Draftversion January14,2016 PreprinttypesetusingLATEXstyleemulateapjv.5/2/11 TWO MASSIVE WHITE DWARFS FROM NGC 2323 AND THE INITIAL-FINAL MASS RELATION FOR PROGENITORS OF 4 TO 6.5 M⊙1 Jeffrey D. Cummings2, Jason S. Kalirai3,2, P.-E. Tremblay4, AND Enrico Ramirez-Ruiz5 Draft version January 14, 2016 ABSTRACT 6 We have observeda sample of 10 white dwarf candidates in the rich open cluster NGC 2323 (M50) 1 0 with the Keck Low-ResolutionImaging Spectrometer. The spectroscopy shows eight to be DA white 2 dwarfs, with six of these having high S/N appropriate for our analysis. Two of these white dwarfs are consistent with singly evolved cluster membership, and both are high mass ∼1.07 M⊙, and give n equivalent progenitor masses of 4.69 M⊙. To supplement these new high-mass white dwarfs and a analyze the initial-final mass relation (IFMR), we have also looked at 30 white dwarfs from publicly J available data that are mostly all high-mass (&0.9 M⊙). These original published data exhibited 2 significant scatter, and to test if this scatter is true or simply the result of systematics, we have 1 uniformly analyzed the white dwarf spectra and have adopted thorough photometric techniques to deriveuniformclusterparametersfortheirparentclusters. TheresultingIFMRscatterissignificantly ] R reduced, arguing that mass-loss rates are not stochastic in nature and that within the ranges of metallicityandmassanalyzedinthisworkmasslossisnothighlysensitivetovariationsinmetallicity. S Lastly, when adopting cluster ages basedon Y2 isochrones,the slope of the high-massIFMR remains . h steepandconsistentwiththatfoundfromintermediate-masswhitedwarfs,givingalinearIFMRfrom p progenitor masses between 3 to 6.5 M⊙. In contrast,when adopting the slightly younger cluster ages - based on PARSEC isochrones, the high-mass IFMR has a moderate turnover near an initial mass of o r 4 M⊙. t s a [ 1. INTRODUCTION quentworkontheIFMRbyanumberofgroups(seeWei- demann2000forreview)resultedinabroadbutsparsely Performing stellar archeology on white dwarfs, by far 1 populatedrelationthatshowedacleartrendwithhigher- the most common stellar remnant, provides valuable in- v mass main sequence stars producing increasingly more formation for not only understanding stellar evolution 3 massivewhitedwarfs. Inthepast15yearstheamountof and mass loss but also galactic evolution. One of the 5 IFMRdatahasgreatlyincreased(e.g.,Claveretal.2001; fundamental relations in the analysis of white dwarfs is 0 Dobbie et al. 2004, 2006a; Williams et al. 2004; Kalirai the initial-final mass relation (hereafter IFMR), where 3 et al. 2005; Liebert et al. 2005; Williams & Bolte 2007; 0 the masses of white dwarfs are compared directly to the Kaliraietal.2007;Kaliraietal.2008;Rubin etal.2008; . zero-age main sequence mass of their progenitors. This 1 Kaliraietal.2009;Williamsetal.2009;Dobbie&Baxter semi-empirical relation is critical to our understanding 0 2010; Dobbie et al. 2012). These newer data retain the of integrated mass-loss over the lifetime of a star and 6 generaltrendofthepreviousIFMRwork,butthescatter how it changes with stellar mass. The IFMR has a vari- 1 inthedataremainssignificant. Thesourceofthisscatter ety of additional applications including predicting Type : maybe attributable to severalfactors including the pos- v Ia supernovae rates (Pritchet et al. 2008; Greggio 2010) siblestochasticnatureofmassloss,effectsfromvariation i and overall stellar feedback in galaxy models (Agertz & X in metallicity or environment, or systematic differences Kravtsov2014), interpreting the white dwarf luminosity r function(Catala´netal.2008),andprovidingatechnique between the studies. One important systematic is the a challenge in defining the ages of the clusters these white formeasuringtheageoftheGalactichalo(Kalirai2013). dwarfs belong to, which creates uncertainty in the de- Analysis of the IFMR began with Weidemann (1977), rivedlifetimes oftheirprogenitorstars. Cummingsetal. where it was shown that the models of the time greatly (2015;hereafterPaperI)begantoanalyzetheimportant underestimated the observed stellar mass loss. Subse- intermediate-mass IFMR (progenitor masses of 3-4 M⊙) from the rich NGC 2099 (M37). This work strength- 1Based on observations with the W.M. Keck Observatory, ened the observational evidence that in this mass range which is operated as a scientific partnership among the Cali- forniaInstitute ofTechnology, theUniversityofCalifornia,and the IFMR is steep, where the final white dwarf mass NASA, was made possible by the generous financial support of increases more rapidly with increasing progenitor mass. theW.M.KeckFoundation. Comparison to the rich population of comparable mass 2Center for Astrophysical Sciences, Johns Hopkins Univer- whitedwarfsinboththeHyadesandPraesepefromKali- sity, 3400 N. Charles Street, Baltimore, MD 21218, USA; [email protected] raietal.(2014)showedstronglyconsistentIFMRs. This 3Space Telescope Science Institute, 3700 San Martin Drive, consistencyalsosuggeststhatacrossthismassrangethe Baltimore,MD21218, USA;[email protected] slightly metal-richprogenitorstarsfrom the Hyades and 4Department of Physics, University of Warwick, Coventry Praesepe ([Fe/H]∼0.15) have no significant increase in CV47AL,UK;[email protected] 5Department of Astronomy and Astrophysics, University of mass-loss rates compared to those in the solar metallic- California,SantaCruz,CA95064;[email protected] ity NGC 2099. 