicrometer SSA particles according965to the parameterizations of Monahan [1986] (Dp80 = 1.6–96620 mm) and Smith et al. [1993] (Dp80 = 2–50 mm),967respectively.968[60] Submicrometer SSA particle flux estimates made969here can be compared with those of Martensson et al.970[2003], applicable for a seawater temperature of 20�C and971Dp80 = 0.04–5.6 mm; and with those of Clarke et al. [2006]972(Dp80 = 0.02–16 mm). The parameterization of Martensson973et al. [2003] is based on laboratory foam droplet measure-974ments. The parameterization of Clarke et al. [2006] is based975on number distributions of nonvolatile particles measured976above a surf zone. The parameterizations of Martensson et977al. [2003] and Clarke et al. [2006] fall within the range of978oceanic submicrometer SSA particle fluxes estimated from979the present results in the size range Dp80 = 60–800 nm (see980Figure 11). Martensson et al. [2003] used salinity 33%981artificial seawater; vg was not reported but was estimated to982be vg <0.06 cm s�1. Direct comparisons between the983parameterization of Martensson et al. [2003] can be made984with the present results for AS33 and vg = 0.090 cm s�1; the985flux of all particles with Dp80 = 200 nm was more than an986order of magnitude lower in the present results. The present987results include a range of bubbling conditions and compo-


13 of 17

XXXXXX

988 sitions at the ocean, our best estimate of submicrometer989 SSA particle flux was calculated as the mean of the UFS100990 experiments conducted at four different vg (see Figure 11).991 Because the best estimate of the total particle flux per992 whitecap foam area, Fw, is larger than that for the AS33993 and vg = 0.090 cm s�1 experiments, the flux estimates994 of Martensson et al. [2003] were comparable to our best995 estimate. This best estimate of total flux, Fo = 980 B996 7.5 cm�2 s�1 forW = 0.002, was comparable to the estimates997 ofMartensson et al. [2003] (Fo = 1300 cm�2 s�1) and that of998 Clarke et al. [2006], Fo = 700 cm�2 s�1.

1000 5. Conclusions

1001 [61] The present experiment was designed to study foam1002 and foam droplet production at sea. Foams were formed by1003 streams of bubbles which had subsurface diameters smaller1004 than 500 mm. Foam droplets produced from the resulting1005 foams had unimodal size distributions with a dN/d logDp

1006 mode at a dry diameter of �100 nm. These results agree1007 with prior laboratory foam droplet experiments [Martensson1008 et al., 2003; Sellegri et al., 2007] and were consistent with1009 the submicrometer SSA particle modes commonly observed1010 in the remote MBL [O’Dowd and Smith, 1993; Murphy1011 et al., 1998; Bates et al., 1998; Nilsson et al., 2001]. The1012 approach used here is a convenient method to generate1013 surrogate SSA particles in the laboratory.1014 [62] The presence of Na and Cl in accumulation mode1015 (Dp = 85–250 nm) particles indicates sea salt makes up a1016 substantial fraction of these particles in the MBL [Huebert1017 et al., 1998; Bates et al., 1998; Murphy et al., 1998].1018 Sodium can comprise a significant fraction of the accumu-1019 lation mode particle mass [Bates et al., 1998]. Particles

1020containing sodium can also comprise a significant fraction1021of accumulation mode particles by number [O’Dowd and1022Smith, 1993; Murphy et al., 1998]. It has been hypothesized1023that the sources of Na inMBL accumulationmode aerosol are1024(1) foam bubble bursting [Clarke et al., 2003; Martensson1025et al., 2003; Sellegri et al., 2007], (2) the tail of coarse mode1026SSA particles [Bates et al., 1998], and (3) the growth of1027Aitken mode particles that contain at least some Na [Murphy1028et al., 1998; Pierce and Adams, 2006]. The present experi-1029ments support the hypothesis that the direct emission of SSA1030particles from foams is also an important source of Na in the1031‘‘accumulation’’ mode. Natural organic matter did not affect1032foam droplet diameter, and therefore does not explain the1033variability in reported mean diameters for submicrometer1034SSA particles.1035[63] Foam droplet flux was affected by up to 40% due to1036natural DOM and POM. This effect was determined from1037two seawater samples collected from the same location, but1038in different seasons. The effect of natural organic matter on1039foam droplet flux was complex; natural organic matter was1040shown to increase or decrease foam droplet flux, an effect1041which depended on (1) POM, (2) bubbling superficial1042velocity, and (3) seasonal variations in the composition of1043DOM. The effects of bubbling flow rate and natural organic1044matter on foam droplet flux variability were estimated to be1045B5 and B1.5, respectively.1046[64] Droplet production was also affected by surrogate1047organic matter. The addition of 1.0 and 10 mg l�1 of oleic1048acid to saline solutions increased droplet production relative1049to natural seawater by approximately a factor of 2. This1050direct correlation between droplet production and oleic acid1051was not observed for natural organic matter. Furthermore,1052foams formed readily on natural seawater but were sup-

