VOL. 103, NO. D13, PAGES 15,971-15,978, JULY 20, 1998
Simultaneous measurements of CO2, CH4, and N20 in air extracted by sublimation from Antarctica ice cores: Confirmation of the data obtained using other extraction techniques T. Grillilk
and F. Slemr
Fraunhofer-Institutffir Atmosph•irische Umweltforschung(IFU), Garmisch-Partenkirchen, Germany
B. Stauffer Physikalisches Institut der Universit•it Bern, Bern, Switzerland
Abstract. A sublimationtechniquehas been developedto extract air samplesfrom polar ice coresfor subsequentsimultaneousmeasurementof severaltrace gasesby frequencymodulatedtunable diode laser absorptionspectroscopy. This extractionand analysis techniqueis shownto be suitableas an extractionmethod for the determinationof
concentrations of thegreenhouse gases CO2,CH4, andN20 in air samples of -1-5 cm3 recoveredfrom ice samplesof 10-50 g. Air samplesfrom the Siple ice core have been analyzedcoveringthe period between 1772 and 1973. In addition, a few samplesfrom two different ice coresfrom Vestok station have been analyzed.Our resultsare in a good agreementwith resultsobtainedby other researchersusingmelting and crushing extractiontechniques.This agreementindicatesthat processesconnectedwith the formation of clathratesin ice under high pressureat greater depths and their destruction after drilling are not affectingthe CO> CH 4, and N•O measurementssignificantly. bubbles at least on the decadal timescaleand for gasesmeasurcd so far. The prcsence of chemical reactions has bcen The compositionof the ancient atmosphereand its global documentedby a detailed intercomparisonof CO?_measurechange during the past 220 kyr can be reconstructedmost ments in ice coresfrom Greenland and Antarctica [Anklin et directlyfrom the analysisof air bubblesentrapped in polar ice al., 1995] which has shownexcessCO• built up in Greenland cores[Oeschgcr and Langway,1989;Jouzelet al., 1993;Raynaud ice cores either by oxidation of organic carbon, the reaction et al., 1993].So far, a changeof mixingratios of CO_•,CH 4, and between ice acidity and carbonate, or both. In grcat depth N•O has been reconstructedfrom different ice cores [e.g., under high hydrostaticpressure,air gets enclosedinto clathNeftel et al., 1985; Barnola et al., 1987; Zardini et al., 1989; rates. Fractionation of gasescould occur if the extractionyield Chappellazet al., 1990;Barnola et al., 1991; Staffelbachet al., of the air samplingis below 100% [Schwander,1989;Anklin et 1991;Jouzelet al., 1993; Chappellazet al., 1993;Raynaudet al., al., 1997] as it is in the casewith extractiontechniquesusing 1993], and its relationshipwith the changingclimate or with mechanical crushing of the ice samples.The consistencyof anthropogenicactivitieshasbeen investigated.Only recently,a trace gas measurementson different ice cores at sites with changeof CO mixingratios during the last 200 yearshasbeen different snow accumulationrates and therefore exposed to different pressures[e.g.,Barnolaet al., 1987;Chappellazet al., reported as well [Haan et al., 1996]. and Siegenthaler,1992;Raynaudet al., 1993] The basisfor theseinvestigationsis the assumptionthat the 1990;Leuenberger compositionof the air extractedfrom ice core samplesrepre- provides an indirect indication that clathrate formation and sents the air composition at the time of the closure of the decompositiondo not cause seriousfractionation of the air bubbles. This assumption implies that physisorption and composition.However, a direct confirmationthat the air comchemisorptionof air componentsat firn grain surfacesand positionis not significantlyaffected by these processesis still fractionationeffectsdue to gravityseparationand thermodif- missing. In a recent paper, Galh;iket al.  reported on the defusionin the firn canbe neglectedor correctedfor and that the compositionof air after enclosurein bubblesis not altered by velopmentof a nondestructivetechniquefor simultaneousdechemical reactions between impurities in the ice. A good terminationof CO2, CH 4, N20, and CO in air samplesof a few agreementof trace gasmeasurementsin air samplesfrom firn cubic centimeters. The technique is based on a frequencymodulated high-resolutioninfrared absorption spectrometer and ice overlapingin time with direct measurementsin the with tunablediodelasers(FM-TDLAS) and is capableof meaatmosphere[Neftelet al., 1985;Etheridgeet al., 1992;Raynaud suringambientmixingratiosof CO2,CH4, andN20 in 2 cm3 et al., 1993;Battle et al., 1996] showsthat physisorptionand of air with a precisionof 1%-2%. The technique has been chemisorptiondo not influence the compositionof the air developedfor the analysisof air samplesextractedfrom 20-50 Copyright 1998 by the American GeophysicalUnion. g ice core samples.A simultaneousdetermination of three or four target gasesin an ice sample of this size representsa Paper number 98JD00686. 0148-0227/98/98JD-00686509.00 significantimprovementin comparisonwith the current ana1.