2 Cummings et al. Expanding beyond Paper I, we now look at the chal- standards were observed each night. lenging higher-mass region (Minitial of 4–6.5 M⊙) of the Three independent LRIS masks were used in 2008 to IFMR by focusing on white dwarfs in younger clusters. observe white dwarfs candidates from NGC 2323. Mask While younger clusters do not provide a broad mass 1 was observed for 40 minutes, Mask 2 was observed for range of white dwarfs, they provide several important 40minutes, andMask3wasobservedfor2hoursand40 advantages. First, the highest-mass white dwarfs are minutes. In 2011, individual longslit observations were the most compact and lowest luminosity, and because performed for five different white dwarf candidates in they form first they remain bright only in the youngest NGC 2323 ranging from 20 minutes to 1 hour and 50 clusters (< 200 Myr). Second, the cooling rates are far minutes. Lastly, in 2015 an additional 70 minutes was more rapid in young and hot white dwarfs and, as a re- acquired on WD11, which is a white dwarf candidate sult, errors in temperature lead to far smaller errors in of interest that only had 20 minutes of observation in bothcoolingageandluminosity. Third,high-masswhite 2011. We have reduced and flux calibrated our LRIS dwarfs(&0.9M⊙)maybe pronetobe ejectedfromtheir observations using the IDL based XIDL pipeline6. Of parent population clusters, either due to dynamical in- ourtotalobservedsampleof10white dwarfscandidates, teractions or potential velocity kicks due to asymmetric eight are DA white dwarfs and WD23 and WD38 have mass loss during their formation (Fellhauer et al. 2003; no clear spectral features. Tremblayetal.2012). Therefore,theprobabilityoffind- Toprovideadditionalhigh-masswhitedwarfsforcom- inghigh-masswhitedwarfsstillwithintheirclusterpop- parison we have taken from the VLT Archive the ob- ulation may decrease with age. These three reasons are servations of seven white dwarf members of NGC 3532, whyyoungerclustersprovidefarmoreadvantagesinan- three white dwarf members of NGC 2287, and four alyzing high-mass white dwarfs with the best precision. white dwarf members of NGC 2516 (Based on obser- In this paper we begin our analysis with the rich, vations made with ESO telescopes under Program IDs: young, and nearby cluster NGC 2323 (M50). Based on 079.D-0490(A);080.D-0654(A);084.D-1097(A);PI:Dob- population analysis, Kalirai et al. (2003) find that it is bie). These observations were performed with FORS1 approximately three times as rich as the Pleiades, mak- and FORS2 using the 600B grism (Appenzeller et al. ing it an excellent environment to search for the rare 1998)giving comparablespectralresolutionto our LRIS remnants of higher-mass stars. Additionally, to expand observations of ∼6 ˚A. Analysis of the parameters and our sample we self-consistently reanalyze publicly avail- membership for the white dwarfs of NGC 2287, NGC able data on all published high-mass white dwarfs from 2516,andNGC3532wereoriginallypublishedinDobbie clusters (Pleiades [M45], NGC 2516, NGC 2287, NGC et al. (2009; 2012). Additionally, from the Keck Archive 3532, and NGC 2168 [M35]) and Sirius B. To further we have taken the observations of 11 white dwarf mem- limit systematics, we also self-consistently analyze the bers of NGC 6128 (Program ID: U49L-2002B; U60L- cluster parameters for all of the parent cluster popula- 2004A; U15L-2004B; U18L-2005B; PI: Bolte). Similar tions based on published high-quality UBV photometry. to our NGC 2323 observations, these observations were To look at the broader picture, we connect these high- performedusingLRISwithamajorityofthemusingthe mass data to the moderate-mass data from Paper I and 400/3400 grism and 1” slits, giving the same character- further analyze the broadercharacteristicsof the IFMR. istics to our data. Analysis of the parameters and mem- The structure of this paper is as follows, in Section 2 bership for the white dwarfs of NGC 6128 were original we discuss the spectroscopic white dwarfobservationsof published in Williams et al. (2009). NGC2323,thepubliclyavailabledatawehaveused,and We have performed our own reductions and analyses our reduction and analysis techniques. In Section 3 we of these data from both the VLT and Keck, but we do discuss the UBV photometry based cluster parameters not redetermine their membership status in this paper. for our six open clusters being analyzed. In Section 4 The VLT data were reduced using the standard IRAF we discuss the cluster membership of our white dwarf techniques for reduction of longslit data, while the Keck candidates in NGC 2323. In Section 5 we discuss the datawerereducedusingthesameXIDLpipeline usedto high-mass IFMR and compare to the intermediate-mass analyzeourNGC 2323data. As a test forour VLT data (3–4 M⊙ progenitors) IFMR from Paper I. In Section 6 reduction,wewereprovidedwiththepublishedspectrum we summarize our study. of white dwarf J0646-203 (NGC 2287-4) (P.D. Dobbie; 2. OBSERVATIONS,REDUCTIONS&ANALYSIS private communication 2014), and we found that there are no meaningful differences in our reduced spectra of Based on the deep BV photometric observations of the same data. Hence, there are no systematics caused NGC 2323(Kaliraietal.2003)withthe Canada-France- byourspectralreductiontechniques,andthesystematic HawaiitelescopeandtheCFH12Kmosaiccamera,asam- differences between our parameters and those presented pleofwhitedwarfcandidatesinNGC2323werespectro- in Dobbie et al. (2012) are due to differences in our ap- scopically observed at Keck I using the Low Resolution plied white dwarf models and fitting techniques. Imaging Spectrometer (LRIS; Oke et al. 1995). In total, For the high-mass LB 1497, from the Pleiades, and 10 of these candidates had sufficient signal to properly Sirius B we have taken the T and log g parameters eff analyze their characteristics. The 400/3400 grism was from Gianninas et al. (2011). We also have taken these used with 1” slits giving a spectral resolution of ∼6 ˚A, parameters from Gianninas et al. (2011) for the super- which provides us the wavelength coverage of ∼3000 to massive GD50 and PG 0136+251 white dwarfs, where 5750 ˚A and a series of 5 Balmer lines (Hβ, Hγ, Hδ, Hǫ, Dobbieetal.