Astronomy & Astrophysics manuscript no. t February 2, 2008 (DOI: will be inserted by hand later) ⋆ Cool carbon stars in the halo: a new survey based on 2MASS 1 2 3 4 N. Mauron , M. Azzopardi , K. Gigoyan , and T.R. Kendall 1 Grouped’Astrophysique,UMR5024CNRS,CaseCC72,PlaceBataillon, F-34095MontpellierCedex5,France e-mail: mauron@graal.univ-montp2.fr 2 IAM, Observatoire deMarseille, 2 Place Le Verrier, F-13248 Marseille Cedex 4, France 4 3 378433 Byurakan Astrophysical Observatory & Isaac Newton Institute of Chile, Armenian Branch, Ashtarak 0 d-ct,Armenia 0 4 Laboratoire d’Astrophysique, Observatoire de Grenoble, Université Joseph Fourier, BP 53, 38041 Grenoble 2 Cedex 9, France n a J Received xxx/Accepted xxx 7 2 Abstract. Wepresent thefirst results of a new surveyfor findingcool N-typecarbon (C) stars in thehalo of the Galaxy. Candidates were first selected in the 2MASS Second Incremental Release database with JHKs colours 1 typicalofredAGBCstarsandKs <13,andsubsequentlycheckedthroughmediumresolution slit spectroscopy. v Wediscovered27newCstarsplusoneknownpreviouslyandtwosimilarobjectsintheFornaxandSculptordwarf 4 galaxies.Wedetermineanddiscussthepropertiesofoursample,includingopticalandnear-infraredcolours,radial 6 velocities,aswellasHαemissionandvariabilitythatarefrequent,allthesecharacteristicsbeingcompatiblewith 5 an AGBC-typeclassification. Surprisingly,of the30studied objects, 8 werefound tohavesmall butmeasurable 1 proper motions (µ) in the USNO-B1.0 catalogue, ranging over 8<µ<21 masyr−1 and opening the possibility 0 that some objects could perhaps be dwarf carbon stars. Yet, a detailed analysis based on comparison with the 4 sample of known carbon dwarfs leads us to consider these µ as incompatible with the broader picture suggested 0 by the other data taken as a whole. So, we adopt the view that all objects are of AGB type, i.e. luminous and / h distant. Because thestream of Sagittarius dwarf galaxy is known to be thedominant source of luminous C stars p in the halo, we chose to determine distances for our sample by scaling them on the 26 known AGB C stars of - o the Sgr galaxy itself, which are found to be, in the Ks-band, ∼ 0.5mag. less luminous than the average LMC C r stars for a given J −Ks colour. The obtained distances of our halo stars range from 8 to 80kpc from the Sun. t Then, examination of position and radial velocities show that about half belong to the Sgr stream. Our findings s a suggest that numerous AGB C stars remain to be discovered in the halo. Long term Ks-band monitoring would : be of great value toascertain distance estimates through the period-luminosity relation, because a large fraction v i of oursample is probably made of Mira variables. X r Key words.Stars: carbon, surveys,galactic halo; Galaxy: stellar content a 1. Introduction the asymptotic giant branch (AGB) evolutionary phase (as is often the case for cool C stars in the galactic disk), Surveysofstellarpopulationslocatedathighgalacticlat- with an R-band magnitude of the order of 15, its high itudeareimportanttocharacterizethehaloandtounder- luminosity (M ∼ −3.5) puts it as far as 50kpc from R stand how the Galaxy formed (see for example Majewski the Sun. Therefore, the luminous C stars constitute valu- 1993,andreferencestherein). Amongthe varioustypes of able probes of the distant halo (e.g. Bothun et al. 1991). stars that have been investigated with this goal, the case Considerableeffortshavebeenaccomplished,andarestill of carbon (C) stars has been the subject of much atten- in progress, in order to find such faint high latitude car- tion for some years. If such a C star is proven to be in bon stars (FHLCs). These rare objects have been discovered using two main methods. The first is by exploiting Send offprint requests to: N.Mauron Schmidt objective-prismplates where C stars have a con- ⋆ Based on observations made at the European Southern spicuous spectral appearance (MacAlpine & Lewis 1978, Observatory,Chile (programs 67.B 0085AB, 69.B 0186A) and Sanduleak&Pesch1988,Gigoyanetal.2001,Christliebet at the Haute Provence Observatory (France) operated by the al.2001).Thesecondmethodusesapreliminaryselection CentreNational deRechercheScientifique,togetherwithdata ofcandidateswithsuitablephotometriccriteria,suchasa from the 2MASS project (University of Massachusetts and very red B−R colour index, as in the APM (Irwin 2000) IPAC/Caltech, USA). 2 N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars survey of Totten & Irwin (1998; hereafter TI98), or mul- sented and analysed in this work. After describing our ticolour properties as in the SLOAN carbon star survey selectionmethod(Sect. 2),the spectroscopicobservations of Margon et al. (2002), with subsequent verification the are reported in Sect. 3. In Sect. 4, we examine the var- ofthe carbonstarnatureofthese candidatesby follow-up ious properties of the sample, including radial velocities, spectroscopy. variabilityandpropermotions.Theseresultsareanalysed One of the most striking results derived from these in Sect. 5 together with a determination of distances and FHLCsurveys,especiallyfromtheAPMone,wasthefact examination of membership to the Sgr stream. The main that the tidal stream of the Sagittarius dwarf