GOLLOK ET AL.: GASESIN AIR EXTRACTED BY SUBLIMATION FROM ICE
lyticaltechniqueswhichrequiresubstantially largerice samples for the same performancein most of the laboratories[e.g., Etheridgeet al., 1988;Machidaet al., 1995].However,to utilize the potential of this analyticaltechnique,an extractiontechnique suitablefor all the target gaseshas to be used.Because of possibleCO2 artefacts,extractionby melting of ice [e.g., Chappellazet al., 1990;Haan et al., 1996]is not suitablefor this purpose[Delmas,1993].Mechanicalcrushingor milling of ice samples[e.g.,Zumbrunnet al., 1982;Barnolaet al., 1987;Etheridge et al., 1988] in devicesmade of metal produced CH 4 contaminationfrom the grindingof metal and specialdesigns were needed to reducethis contaminationto acceptablelevels [Fuchset al., 1991].However,smalleror largerfragmentsof ice are still intact after crushingand can thuscontainundestructed clathrates.The extractionefficiencyby crushingice with clathrates is only 45%-70% efficient [Anklin et al., 1997], so that fractionationis possible.For thesereasonswe concentratedon the developmentof the sublimationextractiontechnique. Wilsonand Donahue  and Wilson recentlyreported the use of sublimationof the ice as a techniquefor the extraction of air samplesfrom ice cores. They applied this techniqueto extract air samplesfor the determinationof the isotopiccompositionof CO2 in ice cores.Their measurements of preindustrialCO2 mixingratios,however,suggestthat substantialamountsof CO2 and possiblyof other trace gasesmay be containedin the ice matrix, inaccessibleto crushingextraction techniques[Wilson,1995]. The comparisonof air sample extractionby sublimationdevelopedby us with other extraction techniquescan contributeto a resolutionof the questions raisedby Wilson [e.g.,Powell, 1994]. In this paper we report on the developmentof the sublimation extractiontechniqueand a few measurementson ice core samplesto demonstrateits viability. The sublimationextraction technique has been developed(1) to complementthe FM-TDLAS technique,(2) to solvethe problemof a possible fractionation of gasesfrom ice where air is enclosedin clathrates (becauseof the incompleteextractionwith the crushing techniques),and (3) to allow new measurements(e.g., gas compositionof air occludedin firn grains).The reliability of the sublimation extraction technique was demonstratedby measurementson the Siple ice core. The resultsof thesemeasurementswere in a goodagreementwith measurements made by other researchersusing melting and crushingextraction techniques.The good agreementof the resultsfrom the Vostok ice core obtained by sublimationwith those obtained by other techniquesindicatesthat processesconnectedwith the disappearanceof air bubbles in ice under high pressureat greater depths and their reappearanceafter drilling do not lead to a significantfractionationof the measuredtrace gases. The agreementof our measurementson Siple and Vostok ice coreswith measurementsby Neftel et al. ,Staufferet al. ,Barnolaet al. ,Blunier,Jouzelet al. , andMachida et al. showsthat the sublimationextraction technique is suitable for simultaneousCO2, CH4, and N20
the extractionapparatusinto the absorptioncell of the spectrometer,and (3) determinationof the traceconcentrations by a frequency-modulated tunablediodelaser(FM-TDLAS) absorptionspectrometer. 2.1. Apparatus for Ice Sublimation
The sublimationextractiontechniqueutilizessublimationof the ice samplein a high-vacuumapparatusat temperatures well belowthe triple point of ice-water-vapor(0øC).The energyneededfor the sublimationis transferredto the ice sample by near infraredirradiation.The releasedwatervaporand the air from the bubblesare refrozenin consecutive cold traps at temperaturesappropriateto separateboth components.The temperatureof the ice during sublimationis monitoredvia the water vapor pressureabove the ice and kept below -20øC (equivalentto 1 hPa vapor pressure)to preventmeltingand the formation of the quasi-liquidlayer on the ice surfacereportedby Elbaum et al.  and Furukawaet al.  at temperaturesbetween -4 ø and -2øC and at the interface betweenthe ice and glass[Furukawaet al., 1993]. Thesemeasuresshouldprevent chemicalreactionsin the liquid phase from producingexcessCO2 from carbonatedust. The sublimationapparatusis shownschematically in Figure 1. The essentialpartsof the apparatusare the samplevesselA in whichthe ice sampleis placedand the ice vapor trap B for freezingout the bulk of the water vapor. Air releasedduring the sublimationflows through the thin tubing reachingthe bottomof the ice vaportrap B via a secondice vaportrap C to a coldfingerwhereit is condensedat 14 K. The apparatuswas installedin a laboratoryat ambient temperature. With the exceptionof the coldfingermadeof stainless steel the apparatusis made almost completelyof glass(Duran). Exceptfor the valve of the calibrationgascylinder,only highvacuumglassstopcockswith teflon o-rings(Fa. Young) are usedin that part of the apparatuswhere the sublimationtakes placeandwhere the water vaporpressureis relativelyhigh.We preferred these materials to stainlesssteel becausea former apparatusbuilt up of ultrahigh-vacuumstainlesssteelcomponentsintroducedhigher levelsof contamination. Both traps B and C are kept at -90øC usinga closedcycle alcoholcryocooler(HAAKE KT 90). The coldfingeris cooled down to 14 K by a two-stageclosedcycle He cryocoolerof Gifford-McMahontype (RGD 210, Leybold).The whole sublimation apparatusis evacuatedby a turbomolecularpump combinedwith a rotary vane pump as a forepump.The pressure is monitoredby two pressuregauges:an ionizationvac-
uumgaugecoveringa pressure rangebetween10-9 and10-2 hPa duringthe evacuationof the apparatusand a capacitance
pressure gauge(LeyboldCM 10,1 x 10-3 - 13hPa)for the
continuousmonitoringof the vapor pressureduringsublimation. The temperatureof the ice sampleis not measureddirectlybut inferred from the measurements of the pressurein the sublimationapparatus. To sustaina high sublimationrate, a sufficientamount of measurements in ice cores. CO measurement has also been energyhas to be transferredto the ice sample.This has been attemptedbut so far has failed becausecontaminationswere achievedby irradiation of the ice sample by four near IR too high. quartzlampswhosetotal power can be adjustedbetween0 and 1200W usinga variableautotransformer.To preventoverheating of the walls of the samplevesselby partially absorbedIR 2. Experimental Section energyand heat from the surroundingair and, consequently, a Measurementsof trace gasesin air extractedfrom ice sam- localmeltingof the ice sample,the annulusbetweenthe lamps ples are made in three steps:(1) extractionof the air sample and the ice samplevesselis flushedby an intensivestreamof from the ice core sample,(2) transferof the extractedair from air cooleddown by liquid nitrogen.
