Observing the products of stellar evolution in the old open cluster M67 with APOGEE

0.87 MB·

by  Clio Bertelli Motta

MNRAS000,1–16(2016) Preprint5January2017 CompiledusingMNRASLATEXstylefilev3.0
Observing the products of stellar evolution in the old open cluster
M67 with APOGEE
Clio Bertelli Motta,1⋆ Maurizio Salaris,2 Anna Pasquali,1 Eva K. Grebel1
1AstronomischesRechen-Institut,ZentrumfürAstronomiederUniversitätHeidelberg,Mönchhofstr.12-14,69120Heidelberg,Germany
2AstrophysicsResearchInstitute,LiverpoolJohnMooresUniversity,146BrownlowHill,LiverpoolL35RF,UK
7
1
0 Accepted2016December12.Received2016December9;inoriginalform2016August9
2

n
a ABSTRACT
J Recent works have shown how the [C/N] ratio in stars after the first dredge-up (FDU) can
4 be used as an age estimator in virtue of its dependence on stellar mass. For this purpose,
precise predictions of the surface chemical composition before and after the mixing takes

] place in the convective envelope of subgiant stars are necessary. Stellar evolution models
R
can provide us with such predictions, although a comparision with objects of known age
S is needed for calibration. Open clusters are excellent test cases, as they represent a single
. stellar population for which the age can be derived through, e.g., isochrone fitting. In this
h
p study,wepresentadetailedanalysisofstarsbelongingtothewell-knownopenclusterM67
- observedbytheAPOGEEsurveyinthetwelfthdatareleaseoftheSloanDigitalSkySurvey
o and whose chemical properties were derived with the ASPCAP pipeline. We find that the
r
t [C/N]abundanceofsubgiantbranchstarsisoverestimatedby∼0.2dexduetoanoffsetinthe
s determinationofthe[N/Fe]abundance.Starsontheredgiantbranchandredclumpareshown
a
nottobeaffectedbythisoffset.Wederive[C/N] =−0.46±0.03dex,whichposesastrong
[ FDU
constraintoncalibrationsof[C/N] asageindicator.Wealsodonotfindanyclearsignature
FDU
1 ofadditionalchemicalmixingprocessesthatsetinaftertheredgiantbranchbump.Theresults
v
obtainedforM67indicatetheimportanceofconductinghigh-resolutionspectroscopicstudies
9
ofopenclustersofdifferentagesinordertoestablishanaccurateage-datingmethodforfield
7
stars.
9
0 Keywords: stars:abundances–stars:evolution–openclustersandassociations:individual:
0 M67
.
1
0
7
1
1 INTRODUCTION constraintsonthemassoftheobjectarenotavailable.Toimprove
:
v RGBstarageestimates,Masseron&Gilmore(2015)employedas
Age-datingoffieldstarsiscrucialinordertoreconstructtheevo-
i age diagnostic the abundance ratio [C/N] measured after the
X lutionary history of the Galaxy. Determining the age distribution FDU
completion of the FDU and at magnitudes fainter than the RGB
throughout the Milky Way requires observations and age-dating
r bumpluminosity, where observations reveal theonset of anextra
a of stars brighter than the main-sequence turn-off, in evolutionary
mixingprocessthatcontinuesuntilthetipoftheRGB(andbeyond
phasesthatcoveralargerangeoftheGalacticevolutionaryhistory.
–see,e.g.,Salarisetal.2015,formoredetails).
Redgiantbranch (RGB)starsareparticularlyuseful forthispur-
pose, for they are bright and span an age range between ∼1 Gyr Stellar model calculations show that at the end of the main
andtheageoftheUniverse,andindeedmodernspectroscopicsur- sequence phase the outer convection zone progressively engulfs
veys(e.g.,theGaia-ESOandtheApachePointObservatoryGalac- deeper regions, reaching layers that had been partially processed
ticEvolutionExperiment–APOGEE–surveys,see Gilmoreetal. by H-burning during the main sequence phase. In these layers –
2012;Holtzmanetal.2015)providesurfacegravity(logg),effec- whether or not the CNO-cycle was the main energy-generation
tive temperature (T ), and photospheric abundances of several mechanism– the C and N abundances had enough time to attain
eff
chemicalelementsforlargesamplesofMilkyWayRGBstars. theCN-cycleequilibriumvalues,implyinganincreaseofNanda
AgeestimatesofRGBstarsarehoweverparticularlychalleng- decreaseofCwithrespecttoaninitialscaled-solarorα-enhanced
ing, since small uncertainties in T (at fixed logg) cause large metalmixture.
eff
uncertainties in thestellar mass, hence their age, if asteroseismic
As a consequence, convection dredges N-enriched and C-
depleted mattertothesurface (FDU),lowering thesurface[C/N]
ratiocomparedtotheinitialvalue.Thisdecreaseof[C/N]afterthe
⋆ E-mail:[email protected] FDUismassdependent, becausewhenthestellarmassincreases,
(cid:13)c 2016TheAuthors
2 BertelliMottaet al.
surfaceconvection attheFDUengulfsalargerfractionoftheto- abundancesofseveralelementsarederivedwiththeAPOGEEStel-
talstellarmass,causingalargerdecreaseofthesurface[C/N]with larParametersandChemicalAbundancesPipeline(ASPCAP)(see
increasing RGB mass – hence with decreasing age of the stellar GarcíaPérezetal.2016;Holtzmanetal.2015).
population.
