19 93MNRAS.2 60. .631B 4 12345 5 2 3 1 Mon. Not.R..Soc.260,631-634(1993) DepartmentofPhysicsandAstronomy,LouisianaStateUniversity,BatonRouge70803-4001,Louisiana,USA M. A.Barstow,T.Fleming,D.S.Finley,D.KoesterandC.J.Diamond SchoolofPhysicsandSpaceResearch,UniversityBirmingham,Edgbaston,BirminghamBI52TT Max--InstitutßrExtraterrestrichePhysik,Giessenbachstrasse,GarchingbeiMünchen,Germany CenterforEUVAstrophysics,UniversityofCalifornia,Berkeley,USA structure inGl91-B2B ROSAT EUVandsoftX-raystudiesofatmosphericcomposition Accepted 1992July28.Received15 DepartmentofPhysicsandAstronomy,UniversityLeicester,Road,LEI7RH first supposed.Comparisonofanearlierphotosphericred- features initsopticalspectrum,withaneffectivetemperature white dwarfssinceitsdiscovery(Giclas,Burnham&Thomas velocities oftheseultravioletlineshadsuggestedthatthe group. However,thepresenceinWEhigh-dispersion the photosphericcompositionisnotassimplemightbe G191-B2B hasbeenonethemostwidelystudiedof but thevelocityofNLTEemissioncoreinHaline material mightbecircumstellar(Bruhweiler&Kondo1981), indicates aphotosphericoriginforthemetalfeatures(Reid shift measurement(Trimble&Greenstein1972)withthe spectra ofhighlyionizedspeciesC,NandSiindicatesthat around 60000Kitisoneofthehottestmembers from thesoftX-ray/EUVobservationsmadebyEXO- & Wegner1988). 1 INTRODUCTION 1966). ClassifiedasaDAstarfromtheabsenceofhelium Koester 1989;Paerels&Heise 1989).Eitherinterpretation analysed theseobservationsusing eitheruniformH/Heor atmosphere (Paerels&Heise 1989).Severalauthorshave SAT observatorywhichrevealasignificantfluxdeficitbelow homogeneous modelrequires aHeabundancethatisnot is areasonableexplanationof theEXOSATdata,but stratified (HontopofHe)models (e.g.Jordanetal.1987; 300 ÂcomparedtothatexpectedifthestarhadapureH consistent with theobservedabsenceofHe linesinthe A furthercluetothecompositionofG191-B2Bcomes © Royal Astronomical Society • Provided by the NASA Astrophysics Data System ABSTRACT fied modelscannotaccountfortheobservedfluxes.Consequently,weconcludethat Previous studiesofthehotDAwhitedwarfG191-B2Bhavebeenunabletodeter- metalsmustbeasubstantialsourceofopacityinthephotosphere hydrogen andheliumatmosphereorfromthepresenceoftracemetalsinphoto- mine whethertheobservedsoftX-rayandEUVopacityarisesfromastratified observatory, whenanalysedinconjunctionwiththeearlierdata,showsthatstrati- sphere. NewEUVandsoftX-rayphotometryofthisstar,madewiththeROSAT Key words:stars:abundances-individual:G191-B2BwhitedwarfsX-rays: G191-B2B. stars. tical grounds. optical andultraviolet(UV)spectra.Furthermore,Venneset al. (1988)haveruledouthomogeneousmodelsontheore- tions ofG191-B2BusingtheROSATobservatory.The optical spectrumandnewmeasurementsoftheHiHe i the longwavelengthbandpassdatafromEXOSAT,an have performedananalysiswhichincludesourROSATdata, better limitsontheallowedatmosphericparameters.We to EXOSAT,allowtheobservedcountratesplacemuch narrower bandpassesoftheROSATinstruments,compared He modelforthephotosphereandshowthatadditional (Kimble etal.1992).WeareabletorejectastratifiedHupon columns madewiththeHopkinsUltravioletTelescope opacity sourcesintheformoftracemetalsarerequired. The ROSATsoftX-rayandEUVall-skysurveywascon- here derivefromthisphaseofthemission.ROSATcarries ducted from1990Julyto1991January.Thedatawepresent been welldescribedelsewhere, butthosedetailsthatare 2 /ÏOSAJOBSERVATIONSOFG191-B2B X-ray telescopehadapositional sensitiveproportional on boardtwocoalignedimaging instruments.Thesehave relevant tothisdiscussionare brieflyreiteratedhere.The counter (PSPC)asafocalplane detectorwhichcoveredthe energy range«0.1-2.4keV (5.2-100 Â;Pfeffermanetal. 