Figure 11. SSA particle flux estimates for two fractional whitecap coverages,W, of 0.002 (light grey), and0.04 (dark grey), characteristic of wind speeds at 10m height,U10, equal to 6 and 15ms�1, respectively. Foamdroplet flux estimates, shown as shaded regions, were extrapolated from the present laboratory results. Thewidth of the shaded regions corresponds to variability in foam droplet production observed due to superficialvelocity (vg) and organic matter composition. The black region depicts where the estimates of foam dropletflux for two values of W overlap. The best estimates of foam droplet flux are shown as dashed lines.


14 of 17

XXXXXX

1053 pressed on oleic acid saline solutions. We conclude that1054 oleic acid is a poor approximation of natural organic matter1055 for studies of foam droplet production.1056 [65] The foam droplet flux measured using this small1057 laboratory apparatus (0.018 m2) was scaled up in order to1058 estimate SSA particle flux for the oceans (3.61 � 1014 m2)1059 based on fractional whitecap coverage, W. We note that the1060 widely used relationship between wind speed and whitecap1061 coverage has an uncertainty of B3–7. The flux of foam1062 droplets in the accumulation mode size range was estimated1063 to be 10 and 100 times greater than the flux of coarse mode1064 SSA particles according to the parameterizations ofMonahan1065 [1986] and Smith et al. [1993], respectively. Previously1066 reported estimates of submicrometer SSA particle fluxes1067 [Martensson et al., 2003; Clarke et al., 2006] fall within1068 the range of foam droplet fluxes estimated here. On the basis1069 of the present experiments, the best estimates of total sub-1070 micrometer foam droplet flux forW = 0.002 (U10 = 6 m s�1)1071 andW = 0.04 (U10 = 15 m s�1) are 980 and 22,000 cm�2 s�1,1072 respectively. These estimates were comparable to parame-1073 terizations based on oceanic measurements; Clarke et al.1074 [2006] estimated the total submicrometer flux to be 700 and1075 16,000 cm�2 s�1 for W = 0.002 and W = 0.04, respectively.1076 The extrapolation of foam droplet flux to the ocean sug-1077 gests that foam droplets are an important source of sub-1078 micrometer SSA particles; thus a significant source of CCN1079 in the remote MBL.

NomenclatureA area of water surface in bubbling columnDb bubble diameter�Db mean bubble diameterDp dry particle diameter

Dp,1 particle diameter at ambient RHDpo wet particle diameter (98% RH)Dp80 particle diameter at 80% RH�Dp;g number geometric mean dry particle diameter�Dpo;g number geometric mean wet particle diameter�Dp80;g number geometric mean particle diameter at

80% RHFo particle number flux per ocean areaFw particle number flux per oceanic whitecap foam

areamd dry mass of solutemo mass of seawater drop at formation (98% RH)m80 mass of seawater drop at 80% RHN particle number concentrationNb bubble number concentrationsi sample standard deviations0 ensemble sample standard deviation

U10 wind speed at 10 m above sea levelW whitecap fractional coveragevg bubbling superficial velocity ( _Vair=A)_V air bubbling airflow raterss density of sea saltrsw density of seawaters standard deviationsg geometric standard deviation

1138 [66] Acknowledgments. We thank Gregory C. Roberts (Scripps Insti-1139 tute of Oceanography) for assistance in collecting seawater samples,1140 J. Shannon Scott (Scripps Institute of Oceanography) for providing the

1141NEOCO data set, Lilian B. Busse (Scripps Institute of Oceanography) for1142providing data from the University of California Coastal Environmental1143Quality Initiative, the SCCOOS formaking their automated shore station data1144at the SIO pier available at www.sccoos.org, and two anonymous reviewers1145for their helpful comments and suggestions. This work was partially1146supported from funding provided by Columbia University Biosphere21147Center. CAT was partially supported by a Phoenix Achievement Reward1148for College Scientists (ARCS) Foundation Scholarship.

