1983ApJ. . .269. .580B metal-rich globularcluster47Tucanae(=NGC104 DBG andbyNC. observational data,withtheaimofobtainingcarbon of CNstrengthsis,infact,bimodal.Analysesthese Dickens, Bell,andGustafsson(1979,hereafterDBG) DDO photometrycarriedoutbyHesser,Hartwick,and conclusion isbasedonlow-dispersionspectroscopyand C0021-723) exhibitarangeofCNbandstrengths.This and nitrogenabundances,havebeencarriedoutby and Freeman(1979)havesuggestedthatthedistribution and NorrisCottrell(1979,hereafterNC). McClure (1976,1977),Hesser(1978),Mallia © 1983.TheAmericanAstronomicalSociety.Allrightsreserved.PrintedinU.S.A. The AstrophysicalJournal,269:580-591,1983June15 inasmuch asitispopulous,relativelynearby,nearly under contractAST78-27879. cluster 47Tueis,ofcourse,idealforsuchstudies (Hesser 1978;HesserandBell1980,hereafterHB).The data dosuggestthatCNstrengthvariationsoccurinthe M< 2.5mag,and,second,low-dispersionspectroscopy andinstarswhichareonlyslightlyevolved for starsdownto—b4.3mag.Thesespectroscopic ter starsare,first,DDOphotometryforwith tory, whichissupportedbytheU.S. NationalScienceFoundation v It isnowquiteclearthattheevolvedstarsin The observationaldataavailableforthefainterclus- Visiting Astronomer,CerroTololo Inter-AmericanObserva- © American Astronomical Society • Provided by theNASA Astrophysics Data System be producedbyastar-to-starrangeofnitrogenabundances~5.Thisresultisinaccordwith previous analysisbaseduponmuchlowerresolutiondata.Therangeinferredissimilartothat with photometricdatatoarguethatsome,ifnotall,ofthedwarfshavesametemperature. required toexplainobservationsofhighlyevolvedstarsin47Tue. (2?~16 mag)inNGC104(47Tue,C0021-723).ArangeofCNstrengthsisobservedamong anddwarfstarsofotherwiseverysimilarspectra.Thespectroscopicresultsarecombined Subject headings:clusters:globular—spectrophotometry—:abundancesPopulationII Spectrum synthesiscalculationsarethenusedtoinferthattheobservedrangeofCNstrengthscould telescope’s Ritchey-Chretienspectrographfor11dwarfs(B-18.2mag)andfivebrighterstars SPECTROSCOPIC ANALYSISOFDWARFANDSUBGIANTSTARSIN47TUCANAE Spectra at~4ÀresolutionhavebeenobtainedwiththeSITvidicondetectoronCTIOm I. INTRODUCTION Dominion AstrophysicalObservatory,HerzbergInstituteofAstrophysics Received 1982August10;acceptedDecember6 Royal Observatory,Edinburgh National ScienceFoundation 1 1 James E.Hesser R. D.Cannon 1 ABSTRACT R. A.Bell AND 580 unreddened, andlocatedathighgalacticlatitude.Fur- enough thatCNbandsarevisibleintheirspectra. thermore, itsmain-sequenceturnoffregionstarsarecool been addressedbyDBGusingphotometry,and lems. Asimilarproblemholdsforothermetal-rich Ardeberg (1978).Pilachowski,Cantema,andWallerstein horizontal branchstarsobservedbyGustafssonand (1980) found[Fe/H]=-1.2fromanalysisofechelle cluster isprobablymoremetal-poorthan[M/H]=-0.5, Pilachowski, Cantema,andWallerstein(1980),using in thatcluster(BellandGustafsson1982). as mightbehoped.Recentworkhasshownthatthe very valuableforstudiesofcarbon andcyanogenabun- clusters suchasM71,butthediscrepancybetweenpho- causes uncertaintyintheanalysisofanumberprob- ferences