Martigetal.(2016) provided asemi-empirical calibrationof
2.1 Membershipanalysis
[C/N] -chemical composition-age relations,basedonasample
FDU
offieldstarswithalsoasteroseismicmassestimates,forwhichthe Foraninvestigationof thechemical composition ofstarsinopen
agewasdeterminedusingmass-agerelationsfromtheoreticalstel- clusters and especially for the study of possible inhomogeneities
larmodels.Theaccuracyoftheagesdeterminedbymeansofthis arisingfromstellarevolution,areliabledeterminationofthemem-
relationisabout40%(Martigetal.2016). bershipprobabilityofeachstarisessential.Infact,onlyifwecan
Onthetheoreticalside,Salarisetal.(2015)presentedanddis- excludecontaminationfromfieldstarswith(almost)absolutecer-
cussedatheoreticalcalibrationof[C/N]FDUasafunctionofmetal- tainty, we can reasonably assume that potential variations in the
licity and age, which allows one to age-date RGB stars with an chemicalabundancesofclusterstarsareaproductofeitherstellar
internalaccuracyofabout∼15%.However,asshownbytheseau- evolutionorofthestarformationhistorywithinthecluster.
thors,differentsetsoftheoreticalmodelsgiveverydifferentresults InordertoselectlikelymemberstarsofM67,wedeveloped
intermsofthe[C/N]FDU-agerelationatagivenmetallicity,espe- acodethatanalysesdifferentpropertiesofthecandidatemembers
ciallyforagesbelow10Gyr.Forthisreason,testsofthetheoretical sample. In the following, we illustrate in detail the membership
[C/N]FDU-agerelationshipsinopenclusterswithwell-determined analysispipeline.
metallicityandageareabsolutelycrucial.
InthispaperweconsidertheoldopenclusterM67(orNGC 1. Thefirststepconsistsinacrossmatch oftheAPOGEEstars
2682). Located at 800-900 pc from the Solar system (see, e.g., in the SDSS DR12 with the central coordinates of M67, which
Sarajedinietal. 2009), M67 is one of the most interesting and we retrieve from Kharchenkoetal. (2013). All stars are selected
beststudiedexamplesamongGalacticopenclusters.Withanage thatliewithinthecluster’soutermostradius(1.03deg),definedby
of3.5-4Gyr(see,e.g.,Sarajedinietal.2009;Bellinietal.2010a; Kharchenkoetal.(2013)asthedistancefromthecentreoftheclus-
Kharchenkoetal.2013),M67isveryoldforanopencluster.More- teratwhichthedensityofstarscannotbedistinguishedfromthat
over,itschemicalcompositionwasshowntobemoresimilartothat ofthefield.
oftheSunthanmostothernearbyfieldstars.Inparticular,thesolar 2. Weexpectthatthememberstarsofagivenopenclustershare
twinM67-1194isconsideredtobethestarmostsimilartotheSun the same tangential velocities (as inferred from their proper mo-
observedsofar.ThesefindingsledtothehypothesisthattheSun tions in RA and Dec) and radial velocities (RV). Thus, after the
mighthaveformedinsideM67,althoughinthisscenario,thepro- spatialconstraintssetbystep1.,weaddressthetangential veloc-
toplanetarydiscor planetarysystemoftheSunshouldhavebeen ities of the stars. In particular, we are interested in the absolute
disruptedduringtheejectionprocess(seePichardoetal.2012). valueofthepropermotion(PM)vector,whichwederivethrough
For M67, spectroscopic values of [C/N] from thevectorialsumofthePMcomponentsinRAandDecprovided
FDU
Gilroy&Brown (1991) were compared by Salarisetal. (2015) by the PPMXL catalogue (Roeseretal. 2010), including the cor-
withtheirtheoretical[C/N] -agerelationforthecluster metal- rectionsbyVickersetal.(2016).Thefollowingrelationholdsfor
FDU
licity. However, the uncertainties in the empirical data were very theabsolutevalueofthePM:
largeandthecomparisondidnotsetanystrongconstraintonsuch
PM = PM2 +PM2 . (1)
relation. Here we use chemical abundances of C, N, O, and Fe tot q RA Dec
taken from APOGEE published as part of SDSS-III data release Plottingahistogramoftheresulting PM forallstarswithinthe
tot
12 (DR12, Alametal. 2015) for a large sample of M67 stars radiusofthecluster,weexpectthemembersoftheclustertoshow
coveringevolutionaryphasesrangingfromthesubgiantbranchto adistributionpeaked at themean PM of thecluster and whose
tot
theredclump.Thesedatadonotonlyprovideabetterestimateof widthisgivenbythePMerrorsfromthePPMXLcatalogueandby
[C/N]FDU –henceastrongerconstraint ontheory–butalsoallow theinternalscatterduetotheprogressivedissolutionofthecluster.
one to follow the [C/N] variation during the FDU, and potential Inordertoexcludeobviousoutliersfromthecalculations,weselect
additionalmixingaftertheRGBbumpphase. afirstrangebyeye(redlinesintheplotsinFig.1,panela).Wethen
The plan of the paper is as follows. Section 2 describes the calculatetheweightedmeanPM anditsstandarddeviationwithin
tot
spectroscopicdataandtheclustermembershipanalysis,whilstSec- thisrange.Allstarslyingwithin2σfromthemeanareselectedand
tion 3 compares the spectroscopic abundances with results from consideredinthenextstepoftheanalysis.
theoreticalmodels,andthe[C/N]FDU-agerelationbySalarisetal. Due tothe very large PMtot errors for some of the starsin the
(2015).Adiscussionandconclusionsclosethepaper. PPMXLcatalogue that wedo not want to include inour sample,
we calculate the mean error for the above selection and exclude
starswhosePM errorislargerthanthemeanerror(∼1.8mas/yr)
tot
+1σ(∼1.2mas/yr).