1986). Although thePSPChasmodestenergy resolution, We havenowobtainednewsoftX-rayandEUVobserva- 19 93MNRAS.2 60. .631B by theWideFieldCamera(Wellsetal.1990).Forsurvey in thisstudyweusejusttheintegratedcountrateband 632 M.A.Barstowetal. for thepost-surveypartofmission,extending observations twobroadbandfilterswereemployed-SI Ä) andP2spans17-24eV(500-730Â).Fig.1showsthe covering therange90-200eV(60-140À)andS2 parison theresponsesofprimethinlexan(3Lx)and coverage tolowerenergies-PIspans56-83eV(150-220 effective areasofallspectralbandsexceptP2.Forcom- 60-110 eV(112-200Ä).Twoadditionalfiltersareavailable times betterspectralresolutionthanthatavailablewith can beseenthattheROSATinstrumentsyieldaboutthree aluminium/parylene (AIP)EXOSATfiltersarealsoshown.It filter responsesshowsthattheyhavesimilarbandpasses. band. Theerrorsquotedarepurelystatistical.Asmalldrift Therefore, weincludetheobservedA1Pcountratein gramme, theyhaveyettobeexecutedsuccessfully.InFig.1a been scheduledintheROSATpointedobservationpro- 48 125),thisamountedtoadecreaseinefficiencyof5and in theWFCresponsewasobservedthroughoutsky (1989). Table1summarizestheobservedcountratesineach data usedhere,publishedintheanalysisofPaerels&Heise comparison oftheEXOSATA1PandROSATWFCPI EXOSAT. taken intoaccountinthesubsequentanalysis. survey. AtthedatewhenG191-B2Bwasobserved(MJD « 0.1-0.4keV(25-100À).TheEUVwavebandisspanned 3 percentinSIandS2filtersrespectivelyhasbeen Figure 1.Effectiveareaasafunctionofphotonenergyforthe thin lexanandaluminium/parylenefilters. ROSAT PSPC,theWFCSI,S2andPIfiltersEXOSAT Although observationswiththePIandP2filtershave Table 1.ObservedG191-B2Bcountrates. Instrument -1-filter WFC SI WFC S2 EXOSAT A1P PSPC © Royal Astronomical Society • Provided by the NASA Astrophysics Data System 1 Count rate(s) 0.060 ±0.008 23.1 ±0.1 0.0403 ±0.0218 2.77 ±0.04 15 2 We includeinouranalysisaveryhighqualityopticalspec- trum ofG191-B2B,withasignal-to-noiseratio300and determined byfittingthehydrogenBalmerß,ôandylines the LickObservatory.Thegravityofstarwasinitially resolution of8ÀobtainedbyDSFusingthe3-mtelescopeat 3 OPTICALOBSERVATIONS formed fortemperatureTandgravityusingastratified with pure-Hmodelatmospheres(e.g.Kidderetal.1992), yielding logg=7.5±0.07(laerror).Asecondfitwasper- free parameters. model withanassumedHlayermass(ra)of10"Malso is showninFig.2.TheresultconsistentwithapureH The 99percentconfidenceintervalobtainedforTandra giving avalue7.5forlogg.Consequently,insubsequent Note thatasraisdecreased,alowertemperature well astheregionaround(undetected)Hen4686-Âline. analyses loggwasheldfixedat7.5withTandmtreatedas atmosphere at«60000K,orwitharangeofTdownto cal spectrumwesimultaneouslyfittedH/?,HóandHyas required toachieveagoodfittheBalmerlines.The somewhat lowerthan56000Kandlogmaslow-14.2. stringent lowerlimitplacedontheallowedHlayermassisa result ofincludingthe4684-Âlineinanalysis. include fulllineblanketing.Adetailedstudy,whichwewill geneous modelswerecalculatedbyoneofus(MAB)using geneous andstratifiedH+Heatmospheres.Thehomo- Our EUV/X-rayanalysiswasperformedusingbothhomo- 4 EUV/X-RAYDATAANALYSIS the NTLEcodetlustyofHubeny(1988).Theydonot model calculations.Thestratifiedmodelswerecomputedby report elsewhere,showsthatattemperaturesabove DK andarefullyblanketed(Jordan&Koester1986;Koe- cantly affectedbythelevelofblanketingincludedin H0 each instrumentresponseandinterstellarabsorption.Pre- routines writtenbyDSFtofoldmodelatmospheresthrough sterl991). 