have studied the sulfur cycle over the Southern Ocean .... cated at the 30 m level of a microwave tower. ... northeast corner of the tower, so that their filters were.
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. D13, PAGES 16,493-16,509,JULY 20, 1998
Filter and impactor measurements of anions and cations during the First Aerosol Characterization
Experiment (ACE 1) B. J. Huebert,S. G. Howell,L. Zhuang,J. A. Heath, M. R. Litchyl, D. J. Wylie2, J. L. Kreidler-Moss 3, S. C6ppicus,and J. E. Pfeiffer Department of Oceanography, University of Hawai'i at MXnoa, Honolulu
Abstract. During the First AerosolCharacterizationExperiment (ACE 1), we measuredthe concentrations of aerosolnon-sea-saltsulfate (NSS), methane sulfonate(MS), ammonium,sodium,nitrate, and a variety of other anionsand cationsfrom the National Center for AtmosphericResearchC-130Q aircraft and the Cape Grim BaselineAir Monitoring Station. We used the data to test a variety of hypothesesconcerningthe sourcesand properties of aerosolin remote environments.
We noted a large gradient in free tropospheric(FT) MS, with much more in the spring (southern)hemisphere,but FT NSS showedno latitudinal gradient. Most specieshad strong vertical concentration gradients, generally with more material
near the biogenicsourcesof the marine boundarylayer (MBL) than in the FT. During the Lagrangian experimentsand in time seriesmeasurementsat Cape Grim, we saw daytime increases of photochemically derived MS, with more constant or decreasingconcentrationsat night. NSS also increasedin the daytime during the Lagrangian experiments. At Cape Grim, we were unable to account for the gravimetric massof particleson impactor substratesusingthe measuredions, which implies that a substantial fraction of the aerosol at that location must be either
organicor mineral matter. Our FT ratios of MS/NSS exhibiteda stronglatitudinal gradient, with the southernmostvalues similar to those found in Antarctic plateau snowfall.
1.
Introduction
Non-sea-saltsulfate(NSS) is principallyformedby gas-
To accurately model the climatic effects of aerosols
to-particle conversion, as sulfur dioxide is oxidized to
NSS by one of severalcompetingmechanisms[Toon et al., 1987; Kerminen and Wexler, 1994]. The natone needs to know both the composition and spatial ural sourceof SO2 is biogenicdimethyl sulfide(DMS) distribution of various aerosolspecies. Very few meagas[Andteac,1986;Pandiset al., 1994],whichcan also surements have been made of either the altitude or latbe oxidizedto a variety of other sulfur species. itude variability of aerosolsin clean parts of the marine Methanesulfonic acidis a ubiquitousproduct[$avoie at•nosphere. There is also very little information about et al., 1989; Quinn et al., 1990; Pszennyet al., 1989; the possibilitythat organicaerosolsconstitutea signifHuebertet al., 1996a].The relationshipbetweenthe reicant fraction of the aerosolmassin remote regions. sultinganionsmethanesulfonate (MS) and NSS varies The remote marine atmosphereis a chemicallydywith latitude and has been used both as a marker for
[Uharlsonet al., 1987,1992;LangnerandRodhe,1991],
namic system, in which particles are created either
DMS chemistryand as a probefor studyingthe temphysicallyfrombubble-breaking [ Woodcock, 1953;Blanperature dependencies of the competingprocesses.The chard et al., 1984] or by gas-to-particleconversion. presenceof MS in ice coreshas beenusedto infer paleobiological productivity[Legrandet al., 1991, 1992; Mulvaneyet al., 1992],eventhoughthe MS/NSS ra1Now at CT Associates,Excelsior, Minnesota. tio dependsstrongly on the efficiencyof oxidation and 2Now at the Geochemical and Environmental Research transport pathways, whosechangesthrough time are Group, Texas A&:M University, College Station. 3Now at the Schoolof Pharmacy,Universityof Washing- poorly known. It is likely, though, that much of the ton, Seattle.
Copyright 1998 by the American GeophysicalUnion. Paper number 98JD00770.
