Table Of Content

(DOI:willbeinsertedbyhandlater) A search for Interstellar absorptions towards Sgr dSph S.Monai1,P.Bonifacio1,andL. Sbordone2,3 ⋆ 1 IstitutoNazionalediAstrofisica-OsservatorioAstronomicodiTrieste,ViaTiepolo1134131Trieste-Italy 2 EuropeanSouthernObservatory,Casilla19001,Santiago,Chile 3 Universita`diRoma2”TorVergata”,viadellaRicercaScientifica,Rome,Italy Abstract.WesearchedforNaID1,D2interstellarabsorptionlinestowards12giantstarsoftheSgrdwarfspheroidaland3 starsoftheassociatedglobularclusterTerzan7,observedathighresolutionwiththeUVESspectrographatthe8.2mKueyen VLTtelescope.Thesestarsarenotanidealbackgroundsourcefortheanalysisofinterstellar(IS)absorptionlinessincethey 5 havebeenobservedintheframeworkofastellarabundancesstudy.Howevertheyaresufficientlyluminoustoallowadecent 0 0 S/Nratioataresolutionofabout43000,i.e. 7kms−1,whichallowsanexploratorystudy.Wedetectedtwodistinctgroupsof 2 ISabsorptions, a“local”group, withradialvelocitiesranging from-43kms−1 to50kms−1,andahigh velocitygroup with radialvelocitiesrangingfrom150km/sto165km/s.LikelythehighvelocitycomponentsareduetogasfallingontheSgrdSph n sincehavebeenobservedonlyin2starswhichareseparatedbyonly1.′5inthesky.Wearguethattheobservationssuggesta a J strippingofgasduetothepassageofSgrthroughtheGalacticdisc. 3 Keywords.InterstellarMedium–galaxies:individual:SgrdSphstars:GlobularClusters:individual:Terzan7 1 v 0 1. Introduction siderably fainter, which implies that their use as background 1 sourcesrequiresamajorinvestmentintelescopetime.Itisthus 0 The Sagittarius dwarf spheroidal galaxy (Sgr dSph) is one of ofparamountimportancetomakethebestpossibleuseofthe 1 the nearest satellites of the Milky Way. It offers an interest- 0 availablespectra. ingopportunityforinvestigatingthestructureoftheinterstellar 5 medium(ISM)intheGalacticHaloand,inprinciple,alongall 0 In general dSph galaxies are considered to be gas poor, / thelineofsightuptothedwarfgalaxy.TheUVESspectraof h as opposed to dwarf irregulars. With regard to the gas con- 12giantstarsinSgrand3giantstarsintheassociatedglobular p tent of the Sgr dSph, pointed HI observations of its central - clusterTerzan7,whichhavebeenrecentlyusedtoinvestigate region indicate that it does not contain a significant amount o the chemical composition of the Sgr dSph (Bonifacioetal., r ofneutralgas(Koribalskietal.,1994).ThesearchforHIwas t 2000, 2004; Sbordoneetal., 2004) give us the opportunity s continuedbyBurtonandLockman (1999), whofoundnode- to perform the first study of this kind. The Sgr giants are, of a tectable emission and providedan upperlimit to the HI mass : course,notidealbackgroundsourcesforthestudyoftheISM, v M(HI)<7000M⊙over18deg2,betweenb=−13◦and−18◦.5. since the spectrum is complicated by the many stellar photo- i This result is somewhat surprising since the chemical abun- X sphericabsorptionlines.TheonlysuitableISMlinesforwhich dances (Bonifacioetal., 2000, 2004) and the colours of the r suchastudyispossible,withtheabovedata,aretheNaIDdou- a Sgrstarsshowthatitcomprisesametal-richyoungpopulation blet lines. The observed stars are K giants and display strong implying that the Sgr dSph was forming stars within the last photosphericabsorptionsintheselines. Howeverthelargera- Gyr.Bonifacioetal. (2004)suggestedthatthepassagesofSgr dialvelocityof Sgr(∼ 140kms−1) guaranteesthatat least all throughtheGalacticdisctriggerstarformationepisodes,how- thelowradialvelocityGalacticabsorptionsarenotaffectedby everthisimpliesthatthereisgasavailableinSgrorthatitcan thecorrespondingstellarabsorptions.Luckily,thespectralre- be stripped from the disc. The “Blue Plume” stars are