30 000KtheemergentEUVandX-rayfluxesarenotsignifi- H dicted countratesarecomparedwiththeobservedvaluesin H H predictions, takingintoaccounttheuncertaintiesin which determinesthequalityofagreementbetweendataand measured countrates.Theprogramsearchesthroughthe each filterandinstrument.Axstatisticisthencalculated H minus thenumberoffreeparameters,mustbegreaterthan range ofallfreeparameterstodeterminethebestfitand of degreesfreedomv,thatisthenumberdatapoints associated errors.Avalidanalysisrequiresthatthenumber the stratifiedcaseappropriate freeparameteristheH is definedbyfractionalabundance withrespecttoHandin ization constant.Inthehomogeneous modelstheHeopacity gravity, heliumopacityandadistance/radius-relatednormal- tures -effectivetemperature,interstellarcolumn,surface emergent EUVandX-rayfluxesforeachofourmodelstruc- or equaltoone.Therearefivevariablesthatdeterminethe from othersources tospecifysomeoftheparameters. Itis layer mass. EUV/X-ray observations,additional informationisrequired To determinetherangesofTandraallowedbyopti- Fits tothesemodelswereperformedusingasuiteofidl H Since wehaveonlyfourindependent datapointsfromthe 19 93MNRAS.2 60. .631B 1 18-2 17-2 possible valueforthealuminium/parylenecountrate(30s). Figure 2.99percentconfidencelimitsonthevaluesoftemperatureTandHlayermassmdeterminedbyfittingstratified model atmospheres totheG191-B2Bopticalspectrum.Alsoshownis99-9percentcontourforEUV/X-rayfitincorporating maximum been determinedfromopticaldata(logg=7.5).Furthermore, magnitude ofthestar(V=11.8)andsurfacegravityhas straightforward tonormalizethespectrausingV- the fineofsightHi(1.76±0.12x10cm)andHei in threebandpasses:3000-Âlexan,4000-Àlexanandalu- perature andHefractionorHlayermass. Kimble etal.(1992).Incorporatingallthisinformationinthe (1.47 ±0.06x10cm)columnshavebeenmeasuredby The effectiveareasforeachEXOSATbandpasshavebeen analysis meansweneedconsideronlytwovariables,tem- H on theeffectiveareasofeachinstrument/filtercombination. minium/parylene. Wehaveonlymadeuseofthealuminium/ parylene dataherebecausethelexanbandscannotreallybe observed withtheEXOSATLEItelescopeandwasdetected determined byPaerelsetal.(1990).G191-B2Bwas for scientificdataanalysisbuthavenotyetbeenpublishedin considered asspectrallyindependentdatapointsduetothe the literature.Inadditiontostatisticaluncertaintiesin dently andtherelevantinstrumentresponsesareavailable substantial overlapwithintheROSATresponses.The measured countratesforG191-B2B,thepredicted ROSAT PSPCandWFChavebeencalibratedindepen- À), thelauncertaintiesin EXOSATcalibrationwere calibrations. Paerelsetal.