Mem.S.A.It.Vol.0,0 (cid:13)c SAIt 2008 Memoriedella Mass segregation in the diffuse outer-halo globular cluster Palomar 14 3 MatthiasJ.Frank1,2,EvaK.Grebel1,andAndreasH.W.Ku¨pper3 1 0 1 Astronomisches Rechen-Institut, Zentrum fu¨r Astronomie der Universita¨t Heidelberg, 2 Mo¨nchhofstrasse12-14,D-69120Heidelberg,Germany n 2 Landessternwarte,Zentrumfu¨rAstronomiederUniversit¨atHeidelberg,Ko¨nigsstuhl12, a D-69117Heidelberg,Germany;e-mail:[email protected] J 3 Argelander-Institutfu¨rAstronomie,AufdemHu¨gel71,D-53121Bonn,Germany 1 2 Abstract.Wepresentananalysisoftheradialdependenceofthestellarmassfunctionin ] A thediffuseouter-haloglobularclusterPalomar14.Usingarchival HST/WFPC2dataofthe cluster’scentral39pc(correspondingto∼ 0.85×r )wefindthatthemassfunctioninthe G h massrange0.55≤m/M ≤0.85iswellapproximatedbyapower-lawatallradii.Themass ⊙ . function steepens withincreasing radius, froma shallow power-law slope of 0.66±0.32 h inthecluster’s centreto aslope of 1.61±0.33 beyond the core radius, showing that the p clusterismass-segregated.Thisisseeminglyinconflictwithitslongpresent-dayhalf-mass - o relaxation time of ∼ 20Gyr, and with the recent finding by Beccari et al. (2011), who r interpretthecluster’snon-concentrated populationofbluestragglerstarsasevidencethat st dynamicalsegregationhasnotaffectedtheclusteryet.Wediscussthisapparentconflictand a arguethattheclustermusthaveeitherformedwithprimordialmasssegregation,orthatits [ relaxationtimescalemusthavebeenmuchsmallerinthepast,i.e.thattheclustermusthave undergoneasignificantexpansion. 1 v Key words. Galaxy: globular clusters – Globular clusters: individual: Palomar 14 – 0 Galaxies:stellardynamics–Stars:formation–Galaxy:halo 1 0 5 . 1. Introduction masssegregation,i.e.moremassivestarsshow 1 a more concentrated radial distribution than 0 Almost all Galactic globular clusters (GC) 3 lower-mass stars, and the cluster appears de- have present-day half-mass relaxation times 1 pletedinlow-massstars. shorterthantheirages(e.g.Harris1996,2010 v: edition). In these clusters, two-body relax- Oneofthefewexceptions,withapresent- i ation has already altered the distribution of day half-mass relaxation time exceeding the X Hubbletime,istheouter-haloGCPalomar14 stars: massive stars, losing kinetic energy to r (Pal14). According to Sollima et al. (2011), a lower-mass stars sink into the cluster’s cen- Pal14hasaprojectedhalf-lightradiusofr = tre, whereas low-mass stars gain energy al- h 46pc, making it the most extended Galactic lowing them to populate orbits further away GC in the Milky Way. Its low mass and large from the cluster’s centre. This is observed as radius implies a half-mass relaxation time of Sendoffprintrequeststo:M.Frank ∼20Gyr.Therefore,nomasssegregationisin- Frank,Grebel&Ku¨pper:MasssegregationinPalomar14 1 tuitivelyexpectedinPal14.Inagreementwith this expectation, Beccari et al. (2011) found 18 that the cluster’s population of blue straggler stars(BSS)isnotcentrallyconcentratedcom- paredtoredgiant(RGB)andhorizontalbranch 20 (HB) stars. BSS in such a diffuseclusterhave most likely formed via mass-transfer in pri- mordial binary systems that have larger to- tal masses than individual RGB or HB stars. g]22 a m Hence, these systems would segregate most [ W quicklyinaGC,suchthatBSSareexpectedto 5 5 5 tracethissegregationprocess.Beccarietal.in- F 24 terprettheirfindingsasevidencethattwo-body relaxationhasnotaffectedPal14yet. Ontheotherhand,Jordietal.(2009)found that the mass function of main sequence stars 26 inPal14isdescribedbyapower-lawdN/dm∝ m−α with a slope of α = 1.3 ± 0.4, i.e. the cluster is significantly depleted in low-mass starscomparedtoaKroupa(2001)initialmass 0.4 0.6 0.8 1.0 1.2 1.4 1.6 function (IMF; α = 2.3 in this mass range). F555W-F814W[mag] This wouldreadily be understood,if the clus- Fig.1. Observed CMD of Pal14 obtained from terweremass-segregated,oralternatively,ifit archivalWFPC2datausingtheHSTPHOTphotom- formed with a IMF already depleted in low- etry package (Dolphin 2000). Artificial star tests massstars(Zonoozietal.2011). wereusedtoestimatephotometricuncertainties(er- In this contribution we present evidence rorbarsontheright)andcompletenesslimits(light