Membership of 23 stars towards the bulge globular clusters NGC 6528 and NGC 6553

a r X i v :a s t r o -p h /0107219v 1 12 J u l 2001Astronomy &Astrophysics manuscript no.(will be inserted by hand later)Membership of 23stars towards the bulge globular clusters NGC6528and NGC 6553⋆P.Coelho 1,B.Barbuy 1,M.-N.Perrin 2,T.Idiart 1,R.P.Schiavon 3,S.Ortolani 4,and E.Bica 51Universidade de S˜a o Paulo,Dept.de Astronomia,CP 3386,S˜a o Paulo 01060-970,Brazil email:pcoelho@usp.br,barbuy@p.br,thais@p.br 2Observatoire de Paris,61av.de l’Observatoire,75014Paris,France email:Marie-Noel.Perrin@obspm.fr 3Observat´o rio Nacional,rua General Jos´e Cristino 77,20921-400Rio de Janeiro,Brazil and Present address:UCO/Lick Observatory,University of California,Santa Cruz CA 95064,USA email:ripisc@ 4Universit`a di Padova,Dept.di Astronomia,Vicolo dell’Osservatorio 5,I-35122Padova,Italy email:ortolani@pd.astro.it5Universidade Federal do Rio Grande do Sul,Dept.de Astronomia,CP 15051,Porto Alegre 91501-970,Brazil email:bica@if.ufrgs.brReceived 30January 2001;accepted 11June 2001Abstract.Low resolution spectra of 23stars towards the bulge globular clusters NGC 6528and NGC 6553are analysed.Radial velocities and atmospheric parameters are derived in order to check their membership in the clusters.Effective temperatures were obtained from photometric data for stars with T eff>3800K,whereas for cooler stars,they were derived from equivalent widths of TiO bands.Calibrations of W(TiO)as a function of stellar parameters based on a grid of synthetic spectra are presented.Metallicities were derived from a comparison of the observed spectra to a grid of synthetic spectra.The sample comprises evolutionary stages from the Red Giant Branch to the Horizontal Branch,with parameters in the range 3200≤T eff≤5000K and -0.5≤log g ≤2.4.The mean metallicities obtained for NGC 6528and NGC 6553are [Fe/H]≈-0.5and -0.7,in both cases with [Mg/Fe]=+0.3;assuming the same overabundance for the αelements O,Mg,Si,S,Ca and Ti,this gives [Z/Z ⊙]=-0.25and -0.45.Membership verification by means of low resolution spectra is a crucial step in preparing targets for high resolution spectroscopy with 8m class telescopes.Key words.Clusters:globular:NGC 6528,NGC 6553;Stars:abundances1.IntroductionThe study of the stellar populations in the Galactic bulge is very important to constrain possible models of galaxy formation.In particular,the determination of the metal-licities and abundance ratios of bulge stars,either from the field or in clusters,provides key information to help decide among the possible scenarios for the history of chemical enrichment of the Galaxy.There are few previous studies of radial velocities and metallicity estimations of bulge stars from low resolution spectra.In the analysis of 400field bulge stars by Sadler et al.(1996),metallicities and [Mg/Fe]values were esti-mated.Minniti (1995a,b)studied the membership of stars towards 7bulge globular clusters.2P.Coelho et al.:Membership towards NGC6528and NGC65536553.An Iron abundance of[Fe/H]=-0.55±0.2and abun-dance ratios[Na/Fe]≈[Al/Fe]≈[Ti/Fe]≈+0.5,[Mg/Fe]≈[Si/Fe]≈[Ca/Fe]≈+0.3were derived.These ratios imply an overall metallicity[Z/Z⊙]≈-0.1.Cohen et al. (1999),analysing high resolution spectra offive red hor-izontal branch stars,obtained a mean metallicity[Fe/H] =-0.16and an excess of theα-element calcium to iron of about0.3dex,which imply an overall metallicity[Z/Z⊙]≈+0.1.Metal abundances of these clusters are also dis-cussed in Barbuy et al.