spheroidal conclusions are finally summarized in Sect. 6. galaxy (Sgr) orbiting the Galaxy could be traced for the firsttimebyconsideringthespatialandkinematicalprop- 2. Selection of candidates erties of the distant coolC stars (Ibata et al. 2001a).The Sgr stream has now been detected through a number of In order to find new FHLC stars, we first considered other methods, such as deep mapping in limited portions all the FHLCs published in the literature and located ◦ of sky of specific populations,e.g., blue horizontalbranch at |b| > 30 . After retrieving their JHK photometry s stars,metalpoorKgiants,turnoffstars,RRLyrvariables fromthe 2MASSSecondIncrementalData Releasepoint- (Dinescu et al. 2002, Dohm-Palmer et al. 2001,Kundu et source catalogue (available when this work started and al. 2002, Vivas et al. 2001, Martinez-Delgado et al. 2001, covering about half of the sky), we plotted them in a Newberg et al. 2002), or with 2MASS selected M giants colour-colour JHK diagram (Fig 1). Very similar dia- s over the whole sky (Majewski et al. 2003). Yet, there are grams,whichinspiredoursearchmethod,havebeen pub- several reasons to pursue the search for cool luminous C lished by Totten, Irwin and Whitelock (2000; TIW), and stars all over the high latitude sky. Firstly, the detection Liebert et al. (2000). It can be seen in Fig. 1 that the of this stream with cool FHLCs currently involves only large majority of FHLCs have an H−K colour of about s ∼ 40 stars, so that enlarging the sample is naturally de- 0.2.Thesestarsarerelativelywarm,presumablyCH-type sirable. Secondly, cool AGB C stars are a population of objects andcome mainly from the Hamburg/ESOsample intermediate age, and consequently their spatial distri- ofChristliebetal.(2001).ThecoolN-typestarswhichwe bution in the Sgr orbits might provide some interesting seek are located atH−K largerthan ∼ 0.3,and appear s informationonthe historyofthemergingprocess.Afrac- toformarelativelywelldefinedlocusuptoH−K ∼1.1, s tion of these AGB C stars may also be Mira variables, although the number of objects is progressively decreas- and help to determine distances of these orbits through ing. The width of this locus is typically ∼ 0.25 mag. This theperiod-luminosityrelation.Finally,roughlyhalfofthe plot also suggests that the C star locus extends up to the cool FHLC stars do not belong to the Sgr stream; their twoobjectsatH−K ∼1.6,andsuchanextensionissup- s originhastobe investigatedandincreasingthesizeofthe ported when one considers a similar diagramshowing the sample studied may possibly reveal other streams. LMCCstarslistedthecatalogueofKontizasetal.(2001) However,inthesearchforFHLCs,onehastotakeinto (see also e.g., Nikolaev & Weinberg 2000, their Fig. 2). account that the C stars in general are of various types At still redder colours, Fig. 1 also shows also two excep- and with diverse evolutionary origins (see Wallerstein & tionally cool N-type stars with H −K ∼ 2.0. These are s Knapp 1998 for a review). Compared to the bright AGB the very dusty C stars IRAS0846+1732, found by Cutri ◦ ◦ stars,onefamilyconsistsoflessluminous,warmercarbon- et al.(1989),for whichl=210 ,b=+35 , J−H =2.37, rich objects presently evolving as clump giants or located H −K = 2.01 and K = 10.71, and IRAS03582+1819, s s ◦ ◦ along the red giant branch, having accreted carbon from found by Liebert et al. (2000), with l = 210 , b = −25 , a more evolved companion (Knapp et al. 2001, Christlieb J −H =2.59, H −K =2.07, and K =9.26. The latter s s ◦ et al. 2001). Moreover, it is now well established that a objectisplottedinFig.1despitehaving|b|<30 because class of dwarf carbon stars exists (dCs; see e.g., Dahn itsheightabovethegalacticplaneisestimatedbyLiebert et al. 1977, Green 2000, Margon et al. 2002, Lowrance et al. to be in the range 6–15kpc. et al. 2003, and references therein). These dCs have very Our method for searching for cool C stars was there- low luminosity, are located within a few hundred parsecs, fore to select in 2MASS objects lying within a distance of havegenerallymeasurablepropermotions,andinfactare ∼ 0.15mag to the median line formed by these template expected to outnumber the C stars of giant type as ob- cases. In order to avoid a large number of ordinary (M- servationsprobe successively fainter magnitudes (Margon type) stars in our selection, we had also to set a limit on 2003). colours,e.g.H−K >0.4,J−H >0.95,meaningthatwe s In this context, we report here on the first results of naturallymissthenumerouswarmbutlessluminousgiant a new systematic search for faint, red AGB C stars at Cstarsthataremuchbetterselectedbyothertechniques, high galactic latitude. Our survey is essentially a near- e.g. through the SLOAN multicolour criteria. Concerning infrared based survey, since our candidates have been se- the limits in galactic latitude, our nominal goal was to ◦ lected fromthe 2MASS SecondIncrementalData Release limit our searchto |b|>30 . However,we also considered ◦ point-sourcecatalogue.About halfofour∼200bestcan- withalowerprioritycandidateslocateddownto|b|∼25 , didates have now been observed spectroscopically, result- especially if they showed an additional favourable prop- ing in the discoveryof 27 new coolFHLCs which are