GOLLOK ETAL.:GASES IN AIR EXTRACTED BYSUBLIMATION FROMICE
Rotary vane pump Glass stopcock
Ice vapour cold trap (c) Viton-O-ring sealing Sample (A)
Ice vapour cold trap vessel (B)
Alcohol bath - 90 øC
l glass •metal
Nupro metal bellows valve Metal sealed connection
Ice sample Alcohol bath
Cold air stream
- 90 øC
Figure1. A schematic diagramof the sublimation apparatus for extraction of air fromicecoresamples.
temperatureof -90øC, and the ice sampleis irradiatedby the infraredlamps.The ice sublimes,and becauseof the large crosssectionof tubingbetweenvessels A andB in comparison withthe smallcrosssectionof tubingleadingto the trap C, the icevaporis almostcompletely trappedin coldtrapB. The total pressureconsistingof the partial pressuresof both the ice vaporandthe releasedair ismonitored.The partialpressureof air is smallerthan that of the ice vapor by -5 orders of magnitude.The intensity of the irradiation is controlled to keepthe icevaporpressure within0.2-0.5 hPa(corresponding -24øC because of the loss of sublimation heat. to ice temperaturesof-36ø--27øC) duringsublimation,thus After evacuationfor at least2 hoursthe pump is shutoff, preventingthe formation of a quasi-liquidlayer on the ice both cold traps B and C are cooleddownto an operating surface.Keepingthe ice vapor pressureat low levelsalsoreA cubeof ice sampleweighing20-50 g is cutwith a bandsaw from the ice core in a cold room at -24øC and placedinto samplevesselA. The vesselwith the sampleis attachedto the apparatus,and the apparatusis evacuated.During the evacuation,ambientair in the vesselis pumpedoff, anda thin layer of the surfaceof the ice samplewith adsorbedgasesis evaporated and pumpedoff aswell. This is an importantcleaning process.The IR quartz lamps are switchedoff during this evacuationso that the ice sampleis cooleddownwell below
GOLLOKET AL.: GASESIN AIR EXTRACTEDBY SUBLIMATIONFROM ICE
ducessubstantiallythe desorptionof CO2 moleculesadsorbed at the system'sinner walls by water moleculesas reported by Zumbrunnet al. . Air releasedin the courseof the sublimationprocedureis carriedby the continuousflow of ice vaporto trap B and then flowsthroughthe thin pipe into cold trap C where the rest of the ice vapor is frozen out. The remainingpartial pressureof
reversethe functionsof vessels A and B sothat the ice sample havingbeen sublimatedinto trap B can be sublimatedback to vesselA by replacingthe alcoholbath by the IR lampsandvice versa.This has the advantagethat blank measurements (contaminationreleasedby the extractionprocedureitself) canbe performedby the sublimationof the originalice sample.In this way the time-consuming preparationof gas-freeice commonly ice vaporafterpassing thistrap amounts to 1 x 10-4 hPa used for this purposeis unnecessary.During "back sublimaisslowly (corresponding to -90øC). The almost dry air is then con- tion"---1-2cm3 (STP)of airwithknowncomposition densedin the coldfinger(stainlesssteeltube,166mm long,ID introducedinto the apparatusto simulate,as closelyas possi14 mm, wall thickness1 mm) at 14 K. Sublimationof 50 g of ble, the sublimationof a natural ice sample containingair natural ice takes 30-45 min and is stronglydependenton its bubbles.The procedureblankswere then determinedby comsurfacearea exposedto the irradiation.In the initial part of the parisonof the composition of the air addedandthe air exposed sublimationprocess,when there is an inhomogeneous temper- to the sublimationprocedure. ature distributionwithin the ice sample,it maybreakin two or more pieceswhich substantiallyacceleratesthe sublimation 2.6. Ice Samples process. Simultaneousanalysisof CO2, CH 4, and N20 was made of air samplesextractedby sublimationfrom ice coresfrom Siple 2.3. Transfer of the Extracted Air Into the Absorption Cell station (75ø55'S,83ø55'W) in west Antarctica. The annual of the Spectrometer mean air temperatureand snowaccumulationrate at this drilAfter completingthe sublimationprocedurethe valveon the ling site are -24øC and 50 g cm-2 yr-•, respectively. The cold fingeris shut.The closedcold fingeris then disassembled drillingwas made by a mechanicaldrill without fluid and was from the sublimationapparatusand warmed up to ambient finishedin November1983reachinga maximaldepthof 200 m temperature.The pressureof the air samplein the coldfinger [Neftelet al., 1985].The ice at the bottomof the drillingis more before expansioninto the absorptioncuvetteamountsto 75- than300yearsold. Air bubblestrappedin the