(1999)and Barbuy(2000).In view of the disagreement between previous deter-minations of[Fe/H],it is important that abundance esti-mations be extended to a larger number of stars of both clusters.In this paper,we determine radial velocities,ef-fective temperatures,gravities and estimations of metal-licities[Fe/H]based on low resolution spectra for23stars towards NGC6553and NGC6528,and verify their mem-bership in these clusters.In Sect.2the observations are described.The ra-dial velocities derived are presented in Sect.3.In Sect. 4the stellar parameters are derived,and synthetic spec-tra are compared to observations to estimate metallicities. Concluding remarks are given in Sect.5.2.ObservationsLow resolution spectra of individual stars of NGC6528 and NGC6553were obtained in1992August and1994 June,at the1.5m ESO telescope at ESO(La Silla).The Boller&Chivens spectrograph was employed.In1992 August the Thompson CCD#18with1024×1024pix-els,with a pixel size of19µm was used.A resolution of ∆λ∼8˚A and a spectral coverage ofλλ4800-8800˚A were achieved.In1994June,the Ford Aerospace FA2048 L,frontside illuminated,uncoated CCD detector(ESO# 24)with2048×2048pixels and pixel size15×15µm was used.The grating#27resulted in a spectral resolution ∆λ∼4˚A and a spectral coverage in the rangeλλ4800-7550˚A.The log of observations is provided in Table1.The stars are identified according to the charts by Hartwick (1975)for NGC6553and van den Bergh&Younger(1979) for NGC6528.Spectra of a given star were co-added by weighting their S/N ratios;thefinal S/N are indicated in Table1.In Figs.1and2are shown the V vs.V−I Colour-Magnitude Diagrams of NGC6528and NGC6553using data obtained with the Hubble Space Telescope(Ortolani et al.1995)where the sample stars are identified.3.Radial velocitiesThe radial velocities were determined by means of three methods,as explained below and reported in Table2.The observed radial velocities derived were transformed to he-liocentric values using the observation dates given in Table 1.Table1.Log of observationsNGC6528I116.10 1.93540006.08.92190I215.73 2.59180016.06.9425540006.08.9275I515.37 2.22270017.06.9420I615.89 3.54270017.06.9440II815.71 2.19210017.06.9420II1415.76 3.47270017.06.9410I2317.19 1.70180016.06.9410I2416.89 1.66450017.06.9420I2516.11 2.09450016,17.06.9440I2715.90 3.08180016.06.9420540006.08.92100I3616.41 1.98210017.06.9420II3915.88 2.30210017.06.9430I4015.93 2.08210017.06.9420I4216.42 2.15210017.06.9430II7015.85 2.36180016.06.9420(a)A Fourier cross-correlation was applied on the pro-gram spectra relative to selected template spectra.As templates,12G,K and M stars were selected from the Jacoby et al.’s(1984)library,which have approximately the same spectral resolution(4.5˚A)of the sample spectra. The templates adopted were the ones closest in spectral type to each of the program stars.The spectra of both sample and template stars were normalized and different regions in the spectra were defined in order to give high-est peaks of cross-correlation for each considered template. The results obtained with this method are given in column 2of Table2.The r.m.s.of the values derived with each template spectrum is of the order of15km s−1.A system-atic effect was identified for the coolest stars,since all of them appeared to show lower velocities when compared to the hotter stars of the same cluster and these values were not considered.(b)Mean shifts between the observed wavelengths of identified absorption lines and laboratory wavelengths were measured(column3of Table2).The r.m.s.of the values derived is of the order of15km s−1.