maynotbetoosurprising,buttheirexistence assumptions madeinthedifferentanalyses,dif- spectra oftwothebrightergiantstars.Inview the valueusedformanyyears.DBGobtained[M/H]= spectroscopy. Thisabundanceisnotyetaswellknown sufficient caliberforthemto drawfirmconclusionson tometric andspectroscopicabundancesisevengreater they concludedthat[C/A]= -0.5.DDOphotometryis the carbonabundances,except fortheAGBstarswhere -0.8 fromsyntheticcoloranalysisofuvbycolorsred The problemoftheoverallmetalabundancehas The spectroscopicdataavailabletoDBGwerenotof 1983ApJ. . .269. .580B 24256 25 26 24 lead toahigherNabundancebeingdeduced.Unfor- Unes ortheCObands.TheDBGphotometricinterpre- was “normal”sincenodatawereavailableforthe[Oi] results mightstillbesystematicallydifferentfromthe dances ofhighprecisionfromthephotometry.The abundance foundforaparticularobjectdoes,ofcourse, factor of3insomecases.Anyreductionthecarbon andaverylownitrogenabundance.Thislatter duced undertheassumptionthatoxygenabundance carbon abundancesfoundfortheAGBstarswerede- tunately, DBGwereunabletodeducecarbonabun- some starswasconsiderablyenhanced,bymorethana in theclusterstarswas“normal,”Nabundance were normal)byafactorcomparabletothatfound DBG values. overdeficiency wouldbereducedifloweroverallmetal for Arcturuswiththosefoundfromhigh-dispersion dances, andDBGshowedthat,ifthecarbonabundance had beenwidelyassumedthatthecyanogenvariations overabundant inN(assumingthatcarbonandoxygen abundance hadbeenused,butitisunclearwhethertheir stars. Assuminganoverallmetalabundanceof[M/H] tained intheNCanalysisofspectratwoAGB spectroscopy. tation wascheckedbycomparisonofthevaluesfound in thegiantstarswereresultofmixingmaterial spectrum synthesistechniques,HBshowedthatsome which areabouttwicethoseoftheCN-weakstars none ofthemarecompletelysatisfactory.Thehypothe- hancement inthedwarfs(cf.Belletal.1981and§IV), mechanisms canbepostulatedtoexplaintheCNen- of thatmechanismontheuppermainsequencerepre- DBG forthegiants,i.e.,nitrogenoverabundancesof slightly evolvedstars(M-4.3mag)in47Tuewere White andCameron1948). from thecenterofstartosurface,butinvocation about afactorof5.Thisresultisrathersurprising.It isotopic abundanceratiois Mg:g 8:1:1; calculations showingthatmixingisunabletoproduce dial innatureseemstobesupportedbystellarevolution sis thattheenhancednitrogenabundancesareprimor- sarily correct)beliefs.Althoughanumberofother sents aradicaldeparturefromlong-held(butnotneces- no evidenceforarangeintheoverallmetalabundance, (Cottrell andDaCosta1981;Lloyd such largeNoverabundances(DaCostaandDemarque magnesium, althoughitispossiblethatenhancementsof nor intheabundancesofotherlightelementssuchas Evans, Smith,andMenzies1982).Thereappearstobe stars withM<-0.6magpossesssodiumabundances the abundancesofMgand Mgwouldbemaskedby a dominantabundanceof Mg(theterrestrial/solar = -0.5,NCfoundonestartohaveslightlyenhanced 1982) aswellbytheobservationthatCN-strong v v A differentresultontheCNabundanceswasob- Using ~16Áresolutionspectraofsevenstarsand © American Astronomical Society • Provided by theNASA Astrophysics Data System