pre- erty suchas very redB−R orJ−K colours,and 6 new s N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars 3 Table 1. Journal of Observations slit width had to be set to 2′′, and the resulting velocity resolution of the spectra is ∼ 90kms−1. AtESO,weusedthe DFOSC focalreducerwhichper- Run# and dates Site # of observed Objects mits both direct imaging and slit spectroscopy. After a 2min image generally taken in the R Bessel filter for 1 2001 Mar 26 - Mar 30 OHP 2, 3, 5 to 10, 14 source identification, the spectrum was obtained through 2 2001 Mar 31 - Apr01 ESO 4, 7, 11, 12, 13 grism #8 which provides a range of 5800 to 8400˚A and 3 2001 Sep 09 - Sep 13 ESO 1, 16, 17, 19, 22 to 28 adispersionof1.2˚Apixel−1 onthedetector,a2148×4096 4 2001 Oct 17 - Oct 22 OHP 15 EEV/MATCCDwith15µmpixels(halfofthe CCDarea 5 2002 Fev 14 - Fev 18 OHP no observations (clouds) is not useddue to the reducer design).The slit width was 6 2002 Aug29 - Sep 03 ESO 17, 18, 20, 21, 29, 30 1.5′′ and the velocity resolution ∼ 120kms−1. 4. Results C stars were found at these low latitudes (more details in In our list of ∼ 200 best candidates, slit spectroscopy Sect. 4). has so far been secured for 97 of them: 30 were found Concerning the limit in brightness, our nominal limit to be C stars,including onethat ismember ofthe Fornax wassetbyK <13,whichcorrespondstotheratherlarge s dwarfgalaxy(#30)andonelocatedinthedirectionofthe distanceof∼150kpcfromtheSun,ifoneadoptsasabasis Sculptor dwarf galaxy (#29) (see Table 2). The 67 other the typical not too red LMC C stars with J −K = 1.6, s objects (not C stars) were found to be mainly M-type gi- which have a mean K of 10.7 (σ =0.4). s ants and will be the subject of future work. After selection in 2MASS which yielded ∼ 1200 ob- The last two objects (#29 and #30) were under con- jects, we excluded the objects that were already known siderationasinterestingcomparisonobjects.Noradialve- andcataloguedintheSIMBADdatabaseasMorCstars, locitycouldbedeterminedbyusfor#29,anditsmember- youngstellarobjects,Ldwarfs,galaxiesorQSOs.Wealso ship to Sculptor needs further observations to be proven. excluded objects with USNOC-A2.0 colour B −R bluer Thisobjectisconsiderablyredder(J−K =3.3)thanthe s than 1.5, when these B and R magnitudes are provided other C stars known in Sculptor (Azzopardi et al. 1986, by the 2MASS database.This is justified by the fact that Aaronson& Olszewski1987)for which J−K is between s many M stars and galaxiesare excluded by this criterion, 0.8and1.12.It appearsveryfaintinthe R-bandPOSS-II while N-type stars are expected to be much redder than image,is invisible onblue plates,butis wellseeninthe I- this limit and have generally B−R ∼ 3. Eventually, we bandUKSTdigitizedimage.ConcerningObject#30,this found 6 new C stars with 2 < B −R < 3 and 2 with starwasalreadynotedbyDemersetal.(2002)asaproba- B−R=1.9and1.6(seebelow).InspectionofPOSSplates ble Fornax carbon star based on its 2MASS near-infrared was also systematically used for further sample cleaning, magnitudes and colors: our spectrum confirms its carbon and numerous supplementary cases of faint, contaminat- nature and provesits membershipthroughradialvelocity ing galaxies were discarded. In addition, the objective- determination. prismplates ofthe FirstByurakanSurveywereexamined All of the C stars found have K < 12.3, with the s byoneofus(K.G.)forrelativelybrightcandidateslocated exception of the Fornax C star at K = 12.68. We also s in the zones covered by this survey, and this allowed the observedasmallsupplementarylistof9 faint(1330◦); Right panel: Colour-colour diagram of the targets for which slit spectroscopy have been obtained (circles). The new carbonstars found in this work are indicated by an overplotted+ sign. Note that the abscissa and ordinate scales differ in the two panels. Here we shall ignore the last two C stars (#29 and The columns J,H,K ,J − K are from the 2MASS s s #30) which are in the Fornax and Sculptor galaxies re- SecondIncrementalrelease.Thetypicalerrorsonthispho- spectively.Itcanbeseenthatofthe28remainingobjects, tometry are ∼ 0.03mag or better. All C stars of our sam- ◦ 22 have been found at |b| > 30 . One star, #5, was erro- ple have J − K > 1.3, which is largely due to our se- s neously rediscoveredandwasknownasFBS 1056+399or lection criteria excluding objects with J −H < 0.95 and APM 1056+4000 (Gigoyan et al. 2001, TI98). Its 7500– H − K < 0.4. Therefore, they are distinctly redder in s 8000˚A spectrum is in Gigoyanet al. (2001), and a 5700– J −K than most of the numerous warm C stars of the s 6600˚A spectrum has been obtained here: it shows Hα in Hamburg-ESOsurvey.Thecolour-colourdiagramofFig.1 emissionand this new spectrumwas used to derivean in- also shows that no candidate redder than J −K ∼ 3.4 s dependent radial velocity measurement which is in very was observed, essentially because C stars or candidates good agreement with that of Gigoyanet al. (2001). redder than this are very rare at high b. In Table 2, the columns B and R, and corresponding B−R index are from the USNO A2.0 catalogue, except 4.2. Spectra for 8 objects which are not present in this catalogue, in which case the data from USNO-B1.0 are given (see the