First Stellar Abundances in the Dwarf Irregular Galaxy Sextans A
FirstStellarAbundancesintheDwarfIrregularGalaxySextansA2
AndreasKaufer
arXiv:astro-ph/0401411v1 21 Jan 2004EuropeanSouthernObservatory,AlonsodeCordova3107,Santiago19,Chile
akaufer@eso.orgKimA.Venn
InstituteofAstronomy,UniversityofCambridge,MadingleyRoad,Cambridge,CB30HA,
UK,andMacalesterCollege,1600GrandAvenue,SaintPaul,MN,55105,USA
venn@macalester.edu
ElineTolstoy
KapteynInstitute,UniversityofGroningen,POBox800,9700AVGroningen,The
Netherlands
etolstoy@astro.rug.nl
ChristophePinte
´EcoleNormaleSup´erieure,45rued’Ulm,F-75005,Paris,France
christophe.pinte@ens.fr
and
RolfPeterKudritzki
InstituteforAstronomy,UniversityofHawaiiatManoa,2680WoodlawnDrive,Honolulu,
Hawaii96822,USA
kud@ifa.hawaii.eduABSTRACT
–2–
WepresenttheabundanceanalysesofthreeisolatedA-typesupergiantstarsinthedwarfirregulargalaxySextansA(=DDO75)fromhigh-resolutionspectraobtainedwithUltraviolet-VisualEchelleSpectrograph(UVES)ontheKueyentelescope(UT2)oftheESOVeryLargeTelescope(VLT).Detailedmodelatmo-sphereanalyseshavebeenusedtodeterminethestellaratmosphericparametersandtheelementalabundancesofthestars.Themeanirongroupabundancewasdeterminedfromthesethreestarstobe[(FeII,CrII)/H]=−0.99±0.04±0.061.Thisisthefirstdeterminationofthepresent-dayirongroupabundancesinSex-tansA.Thesethreestarsnowrepresentthemostmetal-poormassivestarsforwhichdetailedabundanceanalyseshavebeencarriedout.
Themeanstellarαelementabundancewasdeterminedfromtheαelementmagnesiumas[α(MgI)/H]=−1.09±0.02±0.19.Thisisinexcellentagree-mentwiththenebularαelementabundancesasdeterminedfromoxygenintheHIIregions.Theseresultsareconsistentfromstar-to-starwithnosignificantspatialvariationsoveralengthof0.8kpcinSextansA.Thissupportstheneb-ularabundancestudiesofdwarfirregulargalaxies,wherehomogeneousoxygenabundancesarefoundthroughout,andarguesagainstinsitu(“onthespot”)enrichment.
Theα/Feabundanceratiois[α(MgI)/FeII,CrII]=−0.11±0.02±0.10,whichisslightlylowerbutconsistentwiththesolarratio.ThisisconsistentwiththeresultsfromA-supergiantanalysesinotherLocalGroupdwarfirregulargalaxies,NGC6822andWLM.Theresultsofnearsolar[α/Fe]ratiosindwarfgalaxiesisinstarkcontrastwiththehigh[α/Fe]resultsfrommetal-poorstarsintheGalaxy(whichplateauatvaluesnear+0.4dex),andismostclearlyseenfromthesethreestarsinSextansAbecauseoftheirlowermetallicities.Thelow[α/Fe]ratiosareconsistentwiththeslowchemicalevolutionexpectedfordwarfgalaxiesfromanalysesoftheirstellarpopulations.
Subjectheadings:galaxies:abundances,irregular,individual(SextansA)—stars:abundances
–3–
1.
Introduction
Dwarfirregulargalaxies(dIrrs)arelowmass,butgasrichgalaxiesandarefoundratherisolatedandspreadthroughouttheLocalGroup.AlldIrrsdisplaylowelementalabundancesindicatingthatonlylittlechemicalevolutionhastakenplaceoverthepast15Gyrdespiteongoingstarformation.Thelackofstrongstarburstcyclesintheseisolatedobjectsmaybebecauseoflittleornomergerinteraction.Therefore,inthecontextthecolddarkmatterscenariosofhierarchicalgalaxyformationbymergerofsmallerstructures(Steinmetz&Navarro1999),dwarfgalaxiesandinparticulartheisolateddwarfirregulargalaxiescouldbethepurestremnantsoftheproto-galacticfragmentsfromtheearlyUniverse.Hence,thedIrrsarealsooneofthepossiblesourcesforthedampedLyαabsorption(DLA)systemsasobservedinquasarspectraoveralargerangeofredshifts;seee.g.Prochaskaetal.(2003)forarecentcompilationofDLAmetallicitiesover0.5 –4– ofthegalaxy. Unfortunately,mostdIrrsystemsaretoodistantforthedetailedstudyandabundanceanalysesofe.g.theirredgiantbranch(RGB)stars,whichareofgreatimportanceduetotheirlargespreadinage.Hence,todatefewdetailsonthechemicalevolutionofthedIrrgalaxieshavebeenavailable.InthecaseofthedIrrSextansAstudiedinthisworkthetipoftheRGBisfoundatvisualmagnitudesofV≈23(Dohm-Palmeretal.2002)—farbeyondthecapabilitiesofeventhemostefficienthigh-resolutionspectrographsontoday’s8to10-meterclasstelescopes.However,withthesameclassoflatesttelescopesandinstrumentation,thevisuallybrighteststarsofthedIrrgalaxies,i.e.,thebluesupergiants(17.5V20inSextansA)becomenowaccessibletodetailedspectroscopy,whichallowustodeterminethepresent-dayabundancesofαandirongroupelements.Theabundancesoftheαelementsderivedfromtherichspectraofhotmassivestarscanbedirectlycomparedandtiedtothenebularαelementabundancesbecausebothnebulaeandmassivestarshavecomparableagesandthesameformationsites.Thefirstabundancestudiesbasedonhigh-resolutionspectraofA-typesupergiantstarswerecarriedoutforthenearbyLocalGroupdIrrsNGC6822andWLM(Vennetal.2001,2003a).αelementandirongroupabundancescouldbederivedfortwostarsinbothdIrrgalaxies.ForNGC6822andWLMthe[α/Fe]ratiosareingoodagreementwiththesolarratio.Thisresultissomewhatsurprisingsincebothgalaxieshaveongoingstarformationanddifferentstarformationhistoriesasdeterminedfromtheirstellarpopulations(e.g.,Gallartetal.(1996),Mateo(1998),Dolphinetal.(2003a)). TheisolateddwarfirregulargalaxySextansA(=DDO75)withitsdistanceofabout1.3Mpc(Dolphinetal.2003a)islocatedatasimilardistanceasSextansB,NGC3109,andtheAntliadwarfgalaxieswhichformasmallgroupofgalaxiesattheedgeoftheLocalGroup—possiblynotevenboundtotheLocalGroupwhichwouldmakeSextansAamemberofthenearest(sub)clusterofgalaxies(vandenBerg1999).Thepresent-daychemicalcompositionwasstudiedfromtheemission-linespectroscopyofbrightcompactHIIregions(Hodgeetal.1994).TheanalysisbySkillmanetal.(1989)determinesanoxygenabundanceof12+log(O/H)=7.49whichcorrespondsto[O/H]=−1.17ifasolaroxygenabundanceof12+log(O/H)⊙=8.66isadopted(Asplund2003).InarevisedanalysesofthesamedatabyPilyugin(2001),ahighervalueof12+log(O/H)=7.71isfound,correspondingto[O/H]=−0.95.ThestarformationhistoryofSextansAwasfirststudiedindetailbyDohm-Palmeretal.(1997),andmorerecentlybyDohm-Palmeretal.(2002)andDolphinetal.