(c)the code HALO(Cayrel et al.1991)derives radial velocities by comparing the observed spectrum to a gridP.Coelho et al.:Membership towards NGC 6528and NGC 65533Fig.1.Top :HST Colour-magnitude diagram of NGC6528,where the observed stars are indicated as filled cir-cles;Bottom :A map of the cluster scanned from van den Bergh &Younger (1979).of synthetic spectra,using a cross-correlation technique.The grid of synthetic spectra available (Barbuy et al.2001)does not contain stars cooler than T eff<4000K,and for this reason the errors should be higher for velocities of stars cooler than T eff<3700K in which TiO bands are pronounced.Histograms of radial velocities of individual stars (coolest stars excluded)corresponding to each method222120191817161514Fig.2.Top :HST Colour-magnitude diagram of NGC 6553,where the observed stars are indicated as filled cir-cles;Bottom :A map of the cluster scanned from Hartwick (1975).were built.Gaussian curves were fitted to each histogram,from which the radial velocity corresponding to each method was derived,as reported in Table 3together with values from the literature.An example of this procedure is presented in Fig.3for the cross-correlation technique using IRAF.The final radial velocities adopted for the clusters correspond to the mean of the values derived from the three methods.4P.Coelho et al.:Membership towards NGC6528and NGC6553Table 2.Heliocentric radial velocities v r(km s−1)ob-tained with different methods:(a)cross-correlation rela-tive to observed template spectra using IRAF;(b)mean shift in wavelength for a list of selected lines;(c)cross-correlation relative to synthetic spectra using the code HALO(see text).NGC6528I1—262174I2224262208I5—261224I6232246225II8289283271II14—264227I23221251—I24238220202I25265257231I27—237188I36244249229II3930177I40246236244I42197235212II70230263—4.Stellar parameters4.1.TemperaturesThe effective temperatures were estimated from B−V, V−I,V−K and J−K colours,based on the colour vs. T effcalibrations by Bessell et al.(1998),which in turn are based on NMARCS models by Plez et al.(1992)and their grid extensions.These effective temperatures are listed in Table5.V and I colours were obtained with the Hubble Space Telescope(Ortolani et al.1995)and J and K colours were obtained with the detector IRAC2at the2.2m tele-scope of ESO(Guarnieri et al.1998).For NGC6528the colour excesses adopted were E(V−I)=0.68and E(B−V)=0.52(Barbuy et al. 1998).For NGC6553E(V−I)=0.95and E(B−V)=0.7 were adopted(Guarnieri et al.1998).The V−K and J−K colours were dereddened assuming E(V-K)/E(B-V) =2.744and E(J−K)/E(B−V)=0.527(Rieke&Lebofsky 1985).An independent method for the derivation of tempera-tures was based on calibrations of equivalent widths of TiO bands.The indices as defined in Table4were measured onFig.3.Histograms of radial velocities obtained for the stars with the cross-correlation technique using IRAF, where the gaussianfits are presented.The deviant point in the histogram of NGC6528is the star II39,which is probably a non-member.Table3.Radial velocities of NGC6528and NGC6553 given in the literature and values derived in the present work,for which the standard deviation values obtained with the gaussianfits are indicated in parentheses.218 6.41,2164.8-24.532084842128.45189—6160-57143-128236(21)7(14)9248(11)1(16)10217(9)-10(13)11P.Coelho et al.:Membership towards NGC 6528and NGC 65535Fig.4.TiO indices measured on the synthetic spectra asa function of effective temperatures.These measurements correspond to spectra convolved with FWHM =8˚AFig.5.Spectra of three individual stars of NGC 6528with different temperatures.It is clear that the T eff=3350K star has more pronounced TiO molecular bands with respect to