mem- gion adjacent to the Na I D lines is relatively free from ab- bersofSgr(Zaggiaetal.,2004)andthereislittlequestionthat sorptions in the stellar spectrum, these two conditions allow theyaretheMainSequenceoftheyoungpopulation.Stillthe to use these spectra for an exploratory study of the ISM to- questionlingers:wherehasthegasassociatedtothisrecentstar wards Sgr. We further note that the hot stars in Sgr (the so formationgone?ArecentinvestigationofPutmanetal. (2004) called“BluePlume”andtheHBstars),whichwouldbemore hasshownthatgasisassociatedwith thestreamofstellar de- desirablebackgroundsourcesforthestudyoftheISM,arecon- bris found to extend, along the Sgr dSph orbit, for over 150◦ Sendoffprintrequeststo:S.Monai across the south Galactic hemisphere.It thus appearsthat the ⋆ BasedonobservationsobtainedwithUVESatVLTKueyen8.2m tidalinteractionofSgrwiththeMilkyWayresultsinthestrip- telescopeinprogram67.B-0147 pingofgas. Table1.Basicdataonthestarsusedasbackgroundsources Stara α(2000)b δ(2000)b V (V−I)c v σ 0 rad v kms−1 kms−1 432 18 53 50.75 −30 27 27.3 17.55 0.965 160.8 0.3 628 18 53 47.91 −30 26 14.5 18.00 0.928 145.7 0.6 635 18 53 51.05 −30 26 48.3 18.01 0.954 125.2 0.2 656 18 53 45.71 −30 25 57.3 18.04 0.882 135.0 0.3 709 18 53 38.73 −30 29 28.5 18.09 0.917 125.2 0.5 716 18 53 52.97 −30 27 12.8 18.10 0.902 136.0 0.4 717 18 53 48.05 −30 29 38.1 18.10 0.872 142.0 0.4 772d 18 53 48.13 −30 32 0.8 18.15 0.947 143.8 0.8 867 18 53 53.02 −30 27 29.2 18.30 0.933 154.3 0.2 879e 18 53 48.59 −30 30 48.7 18.33 0.965 133.8 0.8 894 18 53 36.84 −30 29 54.1 18.34 0.940 129.5 0.8 927 18 53 51.69 −30 26 50.7 18.39 0.937 144.0 0.3 1282f 19 17 39.36 −34 39 6.4 16.08 1.324 158.2 1.0 1515f 19 17 38.16 −34 29 16.8 16.76 1.156 157.8 1.0 1272f 19 17 37.11 −34 39 11.8 16.62 1.188 158.9 1.0 aStarnumbersforSgrstarsarefromMarconietal.(1998)field1,availablethroughCDS atcdsarc.u-strasbg.fr/pub/cats/J/A+A/330/453/sagit1.dat baccurateto0′.′3(Ferraro&Monaco2002,priv.comm.forSgrandZaggia2004priv.comm.forTerzan7) cTheadoptedreddeningisE(V−I)=0.22 dthisisstar[BHM2000]143ofBonifacioetal. (2000) ethisisstar[BHM2000]139ofBonifacioetal. (2000) f StarnumbersforTerzan7starsarefromthecatalogueofBuonannoetal. (1995)availablethroughCDS atcdsarc.u-strasbg.fr/pub/cats/J/AJ/109/663/table2.dat. Havingthisinmind,ourattentionhasbeenfocusedonthe of a synthetic spectrum, since atmospheric parameters and searchofISMcomponentsatradialvelocitiesclosetothoseof chemical abundances of the background sources are known. the Sgr dSphstars. Thisis certainlya necessaryconditionfor For each star we computed a synthetic spectrum using the gas belonging to Sgr, although radial velocity by itself is not Linux version of SYNTHE (Sbordoneetal., 2004; Kurucz, sufficienttoprovethatthegasisactuallyatthatdistance. 1993) and the same atmospheric parameters and abundances given in Bonifacioetal. (2000), Bonifacioetal. (2004) or Sbordoneetal. (2004), as appropriate. An example of syn- 2. ObservationsandAnalysis thetic spectra is shown in Fig. 2, with this aid we rejected possible identifications of interstellar lines for which one of Details on the observations are reported in Bonifacioetal. thecomponentsappearedtobeseverelyblendedwithastellar (2000), Bonifacioetal. (2004) and Sbordoneetal. (2004). absorptionline.Thewavelength,columndensityandbcoeffi- For the convenienceof the readerwe reportthe basic data on cientwerethereforederivedonlywhenthelinesweredetected thestarsusedasbackgroundsources,inTable1.Thespectral in both componentsand beyonda 3σ leveloverthe synthetic resolutionisabout43000,i.e.thevelocityresolutionisabout7 spectra, i.e. only when the