1149References1150Andreas, E. L., J. B. Edson, E. C. Monahan, M. P. Rouault, and S. D. Smith1151(1995), The spray contribution to net evaporation from the sea: A review1152of recent progress, Boundary Layer Meteorol., 72, 3–52.1153Ayers, G. P., J. P. Ivey, and R. W. Gillett (1991), Coherence between1154seasonal cycles of dimethyl sulfide, methanesulfonate and sulfate in1155marine air, Nature, 349, 404–406.1156Bates, T. S., et al. (1998), Processes controlling the distribution of aerosol1157particles in the lower marine boundary layer during the First Aerosol1158Characterization Experiment (ACE 1), J. Geophys. Res., 103, 16,369–115916,383.1160Benner, R., J. D. Pakulski, M. McCarthy, J. I. Hedges, and P. G. Hatcher1161(1992), Bulk chemical characteristics of dissolved organic matter in the1162ocean, Science, 255, 1561–1564.1163Bigg, E. K., J. L. Gras, and D. J. C. Mossop (1995), Wind-produced1164submicron particles in the marine atmosphere, Atmos. Res., 36, 55–68.1165Bikerman, J. J. (1973), Foams, Springer, New York.1166Biskos, G., A. Malinowski, L. M. Russell, P. R. Buseck, and S. T. Martin1167(2006), Nanosize effect on the deliquescence and the efflorescence of1168sodium chloride particles, Aerosol Sci. Technol., 40, 97–106.1169Blanchard, D. C. (1963), The electrification of the atmosphere by particles1170from bubbles in the sea, in Progress in Oceanography, vol. 1, edited by1171M. Sears, pp. 73–202, Elsevier, New York.1172Blanchard, D. C. (1983), The production, distribution, and bacterial enrich-1173ment of the sea-salt aerosol, in Air-Sea Exchange of Gases and Particles,1174edited by P. S. Liss and W. G. N. Slinn, pp. 407–454, Springer,1175New York.1176Blanchard, D. C. (1989), The ejection of drops from the sea and their1177enrichment with bacteria and other materials, Estuaries, 12, 127–137.1178Blanchard, D. C., and L. D. Syzdek (1988), Film drop production as a1179function of bubble size, J. Geophys. Res., 93, 3649–3654.1180Blanchard, D. C., and A. H. Woodcock (1957), Bubble formation and1181modification in the sea and its meteorological significance, Tellus, 9,1182154–158.1183Blanchard, D. C., and A. H. Woodcock (1980), The production, concentra-1184tion, and vertical distribution of the sea-salt aerosol, Ann. N.Y. Acad. Sci.,1185338, 331–347.1186Bowyer, P. A. (2001), Video measurements of near-surface bubble spectra,1187J. Geophys. Res., 106, 14,179–14,190.1188Campuzano-Jost, P., C. D. Clark, H. Maring, D. S. Covert, S. Howell,1189V. Kapustin, K. A. Clarke, E. S. Saltzman, and A. J. Hynes (2003),1190Near-real-time measurement of sea-salt aerosol during the SEAS cam-1191paign: Comparison of emission-based sodium detection with aerosol1192volatility technique, J. Atmos. Oceanic Technol., 20, 1421–1430.1193Capaldo, K., J. J. Corbett, P. Dasibhatla, P. Fischbeck, and S. N. Pandis1194(1999), Effects of ship emissions on sulphur cycling and radiative climate1195forcing over the ocean, J. Geophys. Res., 400, 743–746.1196Charlson, R. J., J. E. Lovelock, M. O. Andreae, and S. G. Warren (1987),1197Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate,1198Nature, 326, 655–661.1199Chuang, P. Y., D. R. Collins, H. Pawlowska, J. R. Snider, H. H. Jonsson,1200J. L. Brenguier, R. C. Flagan, and J. H. Seinfeld (2000), CCNmeasurements1201during ACE-2 and their relationship to cloud microphysical properties,1202Tellus, 52B, 843–867.1203Cipriano, R. J., and D. C. Blanchard (1981), Bubble and aerosol spectra1204produced by a laboratory ‘breaking wave’, J. Geophys. Res., 86, 8085–12058092.1206Cipriano, R. J., D. C. Blanchard, A. W. Hogan, and G. G. Lala (1983), On1207the production of Aitken nuclei from breaking waves and their role in the1208atmosphere, J. Atmos. Sci., 40, 469–479.1209Clark, C. D., P. Campuzano-Jost, D. S. Covert, R. C. Richter, J. Marine,1210A. J. Hynes, and E. S. Saltzman (2001), Real-time measurement of1211sodium in single aerosol particles by flame emission: Laboratory char-1212acterization, J. Aerosol Sci., 32, 765–778.1213Clarke, A. D., and J. N. Porter (1993), Pacific marine aerosol: 2. Equatorial1214gradients, ammonium, and chlorophyll during SAGA3, J. Geophys. Res.,121598, 16,997–17,101.1216Clarke, A. D., J. L. Varner, F. Eisele, R. L. Mauldin, and D. Tanner (1998),1217Particle production in the remote marine atmosphere: Cloud outflow and1218subsidence during ACE 1, J. Geophys. Res., 103, 16,397–16,409.