ice are, however, 187hPacorresponding to 2-5 cm3 (STP)of air sample. After younger.The mean differencebetweenthe age of the ice and warmingup, the coldfingeris cooleddownto -90øC to freeze the enclosedair is 82 years [Schwanderand Stauffer,1984]. out the remainingwater vapor in the air sample.Finally, the The qualityof the Siple ice core is excellent,and numerous cold finger is attachedto an absorptioncuvetteof the spec- measurementshave been made of the mixing ratios of CO2 trometer,and the air sampleis expandedinto the preevacuated [Neffelet al., 1985;Friedliet al., 1986] and CH 4 [Staufferet al., absorptioncuvetteand analyzedwithin 9 min as describedby 1985] documentingthe increasescausedby increasinghuman Gallak et al. . activitiesduring the past 200 years.So far, no N20 measureThe transfer procedurewas tested by freezing out an air mentshavebeen reportedfor this core.We analyzednine ice sampleof knowncompositionin the cold finger,transferingit samples(31-49 g) from differentdepthscoveringa time peinto an absorptioncuvette,and analyzingit asdescribedabove. riod of 200 yearsfor CO2, CH 4, and N20 content.The oldest No significantdifferences have been observedbetween the air sampledated back to 1772 A.D. and was taken at 176 m CO2, CH4, and N20 concentrationsand their nominal values. below surface,whereasthe youngestsampledated from 1973 A.D. and was taken at 69 m below surface,just below the 2.4.
of the Trace
by the FM-TDLAS Spectrometer
Briefly,the FM spectraare measuredfor the CO2, CH4, and N20 in the air sample. The air is then pumped out of the absorptioncuvette,and the backgroundFM spectraare taken with an empty cuvette. Finally, calibration gas with known CO2, CH4, and N20 mixingratios is filled at the samplepressure into the cuvette, and the calibrationFM spectraare recorded. The CO2, CH4, and N20 mixing ratios are then calculated from the amplituderatios of sampleand calibration FM spectra,both after the subtractionof the backgroundspectrum.
Pressurizedambient air in a cylinderwasusedas a working standard.The CO2, CH4, and N20 mixingratiosin the working standardgaseswere establishedusinga primary standardcertified by Climate Monitoring and DiagnosisLaboratory,Na-
from firn to ice.
In additionto the Sipleice corea few sampleswere analyzed from two differentdrillings(3G and4G2) at Vostok(78ø28'S, 106ø48'E,elevation 3490 m, annual mean temperature and
snowaccumulation rateof -55.5øCand2.3 g cm-2 yr-•, respectively[Loriuset al., 1985]), the Russianstationin central East Antarctica. The purpose of these measurementswas mainly to investigatewhether the sublimationextractionwith almost100% extractionefficiencywill give air with the same mixingratios as other extractionmethods.Two samplesfrom the 3G drilling were taken at 2043.7 m depth below surface. They havebeen dated to be between151,000and 155,000years B. P. [Loriuset al., 1985;Barnolaet al., 1987]corresponding to the end of the penultimate glacial period. Further samples weretakenfrom 2306.1m depthof the 4G2 drilling(maximum depth2546m). Theseair samplesdatebackto ---181,500years B. P., corresponding to the middleof the penultimateice age.
tional Oceanic and AtmosphericAdministration (CMDLNOAA). Mixing ratios of the target gasesin the working standardair were (299.73 m 0.20) ppm for CO2, (1.717 m 3. 0.009) ppm for CH4, and (295.8 m 0.2) ppb for N20.
Measurementspresentedhere were madewith the intention to verify the performanceof the sublimationtechniquefor the The A and B vesselsand the tubingconnectingthem to each extractionof air samplesfrom ice coresfor simultaneous meaother as well as to cold trap C are constructedsymmetrically. surementsof three greenhousegasesusingFM-TDLAS specThe symmetricalconstructionof the apparatusallowsone to troscopy.Consequently, sampleswere analyzedfrom the Siple
GOLLOKET AL.' GASESIN AIR EXTRACTED BY SUBLIMATIONFROM ICE 350
dry extraction m
Figure 2. Temporal changeof CO2 mixing ratios between 1772 and 1973 as obtainedfrom analysesof ice coresfrom Siple drilling (75ø55'S,83ø55'W).Data measuredby Neftelet al.  and Friedliet al.  on the same core using dry extractiontechnique are also shown as well as CO2 data from direct atmospheric measurementsat Mauna Loa reported by Keelinget al. .
ice core, which had been thoroughlycharacterizedin the past. To addressthe questionof fractionationeffects due to clathrate formation, several ice samplesfrom deep parts of the Vostok ice coreswere analyzedas well.