Forclarityofthetext,theatlasofallthespectraisshown Notes of Table 2 for details). These B and R magnitudes in Appendix A. A detailed study of these spectra will be provide only approximate optical photometry with prob- performed in a future paper, and only a few remarks will able errors of ∼0.4 mag, and should also be considered be made here. with caution since many objects are clearly variable, and First, all ESO spectra display a strong rising flux be- objects at |b| < 30◦ may also suffer some interstellar ab- tween6000and7800˚A with a flux ratioof about2 to 15. sorption(seebelow).Howeveritisinterestingtonotethat The OHP spectra have a too small domain to be consid- the magnitude range in R is between 10.5 and 19.8, and ered similarly. This slope is clearly larger for our objects the medianin R is 14.7.For comparison,the carbonstars thanforthewarmergiantordwarfSLOANCstarsshown of the APM survey (see Table 3 of TI98,with 41 stars la- by Margon et al. (2002), for which the 6000–7800˚A flux belled“APM”)haveanRrange10.0–18.0andamedianof distribution is nearly flat. The cool APM stars and the 13.6mag.Themedianofoursample14.7correspondstoa twoN-typeCstarsSDSSJ144631.1-005500andJ1227400- distanceof∼44kpcifoneadoptstheabsolutemagnitude 002751 (previously known as APM1225-0011) shown in ofM =−3.5consideredbyTI98,andifnocircumstellar Margon et al. have spectra very similar with ours. R or interstellar absorption in the red (A ) is assumed (if One notes also that Hα is in emission in 13 of our 28 R A =1 mag., one finds 27kpc). halo objects,i.e., 46%.This fractionis higher but compa- R N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars 5 Table 2.Listofdiscoveredfaintcoolhalocarbonstars.Coordinatesα(in h.,min.,sec.)andδ (in deg.,min,sec.)are from2MASS.l,bareindegrees.B &Rinmag.arefromUSNO-A2.0(±0.4mag.approximatively),exceptforobjects with a Note. JHK are from the 2MASS 2nd Incr. Release database (in mag.; errors ± 0.02-0.03mag. or better) s No. α(J2000) δ(J2000) l b B R B-R J H Ks J-Ks Notes 01 02 11 30.866 −03 49 43.85 165.80 −59.86 18.5 14.9 3.6 12.034 11.007 10.449 1.585 02 09 13 31.865 +19 34 22.64 209.31 +39.79 16.9 11.5 5.4 9.043 7.953 7.363 1.680 03 09 15 05.206 +19 17 37.89 209.81 +40.04 16.8 12.6 4.2 10.203 9.186 8.476 1.727 04 10 15 25.934 −02 04 31.84 244.49 +42.43 20.9 16.3 4.6 14.045 12.861 11.987 2.058 (1) 05 10 59 23.839 +39 44 05.60 177.25 +63.60 16.9 13.3 3.6 11.082 10.080 9.442 1.640 06 11 09 59.686 −21 22 01.15 273.53 +35.65 14.2 10.5 3.7 8.390 7.482 7.001 1.389 07 11 17 19.005 −17 29 15.39 273.18 +39.88 19.4 16.5 2.9 13.464 12.403 11.632 1.832 08 12 09 25.022 +15 16 18.49 261.33 +74.64 17.8 14.2 3.6 11.185 10.277 9.822 1.363 09 12 49 04.767 +13 20 35.51 300.53 +76.20 19.3 14.5 4.8 12.606 11.604 11.136 1.470 10 13 56 02.371 −01 36 26.20 333.93 +57.32 19.4 15.5 3.9 12.915 11.860 11.327 1.588 11 13 59 20.636 −30 23 39.48 319.88 +30.23 - 19.8 - 14.577 13.072 11.798 2.779 (2) 12 15 01 06.923 −05 31 38.70 351.63 +44.74 20.4 16.8 3.6 13.571 12.348 11.506 2.065 (3) 13 15 15 11.063 −13 32 27.93 348.10 +36.43 18.7 17.1 1.6 12.594 11.516 10.785 1.809 (4) 14 15 58 42.227 +18 52 46.86 32.33 +46.37 17.5 14.6 2.9 12.269 11.387 10.966 1.303 15 17 28 25.766 +70 08 29.93 100.83 +32.41 18.1 13.9 4.2 11.551 10.111 9.048 2.503 16 19 42 19.018 −35 19 37.69 4.40 −25.06 18.6 16.7 1.9 12.633 11.142 10.068 2.565 17 19 42 21.315 −32 11 04.19 7.70 −24.13 19.0 14.7 4.3 11.967 10.817 9.981 1.986 18 19 48 50.653 −30 58 31.92 9.43 −25.08 - 17.7 - 12.998 11.215 9.862 3.136 (5) 19 19 53 30.172 −38 35 59.40 1.52 −28.07 19.2 13.9 5.3 11.292 10.088 9.244 2.048 20 20 13 19.435 −23 41 44.26 19.07 −27.92 14.4 11.7 2.7 9.541 8.591 8.109 1.432 21 20 20 27.661 −14 49 27.10 29.05 −26.26 19.4 14.7 3.7 11.849 10.162 8.711 3.138 (6) 22 20 54 54.551 −28 28 56.73 16.76 −38.23 18.6 14.3 4.3 12.407 11.451 10.858 1.549 23 22 05 14.590 +00 08 46.06 60.31 −41.67 18.1 14.3 3.8 10.709 9.503 8.714 1.995 24 22 06 53.669 −25 06 28.28 26.55 −53.17 18.0 15.1 2.9 10.934 9.756 8.922 2.012 25 22 17 09.923 −26 07 03.35 25.64 −55.64 18.4 15.4 3.0 11.056 9.823 8.882 2.174 26 23 17 21.087 −24 11 42.41 35.54 −68.63 17.8 15.5 2.3 13.750 12.750 12.280 1.470 27 23 19 35.533 −18 56 23.79 49.28 −67.38 17.5 14.8 2.7 11.477 10.473 9.958 1.519 28 23 25 31.394 −30 10 56.06 18.64 −70.94 18.4 14.6 3.8 13.409 12.028 11.029 2.380 Two carbon stars in the direction of Sculptor (#29) and in Fornax (#30) 29 00 59 53.680 −33 38 30.77 287.82 −83.24 - 20.2 - 14.877 13.144 11.591 3.286 (7) 30 02 41 03.550 −34 48 05.34 237.84 −65.37 23.3 18.3 5.0 14.445 13.397 12.682 1.763 (8) Notes: (1) B & R are B2 & R2 from USNOC-B1.0 in which R1=18.5; in theAPM catalogue, onefindsR=18.3 and no data for B (2) R is R2 from USNOC-B1.0 in which no otherdata in R or B are given; in APM, R=20.25 and nodata in B (3) B & R are B2 & R2 from USNOC-B1.0 in which R1=16.0; in APM, theobject is blended with neighbours (4) B & R are B2 & R2 from USNOC-B1.0 in which R1=16.7; in APM, R=17.1 B=18.35 (5) R is R2 from USNOC-B1.0 in which R1=15.8 but no data in B is given; in APM, R=18.2 and no data in B (6) B & R are B2 & R2 from USNOC-B1.0 in which R1=16.2; no data in the APMcatalogue for this position (|b| is too low) (7)RisR2 from USNOC-B1.0inwhich nootherdatain RorB