15 of 17

XXXXXX

1219 Clarke, A. D., V. Kapustin, S. Howell, K. Moore, B. Lienert, S. Masonis,1220 T. Anderson, and D. Covert (2003), Sea-salt size distributions from1221 breaking waves: Implications for marine aerosol production and optical1222 extinction measurements during SEAS, J. Atmos. Oceanic Technol., 20,1223 1362–1374.1224 Clarke, A. D., S. R. Owens, and J. Zhou (2006), An ultrafine sea-salt flux1225 from breaking waves: Implications for cloud condensation nuclei in the1226 remote marine atmosphere, J. Geophys. Res., 111, D06202, doi:10.1029/1227 2005JD006565.1228 Day, J. A. (1964), Production of droplets and salt nuclei by the bursting of1229 air-bubble films, Q. J. R. Meteorol. Soc., 90, 72–78.1230 de Leeuw, G., and L. H. Cohen (2002), Bubble size distributions on the1231 North Atlantic and North Sea, in Gas Transfer at Water Surfaces, edited1232 by M. A. Donelan, W. M. Drennan, E. S. Saltzman, and R. Wanninkhof,1233 pp. 271–277, AGU,Washington, D. C.1234 Fitzgerald, J. W. (1975), Approximation formulas for the equilibrium size1235 of an aerosol particle as a function of its dry size and composition and the1236 ambient relative humidity, J. Appl. Meteorol., 14, 1044–1049.1237 Fitzgerald, J. W. (1991), Marine aerosols: A review, Atmos. Environ., 25A,1238 533–545.1239 Garrett, W. D. (1967), Stabilization of air bubbles at the air-sea interface by1240 surface-active material, Deep Sea Res., 14, 661–672.1241 Garrett, W. D. (1968), The influence of monomolecular surface films on the1242 production of condensation nuclei from bubbled sea water, J. Geophys.1243 Res., 73, 5145–5150.1244 Geever, M., C. D. O’Dowd, S. van Ekeren, R. Flanagan, E. D. Nilsson,1245 G. de Leeuw, and U. Rannik (2005), Submicron sea spray fluxes,1246 Geophys. Res. Lett., 32, L15810, doi:10.1029/2005GL023081.1247 Graham, A., D. K. Woolf, and A. J. Hall (2004), Aeration due to breaking1248 waves: Part 1. Bubble populations, J. Phys. Oceanogr., 34, 989–1007.1249 Gras, J. L., and G. P. Ayers (1983), Marine aerosol at southern midlatitudes,1250 J. Geophys. Res., 88, 10,661–10,666.1251 Hegg, D. A., and P. V. Hobbs (1992), Cloud condensation nuclei in the1252 marine atmosphere: A review, in Nucleation and Atmospheric Aerosols,1253 edited by N. Fukuta and P. E. Wagner, pp. 181–192, A. Deepak,1254 Hampton, VA.1255 Heintzenberg, J., W. Birmili, A. Wiedensohler, A. Nowak, and T. Tuch1256 (2004), Structure, variability and persistence of the submicrometre marine1257 aerosol, Tellus, 56B, 357–367.1258 Hoppel, W. A., and G. M. Frick (1990), Submicron aerosol