(2003b)usingHSTVIcolor–magnitudediagrams(CMD).Ahighrateofstarformationhasoccuredinthepast2.5Gyr,withverylittlestarformationbetween2.5and10Gyrs.Ameanmetallicityof[M/H]≈−1.45±0.2isderivedoverthemeasuredhistoryofthegalaxy,withmostoftheenrichmenthavingoccurredmorethan10Gyrago.Onlyslightenrichments(+0.4dex)occuredbetween2and10Gyrfromfittingtheredgiantbranchtoametallicity –5– of[M/H]≈−1.1. Inthispaper,wepresenttheresultsfromadetailedabundanceanalysisofthreeisolatedandnormalAsupergiantstarsinSextansAbasedonhighresolutionspectraobtainedattheESO-VLTusingUVES.Wecomparethestellarresultstothosefromnebularanalysestoexaminethechemicalhomogeneityofthisyounggalaxy.Wealsousetherichstellarspectratocomparetheabundancesofseveralotherelementsnotavailableinthenebularanalysestothosepredictedfromitschemicalevolutionhistory,aswellastoresultsfromotherdwarfgalaxies. 2.Observations Alistofpotentialtargetsforhigh-resolutionUVES(Dekkeretal.2000)spectroscopyattheESOVLTwasselectedfromthephotometriccataloguebyVanDyketal.(1998).Thevisuallybrightest(V<19.8)starswithcolorsrangingfrom−0.10<(B−V)<0.30werechosenfromthecorrespondingCMD.TheselectedtargetswerecheckedforpossiblecrowdingonHST/WFPC2imagesfromDohm-Palmeretal.(1997).Foroneofthetargets,thestarnumber754inthecatalogueofVanDyketal.(1998)(inthefollowingidentifiedas“SexA-754”),alowresolutionKeck/LRIS(Okeetal.1995)spectrumwasobtainedbyKobulnicky(1998),whichconfirmedthatthisstarisamemberofSextansA(cf.below)andanormalAsupergiantsuitedforabundanceanalyses.Almost14hoursofVLT/UVESspectroscopyofSexA-754werecarriedoutinservicemodeinbetweenJanuaryandFebruary2002withinthespecifiedconstraintsofaseeingbetterthan1.2′′,adarkandclearsky,andanairmass<1.5.AdditionalspectraincomplementarywavelengthsettingsintheredwereobtainedinavisitormoderuninApril2002byAK,KAV,CP.InparallelKeck/ESI(Scheinisetal.2002)lowresolutionspectroscopyofotherpotentialtargetswasobtainedbyKAVinFebruary2002.Forthebestcandidates,theESIlow-resolutionspectrawerecomplementedbyhigh-resolutionspectratakenwithUVESintheApril2002visitorrun.AllcandidatesstarswerefirstcheckedfortheirmembershiptotheSextansAgalaxybycomparingthemeasuredradialvelocitiesofthestarstotheheliocentricradialvelocityofSextansAasdeterminedfromopticalandradiospectroscopy:Tomitaetal.(1993)determinedvaluesof325−335km/sfromtheHαemissionlinesofdifferentHIIregions,whichareconsistentwithearliermeasurementsoftheradialvelocitiesoftheHIgas,i.e.,324±3(Huchtmeier&Richter1986)and325±5km/s(Skillmanetal.1988).DuetothelargeheliocentricradialvelocityofSextansA,anyconfusionofourstellartargetswithforegroundstarsisveryunlikely. BasedonthecollectedESIandUVESspectratwoadditionalstars(SexA-1344and –6– SexA-1911)wereidentifiedfor2times12.5hoursofVLT/UVESservicemodeobservations,whichwereexecutedbetweenDecember2002andFebruary2003withinthesameconstraintsasgivenforSexA-754before.ThecandidatestarsSexA-513,SexA-983,andSexA-1456ontheotherhandwerefoundtobenotsuitedforourabundancesanalyseswhicharetailoredforAsupergiantsofanarrowtemperatureandluminosityrangeandthethreestarshadtobediscarded. AcompletelogoftheESIandUVESobservationsisfoundinTab.1togetherwithsomeadditionalinformationontheinstrumentalconfigurationsusedfortheindividualob-servationsandtheenvironmentalconditionsduringtheactualobservations.ThecompletesampleofexaminedSextansAtargetsispresentedinTab.3togetherwiththeradialvelocityandspectraltypeinformationasobtainedeitherduringthedetailedanalysesofthestarsorforthediscardedstarsduringthepre-selectionprocess;thelocationoftheindividualtargetswithinthegalaxyfieldisshowninFig.1togetherwiththepositionsofthefourbrightcompactHIIregionsanalyzedbySkillmanetal.(1989). AllUVESdatawerereducedusingthededicatedUVESpipelinewhichisdescribedbyBallesteretal.(2000)andisavailableasaspecialpackagewithintheESO-MIDAS3datareductionsystem.Optimumextractionwithcosmicrejectionandskysubtractionwasusedtoextracttheindividualstellarspectra.Atypicalsignal-to-noiseratio(SNR)of≈9inthewavelengthregimecorrespondingtotheBbandwasobtainedforasinglespectrumof4500secexposuretime.Thetypically10individualspectraperstarwerethencorrectedforthespectrograph’sresponsefunctiontoobtainasmoothcontinuum,rebinnedtohelio-centricvelocities,andeventuallycombinedwithadditionalcosmicclipping.Thecombinedspectrawereresampledandrebinnedtoa2-pixresolvingpowerofR=20000whichwasusedthroughouttheanalyses.ThereducedresolvingpowerallowstoincreasetheSNRperresolutionelementwithouthavingtosacrificelineprofileinformation.Theprojectedrotationalvelocitiesofthethreestarswerelaterdeterminedtobe>30km/s,i.e.,alllineprofilesremainfullyresolvedatthereducedresolutionof15km/s.TheSNRsobtainedforthecombinedUVESspectraarelistedinTab.2fordifferentwavelengthregionsofinterest.TheobservationsofindividualstarsinadistantLocalGroupgalaxyaspresentedhereareatthelimitofthecapabilitiesoftoday’slargesttelescopesandmostsensitivespectrographs:despitetotalintegrationtimesoftypically12.5hrsperstar,SNRsof30areachievedatbestforaresolvingpowerofR=20000foraV=19.5magstarwhichisdeemedjustsufficientfordetailedabundancesanalysesaswillbeshowninthiswork. InFig.2theHαprofilesofthethreestarsareshown.Toobtainthestellarlineprofiles, –7– thecontributionofthediffuseHαemissionofSextansAwhichwasfoundforallstarswassubtractedaspartofthe’sky’subtractionintheextractionprocessofthespectra.Inspectingtheraw2Dspectra,thediffuseHαemissiondoesshowspatialstructurebutaveragedalongtheslititsdistributionismostlyGaussian.Thediffuseemissionisblueshiftedbyanaverageof−10km/swithrespecttothestellarspectraandconsistentwiththeHIradialvelocityofSextansA(Huchtmeier&Richter1986).TheaverageFWHMoftheemissionis47km/scorrespondingtoa(1σ)velocitydispersionof20km/s.Thisisinexcellentagreementwithathermalvelocity(22km/s)ofawarminterstellarmediumifatemperatureof10000Kisassumed. Inthesky-subtractedstellarHαprofilesinFig.2nostellarline-emissionisvisibleinHαwhichindicatesthattheselectedstarsaresupergiantsoflowerluminosity.Thiswasalreadyexpectedfromthefactthatduringtheselectionprocessonlystarsabout2magnitudesfainterthanthebrightestbluestarsinSextansA(SexA-513)turnedouttobenormalAsupergiants.Interestingly,nobrighter,moreextremeAsupergiants,whicharesupposedtobethevisuallybrighteststars,werefoundinthewholegalaxy.ThesameobservationwasmadeduringthestudyofthedIrrWLM(Vennetal.2003a).TheselectionprocessappliedtofindcandidatestarsforabundanceanalysesinSextansAandWLMwasnotrigorousenoughtoexcludethatthisfindingisaselectioneffectintrinsictoourprocedure.However,thelackofluminousAsupergiantsmightaswellbeanevolutioneffectduetothelowmetallicityofthegalaxy.Similarly,thelackofwindemissionfeaturesinHαasseeninFig.2mightbeduetothelowmetallicitysuppressingtheformationofstellarwindsduetothereducedradiationline-drivingonthemetal-poorgas. Lower-luminositysupergiantsaswerefoundinSextansAarebestsuitedforabundanceanalyes.Higher-luminositystarsaredifficulttoanalyzebecausetheline-formingregionsoftheiratmospheresarenolongerinhydrostaticequilibriumbutarefoundintheaccelerationzoneofthestellarwind.Further,forlower-luminosityAsupergiants,theeffectsduetointrinsicphotosphericandwindvariabilityonthelineprofilesisnegligible(Kauferetal.1996).Equivalentwidthvariationsofweak(Wλ<200m˚A)photosphericlinesasusedfortheabundanceanalysesinthisworkarefoundtobe≤10%(Kauferetal.1997). 3.AtmosphereAnalyses TheatmosphereanalysesofthethreestarsinSextansAhavebeencarriedoutinlinewithpreviousanalysesofAsupergiantstarsintheGalaxy(Venn1995a,b),theMagellanicClouds(Venn1999),M31(Vennetal.2001)andthedwarfirregulargalaxiesNGC6822(Vennetal.2001)andWLM(Vennetal.2003a).Maintainingaconsistentanalysisprocess –8– acrossthedifferentstudiesofAsupergiantstarsisconsideredcrucialtoobtainconsistentandcomparableabundanceresults.Inparticularresultsfromdifferentialanalysescanthenbeconsideredthemostreliablebecausetheuncertaintiesinthedeterminedabsoluteelementalabundancesshallcanceloutatleasttosomeextend. Theatmospherecomputationsarebasedonplaneparallel,hydrostatic,andline-blanketedATLAS9LTEmodelatmospheres(Kurucz1979,1988,1993)whichareconsideredtobeap-propriateforthephotosphereanalysisoflowerluminosityAsupergiants(Przybilla2002).ToaccountforthepossibleeffectsofthelowmetallicityofSextansAonthemodelatmo-spherestructure,allatmospheremodelshavebeencomputedwithametallicityscaledto1/10ofsolarandacorrespondingopacitydistributionfunction.Thevalueof[M/H]=−1waschoseninagreementwiththepresentdayoxygenabundancesasderivedfromtheanal-ysisofHIIregions(Skillmanetal.1989).However,ashasbeenshownlater,theeffectofatmospheremodelswithdifferentmetallicitiesonthederivedelementalabundancesistypicallylessthan0.05dex,i.e.,negligiblewithintheestimatedsystematicerrors. TheanalyseshasbeenrestrictedtotheuseofweakabsorptionlinesonlytominimizetheeffectsoftheneglectedsphericalextensionandoftheneglectedNLTEconditionsintheseextended,lowgravitysupergiantstarsonthemodelatmospheres.Further,weaklinesarepreferredbecausetheysufferlessfromNLTEandmicroturbulenceeffectsinthelineformationprocess.Weaklinesinthiscontextaredefinedaslineswhereachangeinthemicroturbulenceξof∆ξ=1km/sresultsinachangeintheelementalabundanceoflog(X/H)≤0.1dex.Tofulfillthisweak-linecondition,theequivalentwidthofthecorrespondinglinehastypicallytobeWλ<160m˚A.Weak-lineanalysesareachallengeforfaintandmetal-poortargetsasanalyzedinthiswork,inparticularifthegalaxiesarefaraway,thetargetsarefaintandtheSNRofthehigh-resolutionspectraiscomparativelylowasitisthecasehere.TheabsorptionlineshavebeenmeasuredbyfittingGaussianprofilestotheweaklineprofilesandtothestellarcontinuum,thelatterbeingthemainsourceofuncertaintyinthederivedequivalentwidthsWλ.ThecomparativelylowSNRleavesanon-negligibleuncertaintyofabout10%inthedefinitionofthecontinuumlevelandthereforeinthemeasuredWλinthebluepartofthecombinedspectra. ThestellarmodelatmosphereparameterseffectivetemperatureTeff,gravitylogg,andmicroturbulenceξareinthefollowingdeterminedsolelyfromspectralfeaturesandthereforearenotaffectedbythe(badly-defined)extinction. ThetemperatureandgravityparametersofthebestfitsofsyntheticBalmer-lineprofilestothegravity-sensitivewingsoftheHγprofiles(seeFig.3)definearelationintheTeff–loggplaneasshownforeachstarinFig.5.ThetemperatureandgravityparametersforwhichanionizationequilibriumofFeIandFeIIisfounddefinesatemperature-sensitive –9– relationintheTeff–loggplane(foradetaileddiscussionoftheFeINLTEcorrectionsseebelow).TheintersectionofbothrelationsdefinesthepositionintheTeff–loggplaneforwhichboth,thebestfittotheHγprofileandtheionizationequilibriumofFeIandFeIIareachieved.Thisparameterpairisadoptedastheatmosphereparametersofthestar.DuetothelowtemperatureoftheanalyzedstarsandthecomparativelylowSNRofthecombinedspectra(cf.Tab.2)noneoftheweakspectrallinesofMgIIpresentintheobservedspectralrangescouldbeusedtodeterminethestellarparametersfromthemostreliableionizationequilibriumofMgIandMgII.Alsothestrong,easytomeasureMgIIλ4481linehadtobediscardedbecauseofitsknownhighsensitivitytoNLTEandmicroturbulenceeffects:NLTEcorrectionsareoftheorderof−0.3dexforthetemperatureandgravitiesofthestarsanalyzedhere;anuncertaintyofthemicroturbulenceof1km/sresultsinachangeofthecomputedabundancesofthesameorder.Instead,asalreadydescribedabove,theionizationequilibriumofFeIandFeIIwasusedinthiswork.ThisintroducesanadditionaluncertaintytothedeterminationofthestellarparametersbecauseofthesensitivityofFeIlinestoNLTEeffects(Boyarchuketal.1985;Gigas1986;Rentzsch-Holm1996).TheNLTEeffectsonFeIIlinesareconsiderednegligible(Rentzsch-Holm1996;Becker1998).ToquantitativelyincorporatethedependenciesoftheFeINLTEcorrectionsontemperature,gravity,andmetallicityontheFeI/FeIIionizationequilibrium,theresultsfromRentzsch-Holm(1996)presentedinherFigs.4and5wereusedtoderivearelationfortheNLTEcorrectiontobeappliedtotheFeIabundancesforthelimitedtemperaturerangeofinterestinthiswork,i.e.,7000 