aregiven;inAPM,R=20.3andnodatainB;membership to Sculptorrequires supplementary observations and radial velocity determination. (8) B & R are B2 & R2 from USNOC-B1.0 in which R1=16.9; in APM, R= 18.3 and no data in B; this star was previously identified as probable C star by Demers et al (2002) on the basis of its near-infrared photometry (#25 in their Table 1) rable to the result of Maizels & Morris (1990) who sur- suggest that most of our stars are pulsating AGB stars veyed 37 galactic “bright C stars” (presumably of AGB (with a shock wave being the cause of the Hα emission). type, but no details are given on the observed stars or their selection method) and found Hα emission in 14 of them(38%).IntheAPMsurvey,examinationofthespec- 4.3. Radial velocities tra in Fig. 5 of TI98 indicates that 6 of 20 N-type stars haveHαemission(30%),and1 of8CH-type stars(12%). During each run, we observed several times a small num- Incontrast,amongMargonetal.’swarmC stars,5outof ber of template carbon stars with known radial velocities 39 have Hα emission (13%). Therefore the high fraction (see Table 3). In the following, we shall call these stars of Hα emission in our sample and the above comparisons “radial velocity standard”, although, for several reasons, theycannotbeconsideredasclassicalstandardstarswith 6 N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars stable and accurately established velocities. Firstly, it is Table 4. Data from the USNO-B1.0database for objects well known that the photospheric radial velocity of cool withnotnullpropermotions,where∆tistheepochinter- carbon stars may vary with amplitudes of the order of ∼ val, N is the number of detections on the scanned plates, 10kms−1.Secondly,thenumberofindependentradialve- and Probab. is the total motion probability locities available in the literature for a given star is often small. Thirdly, the literature values are occasionally very Object µα cos δ µδ ∆t N Probab. discrepant: for example, in Table 4 of TI98, it is found # (mas/yr) (mas/yr) (yrs) that APM0102-0556 & APM0911+3341 have published values differing by as much as 50kms−1. 10 +24±3 2±3 45 5 0.9 17 −12±10 6±1 35 5 0.7 Therefore,thepublishedradialvelocitiesofthesestan- 18 −6±3 −4±0 33 3 0.7 dards have been considered as a first approximation. For 20 −6±3 14±5 38 5 0.8 each run considered separately, we have cross-correlated 22 −16±2 0±3 36 4 0.9 the spectra of each standard with all the other standards 24 −8±2 6±4 37 5 0.9 observed during the run, and determined best fit radial 25 −12±2 2±5 41 4 0.9 velocities by minimizing the differences between our data 26 −2±3 8±3 41 5 0.9 and the published values. The results, listed in Table 3, appear fairly consistent, especially when comparing the velocities of the standards common to several runs and taking into account the velocity resolution of our exper- andthey arealsowelldetected insubsequentredornear- iments. The global rms scatter (1σ) of the residuals be- infrared surveys (POSS-II, SERC, AAO, ESO), ensuring tween fitted and published values is 12kms−1. time baselines which are of the order of 30 to 50 years. Then, the spectrum of each program carbon star was correlated with the spectra of the standards observed for 4.5. Variability the same run, and the velocities so obtained were averaged (details on the cross-correlation technique can be All objects were examined for variability in available dig- ′ found in, e.g., TI98). In addition to the internal consis- itized sky surveys with 5 size images retrieved from the tency provided by the standards, a further check is pro- USNO website1.Practically,we comparedby eye the ap- vided by two objects that were observedduring two runs: pearanceofourC starswithneighbouringstarsofsimilar for Object #7, we found v = +339 and +345kms−1 brightness on these survey images and on our ESO CCD helio from Run 1 and 2 respectively; for Object #17, we found Bessel-R band images, when possible. One difficulty in v = +135 and +127kms−1 from Run 2 and 6, re- doing so is that the photographicsurveyplates have been helio spectively.Afinalindependentcheckonourvelocityscale exposedwith a varietyofemulsions andfilters.This com- is provided by Object #29 in Fornax, for which we find plicates the comparison, especially because the C stars v = +40kms−1, in fair agreement with the mean ve- are very red and have relatively steep spectra compared helio locity of this galaxy v = +53kms−1 (van den Bergh to neighbouringfieldstars.For example,the availablered helio 2002),sincethedifferenceof13kms−1 represents1.1σ.In plates for a given field may have the following various conclusion,weestimatethattheuncertaintyonourradial emulsionsandfilters:103aE+RP2444(POSSI),IIIaF+ velocities is ∼ 12kms−1 (1σ). RG 610-3 (POSS II), IIIaF + OG590 (AAO-R) or IIIaF + RG630(ESO-R).Consequently,whenexaminingthese plates,wekeptinmindpossiblebandpasseffectsandcon- 4.4. Proper motions cluded the existence ofvariabilityonly whenthe evidence was very strong. For several objects, there are pairs of ThanksfortherecentreleaseoftheUSNO-B1.0catalogue plateswithidenticalemulsion/filtercombinationswithex- (Monet et al. 2003), information on proper motions is posures taken at quite different dates, in which case