2 DATA
3. Instep2,weconsideronlytheabsolutevalueofthePM and
tot
Inthiswork,weanalysestellarspectraobservedwithintheApache thusloseinformationaboutthedirectionofthePM vector.There-
tot
PointObservatoryGalacticEvolutionExperiment(APOGEE,see fore,itmighthappenthatstarshavingbychancethesameabsolute
Majewskietal. 2015) aspart of theSloan Digital SkySurvey III PM astheaveragepropermotionofthecluster,butmovingina
tot
(SDSS-III, DR12, see Eisensteinetal. 2011; Alametal. 2015). differentdirection,areconsideredmembersofthecluster.Inorder
The APOGEE spectrograph is mounted on the 2.5m Sloan tele- tocorrectforthiseffect,weconsiderinthenextstepthedistribution
scopeandoperatesinthenear-infraredrangefrom1.51to1.7µm oftheanglesofthePM vector(seeFigure1,panelb).Alsohere,
tot
withaspectroscopicresolutionofR ∼ 22,500.Stellarparameters wecalculatethemeanweightedbytheerrors(meanerror∼ 9.88
suchasT ,logg,[M/H],[C/M],[N/M],and[α/M],aswellasthe deg) and the standard deviation and select all objects within 2σ
eff
MNRAS000,1–16(2016)
ProductsofstellarevolutioninM67 3
fromthemean.TheresultofthisselectioncanbeseeninFigure1, over the subgiant branch, the lower red giant branch and the
panelc:allstarsmoveinthesamedirectionintheRA-Decplane, red clump, we find M67 to have an average radial velocity of
withpropermotionvectorsofthesamelength(withintheerrors). RV = 33.806±0.528 km/s, a proper motion in RA×cos(Dec)
4. In the next step we repeat the same procedure as in step 2 of PM = −5.906 ± 0.951 mas/yr and in Dec of PM =
x y
and 3, but with the RV of the stars resulting from the ASPCAP −7.175±1.268 mas/yr (using the corrected proper motions from
pipeline (see Figure 1, panel d). Given that the typical errors of Vickersetal.2016)1,consistentwiththeliterature(seeYadavetal.
the ASPCAP-derived RV for stars in the radius of M67 are very 2008; Gelleretal. 2015 and Bellinietal. 2010b). For the metal-
small(∼ 0.05km/s), wedonotputanyconstraintontheseerrors, licity, we obtain [Fe/H] = 0.08 ± 0.04 dex, slightly higher
contrarilytostep1. than other literature values (see, e.g., Tautvaišieneetal. 2000;
5. Oncewehavemadesurethatthekinematicpropertiesofour Shetrone&Sandquist2000;Yongetal.2005;Randichetal.2006;
selection of stars areconsistent withthem belonging tothe same Paceetal.2008;Önehagetal.2014),butstillconsistentwiththem
kinematicgroup,weproceedwiththephotometriccriteria.Forall withintheerrors. Thecluster-wideparametersof M67 havebeen
selectedstarswecanretrieve2MASSmagnitudes.Weuseforthe summarizedinTable1.
distance modulus (m− M) = 9.64 mag, for the reddening E(B- For the stars selected above, we perform an analysis of the
0
V)=0.023 mag and for the age estimate the range between 3.75 elemental abundances affected by stellar evolution, in particular
and4Gyr(fromBellinietal.2010a)tocalculatethecorresponding [C/Fe]and[N/Fe](TablesA.1and A.2containinformationabout
BaSTIisochrones(seePietrinfernietal.2004)withametallicityof theparameters andabundances of the34selected member stars).
[Fe/H]=0.06dex,includingcoreovershootingduringthemainse- These effects are very well illustrated by Fig. 3 and Fig. 4. The
quence.Wethencomparetheisochronesreddenedaccordingtothe [C/Fe]and[N/Fe]abundancesoftheM67membersvaryalongthe
Cardellietal. (1989) extinctionlaw withthe2MASS magnitudes isochrone,startingwiththemainsequenceabundancesonthesub-
onacolour-magnitudediagram.Weexcludetheobjectsthatdonot giant branch and gradually showing theeffect of theFDUon the
cross any of the two isochrones corresponding to an age of 3.75 red giant branch. In addition, the [C/Fe] and [N/Fe] distributions
and4Gyrwithinthreetimestheirerrorinmagnitudeandcolour. exhibittwopeaksrepresentingthepre-andpost-FDUabundances.
Werepeatthisstepinthecolour-magnitudediagramsdescribedby
Jvs(J−H)andK vs(J−K )(seeFigure1,paneleandf).
s s
We should note that by doing this we exclude blue stragglers
3 RESULTS
thatareprobably membersofM67. Forthepurposeofthisstudy
this is not relevant, since blue stragglers due to their anomalous Before comparing the [C/N] abundances of our sample of clus-
origincannotbetakenintoaccountwheninvestigatingtheeffects ter members with theory, we need to assess the precision of the
ofstellarevolutiononthesurfaceabundancesof[C/N]. DR12/ASPCAP measurements. Holtzmanetal. (2015) discussed
6. As a final step, we consider that stars belonging to a given indetailchecksforhomogeneityofthederivedabundanceswithin
open cluster, even in case of inhomogeneities in the abundances individualclustersintheirsample,andtheiragreementwithprevi-
of certain elements, are expected to share the same [Fe/H] abun- ously measured values. Regarding the first point, they examined
dance.Thus,werepeatthesameprocedureasinsteps2.-4.forthe the DR12/ASPCAP abundances under the assumption that clus-
ironabundance[Fe/H]calculatedwiththepipelineASPCAP,and ters are chemically homogeneous for all elements except carbon
consider all stars within 2σ ≈ 0.1 dex from the mean [Fe/H] to and nitrogen, which are expected to have variations in giants be-
bemembersofthecluster(seeFigure1,panelg).Typical[Fe/H] cause of surface mixing of CN-processed material. Any trend of
errorsforthisselectionare∼0.03dex. chemicalabundanceversusTeff (thatappearsforexampleforFe,
We point out that the DR12/ASPCAP results for dwarf stars Si, Ti) has been corrected for, except for C and N. Regarding
have not been calibrated due to the difficulties of the pipeline the agreement with previously measured values, Holtzmanetal.