Teff[K] −0.27 1000 −0.1[M/H]−0.28 (1) ThefirstandsecondtermsdescribethetemperatureandgravitydependentNLTEcorrectionasderivedforsolarmetallicitymainsequence(logg≈4.0)stars.ThethirdtermdescribesthemetallicitydependenceoftheNLTEcorrectionasderivedformetallicitiesintherange[M/H]=±0.5.Duetotheconsiderablylowergravitiesandmetallicitiesofthestarsstudiedhere,thisrelationhastobeusedwithcaution. TotesttherelationforstarsoflowergravityandmetallicityasstudiedinthisworktherelationwasappliedtothedatasetofAsupergiantsintheSMCasanalyzedbyVenn(1999).ThestellarparametersoftheSMCstarsareallbasedonthemostreliableMgIto –10– MgIIionizationequilibrium.TheFeINLTEcorrectionsweredeterminedas∆logǫFeI=log(FeII/H)−log(FeI/H)assumingthattheNLTEcorrectionsforFeIIarenegligible(Becker1998;Rentzsch-Holm1996).ItisfoundthattheaboverelationfitstheSMCdatawellifanadditionalcorrectionof−0.28isadded,cf.Fig.4.ItisimportanttonotethattherelationdescribesratherwellallSMCstarswithFeItoFeIIabundancedifferencesoverthewholerangefrom0.0to0.3dex.Thisconstantadditional(fourth)termintherelationseemssufficienttoabsorbtheadditionalNLTEeffectsduetolowgravityandlowmetallicity,atleastforthenarrowtemperaturerangeof7000−9000K,gravitiesoflogg≈1.5,andmetallicitiesof[M/H]≈−0.7,i.e.,closetothestellarparametersofthestarsstudiedinSextansA.Inthesubsequentatmosphereanalysis,thefinalstellarparametersfortheSextansAstarswerechosentobeconsistentwiththeFeINLTEcorrectionsasgivenbytheaboverelation.Theadoptedfinalcorrectionsare+0.23dexforSexA-754,+0.30dexforSexA-1344,and+0.20dexforSexA-1911andareshowninFig.4,too.Itshouldbefurthernotedthattheremaininguncertaintyof±0.1dexintheFeINLTEcorrectionshasbeentakenintoaccountintheoverallerrorbudgetasestimatedfromtheuncertaintiesofthedeterminationofthestellarparameters(cf.below). MicroturbulenceξhasbeendeterminedfromthelineabundancesofthenumerouslinesofdifferentstrengthfromtheionsFeII,TiII,andCrII.Asinglevalueofξforeachstarwasadoptedsothatnotrendofthecomputedelementalabundanceisfoundwiththeequivalentwidthofthelineswithintheuncertaintyof∆ξ=1km/s.TheestimateduncertaintiesinTeffaredominatedbytheuncertaintiesofthederivedFeINLTEcorrectionsbutalsohavetoaccountfortheuncertaintiesoftheassumptionofnegligibleNLTEeffectsonFeIIintheionizationequilibriumusedforthetemperaturedetermination.TheeffectofachangeoftheFeINLTEcorrectionof±0.08dex(asderivedfromtheresidualsinFig.4)isshowninFig.5forstarSexA-754.Ifweassumeanuncertaintyof±0.05dexfortheLTEassumptionforFeII,weobtainanestimateforthetotaluncertaintyof±0.1dexwhichtranslatesinto∆Teff=±200KasindicatedbythehorizontalerrorbarsinFig.5.Uncertaintiesingravityareestimatedto∆logg=±0.2dexfromvaryingthefitstoHγfordifferentloggatconstantTeff.Theresultingchangeofthesyntheticprofileforavariationof∆logg=+0.2dexisshowninFig.3.Withthedeterminedstellarparametersandelementalabundances(cf.below),asmallsectionofthespectrumbetween4285−4305˚Awassynthesizedtodeterminetheprojectedrotationalvelocitiesvsiniofthethreestars.ThisspectralregioncontainseveralstrongerandisolatedTiIIandFeIIlines.ForagivenresolvingpowerofR=20000or15km/sandthegivenheliocentricradialvelocitiesofthestarstherespectivevaluesofvsiniwereobtainedbyleast-squarefittingofthesyntheticlineprofilestotheobservedspectrawhilekeepingtheelementalabundancesforthemetallinesfixedtothevaluedeterminedbefore.TheresultingfitsareshowninFig.6andthecorrespondingvaluesforvsiniare –11– reportedinTab.4. Thespectraltypesofthethreestarsweredefinedaccordingtotheireffectivetempera-turesaccordingtoSchmidt-Kaler(1982).Itshouldbenotedthatthisspectralclassificationdoesnottakeintoaccountanymetallicitydependenceoftheclassificationcriteriabutisbasedonstarsofsolarmetallicities.TheluminositiesofthethreestarswerethenderivedfromthevisualmagnitudesasgiveninTab.3usingthedistancemodulusofµ0=25.6fromDolphinetal.(2003a)andthe(closetozero)bolometriccorrectionsfromSchmidt-Kaler(1982)correspondingtothespectraltypes.TheresultingspectraltypesandluminositiesaregiveninTab.4togetherwiththestellarradiiascomputedfromtheluminosityandtheeffectivetemperature.Theluminositiesofallthreestarsareconsistentwiththeevolutionarytrackofa12M⊙star(Lejeune&Schaerer2001),i.e.,aZAMSmasscompatiblewithevolvedlow-luminosityAsupergiants.ThisfindingfurtherconfirmstheSextansAmembershipofthethreeanalyzedsupergiantstars. 4.Abundances Thelistofappropriatespectrallinesfortheabundancedeterminationsandthecor-respondingatomicdatawereadoptedfromthepreviousanalysesofVenn(1995a,b,1999);Vennetal.(2001).AtomicdataandtheabundancesderivedfromtheindividuallinesarelistedinTab.6togetherwiththemeasuredequivalentwidthsWλ.Allabundancecalcula-tionsinLTE/NLTEandLTEspectrumsynthesishavebeencarriedoutusingamodifiedversionoftheLINFORcode4.TheresultingaverageelementalabundancesaregivenforeachionizationstageinTab.4.Foreachabundance,twoerrorestimatesaregiven,firstther.m.s.line-to-linescatter,second(initalic)theestimatedsystematicerrorintheelementalabundancefromtheuncertaintiesintheatmosphericparametersTeff,logg,andξ.Thesystematicerrorsoftheaverageabundanceshavebeenestimatedbythevariationoftheindividualstellaratmosphereparameterswithintheirestimateduncertaintieswhilekeepingtheotherparametersfixed.TheindividualerrorestimatesarereportedinTab.5.FromtheerrorestimatesitcanbeseenthatSiII,TiII,CrII,andFeIIprovidethemostreliableelementalabundances.However,abundancesderivedfromtheSiIIionhavebeenfoundtosufferfromalargespreadifderivedfromstarsoverawiderrangeinstellarparameters.Thesampleof10AsupergiantsintheSMCasanalyzedbyVenn(1999)showsaspread –12– inthesiliconabundancesof≈0.5dexprobablyduetoneglectedNLTEeffectsinthelineformationoftheSiIIion.Thethreestarsanalyzedhereshowveryconsistentsiliconabun-dancesbecausethestellarparametersofthestarsareveryclosetoeachother.However,sincetheabsoluteabundanceprobablyisaffectedfromtheuncertaintiesthatwerefoundinotheranalyses,theabundancesfromSiIIwillnotbeconsideredinthefurtherdiscussionofαelementabundances. TheaverageelementalabundancesrelativetothesolarabundancesareshownforthethreestarsinFig.7withtheerrorbarsrepresentingtheestimatedcombinedsystematicerrorsintheabundances.ThecombinederrorswerecomputedasthequadraticsumoftheindividualerrorsfromTab.5. TheabundancesreportedforMgIandMgIIhavebeencomputedusingNLTElinefor-mationbasedonthemodelatombyGigas(1988)whichdeliversresultsingoodagreementwiththemodelatomdevelopedbyPrzybillaetal.(2001).TheNLTEcorrectionsforthein-dividualMglinesarelistedinTab.7andarefoundtobesmallandoftheorderof±0.1dex.AbundancesfromthestrongMgIIλ4481lineblendwerederivedfromfitstotheobservedlineprofileusingspectrumsynthesisinLTE;theresultsarereportedinTab.6.TheNLTEcorrectionsforthislineareexpectedtobelarge(Przybillaetal.2001)andverysensitivetothestellarparametersandinparticulartothemicroturbulence(Przybilla2003).Therefore,noattemptwasmadetocomputetheNLTEcorrectionsandthelinewasnotusedforthisanalysisbutisgivenforreferenceonly. ForstarSexA-754,usefulupperlimitsfortheequivalentwidthsoftheweakMgIIlinescouldbegivenbecauseofthehigherSNRofthecombinedspectrumofthisstar.TheresultingupperlimitsfortheMgIIabundancesareconsistentwiththeabundancesderivedfromtheMgIlinesandthereforefurthersupportthestellarparametersasderivedfromtheFeItoFeIIionizationequilibrium. Theαelementofhighestinterestisoxygen.Unfortunately,noneoftheOIlinesinthe615nmand645nmspectralregionscouldbemeasuredinthecombinedUVESspectraofmoderateSNRbecauseoftheweaknessofthelinesatthemetallicityofSextansA.For[O/H]≈−1,allOIlinesinthisspectralregionareexpectedtohaveequivalentwidthsof<20m˚A,i.e.,clearlybelowthedetectionlimitofthespectra.However,thelowmetallicityofSextansAbringsthenear-infraredOIλλ7771-7775,8446linesofmultiplet1and4intotheweak-lineregimepossiblysuitedforabundancesanalysesdespitetheirknownhighsensitivitytoNLTEeffects(Przybillaetal.2000). Spectraoftherespectivewavelengthregionsareonlyavailableforoneofthestars,SexA-754.Sincethenear-infraredlinesarenon-resolvedmultiplets,fittingofsyntheticLTEline –13– profileswasusedtodetermineLTEoxygenabundances(seeFig.8).TheoxygenabundancesmeasuredinLTEare12+logO/H=9.1±0.3forOIλ7771-7775and12+logO/H=8.1±0.3forOIλ8446.DetailedcomputationsoftheNLTEcorrectionsforthegivenstellarparametersofSexA-754atthemetallicityofSextansAwillberequiredbeforeaccurateoxygenabundancescanbederivedforthisstar.Inanycase,theexpecteduncertaintieswillremainlarge,firstbecauseofthelargerequireddeviationsfromLTEtobringtheoxygenabundancesintothereasonablerangeof12+logO/H≈7.8(assuminganunderabundanceof[α/H]=−1.0assuggestedfromtheotherαelementssiliconandmagnesium),secondbecauseofthelowqualityofthenear-infraredspectra,whichresultsinanon-negligibleuncertaintyinthecontinuumdefinition,whichiscrucialforanaccuratelineprofilemodeling.TheestimatederrorinthederivedLTEabundancesof±0.3dexisduetothisuncertaintyinthecontinuumlocalization.Therefore,currentlynofurtherattempthasbeenmadetodeterminetheNLTEcorrectionsforthenear-infraredlinesofSexA-754.Inthefollowing,elementalabundancesderivedfromMgIwillbeusedinsteadforthediscussionofαelementabundancesinSextansA.EarlierstudiesintheSMC(Venn1999)andNGC6822(Vennetal.2001)haveshownthatagoodagreementbetweentheαelementsoxygenandmagnesiumisfoundfordifferentstarsandsobothelementsmaybeusedtorepresenttheαelementalabundanceofthestar.TheaverageαelementabundanceofthethreestarsinSextansAarefoundtobe[MgI/H]=−1.09±0.02±0.19whichisinexcellentagreementwiththeαelementabundancesfromtheHIIregionsof[O/H]=−1.17asderivedbySkillmanetal.(1989)andtherecentlyrevisedvalue[O/H]=−0.95ofPilyugin(2001). Theiron-groupelementalabundancesofthethreestarsareinexcellentagreementasderivedfromlinesofCrI,CrII,FeI,andFeII.Thegoodagreementoftheabundancesfromthetwodifferentionizationstagesofchromiumisafurtherindicationforcorrectstellarparameter.However,itshouldbenotedthatthisequilibriumisbasedonasingleCrIline.PossibleNLTEeffectsonthislinewhichwerenottakenintoaccountintheabundancecomputationappearthereforetobesmall.TheagreementoftheFeIandFeIIabundancesandhencetheionizationequilibriumisachievediftherespectiveFeINLTEcorrectionsareapplied,i.e.,+0.23dexforSexA-754,+0.30dexforSexA-1344,and+0.20dexforSexA-1911.Asdiscussedbefore,thedeterminationofthestellarparameterisbasedspecificallyonthisNLTEcorrectedionizationequilibrium.TheabundancesofFeIIandCrIIhavebeenderivedfromnumerouslinesandareregardedasthemostreliable:thedependencyoftheabundancesonthestellarparametersissmall(cf.Tab.5)andpossibleNLTEeffectsonthesedominantionizationstagesareexpectedtobesmall. Themeanmetallicityofthethreestarsaresofoundtobe[(FeII,CrII)/H]=−0.99±0.04±0.06which—toourknowledge—makesthesethreestarsthetodatemostmetal-poormassivestarsinanygalaxyforwhichadetailedabundanceanalysishasbeencarried –14– out. Titaniumandscandiumcanbeconsideredaseitherαoriron-groupelements.TheTiabundanceswerederivedfromalargenumberoflinesofthedominantionizationstageofsinglyionizedtitaniumandareingoodagreementwithiron,i.e.,[TiII/FeII]=0.08±0.05±0.08whileScshowsaslightunderabundanceof[ScII/FeII]=−0.20±0.03±0.09.ThismightbeduetothesmallernumberofavailablelinesandhighertemperaturesensitivityofScII.Furthermore,thehyperfine-structuretermsinthelinesofoddelementsoftheirongroupduethepresenceofanuclearmagneticmomenthavebeenneglected.Duetotheseuncertainties,nofurtherdiscussionofthescandiumabundanceswillfollow. 