0148-0227/ 98/ 98JD-00770$09.00
MS in glacialicearrivedvia the freetroposphere (FT), for which few if any MS measurementsexist.
One of the primary reasonsfor choosingto conduct
theFirstAerosolCharacterization Experiment (ACE1) [Bateset al., this issue]in the vicinity of Tasmania 16,493
16,494
HUEBERT
ET AL.: ANIONS
was the long record of high-quality NSS and MS mea-
AND CATIONS
DURING
ACE 1
the hypothesisis disproven,it impliesthat layeringand
surementsby Ayers et al. [1991]and Ayers and Gras horizontal transport are more important than vertical [1991]at Cape Grim. These authorsclearlyshowed mixing in controllingprofilesin someareas. the seasonalvariability of MS and NSS in unpolluted
6. In the clean marine BL, both NSS and MS exhibit
(background)air and demonstratedthe correlationbe- diurnal cycles,increasingin the daytime and decreasing tween these aerosol concentrations
and the DMS source.
at night. Both NSS and MS can be produced by pho-
Berresheimet al. [1990]measuredverticalprofilesof tochemical processes,which would be expected to prosulfur speciesin this area as well. Severalother groups ceed faster in the daytime. Dry removal and dilution have studied the sulfur cycle over the Southern Ocean by mixing would presumably proceed at similar rates [Pszcnny½tal., 1989;Bates½tal., 1992;Andteac½tal., throughout the diurnal cycle, so concentrationswould decreaseat night. If this hypothesisis disproven, it 1995]. This paper reports airborne measurementsin the free impliesthat nonphotochemical processes (suchas the troposphereand the marine boundary layer of a variety oxidationof SO2 by ozoneon sea salt) are important of inorganic anions and cationsfrom the Beaufort Sea parts of the production mechanismsor that the diurnal (76øN) to the SouthernOcean (59øS).We alsoreport variation in the loss or dilution processesis similar to on both time series and size distribution measurements that of the production. of these inorganic speciesand gravimetric aerosolmass 7. The commoninorganicions can explain virtually at Cape Grim, Tasmania. The C-130 transit from Col- all the aerosolmass in the clean marine atmosphere. orado to Alaska to Hobart spanned the dates October If this is true, it means that carbonaceousand min31 to November 15, 1995. The 18 Hobart flights lasted eral aerosolare unimportant when modelingthe aerosol from November 17 until the return transit, which was massand surfacearea. If disproven,though, it means December 15-22. Our Cape Grim measurementsystem that models which only compute NSS concentrations operated from November 15 until December 14. are missingan important part of the radiatively imporWe will usethese measurementsto test the following tant aerosol. hypotheses,which relate to our ability to predict remote aerosol composition and concentrations: 2. Experiment 1. Free troposphericNSS has a stronglatitudinal gra2.1. C-130 Sampling dient, with higher concentrationsin the northern hemiIt is very difficult to measure particles from aircraft sphere. This hypothesiswould be true if most of the FT NSS is derived from anthropogenicSO2, whose emis- becauseinlet lossescan seriouslymodify the ambient sions are concentrated in the northern hemisphere. If distributionasair is conveyed to samplingdevices[Huethe hypothesis is disproven, it implies that a different bertet al., 1990;Baumgardnerand Huebert,1993].We (possiblynatural) sourcecontrolsFT NSS concentra- tried to minimize thesesamplingartifactsby usingtwo
separatesamplingsystemson the C-130: (1) an ex2. The southernmostFT ratio of MS/NSS is similar ternalsampler[Huebertet al., 1996b]that capturesall
tions.
