GEOPHYSICAL RESEARCH LETTERS, VOL. 21, NO. 16, PAGES 1719-1722, AUGUST 1, 1994
Volatility of elemental carbon
S.G.Jennings
•, C.D. O'Dowd
2, W.F. Cooke
3, P.J.Sheridan
4, andH.Cachier
5
Abstract. A volatilitytechnique
wherebyaerosolparticles Volatility Instrumentation
are heatedto the relativelyhigh temperature
of 860 øC is
usedto infer the presenceof elementalcarbonin polluted
The experimental arrangement is the same as that
air massesin the vicinity of the west coastof Ireland. The describedby Jenningsand O'Dowd (1990). The main core
volume of elemental carbon for submicrometre
sized
of the apparatuswas that of a ParticleMeasuringSystem's
particlescontainedin the aerosolis estimatedfrom the fall
(PMS) light scatteringprobe: Active ScatteringAerosol
off in numberconcentration
at a criticalonsettemperature SpectrometerProbe (ASASP-X). The inlet to the probe
of about730 -735 øC, as also obtainedfor laboratory consistedof a quartz tube, capableof being heatedto the
carbonink aerosol. The techniquepermitsdeterminationof
relativelyhightemperature
of 860 øC.
the elementalcarbon volume percentageof the total free
aerosol volume, and an estimation of the abundance of
Laboratory volatility measurements of carbon and
elemental carbon contained within the black carbon fraction
carbonate
of the atmosphericaerosol. Supplementary
black carbon
massconcentrationmeasurements
were obtainedusing a
thermal method and an aethalometerabsorptionmethod.
The work suggeststhat elementalcarboncan be identified
using the volatility techniqueand that it can yield sizesegregated
informationon the fractionof elementalcarbon
in atmosphericaerosol.
aerosols.
The responseof the volatility apparatusto polydisperse
laboratory carbon aerosol was first examined.
Polydispersionsof carbon ink (Staedtler Mars 745) of
concentration1 ml/l in distilledwater were generatedusing
a De Vilbiss M40 glassnebulizerand then storedin a clean
polythenebag before san•pling. Measurementswere made
over 60-minute temperaturecycles, during which the ink
carbonaerosolwas heatedto a temperattire
of 860 øC in
approximately3 minutesand then allowed to cool back
Introduction
close to ambient temperaturefor the remainder of the
heating
cycle. An exampleof the temperature-fractionation
particulateorganiccomponent
and a highly polymeric
fractionreferredto as the blackcarbon(or sootcarbon) curvesobtainedfor carbonink aerosolfor six heatingcycles
component. Atmosphericblack carbonis the dark fraction is shownin Figure 1. The temperaturefractionationcurves
of thecarbonaceous
aerosolwhereit represents
15 to 30% for carbonink are for range 3 (0.09 - 0.195/xm dimneter)
of the total carbonmassonly (Wolff et al., 1982 and and range2 (0.15 - 0.3/xm diameter)of the particlesizing
Cachieret al., 1990). Theblackcarbonappears
to havea probe. The form of the fractionationcurveswere foundto
low organiccontentas indicated
by its relativelylow H/C be largely independentof particle size. The refractory
ratio of the order of 0.4 (T.Kuhlbusch,personal component of the aerosol evaporates at temperatures
Smokeaerosolsare composexl
of two main tractions:the
communication,
1993). Witlfin thisblackcarbonfraction,
pureelemental
carbon(or graphitic
carbon)is likelyto forn•
a significantfraction which up to now has not been
betweenabout730 - 735 øC causingfairly rapidrexluction
in particle concentrationbeyond that temperature. The
volatile propertiesof the following carbonateconstituents:
sodium carbonate, calcium carbonate and annnonium
evaluated.