5.Discussion Forthefirsttime,thepresent-dayirongroupabundanceshavebeendeterminedfromstarsinSextansA.Themeanunderabundancerelativetosolaris[Fe/H]=−1.03±0.05±0.11andhasbeenderivedfromtheFeIIabundanceswhichareexpectedtobethemostreliable.Thisresultiswellsupportedbyabundancesmeasuredforasecondiron-groupelement,namelyCrIIwithameanunderabundanceof[CrII/H]=−0.94±0.06±0.07.Theindividualiron-groupabundancesofthethreestarsfromwhichthemeanabundanceswerecomputedareingoodagreementwithintheestimatedsystematicerrors. ThestarformationhistoryofSextansAderivedbyDolphinetal.(2003b,andref-erencestherein),suggeststhatthemetallicityreached[M/H]=−1.1±0.2only2Gyrago.Themetallicityspreadof±0.2dexistoaccountforthedispersionintheBlueHeliumBurning(BHeB)stars.Themeanstellarmetallicityfoundherefromironandchromium,[(FeII,CrII)/H]=−0.99±0.04±0.06,isconsistentwiththeCMDstellarpopulationsanalysis. Themeanαelementabundancefrommagnesiuminthethreestarsis[α(MgI)/H]=−1.09±0.02±0.19whichisingoodagreementwiththenebularresultsforoxygen,[O/H]=−1.17(Skillmanetal.1989)and[O/H]=−0.95(Pilyugin2001).ItisworthnotingthatthisindependentmeasurementoftheαelementalabundanceconfirmsthelocationofSextansAinthefundamentalmetallicity–luminosityrelationshipfordwarfgalaxies(Skillmanetal.1989;Richer&McCall1995). TheagreementbetweentheαelementabundancesfromAsupergiantstarsandthenebularabundancesisasexpected.Thesestarshaveages≈107Myrandareexpectedtohaveformedoutofthesameinterstellargasasseeninthenebulae.SimilarfindingsarereportedfromtheabundanceanalysesofB-typemainsequencestarsinOrionandtheLMC –15– (Cunha&Lambert1994;Kornetal.2002),respectively,theGalacticabundancegradientsfromBstars(Gummersbachetal.1998;Rollestonetal.2000),andAsupergiantanalysesintheSMC,M31,andNGC6822(Venn1999;Vennetal.2000,2001),respectively.Itshouldbenotedhere,thatthepristinenebularandstellarabundancesofsomeelementscanbealteredbyeffectsofdustdepletionandrotationalmixingwhichcomplicatesaquantitativecomparisonofnebularandstellarabundances(Kornetal.2002).However,inthecaseofthestarsandnebulaestudiedinSextansA,thegoodagreementwithintheestimatederrorssuggestthatthecontributionsduetotheseeffectstothedeterminedabundancesaresmall.OnlythedIrrgalaxyWLMshowsadiscrepancybetweenthestellarandnebularαelementabundances(Vennetal.2003a).ThenebulaeinWLMhave[O/H]=−0.89(Skillmanetal.1989),whereastwoAsupergiantshave[Mg/H]=−0.62,andonestarsuggests[O/H]=−0.21.ThesearesignificantdifferencesthatarecurrentlynotunderstoodandmaysuggestthatWLMisthefirstgalaxytodisplayinhomogeneousmixing. Thespatialdistributionofabundancesinagalaxykeepsanimportantrecordofthere-distributionhistoryoffreshlyproducedelementsandtheirmixingwiththeISM.AscanbeseenfromFig.1,thethreestarsanalyzedinthisworkandthefourbrightcompactHIIregionsexaminedbySkillmanetal.(1989)lieonalineacrossthecentral,visuallybrightestpartofthegalaxy.AdoptingDSexA=1.3Mpc(Dolphinetal.2003a),nosignificantαelementabundancevariationsarefoundoveradistanceof0.8kpcascoveredbythestars(betweenSexA-754toSexA-1344toSexA-1911),andover1.6kpciftheHIIregionsareincluded.Thesamelevelofabundancehomogeneityisfoundfortheiron-groupelementsinthestars.Thehighlevelofhomogeneityofnebularabundancesindwarfirregulargalaxieshasbeendiscussedaspossibleevidenceagainstinsitu(“onthespot”)enrichmentbyKobulnicky&Skillman(1997)andthereforeagainstthegeneralvalidityoftheinstantaneousrecyclingapproximationastypicallyusedinchemicalevolutionmodels.Kobulnicky&Skillman(1997)favourascenarioinwhichthenewlysynthesizedelementsremainineitheradifficulttoobservehot106Koranequallydifficulttoobservecold,dustyphasewhilemixingthroughoutthegalaxy.RecentChandraspectroscopyresultsforNGC1569byMartin,Kobulnicky,&Heckman(2002)favorthehotphaseasthereservoirfornewlysynthesizedandexpelledgas.Theα/Feabundanceratioisakeyconstraintforthechemicalevolutionofagalaxy.αelementsareprimarilysynthesizedinSNeIIwhiletheiron-groupelementsaremainlyproducedinSNeIabutalsoinSNeII.Astarformationburstinagalaxywilltemporarilyincreasetheα/FeratiooftheISMduetothemassivestarsthatquicklyenrichtheISMwithαelements,whilethelowermassstarswillthenslowlyincreasethecontentofiron-groupelementsresultinginaslowdeclineoftheα/Feratioofthegalaxy(Tinsley1979).Gilmore&Wyse(1991)demonstratedtheexpecteddifferencesintheα/Feratiosforgalaxieswithdifferentstarformationhistories,andinparticularnotedthatthesolarα/Feratiomustnot –16– beregardedasanuniversalratio. FromthethreestarsinSextansA,theα/Feratioismeasuredas[α(MgI)/FeII,CrII]=−0.11±0.02±0.10.Thisα/FeratioismuchlowerthaninGalacticstarsofsimilarmetallic-ity(Edvardssonetal.1993;Nissen&Schuster1997).AtthemetallicityoftheseSextansAstars,[Fe/H]≈−1,theGalacticthickdiskandGalactichalostarsoverlapwitha[α/Fe]≈+0.3.Thelowerα/FeratiosintheseSextansAstarsfurtherfollowsthetrendfoundfromtheabundanceanalysesofstarsintwootherdIrrs,NGC6822andWLM(Vennetal.2001,2003a),butextendsthemetallicityrangenowsampledfrom[Fe/H]=−0.5inthosegalaxiesto[Fe/H]=−1.0.AsdiscussedbyVennetal.