latter were not able to completely kms−1attheNaID2linewavelength. reproducethefeatures.Theanalysiswasperformedbymeans We used the verysame spectra used for stellar abundance of the code fitvoigtminuit(E. Caffau, 2004 private com- analysis in the above quoted papers and we refer the reader munication).Thiscode,starting frominitial guessesprovided to those papersfor details on the data reduction.For our pur- bytheuser,performsavoigtprofile(χ2)fit;priortothecom- pose the main problem in our data is that the terrestrial Na I putationofthe(χ2)thecomputedprofileisconvolvedwithan D emission lines, when present, cannot be satisfactorily sub- instrumentalprofile,inourcaseweusedagaussianprofilewith tracted.ThisimpliesthatthelowestspeedlocalIScomponents aFWHMof7kms−1,whichisappropriateforourdata,ascan areaffected,sometimesseverely.However,thecomponentsin beverifiedfromtheprofilesofunblendedThIlinesinthecal- the Sgr dSph velocity range are always clean from the atmo- ibration arc spectrum. The minimum is found by the MINUIT sphericemission.Theheliocentriccorrectionswerealwaysap- routine(James,2001).Thefittingparametersforeachcompo- pliedbeforesummingthespectra. nentarefour:centralwavelength,columndensity,bfactorand Another serious problem is posed by the stellar absorption gf value.OneoftheusefulfeaturesofMINUITisthatanyofthe lines, though these may be reliably identified with the aid Table 2. Column densities, b factors and velocities, for our Table3.atomicparametersusedinthefittingprocedure sample.*meansthatbandcolumndensitywereassumedand notfitted. line λ(Å) fvalue NaID1 5889.9510 0.65459 Star logN b v NaID2 5895.9240 0.32732 kms−1 kms−1 867 13.42 0.50 −17 inFig.2),toproperlyrecoveralltheparametersoftheISlines 14.63 1.57 −2 (central wavelength, b and column density) it would be nec- 716 11.58 6.05 11 11.48 7.33 34 essary to fit simultaneously the stellar and interstellar lines. 927 12.43 6.81 −15 Although this is possible, in principle, using the stellar syn- 12.42 3.70 −7 thetic spectra, we felt that considering the rather low S/N ra- 11.94 1.45 9 tio inthe spectraandtheadditionaluncertaintyintroducedby 635 11.80* 0.60 150 the modelling of the stellar absorptions the results would be 11.50* 0.50 165 affected by a large error,too large in fact to be a useful mea- 432 12.47* 0.93 −16 sure.Instead,intheexploratoryspiritofthepresentwork,we 13.84* 1.30 −5 decided in these cases to assume the values of b and column 13.00* 1.10 7 density and fit only the central wavelength of the IS absorp- 879 12.40* 5.30 −1 tion.Althoughthisprocedureissomewhatarbitraryitisfairly 12.60* 1.40 18 robust. As long as column density and b are fixed arbitrarily, 12.00* 0.20 40 12.00* 0.20 50 but in such a way that the resulting synthetic spectrum is not 628 13.90* 0.20 −4 obviously incompatible with the observed spectrum, the fit- 12.40* 0.68 10 ted centroid of the componentdoes not change by more than 11.50* 1.70 28 1 kms−1. Thisis clearly sufficientforthe exploratorypurpose 11.60* 0.30 36 of the present study. The results of our analysis are reported 11.40* 0.30 44 in Table 2. When b and column density were kept fixed, the 717 12.45* 0.28 −15 symbol*appearsnearthelogNvalues. 