(2006b)usedthree-dimensionalspaceveloc- and H8). These observations were performed on 2008 ities to argue that GD50’s progenitorwas relatedto and December 23 and 24, on 2011 December 27 and 28, and on 2015 February 19. For spectral flux calibration, flux 6 Availableathttp://www.ucolick.org/∼xavier/IDL/ White Dwarfs in NGC 2323 3 TABLE 1 - NGC 2323 White Dwarf Initial and Final Parameters ID Teff logg MWD MV tcool Y2 Mi PARSECMi Mi120 Mi160 S/N (K) (M⊙) (Myr) (M⊙) (M⊙) (M⊙) (M⊙) LikelyWhiteDwarfClusterMembers NGC2323-WD10 52800±1350 8.68±0.09 1.068±0.045 10.36±0.19 1.6+−10..26 4.69+−00..0011 5.07+−00..0022 4.98 4.45 85 NGC2323-WD11 54100±1000 8.69±0.07 1.075±0.032 10.36±0.13 1.3+−00..64 4.69+−00..0011 5.07+−00..0021 4.98 4.45 130 WhiteDwarfsInconsistentwithSingleStarMembership NGC2323-WD21 18200±850 8.26±0.15 0.779±0.096 11.33±0.25 170+60 – – – – 28 −48 NGC2323-WD7 16800±250 7.90±0.05 0.559±0.024 10.92±0.07 112+12 – – – – 122 −11 NGC2323-WD17 19800±300 8.12±0.05 0.694±0.028 10.97±0.07 96+12 – – – – 111 −11 NGC2323-WD12 17100±400 7.88±0.07 0.550±0.037 10.87±0.11 101+17 – – – – 60 −15 LowSignaltoNoiseWhiteDwarfs NGC2323-WD22 24400±1550 8.08±0.22 0.681±0.128 10.53±0.36 33+36 – – – – 13 −16 NGC2323-WD30 13400±900 8.04±0.18 0.629±0.106 11.52±0.28 290+108 – – – – 23 −82 coevalwith the Pleiades cluster. Similarly, but basedon only proper motions, Dobbie et al. (2006b) argued that PG0136+251’sprogenitorislikelyconsistentwithcoeval formationwith the Pleiades. Gianninas et al. (2011)use white dwarf atmospheric models and fitting techniques equivalent to ours, and we adopt their T and log g in eff our analysis of these four white dwarfs. For our spectroscopic analysis we adopted the same analysistechniquesasthosedescribedinPaperI.Inbrief, we used the recent white dwarf spectroscopic models of Tremblayetal.(2011)withtheStarkprofilesofTremblay &Bergeron(2009),andtheautomatedfittingtechniques describedbyBergeronetal.(1992)tofitourBalmerline spectra and derive T and log g. However, for our de- eff rived parameters (mass, luminosity, and cooling age) we expand upon the methods of Paper I because our cur- rent sample has a far broader mass range. For deriving the parameters for white dwarfs of mass less than 1.10 M⊙ we applied our Teff and log g to the cooling mod- els for a carbon/oxygen (CO) core composition with a thick hydrogen layer by Fontaine et al. (2001). For the highest-masswhitedwarfs(>1.1M⊙)wederivedthepa- rameters based on the oxygen/neon (ONe) core models Fig.1.—Acomparisonat1M⊙ ofthecoolingratesforthesev- of Althaus et al. (2005; 2007). eralwhitedwarfmodelsthatwehavediscussed,inadditiontothe BaSTICOcoolingmodelsfromSalarisetal.(2010). Inthe lower Adiscussionofouradoptedmodelsiswarranted. First, panel,thehottesttemperaturesareshownandthesystematicsare wedonotadoptthemorerecentwhitedwarfatmospheric always∼1.5Myrorlessinmagnitude,whichisnotconcerning. At modelsofTremblayetal.(2013)becausethosefocusonly coolertemperatures,allmodelsotherthanthecarboncoremodels on3D modeling of convectiveatmospheres(.14,000K), of Wood (1995) are consistent. This is consistent with the sys- tematic effects of differing core composition, which grow as white and our current analysis looks hot fully radiative white dwarfscoolfurther. dwarfs. Forourcoolingmodels,whilethe Fontaineetal. (2001) models are widely used, we should acknowledge we compare the cooling ages from the Fontaine et al. two limitations they have. First, they assume a 50/50 (2001) CO core models at 1.0 M⊙ to both the CO and carbonandoxygencorecomposition,whichbasedonfull pure-carboncoremodels ofWood (1995)and the BaSTI stellarevolutionarymodelsisnotaccurate(e.g.,Romero CO models from Salaris et al. (2010). The BaSTI CO et al. 2013). The effect of this on the calculated cooling cooling models adopted C/O ratio profiles based on the agescanbeimportant,butfortherelativelyyoungwhite BaSTI scaled solar stellar evolution models. The lower dwarfs we are analyzing this effect remains small. Sec- panel of Figure 1 shows that at high T >45,000 K all eff ond, these cooling models do not begin at the tip of the of the differences between these models never result in asymptotic giant branch (hereafter AGB) and instead coolingagedifferencesofmorethan1.5Myr. The upper begin at a Teff of ∼60,000 K. In comparison, the widely panel of Figure 1 similarly shows that at cooler tem- usedWood(1995)COcoolingmodelsdobeginatthetip peratures 30,000<T <40,000 K the CO models, irre- eff of the AGB and they do not adopt a simple 50/50 car- spective of their adopted C/O ratios or starting points, bon and oxygen core composition7. Unfortunately, their all give strong agreement in cooling ages, but the sys- CO core models have an upper mass limit of 1.0 M⊙, tematic effects introducedfrompure carboncoremodels which limits their application in our analysis but they (∼10 Myr) are now clearly seen. provide an important test for systematics. In Figure 1 Forultramassivewhitedwarfs,the massatwhichthey transitiontoONewhitedwarfsremainsuncertain,andit 7 Equation1ofWood(1995)givestheirC/Oratiorelation. also likely depends on metallicity (Doherty et al. 2015). 4 Cummings et al. Herewehaveadoptedasomewhatconservative1.10M⊙, 3.1. Color-Color Analysis butwenotethatthemodelsofGarcia-Berroetal.