the T eff=4100K one.f [Fe/H]2+g [Fe/H]log T eff),valid in the range 2500≤T eff≤5000K,-0.5≤log g ≤2.5and -0.5≤Fe/H ≤0was derived.The coefficients of the formula above are shown in Table 6for convolutions of ∆λ(FWHM)=4˚A and 8˚A .Table 4.TiO indices used in the derivation of effective temperatures.The definition of these indices is such as to avoid regions of strong telluric lines.TiO26033.6-6050.66300.0-6455.06525.0-6538.0TiO36525.0-6539.06617.6-6860.07036.0-7046.6TiO47036.0-7046.67053.0-7163.07534.2-7546.8Effective Temperatures (K)StarV −I B −V J −K V −K TiO2TiO3TiO4FinalNGC 6553III 24828—48094559—384237763800III 34015—422139393782380037753800III 173606—396636114052369835033750II 513885—3966385632813206—3250II 52————4241370337523900II 853912—41403835———3950II 943506—398434324802440838323650II 95———————65006P.Coelho et al.:Membership towards NGC6528and NGC6553 Table6.Coefficients of polynomialfits of W(TiO)indicesas a function of stellar parameters for2500≤T eff≤5000K,-0.5≤log g≤2.5and-0.5≤[Fe/H]≤0a(constant)81.4981.8955.7056.51-175.11-158.38b(log(T eff))-39.44-39.71-22.93-23.53105.3395.92c(log g)-0.04-0.04-0.04-0.05-0.06-0.06d([Fe/H])9.289.14 3.40 3.21-7.42-5.10e(log(T eff)2) 4.75 4.79 2.17 2.27-15.66-14.34f([Fe/H]2)-0.47-0.45-0.40-0.38-0.45-0.08g([Fe/H]×log(T eff))-2.56-2.51-0.87-0.82 2.16 1.544.2.GravitiesGravities were derived using the classical relation log g∗= 4.44+4log T∗/T⊙+0.4(M bol-M bol⊙)+log M∗/M⊙, adopting T⊙=5770K,M∗=0.8M⊙and M bol⊙=4.74 cf.Bessell et al.(1998).For deriving M bol∗we used the distance modulus adopting a total extinction A V=2.43 for NGC6553and A V=1.8for NGC6528(Barbuy et al. 1998).The bolometric magnitude corrections were taken from Bessell et al.(1998).The resulting M bol∗and gravities are given in Table7. Taking into consideration the errors due to uncertainties in T effand M bol thefinal error in log g is estimated to be of±0.5dex.4.3.MetallicitiesSpectrum synthesis calculations were used tofit the ob-served spectra.The calculations of synthetic spectra were carried out using the code described in Barbuy et al. (2000)where molecular lines of MgH A2Π-X2Σ,CH A2∆-X2Π,CN A2Π-X2Σ,C2Swan A3Π-X3Πand TiOαC3∆-X3∆,γA3Φ-X3∆andγ’B3Π-X3∆systems are taken into account.For atomic lines the laboratory oscillator strengths by Fuhr et al.(1988),Martin et al.(1988),Wiese et al.(1969), and laboratory values compiled by McWilliam&Rich (1994)were adopted whenever available,otherwise they were taken fromfits to the solar spectrum(see discussion in Barbuy et al.1999).ATLAS9and NMARCS models were employed.A grid of models using the ATLAS9code(Kur´u cz1993)was cre-ated adopting a mixing length parameterα=0.5(see Barbuy et al.2001).NMARCS photospheric models for giants by Plez et al.(1992)and their unpublished ex-tended grids were employed(see more details in Schiavon &Barbuy1999).The metallicities were obtained based on two methods, both using synthetic spectra:(i)The observed spectra were compared to synthetic spectra in the rangeλλ5000–7500˚A.The metallicities Fig.6.Analysis of NGC6528I-1employing the code HALO:Top:observed spectrum(dotted line)and syn-thetic spectrum(solid line)computed with T eff=4400K, log g=1.7,[Fe/H]=-0.4and[Mg/Fe]=+0.30(FWHM =8˚A);Bottom:Residualflux relative to the template synthetic spectrum.were estimated by interpolating between synthetic spec-tra of[Fe/H]=0.0,–0.3,–0.5and–0.6,in all cases assum-ing[Mg/Fe]=+0.3,and the temperatures and gravities determined in Sects.4.1and4.2.(ii)Comparisons with a grid of synthetic spectra in the wavelength regionλλ4600-5600˚A,using the differences method as described in Cayrel et al.