DWARF ANDSUBGIANTSTARSIN47TUCANAE -1 work weretoseeif(a)thenewdataconfirmed observations offaint47Tuestars.Theobjectivesthis ter-American Observatoryhavemadeitpossibletoob- results ofHB;(b)moreextremeexamplesCN ing thenightsof1980September8-11(UT),ina having verystrongCNasjudgedfromDDOphotome- calcium, usingtheHandKUnes.Wealsoobtained available toHB,weundertookafurtherprogramof tain spectrawithmuchhigherresolutionthanthose /im (270)slit;KPNOGratingLabgratingnumber1 manner identicaltothatdescribedbyHesserandHarris on theabundanceofcarbon,usingGband,and strengths couldbefoundbyanextensionoftheHB With theexceptionoffirstnight,whenmost Ritchey-Chrêtien focusoftheCTIO4mtelescopedur- try, andfourotherfaintsubgiants. survey; and(c)itwerepossibletoobtaininformation look” facilitiesatthetelescopeindicatedthatthese by takingtwo45minuteexposuresofeach.“Quick was inwholeorpartduetonoisetheirlowerresolu- were clearformostoftheobservingtime. (632 Unesmm)inthefirstorderblue(angle59.°20); (1981). Theinstrumentalconfigurationconsistedof300 the SITvidiconcameraonspectrographat spectra ofthestarHH1-9038,asubgiantsuspected wished tosearchforstarsexhibitingmoreextremeCN near 3883A,didappeartobesimilarHB’s.Wealso etal. 1979)witha250mmfocallength,f/1.4camera. quently reobservedstars1-9004,3-2153,and3-2195 of HB’sstarstoseeifthededucedrangeinCNstrengths observations ofthefivebrightersubgiantstarswere and anRCA4804UVtransmittingSITvidicon(Atwood run, onlyasinglespectrum,sometimesatlargerhour differences thanthoseof1-9004and3-2153.Forthis secured, theseeingwasbetterthan275andskies The locationsofthestarsin clusterC-Mdiagramare survey aspect,carriedoutonthefinaltwonightsof spectra, particularlyinthecrucialCN(0,0)bandregion relevant photometricdata,thejournalofspectroscopic Hesser andHartwick’s(1977)studyofthecolor-magni- . angles and/orbrieferexposuretimes,wasobtainedper tion data.Onourfirstnightofgoodseeingweconse- tra wereobtained. observations, andtheradialvelocitiesderivedfromthem. tude (C-M)diagram,islistedinTable1,togetherwith analysis ontheturnoffregion starsforwhichtwospec- given inFigure1.Inthispaper, wewillconcentrateour Since instrumentaldevelopmentsatCerroTololoIn- We wishedtoreexamine,athigherresolution,several Spectra with-4Àresolutionwereobtainedusing The totalsampleofstarsobserved,selectedfrom II. OBSERVATIONS 581 1983ApJ. . .269. .580B 12 km s';nosuchdifferencewasapparentforNGC288. weighted averagesfromearliervelocitystudiesrevealed NGC 104and6752withWebbink’s(1981) an averagesystematicdifference,t>-v=38 mean clusterradialvelocityfromourobservationsof (C0050—268; dechnation—27°).Comparisonofthe 6752 (Cl906—600;declination-60°)andNGC288 582 understood, conversationswiththe CTIOstaff,particularlyDr.B. While theoriginsofthesezero-point differencesisnotcompletely During therun,datawerealsosecuredforstarsinNGC and thatathirdrunin1981Aprilproduced anullzero-pointerror. sign wasfoundduringthe1980June run(HesserandHarris1981) the methodsdescribedearlier(HesserandHarris1981). cxWebbink 2 Radial velocitieswerederivedforourstarsfollowing Wenoteforcompletenessthata similar differenceofopposite © American Astronomical Society • Provided by theNASA Astrophysics Data System high velocity. 