(1997) In our cluster photometric analysis we first make use arguethatitmaybeaslowas1.05M⊙. Reassuringly,in of color-color diagrams (B-V vs U-B), which provide thismassrangetheCOandONecoolingagesatconstant direct photometric information on the cluster redden- mass are consistentfor such young white dwarfs,but for ing. The photometric metallicity can also be derived these white dwarfs the dependence of the mass-radius but it is quite sensitive to systematics in U magnitude, relationship on core-composition is very important. For a concern considering the typically more complex stan- example, applying the gravities of white dwarfs in the dardization process for U magnitudes and the varying mass range of 1.05 to 1.10 M⊙ (based on CO-core mod- sources of our photometry. Therefore, in the case of the els) to the ONe models derives masses ∼0.05 M⊙ lower Pleiades, NGC 2168, and NGC 2516 we consider more and places them all below 1.05 M⊙. Therefore, it is ap- detailed spectroscopic metallicities, but for NGC 2287, propriate to adopt 1.10 M⊙ as the transition mass. NGC 2323, and NGC 3532 we will simply adopt solar In Table 1 we present our white dwarf parameters for metallicity. However,we note that these adopted metal- the eight DA white dwarfs from NGC 2323. For clarity licitiesdoshowstrongconsistencywiththeobservedpho- in Table 1 we have distinguished between members and tometry. Ourcolor-coloranalysisadoptstwotechniques, nonmembers, where membership is based on our com- the firstis semi-empiricalandbasedon the Hyades fidu- parisonsofmodelbasedandobservedphotometryinad- cialandthesecondisbaseddirectlyontheY2isochrones, dition to comparisons of cooling ages to the cluster age which reachto higher masses than those available in the (seeourdetaileddiscussionofmembershipinSection4). Hyadesfiducial. Forbothmethodsthereddeningrelation Additionally, we have consideredS/N (per resolutionel- adopted is that of Cardelli et al. (1989) and the metal- ement) and the resulting errors given in Table 1. The licity correction is based on that of the Y2 isochrones. WD22andWD30 spectrahavelowS/Nandmasserrors The methods using the Hyades fiducial have been devel- greater than 0.1 M⊙. Therefore, their parameters are opedinDeliyannisetal.(inprep),wherethefiducialwas presented for reference but have been cut from our final derived from single-star cluster members (see Perryman analysis. This is because their membership determina- et al. 1998). The Hyades UBV photometry of Johnson tionsareunreliable,buttheirlowmassesdosuggestthey & Knuckles (1955) was adopted with a cluster [Fe/H] of are field white dwarfs. +0.15 and E(B-V) of 0. The Pleiades provides a good example for our color- 3. CLUSTERPARAMETERS color analysis techniques. In the left panel of Figure 2 we have plotted the photoelectric UBV photometry With IFMR analysis, the parameters of the star clus- from Johnson & Mitchell (1958) with several reddening ters are as critical as the white dwarfs themselves. This curvesbasedontheHyadesfiducial,andwehaveapplied is particularly true for the highest-mass white dwarfs, a metallicity of [Fe/H]=0.01 (Z=0.0185) to match the where the derived masses of their progenitors change reddening insensitive region where all of the reddening rapidly with evolutionary time. Uniform photometric curvesintersectnearB-Vof0.6. Fittingbyeyetheblue- data sets are not available for these six clusters. But wardcolor-colortrend we find that a reddening curve of in Paper I we showed that for NGC 2099, the adopted E(B-V)=0.03±0.02 matches the Pleiades UBV photom- isochrones and fitting techniques had as large, if not etry the best. Both this reddening and metallicity are larger, of an effect on its cluster parameters as did any consistentwith the typically derived values and spectro- systematics between the cluster’s different photometry scopic analyses of the Pleiades (e.g., [Fe/H]=0.01±0.02 sets. This is best illustrated by the systematic effects on Schuler et al. 2010; [Fe/H]=0.03±0.02±0.05 Soderblom derived cluster ages, where the Yi et al. (2001; hereafter et al. 2009) We also note that the Hyades fiducial ends Y2) and the Ventura et al. (1998) isochrones both gave at B-V∼0.1, where the older Hyades turnoff occurs. our final adopted age of 520 Myr for NGC 2099. The In the right panel of Figure 2 we fit the higher-mass PARSEC version 1.2S isochrones (Bressan et al. 2012) starsbluerthanB-V=0.0withthreedifferent135MyrY2 derived a comparable age of 540 Myr, and lastly the Gi- isochrones of differing metallicity. All three metallicities rardi et al. (2000) and Bertelli et al. (1994) isochrones fitareddeningofE(B-V)=0.03intheseblueststars. This bothgavesignificantlyyoungeragesof445Myr. Between demonstratesthatthesehigher-massstarscreateanearly these isochrone sets there is nearly a 100 Myr range of lineartrendthatisinsensitivetovariationsinmetallicity. derived ages when using identical photometry, and fur- We note that the position of this blue linear trend is ther systematics can be introduced based on how the also insensitive to cluster age, where as we look at older isochrones are fit to the data. clusters the trend only shortens in length and does not In this paper we have redetermined as uniformly as shift its position. Therefore, this linear trend’s position possible the reddenings, distance moduli, and ages for provides a reliable reddening measurement independent these six clusters using available high-quality UBV pho- of all other cluster parameters. tometry that covers up to the full turnoff. The Y2 In the upper left panel of Figure 3 we compare di- isochrones provide our final adopted cluster ages, but rectly our Hyades fiducial fit and our Y2 fit for the to broaden our results we also determine ages with the Pleiades. Therearesystematicdifferencesthatarenoted PARSEC isochrones. These two isochrones give only intheB-Vrangeof0.1-0.3,whereitappearsthattheY2 slightly different cluster ages, but in these younger clus- isochrones are too blue in U-B relative to both the data ters the masses of the progenitor stars have a far more and the Hyades fiducial. However, when adopting a Y2 significant dependence on evolutionary time, and even a isochronewithanageconsistenttotheHyades(650Myr; 20 Myr systematic has a significanteffect on our results. notshown)theisochroneisnearlyidenticaltothatofthe White Dwarfs in NGC 2323 5 -0.6 -0.6 -0.4 -0.4 -0.2 -0.2 0 0 0.2 0.2 0.4 0.4 0.6 0.6 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 B-V B-V Fig.2.