in analysing stars with high log(g). Therefore, in step 6 we ex- (2015)concludethattheremaybenon-negligblesystematicoffsets
clude potential cluster members on the main sequence and turn- betweeen DR12/ASPCAP [X/H] abundances and literature abun-
off of M67 from our selection because they do not have entries dancesforsomeelementsatthelevelof0.1-0.2dex.Assuggested
intheDR12/ASPCAPcatalogueofcalibratedabundances.Forthe byHoltzmanetal.(2015),weuseinourwork[X/Fe]abundances
presentstudythisisnotveryrelevant,sinceourpurposeisnotto bysubtracting[Fe/H] fromthe[X/H] abundances. Thisshouldat
obtain a complete sample of M67 members but rather to select a leastpartiallycanceloutpossibletrendsandoffsetsintheresults.
numberofhighlyprobablemembersthatallowustoinvestigatethe It istherefore crucial totest theCN abundances used in our
effects of stellar evolution on the chemical composition of stars. analysis against independent, high-resolution spectroscopic mea-
Therefore, the sample resulting from our pipeline contains stars surements.
from the subgiant and giant branch, which is perfect in order to
analysetheeffectsoftheFDUonthesurfaceabundancesof[C/N].
3.1 Comparisonwithhigh-resolutionspectroscopy
After the cluster members selection, we cleaned the sample
Wefoundthattherearenotmanypublishedhigh-resolutionspec-
frombinariesknown fromtheliterature.Wematched our sample
troscopic estimates of C and N in M67 stars that can help us.
with those of Yakutetal. (2009) and Gelleretal. (2015). We did
Our comparisons will make use of the abundances of turn-off
notfindanystarsincommonwithYakutetal.(2009),butwedid
stars determined by Shetrone&Sandquist (2000), and the abun-
find two binaries from Gelleretal. (2015) among our members,
dances of red clump and post bump red giant branch stars by
asshown inFig.2.Weexclude thesetwo starsfrom our sample,
sincebeingunresolvedbinaries,wecannotusetheirabundancesto
investigatetheeffectsofstellarevolutiononthesurfacechemistry 1 Repeating themembershipanalysis withthenon-corrected propermo-
ofsingleclusterstars. tions fromPPMXLwefind PMx = −7.754±0.952mas/yrand PMy =
Averaging over the resulting sample of 34 stars distributed −5.613±1.276mas/yr.
MNRAS000,1–16(2016)
4 BertelliMottaet al.
60
a) b)
50
60
40
Number30 Number40
20
20
10
0 0
0 10 20 30 40 50 0 50 100 150 200 250 300
PM [mas/yr] θ [deg]
13.0
c) d)
40
12.5
Dec [deg]12.0 Number2300
11.5
10
11.0
131.5 132.0 132.5 133.0 133.5 134.0
RA [deg] 0
10 20 30 40 50 60
6 8 10 12 RV [km/s]
PM magnitude
4
e) 4 f)
6
6
8
J [mag]10 Ks [mag]180
12
12
14
14
16
−0.2 0.0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
(J−H) [mag] (J−Ks) [mag]
14
g)
12
10
mber 8
Nu
6
4
2
0
−0.05 0.00 0.05 0.10 0.15
[Fe/H]
Figure1.Themembershipanalysisstepbystep.Panela:histogramofPMtot(meanerrorPMerr∼1.8mas/yr);panelb:histogramofthepropermotionangle
(θerr ∼ 9.88deg);panelc:representationofthepropermotionvectoroftheselectedstarsintheRA-Decplane;paneld:histogramoftheradialvelocities
(RVerr∼0.05km/s);paneleandf:colour-magnitudediagramforJvs.(J−H)andKsvs.(J−Ks),therejectedstarsarehighlightedinred;panelg:histogram
ofthe[Fe/H]abundancesoftheselectedstars([Fe/H] ∼0.03dex).Thereddashedlinesrepresenttheintervaltakenintoaccountforthecomputationofthe
err
meanvalueandstandarddeviationofeachdistribution.
MNRAS000,1–16(2016)
ProductsofstellarevolutioninM67 5
Table1.GeneralparametersofM67.ThecentralcoordinatesaretakenfromKharchenkoetal.(2013).Thequotedradialvelocity,thepropermotions,and
themetallicityarethemeanvaluesforthesampleofclustermembersselectedinthisstudy.Theturn-offagerange,aswellasthereddeningandthedistance
modulusaretakenfromBellinietal.(2010a).
RA Dec RV e_RV PM_x e_PM_x PM_y e_PM_y Age E(B-V) (m−M)0 [Fe/H] e_[Fe/H]
[hms] [dms] [km/s] [km/s] [mas/yr] [mas/yr] [mas/yr] [mas/yr] [Gyr] [dex] [dex]
08:51:23.4 +11:48:54 33.806 0.528 -5.906 0.951 -7.175 1.268 3.75-4.00 0.023 9.64 0.08 0.04
(J − K ) <0.47). Assessing whether this consistency between
s 0
turn-offandmoreevolvedsubgiantbranchvaluesisaconfirmation
oftheaccuracyofDR12/ASPCAPabundances alsoforthehotter
4
APOGEE data stars in the M67 sample requires a brief discussion about the
Binaries from Geller et al. (2015)
effect of atomic diffusion on the surface abundances of these
6 objects. Atomic diffusion (which usually denotes the combined
effect of gravitational settling, thermal and chemical diffusion,
g] 8 and radiative levitation) can alter the surface abundances of M67
ma stars during the main sequence phase (causing either an increase
Ks [10 or a decrease, depending on the ratio between local gravity and
radiativeaccelerationofthevariousions).Themaximumvariation
comparedtotheinitialabundancesisreachedaroundtheturn-off.