to ratios found in Antarctic ice cores. If this is true, it
particles that enter its tip either on the extractable in-
supportsthe idea that the FT is the transportpathway let cone or on a filter and (2) a pair of single-stage for most ice core ions, so that the commonlydiscussed impactors which derived air samples from a new comboundary layer ratios of MS/NSS may be of limited munity aerosolinlet (CAI). The CAI is designedto devalue for interpreting ice core data. If it is disproven, celerateair (relativeto the aircraft) in a straightline we must rethink the mechanismsby which ions reach from 100to 10 m/s beforeit is broughtvia curvedpickthe ice. off tubesthroughthe fuselageto samplinginstruments. 3. MS/NSS ratiosare muchsmallerin the FT than There is no questionthat somelarge ambientparticles the marine boundarylayer (MBL). This hypothesisde- are still lost in this system, so that our impactor and rives from the idea that SO2 (potentially from long- other instrumentsdo not seethe full rangeof supermirangetransport)is the major sourceof FT NSS.If the cron particles. Post ACE I experimentsare underway hypothesisis disproven,it impliesa largerrolefor DMS to quantify these losses. Our C-130 impactorsystemis shownschematicallyin in controlling FT NSS concentrations. 4. The NH4 +/NSS neutralization ratiois greaterin Figure1. Fourhundredliters/min of air fromthe CAI is the MBL than the FT. If the source of NH3 is at the split into two large-diameternickel-coatedcoppertubes surface while FT NSS is derived from distant SO2, one which deliverthe air to a pair of identicalsingle-stage would expect more completeneutralizationin the MBL impactor systems. Each impactor can be used either to than the FT. If this hypothesisis disproven,it implies (1) split its particlesinto supermicronand submicron that local B L sourcesmay have a greater relative im- fractions(on a 90 mm Tedlar substrateand a 90 mm portance for producingFT NSS. I •umpore size GelmanTeflonfilter, respectively)or 5. All speciesconcentrationsdecreasewith altitude. (2) collecta bulk filter-onlysampleusinga TeflonfilThis is the simple conclusion from one-dimensional ter without the impactor jet plate and substrate. The modelingin which the sourcesare at the surface. If impactor was designedwith a I /•m cut size and was
HUEBERT ET AL.: ANIONS AND CATIONS DURING ACE 1
100 m/s
CAI
16,495
10 m/s
K/I//////////9
1 •m impactor
Bypass
Total
filters Ball valve
Gate valve
Pump
A Flow sensors
Figure 1. A schematicview of the samplingequipmentused on the C-130 aircraft. The two identicalsamplehousingscouldbe usedfor either a filter-onlyor an impactor/filtercombination. The bypassplumbing was usedto maintain flow through the pickofftube when sampleswere not being exposed.With the exceptionof the polyvinyl chlorideball valves(about 10 cm long), all tubing upstream of the samplerswas conductive.
built specificallyfor this application. Flow rates were measured with thermal mass flowmeters.
All concen-
During the flights from Hobart, one impactor was
usedfor our ion chromatography(IC) analyses,and the
trations are reportedas parts per trillion (ppt) molar other was used to collect samplesfor the analysis of ormixingratio (10•2x molesof analyteper moleof air). ganic aerosolsby a California Institute of Technology Becauseof logisticalconsiderations, we usedthe filter- (CIT) group. Those CIT resultswill be publishedseponly option for most FT and transit flight samples.Un- arately. On transit flights, we often used the identical fortunately, the filter-only data are not directly com- impactors to collect either replicate samples or simulparable to the impactor data: side-by-sidecomparisons taneous filter-only and impactor samplesfor quality asshowedthat a lone filter usually held more analyte than the sum of a substrate and filter in the impactor con-
surance
cle mass in the range of 3-8 /•m diameter. However, tests with monodisperse aerosolsfrom a vibrating ori-
ramp.
purposes.
The external sampler (whosediffusercone deposits figuration. From our laboratory tests and the second and filter are both analyzed[Huebertet al., 1996b])proCommunityAerosolInlet Evaluation (CAINE-2) field vided one reliable total aerosol concentration on most program(B. J. Huebert, S. G. Howell,and L. Zhuang, flights. Unfortunately, its detection limits were someunpublisheddata, 1997), it is clear that the CAI and times higher than desired because of the necessityto our inlet plumbing passa substantial amount of parti- manipulate the cone in the environment of the airport
fice aerosolgeneratorreveal that particles in the 3.58/•m range are caught in the holesof the impactor jet plate, from which they cannot conveniently be recovered. Our "supermicron" concentrations therefore actually represent the 1-3.5/•m diameter range, and our impactor "total" concentrationsprobably miss most of the ambient mass above 3.5/•m. Filter-only collections,
however,do containan (as-yet-unquantified) fractionof the ambient 3.5-8/•m mass.