size distribu-1259 tions measured over the tropical and South Pacific, Atmos. Environ., 24A,1260 645–659.1261 Hoppel, W. A., J. W. Fitzgerald, G. M. Frick, and R. E. Larson (1989),1262 Atmospheric aerosol size distributions and optical properties found in the1263 marine boundary layer over the Atlantic Ocean, Tech. rep., Naval1264 Research Laboratory, Atmospheric Physics Branch, Space Science Divi-1265 sion, Washington, D. C.1266 Horne, R. A. (1969), Marine Chemistry, John Wiley, Hoboken, N. J.1267 Huebert, B. J., S. G. Howell, L. Zhuang, J. A. Heath, M. R. Litchy,1268 D. J. Wylie, J. L. Kreidler-Moss, S. Coppicus, and J. E. Pfeiffer (1998),1269 Filter and impactor measurements of anions and cations during the First1270 Aerosol Characterization Experiment (ACE 1), J. Geophys. Res., 103,1271 16,493–16,509.1272 Jaenicke, R. (1993), Tropospheric aerosols, in Aerosol-Cloud-Climate1273 Interactions, edited by P. V. Hobbs, pp. 1–31, Elsevier, New York.1274 Jennings, S. G., and C. D. O’Dowd (1990), Volatility of aerosol at Mace1275 Head, on the west coast of Ireland, J. Geophys. Res., 94, 13,937–13,948.1276 Katoshevski, D., A. Nenes, and J. H. Seinfeld (1999), A study of processes1277 that govern the maintenance of aerosols in the marine boundary layer,1278 J. Aerosol Sci., 30, 503–532.1279 Kolovayev, P. A. (1976), Investigation of the concentration and statistical1280 size distribution of wind-produced bubbles in the near-surface ocean1281 layer, Oceanology, 15, 659–661.1282 Kreidenweis, S. M., L. M. McInnes, and F. J. Brechtel (1998), Observations1283 of aerosol volatility and elemental composition at MacQuarie Island1284 during the First Aerosol Characterization Experiment (ACE 1), J. Geophys.1285 Res., 103, 16,511–16,524.1286 Lamarre, E., and W. K. Melville (1992), Instrumentation for the measure-1287 ment of void-fraction in breaking waves: Laboratory and field results,1288 IEEE J. Oceanic Eng., 17, 204–215.1289 Laskin, A., D. J. Gaspar, W. H. Wang, S. W. Hunt, J. P. Cowin,1290 S. D. Colson, and B. J. Finlayson-Pitts (2003), Reactions at interfaces as1291 a source of sulfate formation in sea-salt particles, Science, 301, 340–344.1292 Lewis, E. R., and S. E. Schwartz (2004), Sea Salt Aerosol Production:1293 Mechanisms, Methods, Measurements, and Models, AGU, Washington,1294 D. C.1295 Martensson, E. M., E. D. Nilsson, G. de Leeuw, L. H. Cohen, and H. C.1296 Hansson (2003), Laboratory simulations and parameterization of the pri-1297 mary marine aerosol production, J. Geophys. Res., 108(D9), 4297,1298 doi:10.1029/2002JD002263.