The aerosolvolatilitywork to date (inchidingthat of
Pinnicket al (1987),Clarkeet al (1987)),haspermitted
inferenceof at least three main atmospheric
aerosol
carbonatewere alsoinvestigatedin the laboratoryand their
temperaturefractionationcurves are shown in Figtare 1.
constituents:
sulphuric
acid,ammonium
sulphate
andsodium
Temperature FractlonatlonCurves
chloride.Thisletterdescribes
theextension
of thevolatility
techniqueto thatof elementalcarboncarbonaceous
aerosols.
1Department
of Experimental
Physics,
University
College
Galway, Ireland
2Department
of Pure and AppliedPhysics,UMIST,
Manchester, UK
3Environment
Institute,
CECJoint
Research
Centre,
lspra,
Italy
4Climate
Monitoring
andDiagnostics
Laboratory,
NOAA,
Boulder, Colorado
5Centre
desFaibles
Radioactivites,
Laboratoire
mixteCNRSCEA, Gif sur Yvette, France.
Copyright1994by theAmericanGeophysical
Union.
TEMPERATURE (deucescelsius)
Figure I Laboratoryaerosoltemperaturefractionationcurves
for sodium carbonate,carbon ink, sodiumchloride, calcium
Paper number 94GL01423
carbonate and ammonium
0094-8534/94/94GL-01423503.00
1719
carbonate.
1720
JENNINGS
ET AL.'
VOLATILITY
OF ELEMENTAL
4
Rapid fall off in particleconcentration
occurredfor sodium
ß ' ' ß 11
carbonate, calcium carbonate and anunonium carbonate at
The temPerature-fractionation
curve,obtained
for sodium
'
ß
ß ß '
2
chloride, using the stone volatility apparatus, is also
includedin Figßire1. Volatilizationwith restlltantfalloff in
concentrationbegins at about 600 øC.
This allowed
differentiationto be madebetweensodiumchloridetogether
with potentialatmosphericaerosol carbonateconstituents
with that of elemental carbon.
The laboratory
of carbon are used as a reference
'1
3
temperatures
of about700øC,380øCand90øCrespectively.
measurements
CARBON
data set
with which to compareambientaerosolfractionatßon
curves
in order to infer the presenceof elementalcarbonas an
atmospheric
aerosolconstituent
mainlyin thesubmicrometre
o
10
z
•0
-2
.10-3
range.
in
Volatility of carbonaceous aerosol
atmosphere
the
Two sets of field measurementsare presentedin tiffs
study. The first set of measurements
were madeaboarda
Figure 3 Aerosol numbersize distributionat Mace Head on
6 April 1989 for a rangeof temperatures.
orderto minimizeparticleloss)to the inletof the particle
counter.
German researchvessel, Friedrich Heinke off the west coast
Backtrajectories
indicated
thatthe air masshadrecently
of Ireland, (53ø.07'N, 9ø.39'W). The particlevolatility
traversedoverthe U.K. andIrelandtransporting
pollutants
equipmentwas placedabovethe bridgeat a heightof 18 m to the samplinglocation. An example of the aerosol
abovethe water level. The equipmentwas situatedoutside
volatilitydata taken is shownin Figßire2(A). Aerosol
sothattheaerosol
wassampled
as closeaspossible
(in concentration
per cc in four particlesize rangesis plotted
versus temperaturefor the period April 6 1989. The
ambient aerosol particle concentrationlevels remained
relativelyconstant
overtheeight-hour
period. Thedatahas
been
averaged
over
the
eight
hourly
cycles,
sothata single
• I0 •
temperaturefractionatßon
curve resultsfor each particle
range. The more rapid fall-off in particleconcentration
at
around
730-740
øC
for
the
three
lower
size
ranges
resemble
• I0•
closelythe laboratorythermal responsedata for carbon
whichis superimposed
on the field data in Figure2(A).
The strong resemblancesuggeststhat this fraction is
• oo
u
Z
z
I It
o
I I Ill
I O0
I I III
I_|
200
300
I I_111
ils
iI
•OO
iii
500
TEHPERATURE(degrees
,
i Ii
iii
I El
GO0
700
!1
I i 11
8OO
!