2.2.
Cape Grim Sampling
At Cape Grim we operated both bulk filter samplers and cascade impactors. The filter samplers were located at the 30 m level of a microwavetower. Twentyfour open-faced 47 mm filter holders were mounted face
downin three plasticboxes(for rain protection)at the northeast corner of the tower, so that their filters were roughly level with the free-stream air blowing past the
16,496
HUEBERT ET AL.: ANIONS AND CATIONS DURING ACE 1
boxes. An automated system exposedthe filters in se- and unloaded in an ammonia-free glove box. Teflon quenceand recordedthe flow rate of air through each filters were washed with ethanol and deionized water filter from a Kurz thermal mass flowmeter. We exposed prior to the experiment to reduce their blank variabila seriesof eight 3 hour filters each day throughoutthe ity. The analytical proceduresand columnswere idenentire
tical to those used in the Atlantic
month.
Stratocumulus
Tran-
As a quality control check, we exposeda secondset of filters for either 6 or 24 hours. If the averageof two
sition Experiment/Marine Aerosoland Gas Exchange (ASTEX/MAGE) experiment[Huebertet al., 1996b], 3 hour filters was not within 20% of the simultaneous exceptthat during ACE I we extendedour cation chro6 hour filter, the data were scrutinized for contamina- matograms to quantifyMg++ and Ca++ ions. tion and often discarded.
Since the tower could not be
climbedduring high wind periods,somefiltersremained 3. Results in place longer than others, making it hard to precisely Before assessingour hypotheses,we first presentthe match each sample to an appropriate field blank with data derived from the aircraft and Cape Grim in graphan identical history. This caused the most difficulty ical and tabular form.
for sea salt: the Na + on 12 simultaneous blanks var-
ied by a factor of 5, but the blank standard deviation
still representedonly 10% of the averageNa+ loading on that day's exposedfilters. The NSS blank standard deviation that day was about 30% of the 3 hour filter
3.1.
C-130 CAI Efficiency
A major concern with airborne aerosol data is the extent
to which
inlet
losses have reduced
concentra-
loadingon this very clean baselineday. On the basis tions relative to the ambient ones. Figure 2 compares of our quality control checks,a singleset of the best simultaneous external samples with samples from the quality filter data has been prepared,enteredinto the impactor (that got its air from the CAI). For MS and NSS (Figures 2a and 2b), the differencesare modest ACE 1 archive, and used for the analysesbelow. We also operated two micro-orificeuniform deposit and understandable: the external sampler captured 10 impactors (MOUDIs) [Marpl½et al., 1991;Howell,1996] to 20% more material than the impactor, suggesting in the Cape Grim laboratory. They receivedair from lossesof that magnitude in the CAI and impactor jet + (Figure2c),the distribution is almost the 10 m Cape Grim Community AerosolInlet system, plate.ForNH4 which warmed the air to 50% humidity. Both used random. The points to the right of the line may be due aluminum collectionsubstratesand Teflon (Millipore to ammonia contamination of the external sampler as I pm pore size FALP) backupfilters. One was used it was bolted and unbolted from the airplane, but the to routinely expose 24 hour samplesfor inorganicion presenceof pointsto the left suggeststhat our impactor analyses,which we report below. The other was used samplesmay at times be high by as much as 200 ppt.