(1991)and Barbuy et al.(2001),are carried out.In this method,the observed spectrum is divided by a reference synthetic spectrum. The resulting signal can be expressed as a linear combi-nation of variations in temperature,gravity and metallic-ity.In conjunction with the grid of synthetic spectra,it is possible to establish the differences in T eff,log g and [Fe/H]between the program star and the reference syn-thetic spectrum through a perturbation method.The grid covers the range4000≤T eff≤7000K,0.0≤log g≤5.0, -3.0≤[Fe/H]≤+0.3,and[Mg/Fe]=0.0and+0.4.Fig. 6shows thefit to NGC6528I-1.In Table7are listed the temperatures,gravities,[Fe/H] and[Mg/Fe]obtained with methods(i)and(ii).Note that method(ii)tends to give lower metallicities relative to method(i).This may be due to limitations of the grid of synthetic spectra,which is being extended to cover wider ranges of parameters.The stars NGC6528I-5,NGC6553III-2and II-95 are probable non-members,given that their atmospheric parameters are incompatible with their location in the Colour-Magnitude Diagrams of the clusters.The star II-51appears to be too cool(Table7)with respect to its CMD locus(Fig.2).However,considering that it could be a red variable,this star is tentatively classified as a possible member.Mean metallicities of[Fe/H]=-0.5for NGC6528and [Fe/H]=-0.7for NGC6553were estimated from gaus-P.Coelho et al.:Membership towards NGC6528and NGC65537Table7.Bolometric magnitudes,atmospheric parameters and method employed(see Sect.4.3).M bol values were not reported for probable non-member stars.NGC6528I1-0.84400 1.7-0.40.3i4400 1.7-0.40.3iiI2-2.037000.9-0.50.3iI5–32500.7-0.50.3iI6-3.035500.4-0.60.3iII8-1.54000 1.2-0.50.3i40000.5-0.60.3iiII14-3.132500.2-0.50.3iI230.64800 2.400.3i4800 2.4-0.40.2iiI240.44800 2.300.3i4800 2.3-0.30.1iiI25-1.04100 1.4-0.60.3i4100 1.4-1.10.2iiI27-2.533500.5-0.30.3iI36-0.64200 1.7-0.60.3i4250 1.0-0.80.4iiI40-1.24100 1.4-0.40.3i4100 1.4-0.70.3iiI42-0.84050 1.5-0.40.3i4050 1.5-1.20.2iiII70-1.63900 1.1-0.60.3i sianfits to the histograms of metallicities given in Table7 (typical standard deviations are of0.2dex).These metal-licities,together with the magnesium excess of[Mg/Fe]≈+0.3,and assuming that all otherαelements show an excess of+0.3dex relative to iron,result in overall metal-licities of[Z/Z⊙]=-0.25and-0.45for NGC6528and NGC6553respectively.5.ConclusionsThe study of individual stars in globular clusters along their evolutionary stages is of prime importance for an improved understanding of stellar evolution.Low reso-lution spectroscopy provides a means for the study of a large number of stars.In the present work we have mea-sured radial velocities and estimated metallicities in23 stars towards the globular clusters NGC6528and NGC 6553,which allows us to identify member stars.We also obtained their atmospheric properties to afirst approx-imation.This is an important step before applying ef-forts to obtain high resolution spectroscopy with8m class telescopes.The method presented here is also of interest for last generation multi-object instruments such as VLT-VIMOS.The stars were analysed by comparisons between their observed spectra and a grid of synthetic spectra.TiO equivalent widths were used to estimate effective temper-atures of stars cooler than T eff≤3800K and a calibration of equivalent widths of TiO bands as a function of atmo-spheric parameters is presented.Mean values of heliocentric radial velocities of v r=234 km s−1for NGC6528and v r=-1km s−1for NGC6553 are derived.Regarding membership,among the23stars observed we concluded that4of them are probable