—Color-coloranalysisofthePleiades. Intheleftpanel wehaveplotted threeHyadesfiducial curvesof[Fe/H]=0.01(Z=0.0185) withdifferingreddening(GreenE(B-V)=0.0,RedE(B-V)=0.03,MagentaE(B-V)=0.1)tothePleiadesdata. ThisfindsthatnearB-V=0.6, where all three reddening curves intersect, the photometry is not dependent on reddening. Therefore, the photometric metallicityis first matched by fitting this region to the data. Of the three curves a reddening of E(B-V)=0.03 (red curve) provides the best fit in the regionspanning B-V of 0.1-0.5. In the rightpanel we have focused on the hottest stars (B-V<0.0)and plot three Y2 isochrone curves of E(B-V)=0.03 with differingmetallicity (Green Z=0.0125, Red Z=0.0185, and Magenta Z=0.0245). These models extend farther into the blue,andthisdemonstrates that whiletheotherregions ofthediagramaresensitivetometallicity,thesehigher-massstarcolorsareonly dependent onreddening. Hyades fiducial, so at these younger ages the isochrones tifies the richest group of cluster stars with consistent appear to overestimate the U flux at this intermediate reddening. Inthe Pleiadesshowninthe upper-leftpanel color range. At all other colors this does not seem to be of Figure 3, we mark in red the high-mass stars consis- aconcernbecause,reassuringly,boththeHyadesfiducial tent with the trend, and several stars deviate from this and the Y2 isochrones find the same reddening in their trend that are likely peculiar. However, we do acknowl- respectiveregions,andthey alsoagreein the regionsen- edgethatthebrighteststarinthePleiades,Alcyone,does sitive to metallicity (B-V∼0.6) and redder. In regard notdeviatefromthistrend. Alcyoneisamultiplesystem to the PARSEC isochrones, they are not independently andhasseveralpeculiarcharacteristicslikespectralemis- considered in our color-color analysis because their U-B sion and rapid rotation (Hoffleit & Jaschek 1991) and is colors do not fit the observations. commonly referred to as both a blue straggler and Be A key advantage to color-color analysis is that it can dwarf. This suggests that while this color-color method be used to clean young cluster turnoffs. Dating younger doesidentifymanyproblematicstars,itdoesnotremove clusters is prone to several difficulties, including that all peculiar stars. turnoffs are relatively sparse even in the richest clusters In the upper right panel of Figure 3 we have simi- and that these higher-mass turnoff stars have high bi- larlyplottedtheNGC2323UBVphotometryfromClaria narity fraction (Kouwenhoven et al. 2007). Addition- etal.(1998). Wefittheblueststarsandseethatindepen- ally,manyhigher-massstarsfallintothepeculiargroups dentofanassumedmetallicity,wefindalargereddening of blue stragglers or Be stars. For example, Mermilliod of E(B-V)=0.23±0.06at (B-V) =0, where we deem this 0 1982a, 1982b, and Ahumada & Lapasset (2007) found reddening large enough to account for the color depen- that several of the brightest stars in the clusters being denceofreddening(seeFernie1963andourdiscussionin analyzed in this paper are blue stragglers that are far Paper I). When we have adopted a color dependent red- too blue or too bright, if not both, for their age. Lu deningwewilldefinethereddeningat(B-V) =0. Again, 0 et al. (2011) modeled the formation of blue stragglers our selected final turnoff stars are shown in red. on the short time scales necessary in these young clus- OurfouradditionalclustersarealsoshowninFigure3. ters andfound that they can rapidly be createdthrough InNGC 2516,we haveusedtwo UBVphotometric stud- binary mass transfer. Identifying these peculiar stars ies, Dachs (1970) for the brightest stars and Sung et al. can greatly improve the fit of the turnoff ages of these (2002) for the fainter stars shown in blue. For the mod- younger clusters. Mermilliod 1982b do find that when erately large reddening we fit a E(B-V) of 0.10±0.03 at plottedincolor-colorspaceBestarsdeviatefromtheap- (B-V) =0 and a [Fe/H]=0.065,and as with the Pleiades 0 proximatelylineartrendofthe“normal”high-massstars. this is consistent with the typically adopted parameters Additionally, several of our clusters may have variable and our spectroscopic analysis (Cummings 2011). For reddening, and fitting this high-mass linear trend iden- NGC 2287 we have used the UBV photometry of Ianna 6 Cummings et al. -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 -0.6 -0.6 -0.4 -0.4 Pleiades NGC 2323 -0.2 -0.2 0 0 0.2 0.2 0.4 0.4 NGC 2516 NGC 2287 -0.5 -0.5 0 0 0.5 0.5 NGC 3532 NGC 2168 -0.5 -0.5 0 0 0.5 0.5 -0.2 0 0.2 0.4 0.6 0.8 -0.2 0 0.2 0.4 0.6 0.8 B-V B-V Fig. 3.— Our color-color analysis of all six of our clusters. The full data sets are displayed in black, with the NGC 2516 data being supplementedbyadditionaldatainblue. OurY2color-colorfitsareshowningreen,andthecomparableHyades-fiducialbasedfitsadopting the samemetallicitiesand reddenings areshown inmagenta. Weusethe Y2 relations to identifythe mostreliableturnoff starsshown in red,whichwewilluseinourclusterageanalysis. SeeTable2andthetextforourphotometricsources. etal.