(1990)estimatethat,inthespectral rates aresubjecttolargeruncertaintiesintheeffectivearea range forwhichG191-B2Bwillhavesignficantflux(>200 value of20percent.TheROSAT 1acalibrationerrorsare estimating the true countrateinthealuminium/parylene height distributionthatintroduced significantuncertaintyin spectral shapeofG191-B2B resultedinadetectorpulse 15-20 percent.Forthepurpose ofthisworkweadopta « 10percent.Paerels&Heise (1989)reportedthatthesoft The fittingtechniquedescribedaboveutilizesinformation © Royal Astronomical Society • Provided by the NASA Astrophysics Data System -j tjjo 13.5 o -12.0 -13.0 -12.5 -14.0 -14.5 -15.0 50 52545658606264 ROSATEUV andsoftX-raystudiesofG191-B2B633 -1 1 3 3 6 e filter, sinceasignificantnumberofeventsmayhavefallen below thelowerelectronicdiscriminatorthreshold.Paerels conservative weallowforalossashigh30percentand at theirpublishedcountrateof23sforthe & Heiseappliedamodestcorrectionforthiseffecttoarrive count rateupto30s“. aluminium/parylene bandpass.Inordertobeevenmore model atmospherescannotexplaintheobservedEUVand formed thespectralfittingwithhomogeneousmodels. Allowing thetemperaturetovaryfreelythroughoutfull However, forcomparisonwiththisnewanalysisofthe soft X-raycountratesinearlierwork(Barstowetal.1992). degrees offreedomtheprobabilityexceedingavalue model grid)withaminimumvalueof°f6.63.Withtwo range ofthemodels(20000-80000K),bestfitisfound well outsidetherangeallowedbyopticalfineprofiles(99 ^ =6.63isapproximately0.0001.Thistemperaturefies at 7=37500KandHe:H=1.0x10'(thelimitofthe ROSAT data,usingstratifiedmodels,wehavealsoper- features intheopticalandUVbands.Wehavealsoexplored He abundancerequiredwouldgiverisetoobservable per centconfidencerange55000-61000K)andthehigh T by fixingtheeffectivetemperature(v=3)atanumberof the degreeofagreementwithinopticaltemperaturerange points andrepeatingthefitswithHeabundanceasonly free parameter.Wefindthatthe valueofXrneverfallsbelow ruled out. 2.3 x10andsothehomogeneous modelscanbecompletely based ontheminimum valueobtainedforXr inlhregionof mass. Theacceptabilityofthe fittotheEUV/X-raydatawas determining asafunction oftemperatureandHlayer We havealreadydemonstratedthathomogeneousH+He We havetestedhowwellthestratified modelsfitthedata, 19 93MNRAS.2 60. .631B -1 -9 _5 - 6 -5 634 MABarstowetal. that theprimarydeterminantofqualityfitwas freedom, theprobabilityofthisbeingagoodfitisformally the parameterspaceallowedbyopticalfits.Wefound consistent withtheopticaldata,was20.Withtwodegreesof lower end(20counts)theminimumforEUVfit, assumed valueofthealuminium/parylenecountrate.At probability ofonly0.2percent.The99centconfidence optical 99percentlimits. rate of30counts"^theminimum%l(6A)correspondstoa This workallowsastratifiedinterpretationtobeexcluded As pointedoutbyBarstowetal.(1992)anddiscussedearlier contour forthisfitisshowninFig.2andcomparedwiththe 2 x10.Evenatthemaximumallowedaluminium/parylene believe thattheevidenceagainstastratifiedatmosphere with greaterthan99percentconfidence.Consequently,we parylene countrate-areanessentialpartoftheanalysis. determined temperatureandtheEXOSATaluminium/ other sources-theinterstellarcolumninformation,optically instruments. However,thesupplementarydatadrawnfrom narrower bandpassesachievedwiththeWFCandPSPC count ratesalone.Thisismainlyaconsequenceofthe in thispaper,theROSATdataprovideamorestringenttest of thestratifiedatmosphericstructurethanEXOSAT 5 CONCLUSION explain thecombinationofROSAT,EXOSATandoptical containing H+Heonlyisnowoverwhelming.Itclearthat borne .Theirlimitonthesizeofedgesuggests phere anymoredeeplywiththepresentdata.Althoughtrace observations, givenourknowledgeoftheinterstellarcolumn. additional opacityintheformoftracemetalsisrequiredto present norwhetherHemaystillcontributetothetotal metals