to provide multi-day samplesfor gravimetric analysis of the collectedmass(alsoreportedhere) and organic aerosolspeciationby the CIT group (to be published elsewhere).Masssamplingwasinitiated or terminated basedon wind directionor condensation nucleus(CN) counter evidenceof air mass changes. Although these impactors were not operated exactly at their design flowrate, we have correctedtheir cut sizesfor this discrepancy,which was at most 17%. The aluminum MOUDI substrates for gravimetric analysis were purchasedfrom MSP Corp., baked at 500øC for 12 hours, and then wrapped in aluminum foil and stored in prebakedglasscontainers.They were dried over silica gel for 12 hoursprior to weighing,both before and after exposure. Both sets of weighingswere conductedat Cape Grim, usinga Mettler ToledoModel UMT2 microbalance. The uncertainty in the gravimettic mass, based on both a procedural blank and repet-
The sodiumplot (Figure 2d) is very informative: the slopeis 0.18, with an R2 of 0.8. That meansthat only about 1/5 of the ambientseasalt modegot asfar asour impactionsubstrate.Roughly4/5 waslost in the CAI
or the impactor jet plate. Of course,the exact fraction depends on the size and shape of the sea salt size dis-
tribution, which changesfrom flight to flight. We have electedthereforenot to try to correctthe Na+ data for this inlet
3.2.
The
loss.
C-130
Data:
variation
Variations
of FT
and MBL
With
Latitude
concentrations
with
latitude is shownin Figures 3 and 4. Statistics for the entire project and for the flights from Hobart are tabulated in Table 1. To compute the means in Table 1 we inserted 0.5 times the detection limit for those samples below the detection limit. Note that the open downpointing triangles in all figuresrepresentthe detection itive weighingof substrates, is at most0.2 pg/m3 on limit for sampleswhose concentrationswere below the eachstage. A larger sourceof error comesfrom the es- detection limit. Ninety-sevenpercent of the free tropotimation of total air sampled,which was at most 10%. sphericsampleswere collectedusing a filter only, but With the exception of a few Cape Grim samplesfrom only 7% of the boundarylayer sampleswerefilter only, late in the experiment and transit samplesfrom the so that we could separate the submicron mode in the C-130's return to the United States, all samples were MBL from sea salt. 3.2.1. FT. It is notable that in the FT only MS extracted and analyzed in a portable laboratory located at the Hobart Airport. All media were loaded (and its ratio to NSS) showeda significantlatitudinal
HUEBERT
ET AL.' ANIONS AND CATIONS DURING
ACE 1
16,497
25
20
rolo
E 5
0
5
10
15
20
25
0
100
External,ppt
200
300
External,ppt 15000
lOOO
d)Na
800
400 2o0
0
200
400
600
800
lOOO
0
5000
External, ppt
10000
15000
External,ppt
Figure 2. A comparison of (a) MS, (b) NSS,(c) NH4 +, and (d) Na+ concentrations measured simultaneouslyfrom the C-130 impactor and external samplers.
Boundarylayer NSS concentrations(Figure 4b) were apparentlyinfluencedby volcanosnear Hawai'i (20øN) and just southof Fiji (20øS).SO2 southof Fiji reached as high as 1.4 ppbv (D. Thornton, personalcommunication, 1997), which is clear evidencethat the source 11% south of Hobart. of the SO2 was nearby. On a more sensitivescale(FigFreetropospheric NSSandNH4 +, ontheotherhand, ure 4c), however, it is evident that in areas without
gradient(Figure 3). Not surprisingly, the highestMS valueswererecordedin the southernhemisphere,which was experiencingspringduring ACE 1. The MS/NSS ratio varied from lessthan 3% in the most northerly samplesto as much as 11% near the equator and 1-
have remarkably similar distributions in the northern and southern hemispheres, with a few more low concentrations in the south, where we made many more measurements. Sulfuric acid was generally more than
half neutralizedby ammonia(the medianratio was1.2),
butthereweremoreNH4+/NSS ratiosbelow1.0in the southern hemisphere. Twenty percent of the FT ammo-
strong SO2 sources, the NSS is generally more concentrated near the equator than at higher latitudes. Ammonium ion showeda similar pattern (Figure 4d and 4e), with the highest concentrationsin volcanic fume and slightly lower concentrationsnear the poles. Seventeen percent of the B L ammonium values were
undetectable. Surprisingly, though,the NH4+/NSS ra-
nia valueswereundetectable(Figure3e), sowecanonly tio (Figure 4g) did not vary muchoverthe entirerange compute an upper limit to their neutralization ratio. 3.2.2. MBL. The MBL distributions are quite different. MS concentrationspeaked at about 15 ppt in the productive region near the equator and again over
of latitudes. It averaged0.92 south of Hobart and varied around I near the equator. 3.3.