non-members. These are:NGC6528II-39,non-member due to a de-viant radial velocity,and NGC6528I-5,NGC6553III-2 and II-95,non-members due to incompatibilities of atmo-spheric parameters vs.location in the Colour-Magnitude Diagrams.NGC6553II-51could be a non-member,or a red vari-able for which the spectrum was taken during a cool phase.The basic stellar parameters derived show the inter-esting result that there is a trend for member giants of NGC6528to be more metal-poor than the two Horizontal Branch stars NGC6528I-23and I-24,thus reproducing the discrepancy found between analysis of NGC6553gi-ants by Barbuy et al.(1999)and Horizontal Branch stars by Cohen et al.(1999).Given the errors involved in the analysis of low resolution spectra,these results have to be checked with high resolution spectra,and further studies of this discrepancy will be possible only with a homoge-neous analysis of stars ranging from the red giant branch to the HB,and also employing different sets of model at-mospheres all along the evolutionary sequence.In summary,we obtained for NGC6528and NGC6553 metallicities of[Fe/H]=-0.5±0.3and[Fe/H]=-0.7±ing[Mg/Fe]≈+0.3,and assuming that otherαelements show the same excess of+0.3dex relative to iron,the results are[Z/Z⊙]=-0.25and-0.45for NGC 6528and NGC6553respectively.Acknowledgements.We are grateful to one for having carried out part of the observations.We acknowledge par-tialfinancial support from CNPq and Fapesp.P.Coelho and T.Idiart acknowledge respectively the Fapesp Master fellow-ship n o98/07492-4,and Post-Doc fellowship n o97/13083-7.RPS acknowledges support provided by the National Science Foundation through grant GF-1002-99and from the Association of Universities for Research in Astronomy,Inc., under NSF cooperative agreement AST96-1361. ReferencesArmandroffT.E.,Zinn R.,1988,AJ96,92Barbuy B.,Bica E.,Ortolani S.,1998,A&A333,1178P.Coelho et al.:Membership towards NGC6528and NGC6553 Barbuy B.,2000,in The Chemical Evolution of the Milky Way:Stars vs.Clusters,Eds.F.Matteucci,Kluwer Acad.Pub.,in pressBarbuy B.,Renzini A.,Ortolani S.,Bica E.,Guarnieri M.D.,1999,A&A341,539Barbuy B.,Perrin M.-N.,Katz D.,Cayrel R.,Spite M.,van’tVeer-Menneret C.,2001,A&A,submittedBessell M.S.,Castelli F.,Plez B.,1998,A&A337,321Carretta E.,Gratton R.G.,1997,A&AS121,95Cohen J.G.,Gratton R.G.,Behr B.B.,Carretta E.,1999,ApJ523,739Fuhr J.R.,Martin G.A.,Wiese W.L.,1988,Atomic TransitionProbabilities:Iron through Nickel,Journal of Physical andChemical Reference Data,vol.17,no.4Guarnieri M.D.,Ortolani S.,Montegriffo P.,Renzini A.,Barbuy B.,Bica E.,Moneti A.,1998,A&A331,70Cayrel R.,Perrin M.-N.,Barbuy B.,Buser R.,1991,A&A247,108Harris W.E.,1996,AJ112,1487Hartwick F.D.A.,1975,PASP87,77Jacoby G.H.,Hunter D.A.,Christian C.A.,1984,ApJ419,592Kur´u cz,R.,1993,CD-ROM18Martin G.A.,Fuhr J.R.,Wiese W.L.,1988,Atomic TransitionProbabilities:Scandium through Manganese,Journal ofPhysical and Chemical Reference Data,vol.17,no.3McWilliam A.,Rich R.M.,1994,ApJS91,749(MR)Minniti D.,1995a,A&A303,468Minniti D.,1995b,A&AS113,299Schiavon R.P.,Barbuy B.,1999,ApJ510,934Ortolani S.,Renzini A.,Gilmozzi R.,Marconi G.,Barbuy B.,Bica E.,Rich R.M.,1995,Nature377,701Plez B.,Brett J.M.,Nordlund˚A,1992,A&A256,551Rieke G.H.,Lebofsky M.J.,1985,ApJ288,618Rutledge G.A.,Hesser J.E.,Stetson P.B.,Mateo M.,SimardL.,Bolte M.,Friel E.D.,Copin Y.,1997,PASP109,883Sadler E.,Rich R.M.,Terndrup D.M.,1996,AJ112,171van den Bergh S.,Younger F.,1979,AJ84,1305Wiese W.L.,Martin G.A.,Fuhr J.R.,1969,Atomic TransitionProbabilities:Sodium through Calcium,NSRDS-NBS22Zinn R.,1985,ApJ293,424Zinn R.,West,M.J.,1984,ApJS55,45。