900
ceLclus)
ß
,
composedpredominatelyof elementalcarbonaerosol. It is
seenthat the largestparticle size fraction is unaffectedat
these high temperattires. Fluctuations in the number
concentration
of the largersizedparticlesis likelyto be due
to statisticalnoise,particularat the elevatedtemperatures
(greaterthan about400 øC) due to the low concentrations
encountered.
....
.
ß.
.
'
--..-
015-03
0 2,1 - 0 8,I Ism
- -O fl - 3 0
CAllOff INI
Volatility measurementsmade at the Mace Head
atmospheric
researchstationover the period27-28 May
1992 are shown in Figure 2(B). The period was
characterised
byhighpressure
centred
in northwestEtarope
which brought continentalpolluted air massesfrom an
easterly direction to the Mace Head site. The ambient
aerosol concentration remained
ß
constant over a six hour
periodandan averagedtemperature
fractionation
curveover
the six cyclesfor each particle size interval is shownin
Figtire2(B). A laboratoryresponsecurvefor carbonink
..
ß
aerosolis also superimposed
on the data.
The gradual
continuous
decreasein particleconcentration
betweenabout
~
c•o
i
200
i
400
!
coo
!
!
!
0oo
TEHPERATURE( de9rwes ceLclus)
Figure 2(A) Temperaturefractionatßon
curve for aerosolat
KilleanyBay, AranIslands,off thewestcoastof Ireland,April
6, 1989 for the temperaturerangefrom ambientto 860 øC.
01) Temperaturefractionatßon
curvefor aerosolat Mace Head
atmosphericresearch station, on May 27, 1992, for the
temperaturerange from ambient to 860 øC.
300 and700 øC (seenalsoin Figure2 (A)) maybe dueto
the volatilization of organic carbon and/or soot carbon
(there is no clear cut distinction between the two
components,
Cachieret al (1989)). Theobserved
sharpfall
off at about730øCfor the two lowestsize rangeswe
attribute to the volatilization of elemental carbon. The minor
peaksappearingat around430 -500 øC may be due to (a)
statistical
fluctuations
at thoserelativelylowparticlenumber
JENNINGS ET AL.' VOLATILITY
OF ELEMENTAL CARBON
1721
1000
9OO
1C
•?;jii:::?•-----%,
::::";•:'%-•.
..............
'•::::•:•
...................
;::":::'
'."•?•..•
....L.......
.;.;.•
............
-•Q•:--';:e•:•:.•::-'5•-•:•:::•:.•
........
•.
•.•:::-:•:••::::'"2'%?•'/•)
'•'•"•'"'"•••••:••••'•
'
' .•.-.•
•:;;';'.......•::•...
'w:e•
.:•-• ;•:-::•:•:?.
•7•-:;•
-::•:-•'"
'....:-.'-;-•='•:•:A•::;%.,..•3:.......:
;:•:.::::•:::::.:::•:•..::•--•r•
+•-
Fibre 4 An el•tron microcope
photomicrograph
of aerosol
pa•iclescoll•t• at MaceHead,27 May 1992. Pa•icleA
•ge aggregate
pa•icle rich in Na andtraceelements;
6OO
5OO
0 $1
:•oo/ill
^,• fi
0
0
1
2
3
-
4
Cu
•
6
7
6
g
10
ENERGY (keV)
little caren. Pa•icle B- Carbon-richaggregatewith lots of
c•s• trace elements. Pa•icle C - Small, sphericalC-rich
Figure 5 An X-ray sp•trum of a C rich aggregate
particle,
particle,probablya combustion
sphere. ParticleD
identifiedas particleB in Figure4.
Carbonaceous
chainaggregate-mostlyC.
conjunctionwith a Bendix (Unico) 240 cyclone to collect
concentrations
or (b) somecharringof organiccarbonanti ambient aerosol. A flow of 113 Uminute through the
possibly
of blackcarbon.Evenif charring
doesoccurwith cycloneyielded a particlediametercut off size of 2.4 •m
andLippman,
1977).Blackcarbon
content
Oxg
m-3)
production
of moreblackcarbonor elemental
carbon,the (Chan
resultant
increase
in particlenumberconcentration
- of not was determined using an analytical thermal method
morethana fewparticles
cm-3 (Figure
2(B))- will not
affect the number concentration of the inferred elemental
previouslyreported(Cachier et al, 1989).