(1987)andfitaE(B-V)of0.035±0.025withanas- full B-V color range. sumed solar metallicity. For NGC 3532 we have used the UBV photometry of Fernandez & Salgado (1980) 3.2. Color-Magnitude Analysis and fit a E(B-V) of 0.04±0.025 with an assumed solar In Figure 4 we display our by eye turnoff age fits with metallicity. Lastly,forNGC2168wehaveusedtheUBV Y2 isochrones in greenwhen adopting from Figure 3 the photometry of Sung & Bessell (1999). We derive a E(B- reddenings, metallicities, and the cleaned turnoff stars V) of 0.25±0.04 at (B-V) =0 for this cluster and adopt 0 shownin red. In magenta we similarly display the PAR- a metal-poor [Fe/H] of -0.143 (Steinhauer & Deliyannis SECisochronefits,wherewehaveadoptedthesamered- 2004), which provides a reliable fit to the bluest stars denings, distance moduli, and metallicities (Z). In Table andfollows the U-B rangewellbut it appearsthat there 2 the derived cluster parameters and the photometric isasystematicshiftinB-Vwherethe starsbecomingin- sources are listed. We note that in the youngest clus- creasingly too blue at redder colors. Otherwise, changes ters the PARSEC isochrones systematically derive ages in either adopted metallicity or reddening cannot fit the 25Myryounger,whileinthe320MyrNGC3532theyde- White Dwarfs in NGC 2323 7 0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5 -4 -4 -2 -2 0 0 2 2 4 4 6 6 -2 -2 0 0 2 2 4 4 6 6 0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5 B-V B-V B-V Fig.4.— Our color-magnitude diagrams for our six clusters. The full data sets are shown in black, with several clusters having supplemental data showninblue. Ourfinal turnoffstarsselected inFigure3areagainshownhereinred,and weshow our Y2 isochrone fitsingreenandourPARSECisochronefitsinmagenta. SeeTable2andthetextforourparameters andphotometricsources. rive ages 5 Myr older, and lastly from Paper I the PAR- parameters because they more successfully fit both the SEC isochrones derive an age 20 Myr older in the 520 color-colordata and the the main sequence features. Myr old NGC 2099. Therefore,these two isochronesnot These cluster ages provide several advantages over onlyhavechangingsystematicsatdifferingages,butthey adopting literature values. 1) The ages are based on are in opposite directions in young versus older clusters. a uniform system of isochrones, while literature values Thepossiblecausesofthesystematicsbetweenthesetwo adopt wide ranging models that have systematic differ- isochrones include the differences in their adopted opac- ences thatbecome more pronouncedatyoungerages. 2) ities, equations of state, and solar compositions. While The fitting techniques applied are by eye but consistent, the Y2 isochrones do not consider evolution past the tip while fitting techniques for literature values can greatly of the red giant branch (RGB), in these typically young vary. 3) The difficulty of peculiar turnoff stars are ad- clusters we cannot reliably fit the giants because their dressedinasystematicway,whiletheirconsiderationcan populationsareverysparseorthey havenogiantsatall. haveimportantdifferencesintheliteraturevalues,ifthey Therefore,wehavechosentheY2 isochronesforourfinal are consideredat all. We will not comment on the abso- 8 Cummings et al. TABLE 2 - Open Cluster Parameters Cluster E(B-V)a (m-M)0 [Fe/H] Y2 Age(Myr) PARSECAge(Myr) PhotometricSources Pleiades 0.03±0.02 5.67±0.10 +0.01 135±15 110±15 1 NGC2323 0.23±0.06 10.0±0.15 0.00 140±20 115±20 2,3 NGC2516 0.10±0.03 8.20±0.12 +0.065 170±20 150±20 4,5 NGC2168 0.25±0.04 9.66±0.10 -0.143 190±20 170±20 6 NGC2287 0.035±0.025 9.52±0.12 0.00 220±30 205±30 7,8 NGC3532 0.04±0.025 8.46±0.14 0.00 320±20 325±20 9 TABLE 2 a)Forreddeningsof0.10orlargerwehaveadopted thecolordependent reddeningrelationofFernie(1963) andgivethe derivedreddeningsatacolorof(B-V)0=0. Iftherearemorethantwophotometricsources,theprimarysourceislistedfirstandthe secondarysourceisonlyforfaintstarsbeyondthephotometriclimitoftheprimary. (1)Johnson&Mitchell(1958); (2)Clariaetal. (1998); (3)Kaliraietal.(2003); (4)Dachs(1970); (5)Sungetal.(2002); (6)Sung&Bessell(1999); (7)Iannaetal.(1987); (8)Sharma etal.(2006); (9)Fernandez&Salgado(1980). TABLE 3 - Membership Data ID α δ MV V (B-V)0 B-V tcool (J2000) (J2000) (Model) (Obs.) (Model) (Obs.) (Myr) NGC2323LikelySingleStarWhiteDwarfMembers NGC2323-WD10 7:02:41.02 -8:26:12.8 10.36±0.19 20.62±0.009 -0.303±0.002 -0.020±0.060 2.9+−11..90 NGC2323-WD11 7:03:22.14 -8:15:58.7 10.36±0.13 20.67±0.008 -0.305±0.002 0.050±0.025 2.6+−10..17 WhiteDwarfsInconsistentwithSingleStarMembership NGC2323-WD21 7:02:08.56 -8:25:48.0 11.33±0.25 21.91±0.026 -0.013±0.022 0.149±0.158 170+60 −48 NGC2323-WD7 7:02:47.87 -8:35:56.4 10.92±0.07 19.50±0.004 -0.005±0.007 0.078±0.004 112+12 −11 NGC2323-WD17 7:03:12.62 -8:30:53.6 10.97±0.07 20.96±0.010 -0.057±0.007 0.107±0.015 96+12 −11 NGC2323-WD12 7:03:17.79 -8:19:59.7 10.87±0.11 20.52±0.007 -0.014±0.010 0.116±0.079 101+17 −15 LowSignaltoNoiseandFeatureless WhiteDwarfs NGC2323-WD22 7:02:33.93 -8:31:04.3 10.53±0.36 21.89±0.024 -0.135±0.025 0.180±0.013 33+36 −16 NGC2323-WD30 7:03:22.56 -8:29:25.7 11.52±0.28 22.04±0.028 0.111±0.026 0.485±0.029 290+108 −82 NGC2323-WD38 7:03:31.87 -8:28:25.0 – 22.93±0.062 – 0.645±0.058 – NGC2323-WD23 7:03:39.85 -8:28:16.7 – 21.42±0.016 – 0.375±0.044 – luteaccuracyofthe variousisochronemodelages,butin tudesandcolorsinFigure5toderiveaneffectivedistance this studyuniformity andprecisionis the goal. We must modulus and reddening for each white dwarf. We define also reiterate two remaining limitations with our cluster our 1σ color and magnitude errorsby adding in quadra- parameters. First, all of our photometries are primarily turetherespectivemodelfittingerrors,theobservational from differing groups, which still may leave important