12
During the following subgiant branch phase the deepening of
surface convection slowly restores the surface values back to
14 essentiallytheinitialabundances(see,e.g.,Cassisi&Salaris2013,
andreferencestherein).
0.2 0.4 0.6 0.8 1.0 1.2
(J−Ks) [mag] Michaudetal.(2004)havecalculatedmodelsandisochrones
for M67 stars including the effect of atomic diffusion. Consider-
ing for example the surface Mg abundance, which behaves like
O (Michaudetal. 2004), these models predict a decrease during
Figure 2. Our selection of members for M67 plotted on the BaSTI
themainsequencephase,reachingaminimumaroundtheturn-off,
isochroneofage3.75Gyr.Thegreencirclesrepresentthememberslabeled
about 0.10 dex lower than the initial value, and then an increase
asbinariesinGelleretal.(2015).Thesestarsareexcludedfromallfurther
alongthesubgiantbranchuntilthesurfaceabundancesarerestored
investigations.
toalmosttheinitialvaluesalongtheredgiantbranch.Ontheother
handalsotheFeabundancehasasimilarbehaviour,reachingamin-
imum at the turn-off about 0.08 dex lower that the initial value.
Tautvaišieneetal. (2000) (see Fig. 5 and Table A.3 for details
Abundance ratios with respect to Fe would therefore display an
about the selected stars). Assuming that the surface abundances
evensmallervariationalongtheseevolutionaryphases.
are unchanged between the turn-off and the start of the FDU,
theresultsofShetrone&Sandquist(2000)shouldmatchthesub- Önehagetal. (2014) presented a differential chemical abun-
giant pre-FDU DR121/ASPCAP abundances. Fig. 6 shows the dancestudyofturn-offandmain-sequencestarsrelativetohotsub-
DR12/ASPCAP[C/Fe],[N/Fe]and[O/Fe]valuesasafunctionof giantbranchstarsinM67(theyincludeC,O,andFeamongother
colour (J−K ) , compared to the two independent sets of abun- elements, but not N), which showed differences of their various
s 0
dance determinations mentioned above. The colours of the stars measured[X/H]abundanceratiosoftheorderof0.02dex,suggest-
sampled by Shetrone&Sandquist (2000) and Tautvaišieneetal. ing that diffusion is active (or maybe partially inhibited by some
(2000)arefoundbymatchingthecoordinatesofthestarswiththe competing mechanism) although the evidence is not compelling,
2MASS catalogue. We consider also the oxygen abundances for astheydiscussed. Their sampleof subgiants (about 500 K hotter
reasonsthatwillbecomeclearerinthecourseofthisdiscussion. than the hotter DR12/ASPCAP subgiants) has average values of
The average red clump [C/Fe], [N/Fe], and [O/Fe] abun- [C/Fe]=-0.04±0.05dexand[O/Fe]=-0.04±0.07dex,whichwithin
danceratiosdetermined byTautvaišieneetal.(2000)areequal to theerrorsagreewiththeDR12/ASPCAPmeansubgiantpre-FDU
−0.18±0.02,0.26±0.06and0.01±0.05dex(theerrorbaristhe abundances. Asa conclusion, thereisno strong evidence for any
σdispersionaroundthemean),respectively.Withintheassociated sizeable variation of [C/Fe] and [O/Fe] between turn-off and the
errors they are consistent with the average [C/Fe]=−0.15±0.04, onsetoftheFDU.Theaverage[C/Fe]and[O/Fe]ratiosdetermined
[N/Fe]=0.29±0.02,and[O/Fe]=0.00±0.02dex,whichweobtain byShetrone&Sandquist(2000)forasampleofturn-offstars,and
from the DR12/ASPCAP abundances of the RC. Notice that the thosedeterminedbyÖnehagetal.(2014)forasampleofsubgiant
colours (and T ) of redclump starsoverlap with thecolours of branch stars (hotter than DR12/ASPCAP ones) are both consis-
eff
redgiantbranchstarsattheFDUcompletion. tent withintheerrorswithDR12/ASPCAPpre-FDUabundances.
Regarding the turn-off abundances from On the basis of these comparisons with red clump, turn-off and
Shetrone&Sandquist (2000), the average [C/Fe] and [O/Fe] hotsubgiantbranchabundancesfromindependenthigh-resolution
abundance ratios are equal to −0.01±0.10 and −0.02±0.08 spectroscopy,thereisnoindicationthatDR12/ASPCAP[C/Fe]and
dex. These values do compare well, within the 1σ errors, [O/Fe]abundancespre-andpost-FDUneedzeropointand/ortem-
with the DR12/ASPCAP averages [C/Fe]=0.04±0.07 dex and peraturedependentcorrections.
[O/Fe]=0.02±0.02dexobtainedforpre-FDUobjects(objectswith However,thesituationfor[N/Fe]intheDR12/ASPCAPpre-
MNRAS000,1–16(2016)
6 BertelliMottaet al.
14
6 0.10
12
0.05 10
8
Ks [mag]10 −0.00.005[N/Fe] Number 68
−0.10 4
12
−0.15 2
14 0
0.2 0.4 0.6 0.8 1.0 1.2 −0.4 −0.2 0.0 0.2 0.4
(J−Ks) [mag] [N/Fe]
Figure3.Leftpanel:Colour-magnitudediagramofthememberstarsofM67withanoverplotted3.75Gyrisochrone.Thesymbolsrepresentingthestarsare
colour-codedwiththeir[N/Fe]abundance.Theeffectofthefirstdredge-upcanbeseenclearlyatthetransitionbetweenthesubgiantandthegiantbranch.The
NproducedbytheCNOcycleisbroughttothesurface,thusenhancingitsabundance.Rightpanel:Histogramshowingthe[N/Fe]abundancedistributionof
thememberstarsofM67(meanerror[N/Fe] ∼0.08dex).Thedistributiondoesnotrepresenttheexpectationsforthechemicalcompositionofthemain
err
sequenceofanopencluster,i.e.asinglepeakwithasmallvariation.Theabundancedistributionofthesubgiantsandgiantsshowsinsteadtwopeaksin[N/Fe].