1299Marty, J. C., A. Saliot, P. Buat-Menard, R. Chesselet, and K. A. Hunter1300(1979), Relationship between the lipid compositions of marine aerosols,1301the sea surface microlayer, and subsurface water, J. Geophys. Res., 84,13025707–5716.1303Mason, B. J. (1957), The oceans as a source of cloud-forming nuclei,1304Geofis. Pura Appl., 36, 148–155.1305Mason, B. J. (2001), The role of sea-salt particles as cloud condensation1306nuclei over the remote oceans, Q. J. R. Meteorol. Soc., 127, 2023–2032.1307McInnes, L. M., D. S. Covert, P. K. Quinn, and M. S. Germani (1994),1308Measurements of chloride depletion and sulfur enrichment in individual1309sea-salt particles collected from the remote marine boundary layer,1310J. Geophys. Res., 30, 869–884.1311McInnes, L. M., P. K. Quinn, D. S. Covert, and T. L. Anderson (1996),1312Gravimetric analysis, ionic composition, and associated water mass of the1313marine aerosol, Atmos. Environ., 30, 869–884.1314Meszaros, A., and K. Vissy (1974), Concentration, size distribution and1315chemical nature of atmospheric aerosol particles in remote oceanic areas,1316J. Aerosol Sci., 5, 101–109.1317Millero, F. J., and M. L. Sohn (1992), Chemical Oceanography, CRC Press,1318Boca Raton, Fla.1319Monahan, E. C. (1986), The ocean as a source for atmospheric particles, in1320The Sea (Marine Chemistry), edited by P. Buat-Menard, pp. 129–163,1321Springer, New York.1322Monahan, E. C., and M. Lu (1990), Acoustically relevant bubble assem-1323blages and their dependence on meteorological parameters, IEEE J.1324Oceanic Eng., 15, 340–349.1325Monahan, E. C., B. D. O’Regan, and K. L. Davidson (1980), Marine1326aerosol production from whitecaps, in The influence of whitecaps on1327the marine atmosphere. Annual report for the year ending 30 June13281980, edited by E. C. Monahan, B. D. O’Regan, and D. M. Doyle,1329University College, Galway, Ireland.1330Monahan, E. C., K. L. Davidson, and D. E. Spiel (1982), Whitecap aerosol1331productivity deduced from simulation tank measurements, J. Geophys.1332Res., 87, 8898–8904.1333Morelli, J., P. Buat-Menard, and R. Chesselet (1974), Production experi-1334mentale d’aerosols a la surface de la mer, J. Rech. Atmos., 8, 961–986.1335Murphy, D. M., et al. (1998), Influence of sea-salt on aerosol radiative1336properties in the Southern Ocean marine boundary layer, Nature, 392,133762–65.1338Nguyen, B. C., B. Bonsang, and A. Gaudry (1983), The role of the ocean in1339the global atmospheric sulfur cycle, J. Geophys. Res., 88, 903–914.1340Nilsson, E. D., U. Rannik, E. Swietlicki, C. Leck, P. P. Aalto, J. Zhou, and1341M. Norman (2001), Turbulent aerosol fluxes over the Arctic Ocean 2.1342Wind-driven sources from the sea, J. Geophys. Res., 106, 32,139–134332,154.1344O’Dowd, C. D., and M. H. Smith (1993), Physicochemical properties of1345aerosols over the Northeast Atlantic: Evidence for wind-speed-related1346submicron sea-salt aerosol production, J. Geophys. Res., 98, 1137–1149.1347O’Dowd, C. D., M. H. Smith, I. E. Consterdine, and J. A. Lowe (1997),1348Marine aerosol, sea-salt, and the marine sulphur cycle: A short review,1349Atmos. Environ., 31, 73–80.1350O’Dowd, C. D., E. Becker, and M. Kulmala (2001), Mid-latitude North-1351Atlantic aerosol characteristics in clean and polluted air, Atmos. Res., 58,1352167–185.1353Paterson, M. P., and K. T. Spillane (1969), Surface films and the production1354of sea-salt aerosol, Q. J. R. Meteorol. Soc., 95, 526–534.1355Pierce, J. R., and P. J. Adams (2006), Global evaluation of CCN formation1356by direct emission of sea salt and growth of ultrafine sea salt, J. Geophys.1357Res., 111, D06203, doi:10.1029/2005JD006186.1358Pilson, M. E. Q. (1998), An Introduction to the Chemistry of the Sea,1359Prentice-Hall, Upper Saddle River, N. J.1360Quinn, P. K., D. J. Coffman, V. N. Kapustin, and T. S. Bates (1998),1361Aerosol optical properties in the marine boundary layer during the First1362Aerosol Characterization Experiment (ACE 1) and the underlying1363chemical and physical aerosol properties, J. Geophys. Res., 103,136416,547–16,563.1365Reid, J. S., H. H. Jonsson, M. H. Smith, and A. Smirnov (2001), Evolution1366of the vertical profile and flux of large sea-salt particles in a coastal zone,1367J. Geophys. Res., 106, 12,039–12,053.1368Resch, F., and G. Afeti (1991), Film drop distributions from bubbles bursting1369in seawater, J. Geophys. Res., 96, 10,681–10,688.1370Schwartz, S. E. (1996), The whitehouse effect—Shortwave radiative1371forcing of climate by anthropogenic aerosols: An overview, J. Aerosol1372Sci., 27, 359–382.1373Sellegri, K., C. D. O’Dowd, and S. G. Jennings (2007), Surfactants and1374sub-micron sea-spray generation, J. Geophys. Res., 111, D22215,1375doi:10.1029/2005JD006658.1376Sharp, J. H. (2002), Analytical methods for total DOM pools, in Biochem-1377istry of Marine Dissolved Organic Matter, edited by D. A. Hansell and1378C. A. Carlson, pp. 35–58, Elsevier, New York.