12.60* 0.33 −5 Afewcommentsareinorderhere.Theavailableresolutionis 12.45* 0.58 14 somewhatlimitedforadetailedanalysisofthekinematicstruc- 772 13.20* 2.08 −2 tureoftheISM.Itisquitelikelythatinhigherresolutionobser- 12.46* 3.43 17 vationsthe“components”wefittedsplitintoseveralnarrower 11.71* 2.11 26 656 12.10* 1.70 −17 components.The numberof componentsto fit to a givenfea- 13.40* 1.90 −5 ture issomewhatarbitraryand,forthe abovereason,maynot 11.50* 0.67 160 haveparticularphysicalmeaning.Weusedthecriteriontouse 709 12.64 0.94 −16 theminimumnumberofcomponentsinthefit. Itisthusclear 13.15 2.41 −2 thatwedonotgiveparticularphysicalsignificancetothecol- 12.18 3.63 15 umndensitiesandbfactorsobtained,keepinginmindthatthey 14.53 0.00 36 reflectthe combinedeffectofseveralgasclouds.Inparticular 894 11.80* 1.40 −16 bfactorsarenotverysignificant;mostofthevalueswefindare 12.40* 2.75 −5 wellbelowourinstrumentalresolution,andthus,notverywell 11.80* 1.70 8 constrained.WestressherethattheFWHMofallthedetected 12.60 0.40 20 1282 11.87 2.82 −42 featuresisabovetheinstrumentalresolution.Averylowfitted 13.72 2.29 1 value of b simply reflects the fact thatthe FWHM of the fea- 1515 11.60 2.04 −40 ture is only very slightly larger than the instrumental profile, 12.70* 3.90 0 howeverattheS/Nratiosofthepresentdataallthevaluesofb 1272 12.38 2.33 −43 upto7kms−1 providealmostequallyacceptablefits.Whatis 13.79 2.34 0 significantisthatwedonotobservebfactorsinlargeexcessof ourinstrumentalresolution.Thisallowstoconcludethatthere isnostrongturbulencewithinthecloudsandthatthevelocity parameterscanbeheldfixed,evenallfour,inwhichcasethere dispersionofgroupsofcloudswhichclusterinvelocityspace isnofitting,theroutinesimplycomputesalinefortheinputpa- (our“components”)islessthanourinstrumentalresolution,i.e. rameters.Inourcasethegf values,whicharewellknownfor 7kms−1. the Na I D lines, were of course always keptfixed (see Table 3). 3. Resultsanddiscussions Atthisrelativelylowresolutionthereisnoneedtotakeinto accountthehyperfinestructureoftheNaIDlines,whichsim- FromTable2wecaninferthepresenceoftwodistinctgroups ply results in a slightly larger b value. In the cases in which ofIScomponents:onewhichwecall“local”,rangingfrom-43 the detected IS lines are contaminated by the stellar absorp- kms−1 to 50 kms−1, and a high velocity group ranging from tions(clearexamplesarethehighvelocitycomponentsshown 150kms−1 to165kms−1. Fig.1.MapoftheobservedstarsinSgrdSph,thestarsinsquaresarethoseinwhichhighvelocityabsorptionshavebeendetected, thecross(×)indicatesthecenterofthefield,labeledbyitscoordinates.Scalesareinarcminutes. 3.1.Thelocalcomponents bvalues(wherereliable)arenotashomogeneousasthoseob- servedinthedirectionofTerzan7. In most cases the sky contaminationspreventus from precise Othercomponentsbehavein thesame ubiquitousmanner, column densities or b measures, neverthelessthe velocity de- temptatively we can identify the following ones: 9-15 kms−1, terminationsarereliable. 18-28kms−1,34-36kms−1and40-45kms−1thelattertwocom- Weunderlinethepresenceofthe≃-43km/sand≃0kms−1 to- ponent are present towards 3 stars. Other absorptions are ob- wardsallthe Terzan7stars, whichare probablyduetoa local served towards less than 3 lines of sight, their distances are cloudcoveringthesmallangularspreadoftheselinesofsight. probably greater then the previous ones, but we cannot infer Thecolumndensitiesandbvaluesareconsistentwiththissce- any more aboutthem. The evidentabsorption features of star nario. #635havenotbeenfittedduetosevereskycontaminationsof The component around –15 kms−1 is also quite ubiquitous, bothD1andD2lines. we find it towards 7 out of 12 lines of sight towards the Sgr dSph. The RedfieldandLinsky (2000) model for the G and Local Interstellar Cloud in these directions, gives velocities 3.2.Thehighvelocitycomponents of –24.27 kms−1 and –22.2 kms−1 respectively, at our reso- lution these two components would appear as a unique fea- As already said, this is the first investigation towards the Sgr ture at ∼ −23 kms−1. It is not clear whether the component dSph, therefore