that Pal14 is mass-segregated based on an anddarkgreylinesatthefaintandbrightendscor- analysis of radial dependence of the cluster’s respondto80%and50%completenesscontours,re- stellar mass function (Section 2) and discuss spectively).Theisochrone(thickgreycurve)corre- two possible scenarios that can reconcile our spondstoanageof11.5Gyr,[Fe/H]=−1.5dexand results with the cluster’s large relaxation time [α/Fe]=+0.2dexassumingadistanceof71±1.3kpc andreddeningof E(F555W−F814W) = 0.06mag scaleanditsnon-segregatedpopulationofBSS (Jordi et al. 2009). To calculate the mass function (Section3). we used stars within the colour limits represented bythingreycurvesonbothsidesoftheisochrone. 2. DataandAnalysis We used deep V and I band archival observedmagnitudesofthesestarsinorderto HST/Wide-Field Planetary Camera 2 infertheirmasses. We correctedforthe radial (WFPC2) imaging of Pal14 (program variation of the photometric completeness, as GO 6512, PI: Hesser) to obtain the colour- well as the inhomogeneous coverage of the magnitude diagram (CMD) shown in Fig. 1. clusterbytheWFPC2pointing,andcalculated The overlaid isochrone is taken from the the maximum likelihood power-law represen- Dotter et al. (2008) library and corresponds tation of the mass function in different radial to an age of 11.5 Gyr, [Fe/H]= −1.5dex and ranges. The cluster’s observed mass function [α/Fe]= +0.2dex. To derive the stellar mass inradialbinscontainingeachonefourthofthe function, we followed the basic procedure observed stars is shown in Fig. 2. The dotted described in Frank et al. (2012): we selected curves correspond to raw star counts in ten stars within the colour limits shown as thin evenly spaced mass bins from 0.54 to 0.82 grey lines in the CMD and interpolated the M ,correspondingtothe∼70%completeness ⊙ masses tabulated in the isochrone file to the limit at the faint end to the tip of the RGB. 2 Frank,Grebel&Ku¨pper:MasssegregationinPalomar14 Dashed curves represent the star counts after 0.00<r≤0.44arcmin s correctionforgeometriccoverage,solidcurves ar256 st after additionally correcting for photometric of128 completeness. Thick grey lines show the er b m 64 best-fitting power-law. The mass function u is well described by a single power law at N 32 α=0.66±0.32 0.44<r≤0.68arcmin all radii and a trend of an increasing power s ar256 law slope with increasing radius is apparent. st This trend is seen more clearly in the finer of128 er radial subdivision of Fig. 3, which shows the mb 64 best-fittingmassfunctionslopeαasafunction Nu α=0.47±0.32 32 of radius. The mass function steepens with 0.68<r≤0.99arcmin s increasing radius, ranging from α < 1 within ar256 st the cluster’s core radius to a slope almost of128 compatible with the Kroupa α = 2.3 in the er b outermostradialbinatr =1.6arcminor33pc. um 64 Aconstantmassfunctionslopeasafunctionof N 32 α=0.86±0.33 0.99<r≤1.89arcmin radiusisexcludedatthe98%confidencelevel. s ar256 The lack of low-mass stars in the cluster’s st centre compared to larger radii clearly shows of128 er thattheclusterismass-segregated. mb 64 u N α=1.61±0.33 32 0.52 0.56 0.60 0.64 0.68 0.72 0.76 0.800.84 3. Discussion m[M⊙] Findingmass-segregationinPal14isinappar- Fig.2. The mass function in four radial ranges in ent conflict with its present-day half-mass re- order of increasing distance from the cluster cen- laxation time of ∼ 20Gyr. If the cluster had tre. Dotted lines correspond to the raw observed a similar structure and therefore a similar re- star counts, dashed lines to the star counts after laxationtime-scalethroughoutitslifetime,the correction for geometric coverage and solid lines observed mass segregation would have to be to thestar counts additionally corrected forphoto- primordial,aswassuggestedbyZonoozietal. metric incompleteness. The best-fitting power-law (2011). In this case, it is likely that the clus- massfunctionsareshownasthickgreylinesandthe terspentmostofitslifetimeinthelow-density best-fittingslopesαarereportedthebottomofeach environmentofanonlyrecentlyaccreteddwarf panel.Inallradialrangesthemassfunctioniswell describedbyapower-law. galaxy(cf.Sollimaetal.2011;C¸alıs¸kanetal. 