6 0.10 15
0.05
8
Ks [mag]10 −0.00.005[C/Fe] Number10
−0.10 5
12
−0.15
14 0
0.2 0.4 0.6 0.8 1.0 1.2 −0.4 −0.2 0.0 0.2 0.4
(J−Ks) [mag] [C/Fe]
Figure4.Leftpanel: A3.75Gyrisochrone isshownoverplotted onthemembersofM67.Thememberstarsarecolour-coded according totheir[C/Fe]
abundance.IncontrasttoN,theabundanceofCdecreasesduringtheCNOcycle,andasaconsequencethesurfaceabundanceof[C/Fe]afterthefirstdredge-
upisdepleted.Rightpanel:Histogramofthe[C/Fe]abundanceofthememberstarsofM67(meanerror[C/Fe] ∼0.06dex).Alsointhiscase,twopeaksin
err
theabundancedistributionarevisible,althoughnotasdistinctlyasinthecaseof[N/Fe].
FDUsampleisdifferent.ItisclearfromFig.6thatbeforetheon- giantbranch, withoutmodifyingbothCandO.Finally,giventhe
set of the FDU [N/Fe] is not constant along the subgiant branch, approximatelysolarmetallicityofthecluster,andthescaled-solar
as expected from the behaviour of [C/Fe] and [O/Fe], but rather valuesofCandO,itisveryplausibletoassumealsoauniformso-
displays a strong trend with colour (hence T ), i.e., [N/Fe] de- lar[N/Fe]forthepre-FDUabundances.Avalue[N/Fe]=0.0before
eff
creasessteadily withdecreasing colour along theSGB.Themin- theFDUwouldalsobeconsistentwiththeconstraintposedbythe
imum[N/Fe] isequal to∼ −0.3dex, alargedepletion compared abundancesofShetrone&Sandquist(2000).
tothesolarratio.TheabundancesofShetrone&Sandquist(2000)
InsupportofthisinterpretationoftheDR12/ASPCAPresults
arenotveryaccurateandfortwoobjectsonlyupperlimitscouldbe
we refer to Masseron&Gilmore (2015). Studying field subgiant
determined.Nevertheless,theirdataseemtoberulingoutthelow
starsinthethinandthickdiscwithinAPOGEE,theyfoundanoff-
DR12/ASPCAPabundances(Önehagetal.2014donotmeasureN
setof−0.2dexinthe[N/Fe]abundanceatsolarmetallicity,simi-
abundances). In addition, it is hard to imagine a physical mech-
lartowhatweobserveinourselectionofmemberstarsforM67.
anism that changes only the surface N abundance along the sub-
Thissuggeststhatwearedealingwithasystematicprobleminthe
MNRAS000,1–16(2016)
ProductsofstellarevolutioninM67 7
0.4
6 SBG & lower RGB SBG & lower RGB
upper RGB & RC upper RGB & RC
Shetrone & Sandquist (2000) Shetrone & Sandquist (2000)
Tautvaisiene et al. (2000) 0.2 Tautvaisiene et al. (2000)
8
ag] e]
Ks [m10 [C/F 0.0
−0.2
12
14 −0.4
0.2 0.4 0.6 0.8 1.0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
(J−Ks) [mag] (J−Ks) [mag]
0
Figure5.TheplotshowsourselectionofM67membersfromAPOGEE 0.4
togetherwithstarsanalysedinotherworks.Inparticular,weconsiderthe
turn-off stars studied by Shetrone&Sandquist (2000) and the giant and
clumpstarsfromTautvaišieneetal.(2000).
0.2
APOGEEdata,oroftheDR12/ASPCAPpipeline,andnotwitha
e]
peculiarity of M67. What still remains to be clarified is whether N/F 0.0
this problem is confined to the subgiant branch or whether it af- [
fectsallstarswithinAPOGEE.DuetothechangeofthesurfaceC
and N during and after the SGB,it ismoredifficult to determine
−0.2 SBG & lower RGB
deviationsoftheobservedabundancesfromthepredictedonesfor upper RGB & RC
Shetrone & Sandquist (2000)
fieldstarsofwhichwedonotknowthemass.Masseron&Gilmore
Tautvaisiene et al. (2000)
(2015)donotinvestigatetheproblemfurtherandonthebasisofthe
observed depletion in the SGB they shift the [N/Fe] abundances −0.4
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
of the entire sample by +0.2 dex. While for the investigation of (J−Ks) [mag]
0
Masseron&Gilmore(2015)thisdoesnotaffecttheresults,forour
studyitisimportanttoinvestigatewhetherthissystematicoffsetis
real.
0.4
Wehaveshownthatthe[N/Fe]abundancesofoursampleof
redgiantbranchandredclumpstarsfitverywellthoseofstarsin-
dependently observed and analysed by Tautvaišieneetal. (2000).