arerequired,wecannotdeterminewhatspecies the atmosphericstructureofG191-B2B.Hence,compo- when combinedwithourresult,thatHehasaminimalrolein the minimumHlayermassthisimpliesbutitdoessuggest, than «3xl0.Adetailedanalysisisrequiredtodetermine opacity. Wilkinson,Green&Cash(1992)reporttheabsence that, assumingahomogeneousmixture,theHe/Hratioisless of a228-ÀHenedgeinspectrumobtainedbyrocket- where anEUVspectrumobtainedbyEXOSAThasalready found inFeige24(e.g.Vennes1992).Recently,co-addition line strengthsinG191-B2Bthatareequivalenttothose (Vennes etal.1988). demonstrated theexistenceofphotospherictracemetals sition ofG191-B2BmustbesimilartothatFeige24, Feige 24,findingsimilarvaluesofFe/H,between3x10 have measuredtheabundanceofFeinbothG191-B2Band of IUEspectrahasrevealedadditionalfeaturesduetoOiv, complete story.Thespectralslope ofFeige24,asviewedby and 1.5X10,forbothstars.However,thisisnotthe O v,Fevandviinbothstars(Bruhweiler&Feibelman 1992; Tweedy1992).Morerecently,Vennesetal.(1992) It isnotpossibletoprobethecompositionofatmos- Bruhweiler &Kondo(1981)reportCiv,NvandSiiv © Royal Astronomical Society • Provided by the NASA Astrophysics Data System The observationofopacitydifferencesinstarsverysimilar Vauclair &1986).Ultimately,wewillrequire preparation) despitethesimilarityofUVlinestrengths. gating themechanismofradiativelevitation(e.g.Morvan, effective temperatureisofenormousimportanceininvesti- implying agreaterlevelofmetalopacity(Barstowetal,in the emergentEUVflux.Wecanonlyobtainthisusing detailed informationregardingtheopacitysourcesthataffect ROSAT, fallsoffmuchmoresteeplythaninG191-B2B, ability ofsuchdatafromtheEUVEmission. spectroscopic observationsandwelookforwardtotheavail- ACKNOWLEDGMENTS Advanced FellowshipandStudentshiprespectively.DSFis The authorswishtothanktheROSAThardwareandground REFERENCES NSF grants. supported byaNASALTSAgrantandDK opportunity ofperformingthiswork.MABandCJD Barstow M.A.,FlemingT.DiamondC.J.,FinleyD.S.,1992,in acknowledge thesupportofUKSERCthroughan system teams,withoutwhomwewouldnothavehadthe Bruhweiler F.C,KondoY.,1981,ApJ,248,1981 Bruhweiler F.C,FeibelmanW.A.,1992,ApJ,inpress Jordan S.,KoesterD.,Wulf-MathiesC,BrunnerH.,1987,A&A, Jordan S.,KoesterD.,1986,A&AS,65,367 Hubeny I.,1988,ComputerPhysicsCommunications,52,103 Koester D.,1991,inMichadG.,TutukovA.,eds,Proc.IAUSymp. Giclas H.L.,BurnhamR.,Jr,ThomasN.G.,1966,Lowell Kidder K.,HolbergJ.B.,BarstowM.A.,TweedyR.W.,Wesemael Koester D.,1989,ApJ,342,999 Kimble R.A.etal.,1992,ApJ,inpress Paerels F.B.S.,HeiseJ.,1989,ApJ,339,1000 Morvan E.,VauclairG.,S.,1986,A&A,163,145 Tweedy R.W.,1992,PhDthesis,Univ.Leicester Paerels F.B.S.,BrinkmanA.C,denBoggendeJ.F,deKörte Vennes S.,1992,ApJ,390,590 Trimble V.,GreensteinJ.L.,1972,ApJ,177,441 Reid N,WegnerG.,1988,ApJ,335,953 Pfefferman E.etal.,1986,Proc.SPIE,733,519 Vennes S.,FontaineG.,WesemaelF,1989,inWegnered.,White Vennes S.,PelletierC,FontaineG.,WesemaelF,1988,ApJ,331, Wells etal.,1990,Proc.SPIE,1344,230 Vennes S.,ChayerR,ThorstensenJ.R.,BowyerShipmanH.L., Wilkinson E.,GreenJ.C,CashW, 1992,ApJ,397,L51 Heber U.,JeffreyC.S.,eds,TheAtmospheresofEarlyType Stars. Springer-Verlag,p.314 Observatory Bulletin,No.132 tion. Kluwer,Dordrecht,p.435 F, 1992,ApJ,394,288 P. A.J.,Dijkstra1990,A&AS,85,1021 185,253 145, EvolutionofStars:ThePhotosphericAbundanceConnec- Dwarfs. Springer-Verlag,p.363 876 1992, ApJ,392,L27