An estimate
calculated
from
of the mass due to elemental
the differential
number
carbon
was
size distributions
carbonby morethana few per cent. Figure3 showsthe
change in particle number size distributionswith obtainedby the ASASP-X probe, assunfinga particular
temperature
for the dataof Figure2(A). A significantdensity. Whilst the density of elemental carbon is
reduction
inparticle
number
concentration,
forparticle
radii dependent
onmorphology
andsurface
composition,
a vah•e
uptoabout
0.2/.tn•,
between
temperatures
710øCand
860of2 gcm
-3isused
here
inlinewith
values
quoted
inthe
øC is observed,
whichwe attributeto the presence
of literature. A sununary
of estinmted
massconcentration
elemental
carbon.Thevoh•me
andpercentage
of thetotal variesforblackcarbon
andelemental
carbon
usinga range
volume
fortheinferred
elemental
carbon
is0.28/.tm
3 cm-3 of techniques
is shown
in Table1 forMaceHeadaerosol
and4.2%respectively.
Theresidual
aerosol
volume
at860 dataofMay27-28,1992.Theaethalometer
(Hansen
etal,
øCwhich
ispossibly
ofcrustal
origin
was0.4%ofthetotal 1984)values
depend
onthevalidity
foruseof a constant
volume.
Thepercentage
volatilized
between
315øCandvalue
(19m2g
-1)fortheattenuation
cross-section,
a.
710øCof thetotalvolume
was7.1%.
However,
Liousse
et al (1993)findvariations
in a which
Supplementary
measurements
were
available
fortheMacearelikely
todepend
onthesource,
ageandtype
ofnfixin•
Head
data
fortheperiod
27- 28May1992.
The
aerosol
ofthecarbon
aerosol.
They
obtain
aavalue
of12m2g
-•
wascollected
ona formvarsubstrate
withsiliconmonoxidefor suburban
aerosol.A similarlowering
in valueof 0'for
coating
andsupported
bytransmission
electron
nficroscope
theMaceHeadpolluted
episodes
of May27-28,1992
(TEM)tabbed
grids.Substrates
werefixedtothethreewould
hnprove
thedegree
ofagreement
between
aethalomestages
of a PIXIECorporation
cascade
hnpactor,
- operating ter inferredblack carbonmassanti thatdetemfinexlusingthe
thermal method.
at 1 lpm flow rate.A photomicrograph
of a grid sample,
shownin Figure 4, inch•deagglomerations
of sphende An estimateof the uncertaintyin the ASASP-X data is
particles. An X-ray spectrum
of an aggregate
particle obtainedthroughcomparisonof its responsecharacteristics
(whichis indicatedby the letterB in Figure4) showsa for sphericalparticlesof polystyrenelatex (which closely
considerableamount of carbon, considerablyabove the
blanklevel, in Figure5 indicating
thatcarboncomprises
a
major massfractionof that particle.
approximatesthat of the manufacturer'scalibration) with
that of carbon. Despitea wide rangeof complexrefractive
The elevated index"m" valuespurportedfor carbon,a valueof m=2
potassium
K+ inFigure5 ischaracteristic
ofcombustion
or - 1.0i basedon a rangeof likely refractiveindexvaluesis
responses'
comparison
showsthatthe
biomass
burningaerosols
(Cachieret al (1991)). An in- used. Thetheoretical
housefibrousglass47 mm filter samplerwasalsoused in ASASP-X will oversizecarbonparticleshavingradii less
Table
1. Comparison
ofBlack
Carbon
andElemental
Carbon
Mass
Concentration
(ngm-3)Values,
using
Different Methods,at the Mace Head AtmosphericResearchStation.