2012), or otherwise it is puzzling how such a diffuse cluster can have survived in the tidal fieldoftheGalaxy.Thattheclusterisaffected bytheGalactictidalfield,evenatitscurrentre- orbit about the Galaxy, similar to the expan- mote location (at a Galactocentric distance of sion of Palomar5 due to disk shocks that has 66kpc),isevidencedbyitstidaltails(Jordi& beensuggestedby(Dehnenetal.2004).Given Grebel2010;Sollimaetal.2011). its present Galactocentric distance of 66kpc, Alternatively, the cluster may have been Pal14 would have to be on a highly eccentric significantly morecompact(by a factorof ∼2 orbitin orderto comesufficientlyclose to the in the projected half-light radius rh) in the Galactic centre to be affected by tidal shocks. past, implying a previously much shorter re- This scenario could not only explain the ob- laxation time scale trh of a few Gyrs (trh ∝ servedmasssegregation,butalsothe cluster’s r3/2;Spitzer&Hart1971).Suchanexpansion large physicalsize, whose light profile in this h could have been caused by tidal shocks dur- case maybesignificantlyinflatedbyunbound ing pericenter passages of the cluster on its stars(Ku¨pperetal.2010). Frank,Grebel&Ku¨pper:MasssegregationinPalomar14 3 2.5 havebeenexpelledfromtheclustercentre,re- rc Kroupa sulting in a more extended radial distribution 2.0 ofBSSinthecluster. α e p slo1.5 Acknowledgements. This work was partially sup- on ported by Sonderforschungsbereich 881, “The ncti1.0 Milky Way System” (subprojects A2 and A3) of u f the German Research Foundation (DFG) at the s as0.5 University of Heidelberg. M.J.F and A.H.W.K. m kindly acknowledge support from the DFG via 0.0 Emmy Noether Grant Ko 4161/1 and project KR 1635/28-1,respectively. 0.0 0.5 1.0 1.5 radius[arcmin] References Fig.3. Thebest-fittingmassfunctionslopeandits uncertainties as a function of radius. The core ra- Beccari, G., Sollima, A., Ferraro, F. R., et al. diusr oftheclusterisindicatedbythedottedline. c 2011,ApJ,737,L3 Atrendofincreasingαwithincreasingradiusisob- Bonnell, I. A., Bate, M. R., Clarke, C. J., & viousandissignificantatthe98%level. Pringle,J.E.2001,MNRAS,323,785 C¸alıs¸kan, S¸., Christlieb, N., & Grebel, E. K. 2012,A&A,537,A83 Regarding the non-segregated population Dehnen, W., Odenkirchen, M., Grebel, E. K., of blue stragglers compared to HB and RGB &Rix,H.-W.2004,AJ,127,2753 stars,ourdataconfirmthefindingsof(Beccari Dolphin,A.E.2000,PASP,112,1383 et al. 2011): the radial distribution of BSS is Dotter,A.,Chaboyer,B.,Jevremovic´,D.,etal. not statistically different from that of HB or 2008,ApJS,178,89 RGB stars (Frank et al. , in prep.). While the Frank, M. J., Grebel, E. K., & Ku¨pper, small numberstatistics (∼ 25 blue stragglers) A.H.W.,inprep. advise caution in the interpretation, both of Frank,M.J.,Hilker,M.,Baumgardt,H.,etal. the evolutionaryscenariosfor Pal14 sketched 2012,MNRAS,423,2917 above can potentially be reconciled with a Harris,W.E.1996,AJ,112,1487 non-segregatedpopulationofBSS.Aplausible Jordi, K. & Grebel, E. K. 2010, A&A, 522, mechanism leading to primordial mass segre- A71 gation is the ‘competitive accretion’ scenario Jordi,K.,Grebel,E.K.,Hilker,M.,etal.2009, (Bonnellet al. 2001),in which protostarsthat AJ,137,4586 reside in the cluster’s centre, where the den- Kroupa,P.2001,MNRAS,322,231 sityofgasishigher,canaccretegasmoreeffi- Ku¨pper,A.H.W.,Kroupa,P.,Baumgardt,H., ciently, and consequently tend to have higher &Heggie,D.C.2010,MNRAS,407,2241 masses. In this picture, it seems conceivable Sollima, A., Mart´ınez-Delgado, D., Valls- thatstarsaregenerallysegregatedbymass,but Gabaud, D., & Pen˜arrubia, J. 2011, ApJ, thatthedistributionofbinariessuchastheBSS 726,47 progenitors is not necessarily more centrally Spitzer,Jr., L.& Hart, M. H. 1971,ApJ, 164, concentrated. If on the other hand the clus- 399 ter was once significantly more compact, it is Zonoozi,A.H.,Ku¨pper,A.H.W.,Baumgardt, possiblethatnotallBSSoriginatefrommass- H.,etal.2011,MNRAS,411,1989 transferinprimordialbinaries,butthata frac- tionofthemformedincollisionsinthe–then denser – cluster centre. If three or more stars were involved in these close encounters (e.g. two binary systems) the resulting BSS would have receivedinitial velocitykicksand would