the finding of high velocity IS absorptions around−15kms−1foundbyuscouldbeindeedcompatiblewith could revealthe presence of gas in it, or in the Galactic Halo the RedfieldandLinsky (2000) model. As mentioned above, ontheothersideofGalacticcentre. we had a heavy contamination by atmospheric NaI emission Only three components of velocities 150, 160, and 165 lines, and considering that the heliocentric corrections are kms−1 havebeenfoundtowardsonlytwolinesofsight,asre- around–8,–10kms−1,apossiblesystematicerrorinourveloc- portedin Fig. 1. Inparticular,the 150and165kms−1 are re- itymeasure,duetoanincorrectsubtractionoftheatmospheric vealedtowardsstar#635asshowninFig.2.Thenonubiqui- emission, leaves open the question of the identification with tous natureof these absorptionsshouldindicate the greatdis- the above components predicted by the RedfieldandLinsky tance of the absorbinggas, given the relative angular clumpi- (2000)model. nessofoursample,andinparticularofthesetwostars.Nearby Another component is revealed between –7 and –2 kms−1. clouds should cover more than 2 or 3 lines of sight, but, at This velocity spread is not very significant, since our veloc- the other side of our Galaxy, their angular sizes could not. In ityresolutionis≃7kms−1,perhapsitcouldberelatedtothe0 particular, the nearest star to # 635 is # 927, which shows no kms−1 componenttowardTerzan7.Thecolumndensitiesand detectableISabsorptionsatthesevelocities,itsangularsepara- Fig.2.Spectraofstarsshowinghighvelocitycomponents.Thedashed-dottedlineisthesyntheticISfitwhereasthecontinuous lineisthesyntheticstellarspectrum.Thelocalcomponentsof#635areseverelycontaminatedbytheskyandnofitwasreliable. tioncorrespondstoalinearseparationof≃2.9pcatthedistance that the gas is at the same distance as Sgr. In the first place of Sgr dSph. The non detection could be explained by three it shows that there is a small amount of gas in Sgr, possibly causes: clumpyandconsistingofmanysmallclouds.Thisgasmaybe theresidualofthegaswhichwasusedupinthelaststarforma- 1. since the radial velocity of star # 927 is 18.8 km/s higher tion episode of Sgr, which gave birth to the youngmetal-rich than that of the # 635, it is obvious from Fig.2 that the populationfoundbyBonifacioetal. (2004),therestbeingnow stellar features could mask completely the IS ones in this intheformofstarsorlostalongtheorbit.Inthesecondplace velocityrange. sincethisgasischemicallyenrichedand“falling”towardsSgr 2. sincefromtheanalysisofBonifacioetal. (2004)the#927 it could well be gas which has been stripped from the Milky is moremetalrichthan# 635bya factoroftwo, itshows Way. If ateach passage Sgr is actually capable of“refueling” stronger stellar absorptions just around these components with metal enriched gas stripped from the Galactic disc, this andthisprobablycould,byitself,preventusfromdetecting couldbeameanstomaintainthestarformationepisodesovera themeveniftheywerepresent; verylongperiodoftime,astestifiedbytheverylargemetallic- 3. thecloudsareveryclumpy,ofsizelessthan2.9pc. ityspreadofSgr(−3≤[Fe/H]≤0.1Zaggiaetal. 2004),even Arguments1) and 3) apply also to star # 628, which falls ex- without a large reservoir of local gas. Moreover it could ex- actly between the # 635 and # 656, and has a larger angular plaintheanomalouslyhighmetallicityofSgr,comparedtoits distance than the angular distance of star # 927 from star # luminosity;perhapsSgrisnotsochemicallyevolvedbecauseit 635,fromeitherstar#635orstar#656.ThelackofHIdetec- retainsefficientlytheejectaofitsownSNe,butsimplybecause tionsintheSgrdSphdirection,(BurtonandLockman,1999), it “steals” enriched materialfrom the Milky Way. These con- could be well explained by the small angular size of case 3). siderationsarestillhighlyspeculative,howevertheyprovidean Moreover,itisveryinterestingtonotethatthesevelocitiesare importantthrustfor the continuationof such an investigation. always greater than the stellar ones, implying that the gas is Data for more lines of sight using hotter backgroundsources “falling”towardsthem. whichspana rangeofdistancesfromtheGalactic disc to Sgr arebadlyneededtotestthesehypothesis. 