dry air samplebecomesdepletedof CO2 by 53-108 ppm (of initially 299.73 ppm) dependingon the pretreatment of the apparatus.The highest depletion of 108 ppm was observed when the cold fingerwas heated duringthe evacuationperiod before injectingthe air sample.This observationsuggeststhat 3.1. Tests of the Sublimation Procedure COg is adsorbedon the walls of the apparatus.Regardlessof The performanceof the sublimationextractionprocedureis the reasonsfor these artefacts,theseobservationsclearly demcharacterizedmostappropriatelyby its blanksand their repro- onstratethat blank runs with a dry sublimationapparatusare ducibility. Blanks of the procedure were determined as de- not representativefor wet conditionsduring the sublimation scribedearlier. Averagedblanksfrom five determinations(av- procedure.This conclusionled us to the blank determination erageblank _+or)were + 11.0 _+1.4 ppm for COg, +48.8 _+5.8 proceduredescribedabovewhichsimulatesthe conditionsdurppbfor CH4,and+ 10.0+ 1.1ppbfor N20 in 5 cm3 (STP)air. ing the extraction as closely as possibleand which leads to All blankswere positive;that is, the mixingratiosof COg, CH4, reproducibleresultsfor the blank value. and N=O increase during the sublimation procedure. The blanksfor COg, CH4, and N20 representroughly 3% of their 3.2. Measurements on Siple Ice Cores current atmosphericmixing ratios. The COg, CH4, and N20 Figures 2-4 show the results of the COg, CH4, and N20 mixingratios reported below are correctedfor theseblanks. measurementsperformed on the Siple core. The standarddeContaminationof the air samplecausedby the handling of viations_+orof the measurementsobtainedfrom replicateexthe ice sample or by gasesadsorbedat the ice surfaceduring traction and analysesare _+2.0ppm (n = 4) for COg, _+8.5 samplepreparation is eliminated by the initial evacuationof ppb (n : 5) for CH4, and _+1.7ppb (n = 5) for N20. For the whole systemand the sublimationof the outer layersof the comparison,measurementsby other laboratoriesusingdifferice sample.The leakagerate of the sublimationapparatuswas ent techniquesfor extraction and analysisof the air samples determinedfrom the pressureincreasewith the vacuumpump are alsoplotted in Figures2-4. Figure 2 alsoshowsthe annual shutoff andwasfoundto be2.24x 10--6hPas- •, correspond- meansof COg mixingratios measuredat Mauna Loa [Keeling ingto a volumeof 0.0040cm3 (STP)of ambientair duringa et al., 1989]. Other COg and CH 4 measurementson the Siple sublimationperiod of 30 min. Related to the typicalair sample ice core in Figures 2 and 3, respectively,were originally resizeof 2 cm3 (STP),the contamination fromleakagerepre- ported by Neffel et al.  and Staufferet al. . CH 4 sents --•0.2%. In comparisonwith the experimentallydeter- measurementsin Figure 3 were later corrected by Blunier mined blanks the contaminationfrom leakage contributesto  on the basisof six new measurementswith improved the blanksonly to a small extent. millingtechnique[Fuchset al., 1991].So far, no N20 data have The strongestcontaminationsourceis apparentlydesorption been reported for the Siple core. Consequently,Figure 4 shows from the systemswalls and outgasingof the Viton-o-ringsand N20 data reported by Machida et al.  for the Japanese Teflon-o-rings used to seal vesselsA and B and the glass ice corecollectedin 1991at H15 (Antarctica,69ø05'S,40ø47'E, stopcocks.This experienceis in agreementwith that reported 1057 m above sea level, annual mean temperature and accuby Zumbrunnet al. , who found that water vapor dis- mulationrate -20.5øCand26 g cm-2 yr-•, respectively) with places COg and possiblymoleculesof other gasesfrom the a mechanical drill and with no fluid. apparatuswalls. Blank runs of the apparatuswithout ice and Figure 2 showsa goodagreementof our COg measurements without coolingdownthe cold trapsbut with the condensingof obtained by the sublimation extraction procedure and FMaddedair in the coldfingershowed thata 2 cm3 (STP)large TDLAS analysiswith data obtained by milling extractionand
GOLLOK ET AL.: GASESIN AIR EXTRACTED BY SUBLIMATION FROM ICE
dry extractionby milling
Figure 3. Temporalchangeof CH4 mixingratiosbetween1772and 1973as obtainedfrom analyses of ice coresfrom Sipledrilling(75ø55'S,83ø55'W).Data determinedby Staufferet al. andBlunieron the sameice core usingdry extractionby milling are also shown.
analysisby TDLAS analysis[Neftelet al., 1985;Blunier,1992]. Our measurements tend to be 4 ppm higher,but the numberof measurementsis too smallto find any pattern in this tendency. The most probablereasonfor higher CO2 valuesis that the apparatusblank usedfor correctingthe CO2 measurementsis smallerthan the actualone. However,the differenceof 4 ppm is still within the combineduncertaintyof our CO2 blanksof _+1.4ppm, of our CO2 measurements(extractionincluding analyses) of _+2.0ppm(bothaddingup to 2.4ppm),andof the uncertaintyof CO2 measurements in Bern of 2 ppm [Neftelet al., 1985;Friedliet al., 1986].The air sampledatingfrom 1973 isjustfrom belowthe transitionzonewherethe permeabilityof very large ice samples,but not necessarilyof small ones, is zero.This addssomeuncertaintyfor the datingof air from this
determinedusinga millingdevicefor extractionandgaschromatographyfor analysis[Staufferet al., 1985;Blunier,1992]. Figure4 showsalsoa very goodagreementbetweenour N20 dataandthosereportedbyMachidaet al. for Antarctica for the sameperiod of time. Their data were obtainedby milling 400-600 g ice [Nakazawaet al., 1993a,b] and gas chromatographicanalysis. 3.3.