Period
May27 1992
(black carbon)
Aethalometer average
vah•e (black carbon)
Inferred from volatility
data (elemental carbon)
1037_+105+
789 _+ 25'
594*
1389 __ 140
624 _+ 4
720
Thermal method
1021-2017
May 28 1992
0221-1019
+ Estimated
precision
in therangeof 10%(Liousse
et al., 1993). ß Standard
errorbased
onsome40 individual
measurements.
*Inferredelemental
carbonmassmaybe overestimated
by up to a factorof 1.75 (seetext).
1722
•ENNINGS
ET AL.:
VOLATILITY
OF ELEMENTAL
CARBON
than 0.15 /am (over ranges2 and 3) by an averageof Cachier,H., M.P. Bremond,andP. Buat-Menard,Organicand
blackcarbonaerosolsin the remotemarineatmosphere
of
20.7%, (anoverestimation
in massby a factorof 1.75)if a
the NorthernHemisphere,Proceedings
of the International
calibration
basedon thepolytyrene
latexresponse
is used.
Conferenceon Global AtmosphericChemistry(Beijing
The particlesizingprobehas a lower cut off diameterof
1989), Newman and Kiang eds, BrookhavenNational
0.09/an whichprobablyresultsin an underestimate
of the
Laboratory,249-261, 1990.
numberconcentration
of elemental
carbonaerosolpresent.
Theremaybe someunvolatilised
carbonremaining
at the Cachier, H., J. Ducret, M.P. Bremond, V. Yoboue, J.P.
Lacaux, A. Gaudichet,and J. Baudet,Biomassburning
uppermost temperaturesof the volatility apparatus.
aerosolsin a savannaregionof theIvory Coast. In: Global
However,the volumeof residueparticlesis only of the
BiomassBurning, J.S. Levine (ed.), IV[IT Press,174-180,
order of about 10% of the inferred carbon volume.
1991.
Underestimation
of particlemassmay alsooccurdueto the Chan, T., andM. Lippman,Particlecollectionefficienciesof
undersizing
of thehighlyabsorbing
carbonparticles
greater
air samplingcyclones:an empiricaltheory,Env. $ci. and
Technol., 11, 377 - 382, 1977.
than 0.15 /am in radiusthroughthe use of the ASASP-X
probe. The comparisonbetweenthe inferred elemental Clarke, A.D., N.C. Ahlquist, and D.S. Covert, The Pacific
marine aerosol:evidencefor natural acid sulfates, J.
carbonmassfrom theparticlevolatilitymeasurements
with
Geephys.Res., 92, 4179-4190, 1987.
the blackcarbonmassobtainedusingthe thermalmethod
gives a measureof the relative abundanceof elemental Hansen, A.D.A.,
carbon in the black
carbon fraction
of carbonaceous
aerosols. A minimum percentageof about 30% of
elementalcarbonwithin the black carboncomponent
is
obtained,throughusingthe upperlimit correctionto the
inferred mass.
Discussion
aethalometer-
H.
Rosen, and T.
Novakov, The
an instrument for the real-time measurement
of optical absorptionby aerosol particles, $ci. Total
Environ., $6, 191-196, 1984.
Heintzenberg,J., and D.S. Covert, Size distributionof
elementalcarbon,sulphurandtotalmassin theradiusrange
10'6to 10'4 cm,Sci.TotalEnviron.,
36, 289-297,1984.
Jennings,S.G. and C.D. O'Dowd, Volatility of aerosolat
Mace Head, on the westcoastof Ireland,J. Geophys.Res.,
95, 13,937-13,948, 1990.