Thus,wedonotseeanyreasonwhythe[N/Fe]abundancesshould 0.2
beshiftedby0.2dexfortheentiresampleofAPOGEEstars.We
claimthatthereasonforthesystematicoffsetof[N/Fe]abundances
ipnertahtuereAoPfOtGheEEstasrusbggoiainngt rferogmimreedistodubeluteor cthoeloiunrcsr.eTashieng[Nt/eFme-] [O/Fe] 0.0
abundancesintheDR12/ASPCAPpipelinearecalculatedfromCN
molecularlines.Thesebecomeprogressivelyweakerandmoredif-
ficulttoanalyseastheeffectivetemperatureofthestarsincreases, −0.2 SBG & lower RGB
upper RGB & RC
thus leading to untrustworthy results in the subgiant region. We Shetrone & Sandquist (2000)
thereforewillnotconsiderstarsbluerthan(J−Ks) = 0.54mag Tautvaisiene et al. (2000)
0
(theapproximatevalueatwhichtheAPOGEE[N/Fe]abundances −0.4
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
reachthesolarvalue),corresponding toatemperaturehotterthan
(J−Ks) [mag]
0
T ∼5000K.
eff
3.2 Comparisonwiththemodels Figure6.The[C/Fe],[N/Fe],and[O/Fe]abundancesoftheAPOGEEsam-
pleandthosefromtheliteratureareplottedasafunctionofcolour.Thispic-
Afterhavingfoundareasonableexplanationfortheoffsetof[N/Fe] tureconfirmsthattheabundances fromShetrone&Sandquist(2000)and
abundancesinthesubgiantbranchinDR12,wecanproceedwith Tautvaišieneetal.(2000)agreewellwiththeexpectation forsolarvalues
the comparison between the APOGEE data and the predictions beforethefirstdredge-upandwiththevaluesderivedwithDR12/ASPCAP
fromstellarevolutionmodels. afterthefirstdredge-up.While[C/Fe]and[O/Fe]fortheAPOGEEsample
agreeverywellwiththeliteraturevalues,[N/Fe]appearstobedepletedby
∼0.2dexonthesubgiantbranch.
MNRAS000,1–16(2016)
8 BertelliMottaet al.
0.6 0.4
SBG & lower RGB SBG & lower RGB
0.4 upper RGB & RC upper RGB & RC
pMo=s1t .F3D5MUS suntars 0.2 MM==11..3450MMSSuunn
0.2 M=1.40MSun
C/N]−0.0 C/Fe] 0.0
[ [
−0.2
−0.2
−0.4
−0.6 −0.4
0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.3 0.4 0.5 0.6 0.7 0.8 0.9
(J−Ks) [mag] (J−Ks) [mag]
0 0
Figure7.Theplotshowsthe[C/N]abundancesderivedbyDR12/ASPCAP 0.4
asafunctionofthe(J−Ks)0 colourcomparedwithtwomodelscharac-
terised bydifferent masses (solidblack line for1.35M⊙ andgreen dash-
dottedlinefor1.40M⊙).ThelightbluecirclesrepresentstarsoftheRGB
andRCthatareexcludedfromthecomputationofthepost-dredge-up[C/N] 0.2
abundance duetopossibleextramixingintheirinterior. Theorangedia-
mondsarethestarsusedtocalculatethemeanvalueforthepost-dredge-up
[C/N]abundance. N/Fe] 0.0
[
InFig.7,weplotthe[C/N]abundanceobtainedforoursam- −0.2 SBG & lower RGB
upper RGB & RC
pleofM67membersasafunctionof(J−Ks)0colourtogetherwith M=1.35MSun
twodifferentmodelsforstarsofthemasses1.40M and1.35M , M=1.40MSun
⊙ ⊙
[Fe/H]=0.06dex(fromtheBaSTIdatabase).Thesecorrespondto
−0.4
0.3 0.4 0.5 0.6 0.7 0.8 0.9
anageof4and3.75Gyr,respectively, consideredtobetheturn-
(J−Ks) [mag]
off age range of M67 following Bellinietal. (2010a), who also 0
employed the same BaSTI models used in this work. The light-
bluecirclesrepresentstarsthatmighthaveundergoneextramixing
andarethereforenotsuitableforthecomputationofthepost-FDU Figure8.[C/Fe]and[N/Fe]asafunctionofthe(J−Ks)0colourisplotted
[C/N]abundance.Finally,theorangediamondsrepresentthestars togetherwiththerespectivemodelpredictions.WhiletheMSabundanceof
thatweusedtocomputethepost-FDUmean[C/N]abundance,as [C/Fe]derivedwithDR12/ASPCAPisconsistentwithasolarcomposition
wewillexplainlaterinthissection.Weselectedthembasedonthe andwiththemodels,the[N/Fe]abundancebeforetheFDUisdepletedand
doesnotagreewiththemodels.
colour at which the FDU is complete, (J−K ) > 0.6 mag, and
s 0
excluding starsbrighter than thered giant bump luminosity from
themodels,K <8.5mag.
s dictionof [C/N] = −0.4dexfor themodelswithstellarmass
TheDR12/ASPCAPabundancesforstarswith(J−K )>0.54 FDU
s M=1.35M andof[C/N] =−0.43dexforM=1.4M .
magareconsistentwiththepredictionsofthemodelsrepresenting ⊙ FDU ⊙
Fig.9updatesFig.2inSalarisetal.(2015)withthe[C/N]
theagerangeofM67.InFig.8the[C/Fe]and[N/Fe]abundances FDU
ofM67obtainedinthisstudy.Theerrorsareconsiderablysmaller
oftheM67starsandtherespectivemodelsareplottedseparatelyas
andthepost-FDU[C/N] isinbetteragreementwiththemodels.