Our CO2,CH4, andN20 measurements in ice samplesfrom the 3G and 4G2 Vostok drillingsare summarizedin Table 1. It showsalsoCO2 and CH 4 measurements madeby Laboratoire
de Glaciologie et Geophysique de L'Environment(LGGE) in Grenoble[Barnolaet al., 1987;Jouzelet al., 1993]for compar-
ice sample. ison.They usedcrushingand meltingtechniquesto extractair Figure3 showsthat our CH4 measurements obtainedby the samplesfor CO2 and CH 4 determination,respectively.Both sublimationextractiontechniqueand analysisby FM-TDLAS gaseswere measuredby gaschromatography. spectroscopy are in excellentagreementwith the earlier results Consideringthe combineduncertaintyof our and LGGE's
sublimation technique,Sipleice core dry extractionby milling, H15 ice core
Figure 4. Temporalchangeof N20 mixingratiosbetween1772and 1973as obtainedfrom analyses of ice coresfromSipledrilling(75ø55'S,83ø55'W).For comparison, N20 datareportedbyMachidaet al. on the H15 core from Antarctica
are also shown.
GOLLOKET AL.: GASESIN AIR EXTRACTEDBY SUBLIMATIONFROM ICE Table
of the Measurements
on Ice Cores 3G and 4G2 Drilled
Obtainedby the SublimationTechnique(IFU) Laboratorya
3G Drilling (Maximum Depth 2083 m) 187 + 3 (n : 2) 370 _+13
N20 , ppb
272 + 3 (n = 2)
191 _+ 8 345 _+ 23
4G2 Drilling (Maximum Depth 2546 m) 207 + 3 (n : 6) 484 _+36 (n : 3) 199
269 _+4 (n : 5)
Data on the samecoresbut neighboringdepthsperformed by LGGE usingmechanicaldisintegration for CO2 and melting for CH 4 are added for comparison. alFU, Fraunhofer-Institutffir AtmospharischeUmweltforschung;LGGE, Laboratoire de Glaciologieet
G?ophysique deL'Environment. øBarnolaet al. . CChappellaz et al. .
dJouzel et al. .
CO2 and CH 4 measurementsand the slightlydifferent depths of the ice samples,our CO2 and CH 4 data on the 3G core samplesfrom 2043.7 m depth are in excellentagreementwith measurementsmade at LGGE. The agreement of measurements on the 4G2 drilling is slightlyworse but still within the analyticaluncertaintiesof both laboratoriesand the additional uncertaintyimposedby the age differenceof the ice samples. So far, no N20 data is availablefor comparisonwith our results at this depth.
ment of our CH 4 and N20 measurementson Siple ice core with thosemade in Bern [Blunier,1992] and Japan[Machidaet al., 1995] suggeststhat these gasesare also not contained within the ice matrix to a significantdegree. The good agreementof our CO2 and CH 4 measurementson Vostok ice coreswith thosemade at LGGE in Grenoble suggeststhat the change in CO2 and CH 4 concentration due to clathrate formation, if it occurs, is below the combined uncertainty of our and LGGE's measurements.Jaworowskiet al.
[1992, 1994] suggestedthat CO2 measurementsmay be subject 4.
due to clathrate
The good agreement of our CO2 measurementswith those The sublimationtechniqueto extract air samplesfrom ice made by LGGE usingthe milling extractionproceduremakes cores combinedwith FM-TDLAS analysisprovided synchro- this artefact unlikely. nousCO2, CH 4, and N20 measurementson 20-50 g large ice The sublimation technique can also provide information samples.The amount of ice needed for measuring the three about the influence of chemicalreactionsbetween impurities componentsis smaller than most laboratoriesneed at present in the ice on the compositionof the air bubbles. If CO2 is for the measurementof a singlecomponent[e.g., Etheridgeet producedby chemicalreactionsbetween impurities in the ice al., 1988, 1992;Machidaet al., 1995].The multicomponentair [Anklinet al., 1995], the sublimationtechniquewill be able to analysisin small ice samplescan enablea more detailed inves- detect this CO 2 immediately after its production, while the tigation of the leads or lags in the changesof trace gas con- crushingtechniquesdetect it only after the CO2 has entered centrationswith the changeof climate [e.g., Raynaudet al., the air bubbles.Therefore the new extractiontechnique will 1993]. become an important tool for investigatingthe character and Good agreementwas observedbetween CO2 measurements mechanismsof chemicalreactionsleadingto a CO2 surplus. obtained by our sublimation extraction technique and measurementsof other laboratoriesusingthe crushingtechniques. The agreement of CO2 measurementsmade by sublimation Acknowledgments. We thank D. Zardini and H. E. Wagner for the and crushingtechniqueson samplesfrom Siple ice core dating initial help in the developmentof the sublimationextractiontechnique back 200 years suggeststhat the content of CO2 in the ice and LGGE in Grenoble for providingthe Vostok ice core samplesfor matrix, which would be inaccessibleto crushing extraction analysis.We are indebtedto T. Blunier from Universityof Bern and to techniques,is negligible.This observationis contraryto find- D. Raynaud and J.