(2003a),thislowermetallicitynowoverlapstheuppermetallicitiessampledbyredgiantstarsindwarfspheroidalgalaxies,whichalsoshowmuchlowerα/FeratiosthanGalacticmetal-poorstars(Shetroneetal.2001,2003;Tolstoyetal.2003).Inthedwarfspheroidalgalaxies,thelow[α/Fe]ratiosnear[Fe/H]=−1.0aremostlikelyduetothelowerstarformationrates,thusslowerchemicalevolution.ThemetallicityrangeofthedwarfirregulargalaxiesalsooverlapswiththedampedLyαab-sorptionsystems(Nissenetal.2003;Ledouxetal.2002;Pettinietal.1999),whicharealsorecognizedtohavelowα/Feratios.Thismightsuggestthatlowα/Feratiosareagenericeffectatlowmetallicities. ThestarformationhistoriesforSextansA,WLM,andNGC6822aregloballysimilarasderivedfromtheirCMDs.AcomparisoncanbeseeninFig.8ofMateo(1998).Accordingly,allthreedIrrshadsignificantstarformationatancienttimes,>5−10Gyrago,withahiatusatintermediateages,1−5Gyrago,andrecentstarformationeventsinthepast1Gyr.Matteucci(2003)hasrecentlyemphasizedthattheSFHofagalaxyaffectstheabsoluteelementalabundances,butisonlyofminorimportancefortheabundanceratios,thelatterbeingprimarilydeterminedbythestellarlifetimes,initialmassfunctions(IMF)andthestellarnucleosynthesis.Thus,theimpactofdifferentSFHsontheα/Feratioisrelatedtothestarformationefficiency,andmanifestsitselfasashortplateauinthe[α/Fe]ratioatlowmetallicitiesingalaxieswithlowstarformationrates,likeirregulars(eitherinburstsorcontinuous),whiletheplateauismaintainedtohighermetallicitiesforsystemswithhighstarformationrateslikebulgesandellipticals.Thus,ourfindingthatthe[α/Fe]ratiosinthreedIrrswithsimilarSFHsissignificantlylowerthaninthemetal-poorGalacticstarscanbeexplainedinthisscenario.ThethreedIrrswithametallicityrangefrom−1<[Fe/H]<−0.4havealreadyleftthe[α/Fe]plateauandpresentlydisplaythesamelower[α/Fe]ratio.Thatthisratioissimilartosolarsuggeststhattheintegratedstellaryieldsfromastarformationeventisconsistentwiththesolarabundanceratios,probablyduetotheuniversalnatureofstellarlifetimes,IMFs,andstellarnucleosynthesis.Thatthehigh-redshiftDLAabsortionsystemsalsoshowlow[α/Fe]ratiossuggeststhattheyarealsosystemswithlowstarformationrates,butthiscouldbeeitherirregulargalaxiesortheouterpartsofspirals. –17– ArecentstudyofpublishedDLAsystemabundancesincomparisonwithchemicalevolutionmodelsofdifferentgalaxymorphologiesbyCaluraetal.(2003)hasidentifiedtheirregularandspiralgalaxiesasthepossiblyidealsitestocreatetheDLAsystems,whileellipticalscanberuledout. Thevalueofhighresolutionspectroscopyofthebrightestbluestarsinnearbygalaxiesfordetailed,quantitativeabundancestudieshasbeenfurtherdemonstratedinthiswork.Togainfurtherinsightintothechemicalevolutionofthedwarfgalaxiesandhenceintotheearlyevolutionofgalaxieswithlowstarformationrates,requiresmoredetailedabundanceanalysesoftheirstars,particularlytowardsthelowermetallicities. WethankH.A.KobulnickyforthelowresolutionKeckspectrumofSexA-754whichallowedustogetthisworkstarted.AKwantstothanktheESODirectorGeneral’sDis-cretionaryFund(DGDF)forthesupportofthisprojectandtheInstituteofAstronomy,UniversityofCambridge,UKforitshospitalityduringashort-termvisitwhereamajorpartofthisworkwasdone.KAVwouldliketothanktheNSFforsupportthroughaCA-REERaward,AST-9984073,andtheIoACambridgeforsupportduringaoneyearvisit.CPwantstothankESOforthepossibilityofashort-termstudentshipattheESOfacilitiesinVitacura,SantiagodeChile.ETgratefullyacknowledgessupportfromafellowshipoftheRoyalNetherlandsAcademyofArtsandSciences.FurtherwewanttothanktherefereeA.J.Kornforvaluablecommentsonthemanuscript. REFERENCES Asplund,M.,inCNOintheUniverse,eds.C.Charbonnel,D.Schaerer,G.Meynet(San Francisco:ASP),inpressBallester,P.,Modigliani,A.,Boitquin,O.,CristianiS.,HanuschikR.,Kaufer,A.,Wolf,S. 2000,TheESOMessenger101,31Becker,S.R.1998,inBoulder-MunichWorkshopII:PropertiesofHot,LuminousStarsed. I.D.Howarth,ASPConf.Ser.,131,137Boyarchuk,A.A.,Lyubimkov,L.S.,Sakhibullin,N.A.1985,Astrophysics,21,203Calura,F.,Matteucci,F.,Vladilo,G.2003,MNRAS,340,59Cunha,K.,Lambert,D.L.1994,ApJ,426,170 –18– Dekker,H.,D’Odorico,S.,Kaufer,A.,etal.2000,inOpticalandIRTelescopeInstrumen-tationandDetectors,eds.M.IyeandA.F.Moorwood,Proc.SPIEVol.4008,p.534Dohm-Palmer,R.C.,Skillman,E.D.,Saha,A.,etal.1997,AJ,114,2514Dohm-Palmer,R.C.,Skillman,E.D.,Mateo,M.,etal.2002,AJ,123,813Dolphin,A.E.,Saha,A.,Skillman,E.D.2003a,AJ,125,1261Dolphin,A.E.,Saha,A.,Skillman,E.D.2003b,AJ,126,187 Edvardsson,B.,Andersen,J.,Gustafsson,B.,etal.1993,A&A,275,101Gallart,C.,Aparicio,A.,Bertelli,G.,Chiosi,C.,1996,AJ,112,1950Gigas,D.,1986,A&A,165,170GigasD.,1988,A&A,192,264 Gilmore,G.,WyseR.F.G.,1991,ApJ,367,L55 Gummersbach,C.A.,Kaufer,A.,Schaefer,D.R.,Szeifert,T.,Wolf,B.1998,A&A,338,881Hodge,P.,Kennicutt,R.C.,Stobel,N.1994,PASP,106,765Huchtmeier,W.K.,Richter,O.G.1986,A&AS,63,323Kaufer,A.,Stahl,O.,Wolf,B.,etal.1996,A&A,305,887Kaufer,A.,Stahl,O.,Wolf,B.,etal.1997,A&A,320,273Kobulnicky,H.A.,Skillman,E.D1997,ApJ,489,636Kobulnicky,H.A.1998,priv.comm. Korn,A.J.,Keller,S.C.,Kaufer,A.,Langer,N.,Przybilla,N.,Stahl,O.,Wolf,B.2002, A&A,385,655Kurucz,R.L.,1979,ApJS,40,1 Kurucz,R.L.,1988,Trans.IAU,Vol.20B,ed.M.McNally(Dordrecht:Kluwer),168Kurucz,R.L.,1993,CD-romNo.13,Cambridge,Mass.:SmithsonianAstrophysicalObser-vatoryLedoux,C.,Bergeron,J.,Petitjean,P.2002,A&A,385,802 –19– Lejeune,T.,Schaerer,D.2001,A&A,366,538 Martin,C.M.,KobulnickyH.,HeckmanT.,2002,ApJ,574,663Matteuchi,F.,Tosi,M.1985,MNRAS,217,391Matteuchi,F.2003,Ap&SS,284,539Mateo,M.1998,ARA&A,36,435 Nissen,P.E.,Chen,Y.Q.,Asplund,M.,Pettini,M.,2003,A&A,inpressNissen,P.E.,Schuster,W.J.1997,A&A,326,751 Oke,J.B.,Cohen,J.G.,Carr,M.,etal.1995,PASP,107,375Pettini,M.,Ellison,S.,Steidel,C.C.,Bowen,D.1999,ApJ,510,576Pilyugin,L.S.2001,A&A,374,412 Prochaska,J.X.,Gawiser,E.,Wolfe,A.M.,Castro,S.,Djorgovski,S.G.2003,submittedto ApJLetters,astro-ph/0305314Przybilla,N.,ButlerK.,BeckerS.R.,KudritzkiR.P.,VennK.A.2000,A&A,359,1085Przybilla,N.,ButlerK.,BeckerS.R.,KudritzkiR.P.,2001,A&A,369,1009Przybilla,N.2002,Ph.D.thesis,Univ.M¨unchenPrzybilla,N.2003,priv.comm.Rentzsch-Holm,I.1996,A&A,312,966Richer,M.G.,McCall,M.L.,1995ApJ,445,642 Rolleston,W.R.J.,Smartt,S.J.,Dufton,P.L.,Ryans,W.S.I.2000,A&A,363,537Shetrone,M.D.,Cˆot´e,P.,Sargent,W.L.W.2001,ApJ,548,592Shetrone,M.D.,Venn,K.A,Tolstoy,E.,etal.2003,AJ,125,684Sheinis,A.I.,BolteM.,Epps,H.W.2002,PASP,114,851 Schmidt-Kaler,Th.1982,PhysicalParametersoftheStarsinLandolt-B¨ornsteinNumerical DataandFunctionalRelationshipsinScienceandTechnology,NewSeries,GroupVI,Volume2b(Berlin:Springer) –20– Skillman,E.D.,Terlevich,R.,Teuben,P.J.,vanWoerden,H.,1988,A&A,198,33Skillman,E.D.,Kennicutt,R.C.,Hodge,P.W.1989,ApJ,347,875Steinmetz,M.,Navarro,J.F.1999,ApJ,513,555TinsleyB.,1979,ApJ,229,1046 Tolstoy,E.,Venn,K.A.,Shetrone,M.,etal.,2003,AJ,125,707Tomita,A.,Ohta,K.,Saito,M.1993,PASJ,45,693vandenBergh,S.1999ApJ,517,L97 VanDyk,S.D.,Puche,D.,Wong,T.1998AJ,116,2341Venn,K.A.,1995a,ApJ,449,839Venn,K.A.,1995b,ApJS,99,659Venn,K.A.1999,ApJ,518,405 Venn,K.A.2000,McCarthy,J.K.,Lennon,D.J,etal.2000,ApJ,541,610Venn,K.A.,Kaufer,A.,McCarthy,J.M.,etal.2001,ApJ,547,765Venn,K.A.,Tolstoy,E.,Kaufer,A.,etal.2003a,AJ,inpress Venn,K.A.,Tolstoy,E.,Kaufer,A.,Kudritzki,R.P.,2003b,inCarnegieOb-servatoriesAstrophysicsSeries,Vol.4:OriginandEvolutionoftheEle-ments,eds.A.McWilliam,M.Rauch(Pasadena:CarnegieObservatories,http://www.ociw.edu/ociw/symposia/series/symposium4/proceedings.htm)Wheeler,J.C.,Sneden,C.,Truran,J.1989,ARA&A,27,279 –21– Fig.1.