A volatilitytechniqueis usedto infer the presence
and Liousse,C., H. Cachier, and S.G. Jennings,Optical and
thermal measurements of black carbon aerosol content in
amount of elementalcarbon in polluted air masses. It
permitsan in-situ,continuous
and relativelyrapidmethod
differentenvironments:
variationof the specificattenuation
cross-section,
sigma, (o), Atmos. Environ., 27A, 1203of determiningelementalcarbon number and volumetric
1212, 1993.
concentrations
in the atmosphere.It shouldbe pointedout
andG. Fernandez,Volatilityof
howeverthat the techniqueoperatesmosteffectivelyfor Pinnick,R.G.,S.G.Jennings,
aerosols
in
the
arid
Southwestern
United States,J. Atmos.
relatively polluted air masses- possessing
carbon mass
concentration
levels
inexcess
ofapproximately
200ngm'3.
Sc/. 44, 562-576, 1987.
Rosen,H., andA.D.A. Novakov,Opticaltransmission
through
aerosoldepositson diffuselyreflectivefilters:a methodfor
measuringthe absorbingcomponent
of aerosolparticles,
Appl. Opt., 22, 1265-1267,1983.
Stevens, R.K., T.G. Dzubay, R.W. Shaw Jr., W.A.
T•0sis
toaparticle
particle
number
concentration
of
25
- perequivalent
cm3formean
radii
of0.1
- 0.15/zm.
This
approximatethresholdlevel is considerednecessaryto
providea sufficientlyhigh signalto noiseratio, the noise
beingdueto statistical
fluctxmtions
in low particlenumber
concentrations,
in the eventof lesspollutedair masses.
A percentageof betweenabout2 - 4 % of the estimated
volume of elemental carbon to that of total aerosol volume
in the 0.09 - 3.0 tun diametersizerangefoundin thiswork
is in broad agreementwith the finding of some other
workers suchas Stevenset al (1980) andHeintzenberg
and
Covert(1984). The morphological
andX-ray spectraldata
providessupporting
evidencefor the presence
of elemental
carbon and black carbon in the sampledaerosol. In
addition,the separatethermalmeasurement
techniqueof
Cachier et al (1989) on the bulk aerosol permitted
comparisonof black carbon content to be made with
volatility inferredelementalcarboncontent.
McClenny,C.W. Lewis, andW.E. Wilson,Characterization
of the aerosolin the GreatSmokymountains,
Environ.Sci.
Technol.12, 1491-1498, 1980.
Wolff, G.T., P.J. Groblicki, S.H. Cadle, and R.•I. Countess.
Particulatecarbonat variouslocationsin the United States,
In: ParticulateCarbon:Atmospheric
Life Cycle,Wolff G.T.
and Klimisch, R.L., eds., PlenumPress,New York, 297315, 1982.
S.G. Jennings, Department of Experimental Physics,
University College Galway, Ireland. (e-mail:phyjennings
•bodkin. UCG.ie)
C.D. O'Dowd, Departmentof Pure and AppliedPhysics,
UMIST,
Manchester M60
1QD, England. (email:colin.o'dowd@mailhost.
mcc.ac.uk)
W.F.Cooke, EnvironmentInstitute,T.P.460, Commissionof
Acknowledgements. The provisionof air massback the EuropeanCommunities,Joint ResearchCentre, 1-21020
trajectoriesfor the Mace Head Atmospheric
ResearchStation Ispra (Varese),Italy (e-mail:william-cooke•ei.jrc.it)
P.J. Sheridan,Climate Monitoring and DiagnosticsLabby Joyce Harris, CMDL, NOAA, Boulder, CO, USA is
oratory,
R/E/CGI, NOAA, 325 Broadway,Boulder,CO80303,
gratefully acknowledged. We also wish to thank Gunther USA.
Gassmann,chief scientistaboardthe Friedrich Heincke and the
crew, for shiptime and their valuablehelp.
References
H. Cachier, Centre des Faibles Radioactivites, CNRSCEA,91198-Gif sur Yvette, France (e-mail:[email protected].
cnrs-gif.fr)
Cachier, H., M.-P. Bremond,and P. Buat-Menard,Deter-
minationof atmospheric
sootcarbonwith a simplethermal (ReceivedJuly29, 1993;revisedDecember10, 1993;accepted
method,Tellus, 4lB, 379-390, 1989.
March 18, 1994.)