-M. Barnola from LGGE in Grenoble for discussionsof the experimentaldetails and exchangeof data for intercomingsby Wilson but in line with the good agreementof paringwith our results.We would alsolike to thank to W. Seller for his measurementsin ice with the direct measurementsof CO2 in encouragementof this work and two anonymousreferees for their air [Neftelet al., 1995; Etheridgeet al., 1992; Raynaudet al., constructivereviews of the original manuscript.The work has been 1993]. The major difference between our sublimation tech- funded by the European Communityunder the contractEPOC-CT910033 (DSCN) and by the German Ministry for Education and Renique and that of Wilson  is the collection of the ex- search under contract 07 KTF 95. tracted air sample. We freeze out the air sample at 14 K whereasWilsonisolatesCO2 from the air sampleby first freezing out CO2 in a cold trap at the temperature of liquid References nitrogen and subsequently collectingthe air sampleon a molecular sieve kept at the temperature of liquid nitrogen. A Anklin, M., J.-M. Barnola, J. Schwander,B. Stauffer,and D. Raynaud, further
the two methods
is the determina-
tion of blank values.Wilson does not add any air in the sublimation vesselfor his blank measurements.The good agree-
Processes affectingthe CO2 concentrationsmeasuredin Greenland
ice, Tellus, Set. B, 47, 461-470, 1995. Anklin, M., J. Schwander, B. Stauffer, J. Tschumi, A. Fuchs, J•-M.
Barnola, and D. Raynaud, CO2 record between 40 and 8 kyr B. P.
GOLLOK ET AL.: GASESIN AIR EXTRACTED BY SUBLIMATIONFROM ICE
from the Greenland Ice Corp Project ice core,J. Geophys.Res.,102, 26,539-26,545, 1997.
Barnola,J.-M., D. Raynaud,Y. S. Korotkevich,and C. Lorius, Vostok ice core provides160,000-yearrecord of atmosphericCO2, Nature, 329, 408-411, 1987.
Barnola,J.-M., P. Pimienta, D. Raynaud, and Y. S. Korotkevich,CO2climate relationship as deduced from the Vostok ice core: A reexamination
based on new measurements
and on a re-evaluation
the air dating, Tellus,Set. B, 43, 83-90, 1991. Battle, M., et al., Atmosphericgasconcentrations overthe pastcentury measuredin air from firn at the south pole, Nature, 383, 231-235,
Analysisof observationaldata, in Aspectsof ClimateVariabilityin the Pacific and WesternAmericas,Geophys.Monogr.,vol. 55, edited by D. H. Peterson,pp. 165-236, AGU, Washington,D.C., 1989. Leuenberger,M., and U. Siegenthaler,Ice-ageatmosphericconcentration of nitrous oxide from an Antarctic ice core, Nature, 360, 449-452, 1992. Lorius, C., J. Jouzel, C. Ritz, L. Merlivat, N. I. Barkov, Y. S. Korot-
kevich, and V. M. Kotlyakov,A 150,000-yearclimatic record from Antarctic ice, Nature, 316, 591-596, 1985.
Machida, T., T. Nakazawa,Y. Fujii, S. Aoki, and O. Watanabe, Increasein the atmosphericnitrousoxideconcentrationduringthe last 1996. 250 years, Geophys.Res.Lett., 22, 2921-2924, 1995. Blunier, T., Spurengasanalysen an Eisbohrkernen:Konzentrationsver- Nakazawa,T., T. Machida,K. Esumi,M. Tanaka, Y. Fujii, S. Aoki, and O. Watanabe, Measurementsof CO2, and CH 4 concentrationsof air lauf von atmospharischemMethan fiber die letzten 1000 Jahre gemessenan einem Bohrkern ausZentralgr6nlandund Entwicklung in polar ice cores,J. Glaciol.,39, 209-215, 1993a. eines Systemszur Analysevon Methan, Kohlendioxidund Distick- Nakazawa, T., T. Machida, M. Tanaka, Y. Fujii, S. Aoki, and O. Watanabe, Differences of the atmosphericCH 4 concentration stoffoxidan Eisproben,M. S. thesis,69 pp., Inst. for Phys.,Univ. of Bern, 1992. between the Arctic and Antarctic regions in preindustrial/ preagriculturalera, Geophys.Res.Lett., 20, 943-946, 1993b. Chappellaz,J., J. M. Barnola, D. Raynaud, Y. S. Korotkevich,and C. Lorius, Ice-core record of atmosphericmethane over the past Neftel, A., E. Moor, H. Oeschger,and B. Stauffer, Evidence from 160,000years,Nature, 345, 127-131, 1990. polar ice coresfor the increasein atmosphericCO2 in the past two centuries, Nature, 315, 45-47, 1985. Chappellaz,J., T. Blunier, D. Raynaud, J. M. Barnola, J. Schwander, and B. Stauffer, Synchronouschanges in atmospheric CH 4 and Oeschger,H., and C. C. LangwayJr., The EnvironmentalRecordin Glaciersand Ice Sheets:Physical,Chemical,and Earth SciencesReGreenland climate between40 and 8 kyr BP, Nature, 366, 443-445, 1993. searchReport8, JohnWiley, New York, 1989. Delmas, R. J., A natural effect in Greenland ice-core CO2 measure- Powell, C. S., Cold confusion,Sci.Am., 270(3), 13-15, 1994. ments, Tellus, Set. B, 45, 391-396, 1993. Raynaud,D., J. Jouzel,J.