16 of 17

XXXXXX

1379 Sharp, J. H., Y. Suzuki, and W. L. Munday (1993), A comparison of1380 dissolved organic carbon in North Atlantic Ocean nearshore waters by1381 high temperature combustion and wet chemical oxidation, Mar. Chem.,1382 41, 253–259.1383 Sievering, H., J. Boatman, E. Gorman, Y. Kim, L. Anderson, G. Ennis,1384 M. Luria, and S. Pandis (1992), Removal of sulphur from the marine1385 boundary layer by ozone oxidation in sea-salt aerosols, Nature, 360,1386 571–573.1387 Smith, M. H., P. M. Park, and I. E. Consterdine (1993), Marine aerosol1388 concentrations and estimated fluxes over the sea, Q. J. R. Meteorol. Soc.,1389 119, 809–824.1390 Spiel, D. E. (1997), More on the births of jet drops from bubbles bursting1391 on seawater surfaces, J. Geophys. Res., 102, 5815–5821.1392 Spiel, D. E. (1998), On the births of film drops from bubbles bursting on1393 seawater surfaces, J. Geophys. Res., 103, 24,907–24,918.1394 Thorpe, S. A. (1986), Bubble clouds: A review of their detection by sonar,1395 of related models, and of how Kv may be determined, in Oceanic White-1396 caps and Their Role in Air-Sea Exchange of Processes, edited by1397 E. C. Monahan and G. MacNiocaill, pp. 57–68, Springer, New York.1398 Tyree, C. A., and J. O. Allen (2004), Diffusional particle loss upstream of1399 isokinetic sampling inlets, Aerosol Sci. Technol., 38, 1019–1026.

1400Walsh, A. L., and P. J. Mulhearn (1987), Photographic measurements of1401bubble populations from breaking wind waves at sea, J. Geophys. Res.,140292, 14,553–14,565.1403Wilheit, T. T. (1979), A model for the microwave emissivity measurements1404of bubbles and foam, IEEE Trans. Geosci. Electron., GE-17, 244–249.1405Woolf, D. K. (1997), Bubbles and their role in gas exchange, in The Sea1406Surface and Global Change, edited by P. S. Liss and R. A. Duce,1407pp. 173–205, Cambridge Univ. Press, New York.1408Woolf, D. K., P. A. Bowyer, and E. C. Monahan (1987), Discriminating1409between the film drops and jet drops produced by a simulated whitecap,1410J. Geophys. Res., 92, 5142–5150.1411Yoon, Y. J., and P. Brimblecombe (2002), Modelling the contribution of sea1412salt and dimethyl sulfide derived aerosol to marine CCN, Atmos. Chem.1413Phys., 2, 17–30.

��1415O. A. Alexandrova, Civil and Environmental Engineering Department,1416Arizona State University, Tempe, AZ, USA.1417J. O. Allen and C. A. Tyree, Chemical Engineering Department, Arizona1418State University, PO Box 876006, Tempe, AZ 85287-6006, USA.1419([email protected])1420V. M. Hellion, Ecole Nationale Sup’erieure d’Electrochimie et d’Electro-1421metallurgie, Institut National Polytechnique de Grenoble, Grenoble, France.


17 of 17

XXXXXX

American Geophysical UnionAuthor Query Form

Journal: Journal of Geophysical Research - AtmospheresArticle Name: Allen(2006JD007729)

No Query

foam droplets generated from natural and artificial seawaters

Technology

submicrometer ssa particles

submicrometer particles

flux of ssa particles

submicrometer marine

mode dp

aitken dp

foam droplet mode

remote mbl