4. Conclusions Acknowledgements. We are grateful to E. Caffau for providing the The presence of high velocity absorptions towards 2 out of a fitvoigtminuitcodeandtoS.Zaggiaforprovidingaccuratecoor- sample of 12 stars of the Sgr dSph is the main result of our dinatesoftheTerzan7stars.WearealsoindebtwiththerefereeDr.J. analysis. Given the BurtonandLockman (1999) upper limit Linskybecausethearticlehasbeenimprovedthankstohisconstruc- on HI mass over the area, one can derive an upper limit for tivecomments.ThisresearchwasdonewithsupportfromtheItalian the HI column density of logN(HI) ≤ 18.87. We stress the MIURCOFIN2002grant“StellarpopulationsintheLocalGroupasa fact that the 21′ radio beam integratesthe signal overan area tooltounderstandgalaxyformationandevolution”(P.I.M.Tosi). whichisconsiderablylargerthanourfield.TheSgrstarswere initially foundas radialvelocitymembersonan EMMI MOS References frame (Bonifacioetal., 1999) which had a 9′ side. In spite of this it is interesting to derive an estimate of an upper limit Bonifacio P., Pasquini L., Molaro P., Marconi G., 1999, onthecorrespondingNaIcolumndensity.Suchanestimateis Ap&SS,265,541 clearlyquiteuncertain,infactonehastoassumeaNa/Hratio, Bonifacio P., Hill V., Molaro P., Pasquini L., Di Marcantonio an ionizationbalance and a dustdepletion factor. Rather than P.,SantinP.2000,A&A,359,663 relying on theoretical estimates, we can assume the physical BonifacioP.etal.A&A,414,503,2004 stateofthegastobesimilartowhatobservedbyMolaroetal. BuonannoR.etal.1995,AJ109,663 (1993)towardstheMagellanicClouds.Theyfoundavalueof BurtonWandLockmanF.J.A&A,349,7,1999 log(NaI/HI) = −7.79 for IntermediateVelocity Clouds, thus James, F., 1998 MINUIT, Reference Manual, Version 94.1, we could expect logNaI ≤ 11.08. The fact that we assume CERN,Geneva,Switzerland higher column densities, but still compatible with the obser- KoribalskiB.,JohnstonS.andOtrupeckR.,MNRAS,270,43 vations, suggests one out of three possibilities: either our as- 1994 sumedcolumndensitiesaretoohighby∼0.5−0.8dex,orour Kurucz,R.L.1993,CD-ROM13,18 http://kurucz.harvard.edu detected high velocity absorptions arise in a clumpy medium Marconi, G., Buonanno, R., Castellani, M., Iannicola, G., whichpreventsdetectionofHIemissionwitharadiobeamof Molaro,P.,Pasquini,L.,&Pulone,L.1998,A&A,330,453 21′, or the physical conditions in the gas are different from Molaro,P.etal.A&A,274,505,1993 what assumed to derive the upper limit on NaI column den- Putmanetal.Ap.J.,603,L77,2004 sity and lead to log(NaI/HI) > −7.0. Since this galaxy re- RedfieldS.&LinskyJ.L.Ap.J.,534,825,2000 vealsalackofgas,probablystrippedawayduringthelastperi- Sbordone, L., Bonifacio, P., Marconi, G., & Buonanno, R. galacticpassage,itisalsopossiblethattheinteractionwiththe 2004,MSAIt,75,396 MilkyWaycausedsomegastofallontoSgr(seePutmanetal., Sbordone,L.,Bonifacio,P.,Castelli,F.,&Kurucz,R.L.2004, 2004).Althoughourconclusionhastobe consideredprelimi- MSAIS,5,93 nary, since it is based on only two background stars and we Zaggiaetal.MSAIS,5,291 havenoindicationontheactualdistanceoftheclouds,itisin- terestingtoconsidertheimplicationsofthisfinding,assuming