vari- available for all the objects. For 22 objects, the proper ability is much better assessed. In such cases, we noted motion is found to be null, while for 8 objects, a mea- ’var 2r’ or ’var 2b’, corresponding to a pair of red or blue surable proper motion is provided. For these 8 objects, plates, respectively. In one case (Object #17), two CCD details are listed in Table 4, including the time interval BesselRimageswereobtainedduring2differentruns,and between first and last plates on which the objects were differentialphotometricanalysisoftheframesveryclearly detected, andthe total motionprobability as providedby establishesvariability,by0.28±0.02magoveraperiodof USNO-B1.0. ∼ 1year. We checked whether the objects for which a zero As can be seen in Table 5, of the 28 FHLC stars in proper motion is given have been correctly observed in our sample, variability is found for 11 objects. Since our the scanned surveys. Because most of the objects are at methodissensibletoonlylargevariations,itismostprob- ◦ a declination larger than ∼ −30 and bright enough (R able that a larger fraction of objects is actually variable. in the range 14 to 18), they have been well imaged in the The Sculptor C star (#29) is also found to be variable. first POSS-I survey: they are generally well detected in red plates and very often also in blue plates of POSS-I, 1 www.nofs.navy.mil/data/fchpix N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars 7 Table 3. Heliocentric radial velocities for template carbon stars. In columns labelled Run 1, Run 2, ..., Run 6 are listedthe bestfitted radialvelocitiesdeterminedforeachrun,v ,inkms−1.The columnv isthe valuetakenfrom fit pub the literature and adoptedas a first approximationfor the fits (see text and Notes). The column gives fit pub the average difference over the runs (in kms−1). Star Run 1 Run2 Run 3 Run4 Run 6 v Noteon v pub fit pub pub TWOph +21 +28 +10 +14 +6 (1) APM1406+0520 −25 −37 −21 −10 (2) HR3541 = XCnc −4 −3 −1 (3) APM0915−0327 +95 +79 +16 (2) APM0123+1233 −324 −302 −22 (2) APM0418+0122 +19 +27 +33 −10 (2) APM2225−1401 −103 −118 −113 +3 (2) APM0222−1337 −7 −24 −27 +11 (2) RVAqr +6 −4 −10 +11 (2) HD16115 −6 +11 +4 −1 (4) APM2213−0017 −49 −44 −5 (2) APM2111+0010 −195 −208 +13 (2) Notes: (1) v is from the SIMBAD database; from the CO millimeter observations listed by Loup et al. (1993), a center of mass pub heliocentric velocity of +15kms−1 is derived and is in very good agreement. (2) v are from TI98; quoted uncertainties are ≤ 7kms−1; pub (3) TI98 indicate v = −1kms−1 with σ = 12kms−1 (2 measurements); from the data in Loup et al. catalogue, one derives pub −6kms−1 (heliocentric), and v = −3kms−1 was adopted; pub (4) we adopted v = +4kms−1 from TI98, who give σ = 1kms−1 (3 measurements), but v = +16±5kms−1 is given in pub SIMBAD. ConcerningtheFornaxCstar(#30),theevidenceforvari- only 4 of our total of 28 are bluer than J −K = 1.45. s ability was not conclusive from the examined plates, but Therefore,whenoneconsidersthephotometricproperties, variability has been proven by the CCD optical imaging the sample of our C stars is globally very different from survey of Bersier & Wood (2002). the sample of known dCs. Concerning the proper motions, we can also compare known dCs and our C star sample. We retrieved the 5. Discussion USNO-B1.0 data for the 31 dCs, and plotted them in Fig. 2. For 2 dCs (PG0824+289Band WIE93 2048-348), 5.1. Are there dwarf carbon stars in our sample? USNO-B1.0 gives null proper motions. For the first ob- One of the major issues is to estimate distances for our ject, we adopted µα = −28.2 ± 1.4masyr−1, µδ = 0 C stars, and the main problem is to know whether they from Heber et al. (1993), and for the second we adopted are the distant evolved AGB stars which were targeted µα =15±24masyr−1,µδ =−3±24masyr−1fromWarren when defining our selection criteria, or if some of them et al. (1993).More importantly, there are 7 dCs which lie are cool dwarf carbon stars with much lower luminosity, outside of the diagram, because their µα or µδ are larger as is suggested at first sight by the surprising fact that 8 than 100masyr−1 in absolute value. We also plotted the have measurable proper motions. USNO-B1.0dataforourCstars,with20lyingatthe(0,0) Therefore,itisfirstusefultocomparethepropertiesof coordinates(null proper motion)and8 at notnull proper our sample to the population of dwarf carbon (dC) stars. motion. It can be seen that all dCs known have a larger Lowrance et al. (2003) gives an exhaustive list of the 31 µ than all our C stars except #10 . If these C stars were dCs presently known and considers their 2MASS JHK in majority dCs,one wouldexpect largerpropermotions, s data (from the all sky release). They found that 20 dCs since they are in majority brighter than the known ones outofatotalof31aredetectedby2MASS,theundetected and would be statistically closer to us. ones being too faint. Considering the K magnitude, one In a complementary way, instead of a statistical ap- s finds that the majority (15 dCs over 20 detected) have proach,onecanexamineindetailthe 8objectswithmea- K > 12, and the brightest one is at K = 10.48. As for surableUSNO-B1.0propermotion.Three(#10,#22and s s the J−K colour,allhaveJ−K <1.5,andonly 