-M. Barnola,J. Chappellaz,R. J. Delmas,and C. Lorius, The ice record of greenhousegases,Science,259, 926Elbaum, M., S. G. Lipson, and J. G. Dash, Optical studyof surface 934, 1993. melting on ice, J. CrystalGrowth, 129, 491-505, 1993. Etheridge,D. M., G.I. Pearman,and F. de Silva,Atmospherictrace- Schwander,J., The transformation of snow to ice and the occlusionof gases,in TheEnvironmentalRecordin Glaciersand Ice Sheets,edited gas variations as revealed by air trapped in an ice core from Law Dome, Antarctica, Ann. Glaciol., 10, 28-33, 1988. by H. Oeschgerand C. C. LangwayJr., pp. 53-67, JohnWiley, New York, 1989. Etheridge,D. M., G.I. Pearman,and P. J. Fraser, Changesin troposphericmethanebetween1841 and 1978 from a high accumulation- Schwander,J., and B. Stauffer, Age differencebetweenpolar ice and rate Antarctic ice core, Tellus, Set. B, 44, 282-294, 1992. the air trapped in its bubbles,Nature, 311, 45-47, 1984. Friedli, H., H. L. F6tscher, H. Oeschger, U. Siegenthaler, and Staffelbach, T., B. Stauffer, and A. Sigg, CO2 measurementsfrom B. Stauffer,Ice corerecordof the •3C/•2Cratioof atmospheric CO2 polar ice cores:More data from different sites, Tellus,Set. B, 43, 91-96, 1991. in the past two centuries,Nature, 324, 237-238, 1986. Fuchs,A., J. Schwander,and B. Stauffer, A new mill allowsprecise Stauffer, B., G. Fischer, A. Neftel, and H. Oeschger,Increase of concentrationdetermination of methane and most probably also atmosphericmethane recorded in Antarctic ice core, Science,229, 1386-1388, 1985. other trace gasesin the bubble-air of very small ice samples,J. Glaciol., 39, 199-203, 1991. Wilson, A. T., Application of AMS C-14 dating to ice core research, Radiocarbon, 37, 637-641, 1995. Furukawa, Y., and I. Ishikawa,Direct evidencefor melting transition at interfacebetweenice crystaland glasssubstrate,J. CrystalGrowth, Wilson, A. T., and D. J. Donahue, AMS carbon-14 dating of ice: 128, 1137-1142, 1993. Progressand future prospects,NuclearInstrum.MethodsPhys.Res., Sect. B, 52, 473-476, 1990. Furukawa, Y., M. Yamamoto, and T. Kuroda, Ellipsometricstudyof the transitionlayer on the surfaceof an ice crystal,J. CrystalGrowth, Zardini, D., D. Raynaud,W. Scharffe,and W. Seiler, N20 measure82, 665-677, 1987. ments on air extracted from Antarctic ice cores: Implications of Gtillfik, T., H. E. Wagner, and F. Slemr, A high-frequencymodulated atmosphericN20 back to the last glacial-interglacialtransition,J. Atmos. Chem., 7, 189-201, 1989. tunable diode laser absorptionspectrometerfor measurementsof CO2, CH4, N20, and CO in air samplesof a few cm3, Rev.Sci. Zumbrunn, R., A. Neftel, and H. Oeschger,CO2 measurementson 1 Instrum., 68, 230-239, 1997.
Haan, D., P. Martinerie, and D. Raynaud,Ice core data of atmospheric carbonmonoxideover Antarctica and Greenland during the last 200 years, Geophys.Res. Lett., 23, 2235-2238, 1996. Jaworowski,Z., Ancient atmosphere:Validity of ice records,Environ. Sci. Pollut. Res., 1, 162-171, 1994.
Jaworowski,Z., T. V. Segalstad,and N. Ono, Do glacierstell a true atmosphericCO2 story?,Sci. Total Environ.,114, 227-284, 1992. Jouzel, J., et al., Extending the Vostok ice-core record of palaeoclimate to the penultimateglacialperiod,Nature,364, 407-412, 1993. Keeling, C. D., R. B. Bacastow,A. F. Carter, S.C. Piper, T. P. Whorf, M. Heimann, W. G. Mook, and H. Roeloffzen, A three-dimensional
model of atmosphericCO2 transport basedon observedwinds, 1,
cm3 ice samples with an IR laserspectrometer (IRLS) combined with a new dry extractiondevice,Earth Planet. Sci. Lett., 60, 318324, 1982.
T. Gfillfik and F. Slemr, Fraunhofer-Institut ffir Atmospharische Umweltforschung (IFU), Kreuzeckbahnstra13e 19, D-82467 GarmischPartenkirchen,Germany.(e-mail: [email protected]
) B. Stauffer, Physikalisches Institut der Universitat Bern, Sidlerstra13e5, CH-3015 Bern, Switzerland.
(ReceivedAugust25, 1997;revisedJanuary29, 1998; acceptedFebruary24, 1998.)