3-2195 3-2278 3-2467 3-2183 3-2397 3-2384 from Fig.1,photoelectricvaluesarealsoavailableasfollows(topoftabletobottom)intheorderstar,V,B-V\1-9038, 3-2455 3-2262 3-2234 1-9015 wereaffectedbyatemperaturecontrolproblem(cf.HesserandHarris1981),whichundoubtedlyaccountsforthe 14.61, 0.95;1-9403,15.11,0.87;1-9015,15.47,0.85;1-9055,15.53,1-9047,15.54,0.81;1-9004,17.42,0.62. 1-9004 1-9055 1-9015 1-9403 1-9038 b a 2153 9047 Observationson1980Sep8weretakenunderconditionsofpoorseeingandhighhumidity.Inaddition,dataforstar ID numbers(intheform“figurenumber-IDnumber”),K,(B-F)fromHesserandHartwick1977.Forstars pgpg Star III. SPECTROSCOPICANALYSIS a) RadialVelocities (mag) 17.42 17.58 17.55 17.53 17.51 17.53 17.52 17.50 17.48 17.48 17.46 15.54 15.53 15.47 15.11 14.58 3 (mag) 0.57 0.63 0.58 0.59 0.61 0.64 0.62 0.64 0.59 0.62 0.60 0.81 0.85 0.85 0.84 0.88 Summary ofObservationalDatafor47TucanaeStars BELL, HESSER,ANDCANNON b (dd/hh:mm) U.T. (Start) (1980 Sep) 09/04:48 09/03:56 09/07:25 09/08:17 09/09:09 09/06:31 09/05:40 08/08:03 08/09:17 08/08:44 08/04:16 08/02:28 09/09:54 09/03:41 09/03:35 08/03:06 08/01:59 10/09:05 11/09:07 10/03:21 11/03:50 11/02:54 11/07:32 10/08:12 11/02:37 10/03:08 TABLE 1 _1 -1 no convincingevidencefornonmembershipofany purposes, becausetheradialvelocityof47Tueissmall, night), wehavenotattemptedtocorrect forit. brief temperaturecontrollerfailure thatoccurredonthefirst sample. InspectionofthevelocitiesinTable1provides ( -25kms)velocitiessuchasoursmayonlybeused The zero-pointdifferencehasbeenappliedtothevalues night. Largelybecauseourdataare insufficientinnumbertomap occurred duringthe1980September run,particularlyafterthefirst to removeachancehigh-velocityfieldstarfromthe given inTable1,althoughitislargelyirrelevantforour effective wavelengtherrors,assuggestedbyHesserandHarris Atwood, suggestthattheymaymorelikelybeassociatedwith the suspectedproblem(whichmight havelargelyoriginatedina removed fromthecoldboxbetweenruns),ratherthantopossible changes inorientationoftheSITtube(duetoitshavingbeen 1-9038, thatachangeinvelocityscalezero-pointmayhave 1981. Finally,wenote,frominspectionofthevelocitiesforstar Hour Angle(Start) — 14kms(Webbink1981),sothatlow-precision 01:03 E 01:56 E 03:18 W 03:24 W 02:27 E 01:54 E 02:38 W 01:36 W 02:25 W 02:49 E 01:50 W 03:18 W 00:39 W 00:12 E (hh:mm) 02:11 W 03:22 W 02:48 W 01:36 E 03:30 E 03:03 E 02:36 E 04:07 W 02:07 E 02:13 E 02:45 E 03:52 E (min.) Exp. 45 45 40 45 45 45 45 45 45 45 35 45 45 37 20 30 30 30 30 10 10 6 5 5 5 3 (km s) -37 -43 -22 -30 -36 -90 -35 -13 v -26 -25 + 46 + 30 + 6 -1 -2 -1 r,Q 42 61 : 26 15 14 12 7 11 10 11 12 7 8 7 9 9 9 9 9 9 9 Vol. 269 1983ApJ. . .269. .580B immediately confirmsthelowerresolution(~16A) ple; e.g.,3-2183and3-2234,3-21953-2278.)This head] intheformerandnotlatter.