4of20 #26)haveJ−K ∼1.5andK in the range10.8to12.3. s s s s have1.31.4) are much too ered by Whitelock et al. (1999). If one outlier star (the s red to be R-type clump giants (see Table 1 of Knapp et faintest in R) is excluded, one finds for the C stars of Sgr al. 2001, and Ivanov & Borissova 2002) M = −3.1 with 25 objects, σ = 0.64 (if all 26 stars are R Finally the two remaining objects # 24 and#25 form consideredMR =−3.0andσ =0.79).Itcanbenotedthat anastonishingpairofclosetwinobjects,withalltheirpa- ourMR =−3.1is0.4maglessluminousthanMR =−3.5 rameters being almost equal: J−K =2.01 & 2.18, K = adopted by Totten and Irwin (1998), and this last value s s 8.92& 8.88,R= 15.1& 15.4,B−R = 2.9& 3.0,v = would have implied distances 20% larger. helio +9 & +9kms−1. In addition, their angular separation in For each halo C star, extinction in the R band was the sky is as low as 2.5◦. These stars are exactly the kind again found from the Schlegel et al. maps, and distances of kinematically and spatially coherent objects which are derived with MR = −3.1 called dR are listed in Table5. tracers of halo streams! With these characteristics, espe- Thesedistancesareadmittedlyverycrude,duetothe low cially the K magnitude and the colours, it is extremely accuracyofUSNOphotometry(∼0.5mag.error),possible s improbable that they are dCs, despite their proper mo- variability,uncertainty inMR,andespecially the effect of tion (µ ≈ 11masyr−1 with a probability of 0.9 and > 3σ circumstellardustforthereddeststarswithJ−Ks larger significance). than ∼ 2.0. For information, Table 5 indicates the result of this distance scale for the Fornax C star (#30), which Theaboveremarks,togetherwithconsiderationofHα is not too red in J − K . No R is available in USNO- emission and/or variability being seen in 6 of the 8 ob- s A2.0, but two R magnitudes are given in USNO B1.0: jects with non zero µ, lead us to have some doubts on with R = 18.2, one finds 180kpc, while for R = 16.9, these µ measurements. These proper motions are quite 2 1 small (< 25 mas yr−1) and supplementary, independant one finds 96kpc. These estimates are within 30% of the truedistance,135kpc.Inconclusion,thedistancesderived measurementsofµareobviouslyneededtoconfirmthem. with R magnitudes, called d and listed in Table5, are We defer to a future work a deeper investigation of the R probably not better than ∼ 30% in relative accuracy. questionwhetherdistant,veryred,possiblyvariablestars A second and probably surer estimate of distances is might have inaccurate µ measurements in USNO-B1.0. obtainable with 2MASS photometry. Considering again Wetentativelyconcludethatoursampledoesnotcon- the26SgrCstarsofWhitelocketal.(1999)whichallhave tain C dwarfs. Our stars are also too red in J −K to be s 2MASS data, we compared their K magnitudes to the clump giants or stars on the first ascending giant branch s averaged K magnitudes of LMC C stars, this averaging like the Hamburg/ESO objects. Therefore, in the follow- s being done over several different bins in J −K . The Sgr ingsections,we shalladoptthe view thatallourstarsare s genuine distant AGB C stars, and examine their location and LMC have similar mean reddening, E(B−V) ∼ 0.15, which can be ignored in this comparison. We find that, with respect to the Sgr stream. for a given J −K , the C stars are fainter in apparent s K magnitudes in Sgr than in LMC by an averageof0.98 s 5.2. Distances mag.(26objects,σ =0.41).Byadopting(m−M) =18.5 0 and 17.0 for LMC and Sgr respectively, it is derived that, Inordertoestimatedistances,wehaveassumedherethat on average, the C stars of Sgr are less luminous by 0.50 our C stars are similar to AGB C stars located in the mag in the K band than the C stars in the LMC. [this s Sgr dwarf galaxy. This working hypothesis can certainly shift increases to 0.67 mag is one adopts, as Majewski et be criticized, but it is suggested by the fact that half of al.(2003),distancemoduliof18.55and16.9forLMCand the high latitude coolC starspreviously knownbelong to Sgr, and is roughly consistent with their Figure 20 where the tidal debris of this dwarf galaxy (Ibata et al. 2001a; candidate SgrC starsarecomparedto candidate LMC C see also Sect.1). Therefore, we adopt the AGB C stars of stars mean locus.] the Sgr galaxy as templates, and analyse their properties We then used for template K -band luminosity the s below to determine absolute magnitudes. averaged K magnitudes of the LMC C stars corrected s A first estimate of distances can be based on R-band by the above 0.50 mag. In Table5 are explicitly listed magnitudes. Whitelock et al. (1999) published a list of the adopted M for each programstar, and the inferred Ks 26 spectroscopically confirmed C stars in Sgr with mem- distances called d . These distances based on JK are K s bership established on the basis of radial velocities. For presumably more accurate than those based on R magni- all of them, we retrieved the 2MASS data together with tudes, because of better photometric quality, smaller am- the USNO-A2.0 B and R magnitudes. These 26 R mag- plitude in K due to variability and reduced sensitivity to nitudes range from 13.0 to 16.7, with an average of 14.4. dust effects. However, for a given J −K , the scatter in