errors, and in the case of magnitude the distance mod- systematics remaining in our parameter analyses. Addi- ulus errors or in the case of color the reddening errors. tionally, uniformly measured spectroscopic metallicities We selectwhite dwarfs as consistentwith membershipif are also needed to address the metallicity sensitivity in theireffectivedistancemodulusandreddeningarewithin the turnoff isochrone fits. 2σ oftheclusterparametersderivedinSection3. Ofour Lastly, for Sirius B there is no parent cluster that we six high signal white dwarfs, only WD10, WD11, and canself-consistentlyanalyzeforthetotalage,buttheage WD21 have both magnitudes and colors consistent with oftheSiriussystemiswellstudiedandhereweadoptso- membership. However, WD21 has a cooling age longer larmetallicityandtheageof237.5±12.5Myrdetermined than the age of the cluster. Therefore, we do not con- in Liebert et al. (2005). sider it a member but a field white dwarf at comparable distance to NGC 2323. 4. WHITEDWARFMEMBERSHIPINNGC2323 In Figure 6 we show the Balmer line fits of WD10 Cluster membership determination is key in analyzing and WD11, the two cluster members. They both are theformationhistoryofthesewhitedwarfsandapplying highmass at∼1.07M⊙ andhavevery hightemperature themtotheIFMR.InthecaseofNGC2323,wehavethe (>50,000 K) with accordingly extremely short cooling advantagethatbecauseitissuchayoungcluster,anyhot times of ∼3 Myr. Their membership further suggests and young high-mass white dwarfs observed in its field that all of our other observed white dwarfs, which would already have reliable membership. For example, all have both lower masses and longer cooling times in the Sloan Digital Sky Survey sample of field white are not consistent with NGC 2323 membership. In dwarfs, only ∼2.6% of DA white dwarfs have a mass Figure 7 we display the CMD of all observed candidates greaterthan 1 M⊙ and less than 10%of these are young relativetoseveralwhitedwarfcoolingmodelstakenfrom withaT >20,000K(e.g.,Kleinmanetal.2013;Kepler http://www.astro.umontreal.ca/∼bergeron/CoolingModels/ eff et al. 2016). To further confirm cluster membership in (Holberg & Bergeron 2006, Kowalski & Saumon 2006, NGC 2323, however, we have also performed magnitude Tremblay et al. 2011, and Bergeron et al. 2011). and color analysis. In Table 3 we list the observed photometry for these 5. INITIAL-FINALMASSRELATION white dwarfs and also list their cooling age and model With the white dwarf cluster members, a simple com- based MV and (B-V)0. For our membership analysis we parisonof their cooling age to the total cluster age gives directly compare the model based and observed magni- the evolutionary time to the tip of the asymptotic giant White Dwarfs in NGC 2323 9 8 9 10 11 0 0.1 0.2 0.3 0.4 Fig.5.—The upper panel compares the model based and observed whitedwarf magnitudes relativeto the cluster distance modulus of 10.725 (adjusted based on their blue colors). Similarly, the lower panel compares the model based and observed white dwarf B-V colors relativetotheclusterreddeningof0.235(adjustedbasedonthebluecolors). Inbothpanelstherespective1σerrorbarsareshown,which include both the spectroscopic and photometric errors. WD10, WD11, and WD21 are shown at the top and are consistent (within 2σ) withclustermembershipinbothmagnitudeandcolor,butWD21hastoolongofacoolingageforNGC2323andhasbeengroupedwith thethreeadditionalnonmembersdisplayedinthemiddle. Lastly,thebottomtwowhitedwarfsaredisplayedbuthavespectroscopicfitting errorsaboveourcutformembershipanalysis. that the PARSEC models do not include the thermally pulsing-AGBphase,butthisphaseisveryshortanddoes not meaningfully affect the resulting progenitor masses. For the Y2 isochroneswe cannotinfer progenitormasses directly because these isochrones do not evolve beyond the RGB. But we note that while the Y2 isochrones predict slower evolution to the turnoff in these younger clusters, they predict more rapid evolution through the RGB. This results in the total evolutionary time scales to the tip of the RGB being comparable in both model isochronesforallbutthehighestmasses. Forexample,as we reach evolutionary times of 100 Myr or shorter (pro- genitorsof &5.3M⊙), the systematic differences become significant between all three models. For our current analysis, with both our Y2 and PARSEC ages, we use progenitor masses derived from the PARSEC isochrones becauseitwillprovidethestrongestconsistencywithour cluster age fits. Fig.6.—TheBalmerlinefitsforthetwowhitedwarfmembers For NGC 2323-WD10 and WD11, with their cooling of NGC 2323. The Hβ, Hγ, Hδ, Hǫ, and H8 fits are shown from times and our Y2 isochrones based age for NGC 2323 of bottom totop. 140 Myr, the corresponding progenitor masses for both branch (AGB) for their progenitor. Application of this white dwarfs are 4.69 M⊙. (See Table 1, where we also give the progenitor masses based on the cluster age er- time toevolutionarymodelsgivesthe white dwarf’spro- rors [140±20 Myr] and the PARSEC based age of 115 genitor mass. In Paper I we adopted the models of Hur- Myr.) It is quite remarkable that we have two inde- ley et al. (2000) for the progenitor masses in NGC 2099 pendentlyformedhigh-masswhite dwarfsfromthesame (3-4 M⊙). At these masses, the difference between pre- cluster that are so consistent in both initial and final dictedprogenitormassbytheHurleyetal.