(Otherpairs upper-main-sequence stars(M~4.3mag)inorderof of theobservingrunstar1-9038,astrongCN superposition ofthem.Another quiteobviousdifference results ofHesser(1978)andHBthatprovidedthefirst demonstrating thesamedifferencesareseeninsam- decreasing Vmagnitude. somewhat fainter,subgiantstars(1-9403through four spectraaredataobtainedoneachofthenights cluster dwarfs.Thedifferences inthespectraofthese evidence forspectral(CN)differences amongglobular [the wavelengthofthevioletdegradedCN(0,0)band allows evaluationofthedataquality.Spectraother, and McClure(1977).Intercomparisonofthesespectra according totheDDOphotometryofHesser,Hartwick, probabilities comebothfromthegeneraldiscussionsof two starsmaybeappreciated inFigure3,whichisa HB andthesimilarityofspectrathemselves. No. 2,1983 the stars.Additionalargumentsforhighmembership the squaresrepresentcolorsandmagnitudesofmodelatmospheres(giveninTable3). and Hartwick1977isindicatedbythedashedlines.Thedotsarestarsforwhichspectrawereobtained(identificationsinTable1) and 1-9407) arefollowedinFigure2bythespectraof11 1-9004 showsanoticeabledepressionatX-3883À v An atlasofthespectraisgivenasFigure2.Thefirst Examination ofthespectrastars3-2153and Fig. 1.—Aschematiccolor-magnitudediagramfor47Tue.Thetotalwidthofthemain-sequenceandturn-offregionobservedbyHesser © American Astronomical Society • Provided by theNASA Astrophysics Data System b) TheObservedSpectra DWARF ANDSUBGIANTSTARSIN47TUCANAE while, forthefaintturnoff-regionstars,anequallyplau- ple (cf.Table1andFigs.2),onemightarguethat case ofourdatafor47Tuedwarfs,therearetwoaspects idea ofthetemperaturetobeusedforeachstar.In (1982) haveremeasureda sample of-100starsin merits oftheseapproaches, Hesser,Egles,andTiller Hesser 1974;andHartwick1977)showsthe estimate thetemperature.To initiateastudyofthe errors aresuchthatamean colorshouldbeusedto estimating thetemperaturewhenmodelingspectrum to thisproblem. sible argumentmightbeadvancedthatthephotometric sufficiently precisethattheyshouldbeadoptedfor the individualphotoelectricorphotographiccolorsare to beexpectedforthenumberofplatesmeasured.For scatter inB—Vcolorsattheturnoffluminositywhichis spectra (describedin§Hie),itisnecessarytohavean the brighterstarsinspectroscopicallyobservedsam- in thespectraofFigure2,namelychangesaverage differences. and faintgiants,isdueprimarilytotemperature CH GbandandCNstrengthsbetweenthedwarfs First, theavailableC-Mdiagram(Hartwickand Before undertakingthecomparisonwithsynthetic c) TheTemperatureSpreadontheUpperMain Sequence 583 1983ApJ. . .269. .580B obtained oneachoffournights.Furtherdetailsthestarsaregiven inTable1.Somespectralfeaturesareidentifiedatthefootoffirst column. Thegiantsareinthefirst columnandthetopofsecond. Fig. 2.—Anatlasofthespectraobtained,arrangedcolumnwisein orderofincreasingVmagnitude.Thespectraforstar9038were © American Astronomical Society • Provided by theNASA Astrophysics Data System x(Â)—► 1983ApJ. . .269. .580B > lie within0.2magoftheVmagnitudestarsin Table 1.ThethreeVandtwoBdirectplatesusedinthe Hesser andHartwick’s(1977)C-Mdiagram.Thesestars Hesser andHartwickC-Mdiagram,aswelltwomore The resultsaregivenfortheturnoff-regionstarsofour B directplatestakenaspartoftheiroriginal1.5m dispersion hasbeenhalvedforstarswith0.40