—TheSextansAdwarfirregulargalaxy.ThethreeAsupergiantsanalyzedinthisworkareindicatedwithredsquares;otherbluesupergiantsfromthetargetlistareindicatedwithbluesquares.ThefourbrightcompactHIIregionsanalyzedinSkillmanetal.(1989)areshownwithgreencircles. –22– Fig.2.—HαprofilesofthethreeSextansAstarsunderanalysis.AllprofileshavebeenshiftedtothelaboratorywavelengthofHαbytherespectiveheliocentricveloctiesfromTab.3.Thesharp,slightlyblue-shiftedemissionseeninstarSexA-754isduetoimperfectsubtractionofthehighly-structureddiffuseHαemissionfromtheHIgasofthegalaxy.Note,thatnostrongwind-emissionfeaturesarepresentintheHαprofilesindicatingthelowerluminosityand/ormetallicityofthestars. –23– Fig.3.—HγprofilesofthethreeSextansAstarsunderanalysis.Thesolid(red)lineindicatesthefitofthesyntheticprofilestothewingsoftheBalmerlineforthefinalmodelatmosphereparameterswhilethedashed(blue)lineisthesyntheticprofileforthesamestellarparametersbutwithloggincreasedby+0.2dexwhichisconsideredtheuncertaintyinthefit. –24– Fig.4.—TheFeINLTEcorrectionsusedinthisworkhavebeenestimatedusingEqn.(1).ThisrelationwascalibratedforthelowmetallicityandgravityofthestarsinSextansA(redsquares)usingthemeasuredNLTEcorrectionsinstarsintheSMC(bluecircles).ThefigurecomparesthemeasuredcorrectionswiththeestimatesfromEqn.(1). –25– Fig.5.—ThefinalselectedatmosphericparametersforthethreeanalyzedAsupergiantsareindicatedintheTeff−loggdiagrambythe(red)solidcircleswitherrorbarsandweredeterminedbyfitstotheHγprofiles(dashedline)andtheFeI/FeIIionizationequilibria(filledsquares).Inthelabels“Fex”,thexnotestheNLTEcorrectionappliedtotheFeIabundances.ForstarSexA-754theionizationequilibriaforthreedifferentFeINLTEcorrectionsareshowntoillustratetheeffectoftheestimateduncertaintiesonthestellarparameters. –26– Fig.6.—Spectrumsynthesistodeterminetheprojectedrotationalvelocityvsiniofthethreestars.ThelinesofTiIIλλ4290,4294,4300,4301andtheFeIIλ4303linewhereusedtofittheprofiles.DuetothelowervsiniofSexA-1344,theFeIIλ4297linecouldbeaddedtothelistoflinestobefitted. –27– Fig.7.—ElementalabundancesforSexA-754(top),SexA-1344(middle),andSexA-1911(bottom)relativetosolar(dottedline)andtothenebularoxygenunderabundanceof−1.17dex(dashedline).Abundancesderivedfromneutralatomsareindicatedwithopencircles,abundancesfromsinglyionizedatomswithfilledcircles.Theshownerrorbarsaretheestimatedcombinederrorsfromtheuncertaintiesofthedeterminationofthestellarparameters(cf.Tab.5). –28–Fig.8.—Spectrumsynthesisforthethenear-infraredOIλλ7771-5,8446multipletlines.TheoxygenabundancewasfittedusingspectrumsynthesisinLTE.TheuncertaintyoftheLTEabundancesof±0.3dexareduetothebadly-definedcontinuum.ToconvertthemeasuredLTEabundancesintosensibleoxygenabundancesdetailedcomputationoftheexpectedlylargeNLTEcorrections(Przybillaetal.2000)havetobecarriedout(seeSection4). –29– Table1.SextansAKeck/ESIandVLT/UVESobservations Star Date UTCStart λc[nm] Exp.Time[sec] AirmassStart Seeing[′′] –30– Table1—Continued Star Date UTCStart λc[nm] Exp.Time[sec] AirmassStart Seeing[′′] wasrepeatedinthenextexposurebecausetheseeingcon-straintwasviolatedhere Note.—AllESIdatawastakeninechellemodewitha20′′×1.25′′slitcorrespondingtoaresolvingpowerofR=3200.Thewavelengthcoverageis390−1100nm.AllUVESdatawastakenindichroicmode,witha1.2′′slitwidthand2x2CCDbinningcorrespondingtoa3-pixresolvingpowerofR=40000.Thewavelengthcoverageofthedichroicsettingsisfor390+580:330−452,478−575,584−680nmandfor437+840:376−498,650−831,845−1025nm. aObservation –31– Table2.Signal-to-NoiseRatiosofthecombinedUVESspectra Star @400nm Signal-to-NoiseRatioa @500nm@650nm@800nm aFor 2pixresolutionelementsofR=20000 –32– Table3.SextansAsample Stara Va (B−V)aRVhel[km/s] Sp.Type Comment aFrombAs VanDyketal.(1998) derivedinthiswork,cf.Tab.4 –33– Table4.AtmosphericAnalysis Solar SexA-754 SexA-1344 SexA-1911 –34– Table5.AbundanceUncertainties SexA-754 ∆Teff+200K ∆log(X/H) ∆logg+0.2 ∆ξ+1km/s ∆Teff+200K SexA-1344∆logg+0.2 ∆ξ+1km/s ∆Teff+200K SexA-1911∆logg+0.2 ∆ξ+1km/s –35– Table6.LineStrengths,AtomicDataandAbundances λ(˚A) Multiplet Table χ(eV) SexA-754Wλlog(ǫ) SexA-1344 Wλlog(ǫ) SexA-1911 Wλlog(ǫ) Elementlog(gf) –36– Table6—Continued λ(˚A) Multiplet Table χ(eV) SexA-754Wλlog(ǫ) SexA-1344 Wλlog(ǫ) SexA-1911 Wλlog(ǫ) Elementlog(gf) –37– Table6—Continued λ(˚A) Multiplet Table χ(eV) SexA-754Wλlog(ǫ) SexA-1344 Wλlog(ǫ) SexA-1911 Wλlog(ǫ) Elementlog(gf) –38– Table6—Continued λ(˚A) Multiplet Table χ(eV) SexA-754Wλlog(ǫ) SexA-1344 Wλlog(ǫ) SexA-1911 Wλlog(ǫ) Elementlog(gf) –39– Table6—Continued λ(˚A) Multiplet Table χ(eV) SexA-754 log(ǫ)Wλ SexA-1344 Wλlog(ǫ) SexA-1911 Wλlog(ǫ) Elementlog(gf) –40– Table7.MagnesiumNLTECorrections λ[˚A] χ [eV] SexA-754LTENLTE SexA-1344 WλLTENLTE SexA-1911 LTENLTE Levels a log(gf)WλWλ Thelevelsaredescribedwiththeirconfigurationsandterms:configurationn[l],termmeanstheparityisoddandtheabsenceofsubscriptmeanstheparityiseven. a2S+1 [L],thesuperscript 0 因篇幅问题不能全部显示，请点此查看更多更全内容