s An estimate of the extinction to each star was obtained K forLMC Cstarsisoftheorderof0.4to0.5mag(1σ). s from the Schegel et al. (1998) tables, and the extinctions AlthoughpartofthisdispersionisduetoLMCdepthand in the R band, called A , range from 0.23 to 0.48mag. inclination effects that fairly cancel out when an average R 10 N.Mauron, M. Azzopardi, K. Gigoyan, T.R. Kendall: Halo carbon stars is taken, a natural dispersion ∼ 0.2-0.3 mag is probably presentinthe LMCCstars’K luminosities(Weinberg& s Nikolaev 2001). Taking into account possible variability effects for our stars (maybe ∼ ±0.2 mag. in Ks) and the uncertainty on the C stars luminosity shift between LMC andSgr,onefinds thatthe distancesofourprogramstars derived from JK data are probably not better than ±25 s percent (± 1σ). Looking at Table 5, it can be seen that distances de- rivedfromRandfromnear-infraredareofteninfairagree- ment. There are a number of cases where d is obviously R too large comparedto d because the star is particularly K red and is presumably embedded in dust: for stars #13 #16#18#21#24,theratiod /d islargerthan R K a factor of 2 andtheir J−K colours are1.81,2.56,3.14, s 3.14,2.012&2.174,respectively.Forobject#11,novalue of d is given because R=19.8 would lead to an exceed- R ingly large distance (360kpc): its d = 66kpc is clearly K more plausible. In the case of #29 (in Sculptor) and #30 (in Fornax), the scale adopted above is not necessarily applicable, but remains interesting to consider. It leads to distances dK Fig.2. Proper motions of our C stars (filled octagons) as of 50 kpc and 112 kpc, which are 0.57 and 0.83 times given in the USNO-B1.0 catalogue. Twenty objects are smaller than the generally adopted distances of Sculptor located at the (0,0) coordinates with no motion. Empty and Fornax, ∼ 87 and 135 kpc, respectively. While dK octagonsrepresentthe dCs presently known(USNO-B1.0 for #30 is acceptable, the small value of dK for #29 sug- data), but 7 dCs have too large a proper motion to be gests that our rule for NIR distances based on Sgr tem- located within the limits of this diagram. The circle rep- plates underestimates the K-band luminosity of this star resents a motion of 21 masyr−1, that was the 3σ upper by ∼ 1.2mag. Its membership to Sculptor remains to be limit ofTIW,enclosing 48ofthe 50APMC starsstudied definitively established through a radial velocity determi- by these authors. nationthatourspectrumunfortunatelycouldnotprovide. If it actually is member, its M is −8.11, which is sim- Ks ilar to the most luminous C stars in LMC for the same and its multiple wrapped components is not yet avail- J −K colour: in LMC with (m−M) = 18.5, we find s 0 able. Therefore, we have simply compared here our data < M >= −7.42, σ = 0.53 for J −K = 3.3, and the Ks s with predictions ofthe modelby Ibata et al. (2001a)that star’s M is +1.3σ above the mean. Ks already accounts for several observed aspects of the Sgr Finally, if we exclude the 7 stars discussed above for stream and has a (dark matter) halo density flattening which d /d > 2 and the two Sculptor and Fornax ob- R K parameter q =0.9. m jects, one finds that the log ratio x = log (d /d ) has 10 R K TheFig.1ofIbataetal.(2001a)displaysintwoAitoff a mean of -0.025 and a dispersion of σ = 0.108 (N=21 x projection maps the colour-coded heliocentric velocities objects). Adopting d as a reference,a discrepancy of 2σ K and distances of a Sgr stream simulation. Their distances means that R has a typical “error” by ∼1.1 mag., which ∗ are coded as the SLOAN apparent g magnitudes of A- seems in reasonable agreement with the fact that USNO type stars, for which they adopted Mg∗= 1.0 (these A photometry is poor and R may also be variable with a stars are a mixture of blue horizontal branch and blue comparable amount. In the following, the NIR-based dis- stragglers; see Ibata et al. 2001b for more details). Using tances d will be adopted as the surest estimates, which K this absolute magnitude and the NIR distances of Table we recall are based on the Sgr C stars templates. ∗ 5,wederivedacorrespondingg foreachofourpro- A−type ∗ gramstars.This g is the apparentmagnitude of an A−type 5.3. Location with respect to the Sagittarius Stream imaginaryA-type populationif it werepresentatthe distance of the C star. We then compared for each star α, ∗ With positions, heliocentric radial velocities and distance δ, v and g with those of the model stream in helio A−type estimates in hand, we can examine the likelihood of as- Fig.1ofIbataetal.(2001a).Thiscomparison,andthere- sociation of each star with the Sgr steam. Whereas the foreestablishingthemembershiptothestream,isdifficult pathin the sky is relativelywellknown,especially thanks forseveralobjects,especiallywheni)uncertaintiesondis- ∗ to the recent analysis of Majewski et al. (2003) who em- tance or, equivalently, on g are taken into account; A−type ploy 2MASS M-type giants as tracers, an accurate deter- and/orii)theobjectislocatedinaregionwherethemodel mination of the distances and kinematics of this stream streampresentsarelativelylowdensityofparticles(inthe

Cool carbon stars in the halo: a new survey based on 2MASS

Similar Cool carbon stars in the halo: a new survey based on 2MASS