(2000)models mass. Acrosstheirlifetimestheybothlost77.2%oftheir and the PARSEC isochrones are less than 1%. We note 10 Cummings et al. TABLE 4 - Reanalyzed White Dwarf Initial and Final Parameters ID Teff logg MWD MV tcool Y2 Mi PARSECMi S/N (K) (M⊙) (Myr) (M⊙) (M⊙) NGC2287-2 25900±350 8.45±0.05 0.909±0.028 11.01±0.09 76+−190 4.61+−00..1131 4.81+−00..1162 150 NGC2287-4 26500±350 8.71±0.05 1.065±0.027 11.44±0.10 127+−1143 5.50+−00..4300 5.93+−00..5328 130 NGC2287-5 25600±350 8.44±0.04 0.901±0.028 11.02±0.08 77+−190 4.63+−00..1131 4.83+−00..1162 170 NGC2516-1 30100±350 8.47±0.04 0.925±0.027 10.74±0.08 42+−77 4.85+−00..1019 5.19+−00..1142 160 NGC2516-2 35500±550 8.55±0.07 0.981±0.040 10.58±0.13 24+−87 4.61+−00..1018 4.88+−00..1130 90 NGC2516-3 29100±350 8.46±0.04 0.918±0.027 10.78±0.08 48+−87 4.94+−00..1130 5.31+−00..1194 170 NGC2516-5 32200±400 8.54±0.05 0.970±0.027 10.73±0.09 38+−76 4.80+−00..1009 5.12+−00..1131 170 NGC3532-1 23100±300 8.52±0.04 0.950±0.026 11.33±0.08 131+−1132 4.17+−00..1019 4.13+−00..1019 240 NGC3532-5 27700±350 8.28±0.05 0.804±0.028 10.57±0.08 31+−76 3.57+−00..0033 3.55+−00..0033 220 NGC3532-9 31900±400 8.18±0.04 0.752±0.026 10.13±0.07 9.3+−21 3.48+−00..0011 3.46+−00..0011 210 NGC3532-10 26300±350 8.34±0.04 0.838±0.027 10.78±0.08 51+−88 3.66+−00..0044 3.64+−00..0044 200 NGC3532-J1106-590 21100±350 8.48±0.05 0.922±0.031 11.43±0.09 163+−1187 4.46+−00..2117 4.41+−00..2117 110 NGC3532-J1106-584 20200±300 8.52±0.05 0.945±0.029 11.58±0.09 197+−2108 4.91+−00..3276 4.83+−00..3276 120 NGC3532-J1107-584 20700±300 8.59±0.05 0.990±0.028 11.66±0.09 211+−2210 5.16+−00..4394 5.07+−00..4394 180 SiriusB 26000±400 8.57±0.04 0.982±0.024 11.21±0.08 99+−1110 4.69+−00..1152 4.69+−00..1152 - Pleiades-LB1497 32700±500 8.67±0.05 1.046±0.028 10.94±0.09 54+−98 5.86+−00..2296 6.85+−00..5471 170 Pleiades-GD50 42700±800 9.20±0.07 1.246±0.021 11.58±0.15 70+−1143 6.41+−00..7421 8.05+−11..9055 - Pleiades-PG0136+251 41400±800 9.03±0.07 1.186±0.027 11.28±0.15 48+−1121 5.64+−00..4226 6.52+−10..0460 - NGC2168-LAWDS1 33500±450 8.44±0.06 0.911±0.039 10.47±0.11 19+−76 4.27+−00..0075 4.48+−00..0096 95 NGC2168-LAWDS2 33400±600 8.49±0.10 0.940±0.061 10.57±0.18 25+−1130 4.33+−00..1140 4.55+−00..1182 70 NGC2168-LAWDS5 52700±900 8.21±0.06 0.801±0.031 9.49±0.10 1.0+−00..11 4.10+−00..0011 4.28+−00..0011 210 NGC2168-LAWDS6 57300±1000 8.05±0.06 0.731±0.029 9.13±0.11 0.5+−00..11 4.10+−00..0011 4.28+−00..0011 260 NGC2168-LAWDS11 19900±350 8.35±0.05 0.834±0.035 11.31±0.09 149+−1187 7.93+−11..4040 11.60+−∗3.34 100 NGC2168-LAWDS12 34200±500 8.60±0.06 1.009±0.037 10.73±0.12 36+−98 4.44+−00..1019 4.69+−00..1131 100 NGC2168-LAWDS14 30500±450 8.57±0.06 0.988±0.038 10.89±0.12 54+−1110 4.67+−00..1164 4.98+−00..2117 100 NGC2168-LAWDS15 30100±400 8.61±0.06 1.009±0.032 10.98±0.11 64+−1100 4.80+−00..1174 5.16+−00..2128 130 NGC2168-LAWDS22 53000±1000 8.22±0.06 0.807±0.035 9.50±0.11 1.0+−00..11 4.10+−00..0011 4.28+−00..0011 250 NGC2168-LAWDS27 30700±400 8.72±0.06 1.071±0.031 11.16±0.11 78+−1121 5.06+−00..2250 5.49+−00..3257 120 NGC2168-LAWDS29 33500±450 8.56±0.06 0.984±0.034 10.70±0.11 34+−88 4.42+−00..1008 4.67+−00..1120 120 NGC2168-LAWDS30 29700±500 8.39±0.08 0.878±0.048 10.63±0.13 33+−1120 4.41+−00..1130 4.65+−00..1162 60 totalmass. This arguesforthe consistencyofsingle-star tionofY2 agesinthispaper,whichsystematicallyderive masslossathighermasses. Tolookatthesedataincon- olderagesinyoungclusters,wefindthereisnomeaning- text, inTable 4 wepresentthe initial-finalmassdata for fulchangeinslopeathighermassesandthedataappears 30 mid to very high-mass (0.73-1.25 M⊙) white dwarfs to be defined well with a weighted linear relation: that have been self-consistently analyzed from publicly available data. Mfinal =(0.143±0.005)Minitial+0.294±0.020M⊙. In the left panel of Figure 8 we plot the initial and fi- ThelinearnatureoftheIFMRacrosssuchabroadrange nal masses of the two analyzed white dwarfs from NGC (3–6.5M ) is ofinterest. Forexample, this may sug- initial 2323 and the 30 white dwarfs we have reanalyzed from gest we can extrapolate to derive the progenitor of a the literature, adopting the Y2 ages. In the right panel Chandrasekharmass limit white dwarf to be ∼7.75 M⊙, of Figure 8 we plot these same data, but with appli- but the still limited data at the highest masses and the cation of the PARSEC based ages. We also compare remaining uncertainties in the evolutionary models sug- these high-mass white dwarf data to the sample of 31 geststhisisunreliable. Furthermore,intheoreticalmod- intermediate-mass white dwarfs taken from NGC 2099, els a moderate turnoverin the slope of the IFMR is pre- theHyades,andPraesepe(PaperI).Forthehigher-mass dicted near initial mass of 4 M⊙ (e.g., Marigo & Girardi white dwarfs,while there is some dispersionin the trend 2007 and Meng et al. 2008). This predicted turnover is there is only one clear outlier that is from NGC 2168. the result of the second dredge-up, which only occurs in LAWDS11 is an extreme outlier with a far longer cool- stars of ∼4 M⊙ and higher. This diminishes their core ing time than the other members of NGC 2168, giving mass and hence their final white dwarf mass. A compa- it a massive progenitor. For clusters in the rich galactic rable turnover could still be lost in our data’s remaining plane, contamination from common field white dwarfs is scatter, causing further problems with a linear extrapo- expected and likely explains this white dwarf. Proper lation. In Figure 9 we more closely analyzethe residuals motions may be necessary to further constrain its mem- of this linear data fit. bership,butinourcurrentanalysisandIFMRfitswedo Toillustratetheimportanceofouradoptedisochrones, not consider LAWDS11. the right panel of Figure 8 shows a clear turnover in the In our initial analysis from Paper I, we demonstrated IFMRnearaninitialmassof4M⊙ whenadoptingPAR- that the intermediate-mass white dwarfs (0.7–0.9 M⊙) SEC isochrone based ages, consistent with theoretical create a steep IFMR slope. With the continued adop- predictions. We note that while our comparison IFMR