C-130
Data:
Variations
With
Altitude
It is evident from Table I that means and medians the springtimeSouthernOcean. The MS/NSS ratio (Figure 40, however,was dramatically higher to the for every species are higher in the BL than the FT.
south. It was only 3-7% in unpolluted air near the In the case of MS, the BL concentrationsare roughly equator, while the samplessouth of Hobart averaged an order of magnitude higher, while for NSS, the difaround 14% and had a maximum of 28%. Clearly the ference is a factor of 2 or less when volcanic plumes are excluded. Ammonium concentrations demonstrate production of MS is favored at higher latitudes.
16,498
HUEBERT
ET AL-
'
ANIONS AND CATIONS DURING
ACE 1
25
•) MS '
•) MS
20
4
15
v 09 o.v v v
5
v
0 oo t-.cu
b) NSS
o o
150
i
,
15OO
+
+
I
b) NSS
+
•_ 1000
100 .+
50
+.
+
+ i
+ •+
150
500
+
I
I
•
o•o
•+ +
I
c)NH•
+
+
150
x
Vx
100
xxX
x
X X
x
•
x
+ * +
*
-+ •+
50
x
ß
+
I
•
I
o
d) MS/NSS
10
small scale
.+
t•
X X
i
o
•o
[]
..• oo []
[]
v
•;7
+
d)NH4
o
o
x
I
•1"" 150
-øo%o-d o
-40
i -20
-o o
i i 0 20 Latitude
o o
i 40
i 60
x
•xx ' X
!
c•
I
small 4scale
,X
I
free troposphere. Down-pointingtrianglesrepresentthe detection limit for sampleswhose concentrationswere
1-113
20
ß
10 0
g•) NH 4/NS S.
limit.
much less of an altitude difference, with nearly identical BL and FT means on the Hobart flights. This implies slightly higher NSS neutralization ratios in the
FT than BL, by about 20%. MS/NSS ratioswere4-5
i
-60
times larger in the MBL than FT.
Although we cannot be certain what fractionof the sea-saltmode we were able to sample,Na+, CI-, and NO• concentrations have been includedin Table 1. Sodium and chloride are more concentrated in the MBL
by about an order of magnitude, but aerosolnitrate
(whichtendsto be on the sea-saltmodein the MBL)
I
f) MS/NSS
13õ n
13
and(c) NH•+ concentrations and(d) MS/NSSand(e) NH4 +/NSSmolarratiosmeasured fromtheC-130in the •
is nearly identical in the two regimes.
I
e) NH+
50
80
Figure 3. Latitudinal distributionsof (a) MS, (b) NSS,
below the detection
x
100
o
2f•,o ø .070 • o-ø-øo
-60
x
vvV67 F
e)NH I / NSS
1
x
500
V
I
i
1500
•. 1000 5
I
c) NSS
+
x X
50
%t
x
•
Figure 4.
-40
-20
0 20 Latitude
40
i
60
80
Latitudinal distributionsof (a) MS, (b)
all NSS,(c)smaller-scale NSS,(d) all NH4 +, and (e) smaller-scale NH4 + concentrations and(f) MS/NSSand (g) NH4+/NSS molarratiosmeasured fromthe C-130 in the boundary layer. Down-pointingtrianglesrepresent the detection limit for sampleswhose concentrations were below the detection
limit.
HUEBERT ET AL.: ANIONS AND CATIONS DURING ACE 1
16,499
Table 1. Statistics on C-130 SamplesfromtheBL andFT
MS, NSS, NH4+, NO3', CI', ppt ppt ppt ppt ppt Mean 1.1 Median 0.9 Std.dev. 0.9 Minimum