afunctionofthe(J−K ) colour.Alsointhiscasethedataarein FDU
s 0
goodagreementwiththemodels,ifweexcludethe[N/Fe]deple-
tioninthesubgiantrange.WealsocomparedtheDR12/ASPCAP
3.3 Post-FDUextramixing
resultswithmodelsusingdifferentassumptionregardingenvelope
overshootinginthesubgiantbranch.Wefoundthatallmodelsare Asmentionedabove,weexcludebrightRGBstarsandRCstarsto
consistent with the observational data and that we cannot distin- avoid problems related to possible extra mixing episodes that set
guish between them within the errors (see Salarisetal. 2015 for inaftertheRGBbump.Themeanmolecularweightdiscontinuity
detailsabouttheeffectofenvelopeovershootingon[C/N] ). leftoverbytheFDUisexpectedtoinhibitanyextra-mixingbelow
FDU
Weusefivestarstodeterminethemean[C/N]abundance in theconvective envelope, but whenthisdiscontinuity iserasedaf-
M67 after the FDU (orange diamonds in Fig. 7), which we then tertheRGBbump,extra-mixingprocessesarepossible,asshown
comparewiththemodelsfromSalarisetal.(2015)fortheagees- forexamplebytheevolutionoftheC12/C13 ratioinfieldgiantsat
timation of field stars (see Fig. 9). We obtain an abundance of variousmetallicities(Charbonnel&Balachandran2000,andrefer-
[C/N] = −0.46±0.03 dex, which is consistent with the pre- encestherein).Theseextramixingprocessescanalsodecreasethe
FDU
MNRAS000,1–16(2016)
ProductsofstellarevolutioninM67 9
−0.1
−0.2
−0.3
N]
C/−0.4
[
−0.5
upper RGB & RC
post FDU stars
Tautvaisiene et al. (2000)
−0.6 M=1.35MSun
M=1.40MSun
−0.7
0.5 0.6 0.7 0.8 0.9
(J−Ks) [mag]
0
Figure10.Theplotshowsthe[C/N]abundancesforthepost-FDU(orange
diamonds),andupperRGBandRC(light-bluecircles)APOGEEstarsin
M67,aswellasfortheRGBandRCstarsfromTautvaišieneetal.(2000).
Thedataarecomparedtothemodelsforstellarmassesof1.35M⊙ (black
solidline)and1.4M⊙(greendashed-dottedline).
Figure9.AsFig.2fromSalarisetal.(2015):Theplotshowsthetheoret-
icalvalueof[C/N] asafunctionofagefordifferentmetallicities.The
FDU
curvesshowncomputedfordifferent[M/H]:-2.27dex(blackdottedline),
guethatmorehigh-resolutionspectroscopicobservationsofupper
-1.49dex(blueshortdashed),-1.27dex(lightbluelongdashed),-0.66dex
(blackdot-dashed),-0.35dex(greensolid),0.06dex(redsolid)and0.26 RGBstarsinM67arenecessarytoassesstheroleofextramixing
dex(blacklongdashed),respectively. Valuesforthe[C/N] ofseveral insuchstars.
FDU
Galacticopenclustersandhalofieldstarsavailableintheliteraturearealso
plotted(seeSalarisetal.2015fordetails),togetherwiththenew[C/N]
FDU
valueforM67obtainedinthecurrentwork.
4 CONCLUSIONS
[C/N] ratio compared to the FDU value (N increases and C de- Inthisstudyweanalysethepre-andpost-FDU[C/N]abundance
creases),asseen,e.g.,infieldhalostars(Grattonetal.2000). intheoldopenclusterM67anduseitasacalibratorfortheage-
OursampleofM67memberscontainssevenredclumpstars datingoffieldstars(seeSalarisetal.2015).Inordertoassessthe
as well as one bright red giant branch (or possibly AGB) star. accuracy of our analysis, we compare data obtained within the
Fig. 10 shows that the [C/N] abundances of the red clump stars APOGEEDR12surveyandanalysedbythestellarparametersand
are consistent with the post-FDU [C/N] abundance and do not abundancespipelineASPCAPwith[C/Fe]and[N/Fe]valuesfrom
appear to be affected by any extra mixing process. On the other the literature for turn-off (Shetrone&Sandquist 2000), red giant
hand, the[C/N] of the upper RGB star (K < 8.5mag) ishigher branch,andredclump(Tautvaišieneetal.2000)starsinM67.
s
than [C/N] and the value predicted by the models, contrarily We find the [C/Fe] abundances from DR12/ASPCAP to be
FDU
to the drop in [C/N] expected from extra mixing. If we add the in very good agreement with the literature, both before and after
upperRGBandRCstarsindependentlyobservedandanalysedby the FDU. Less straightforward are the results for [N/Fe]. While
Tautvaišieneetal. (2000) tothis sample, we see thispicture con- the values of our sample after the FDU are consistent within the
firmed(see also Fig6in Tautvaišieneetal. 2000): RC starshave errors with the results from Tautvaišieneetal. (2000), the abun-
[C/N] abundances comparable to the [C/N] value, while the dances in the subgiant region appear systematically depleted and
FDU
[C/N]inupperRGBstarsisslightlyhigher(seeFig.10).Thissug- evenshow atrendwithcolour.Takingintoaccount theresultsby
geststhatattheageandmetallicityofM67extramixingafterthe Masseron&Gilmore(2015),weassertthatthedepletionof[N/Fe]
redgiantbumpisnotplayingasignificantroleandthattheremight inthesubgiant region isaproblem common totheentire sample
evenbeadifferentprocesstakingplacethathastheoppositeeffect of APOGEE stars, and not a characteristic of M67. Possibly, the
onthesurfaceabundancesofupperRGBstars. offsetshownbythedataisaresultofthedecreasingprecisionof
We note that the study by Soutoetal. (2016), analysing the the[N/Fe]measurementsinDR12/ASPCAPwithincreasingtem-
chemical composition of six RC and six lower RGB stars in the peratureduetotheweakeningofthemolecularbandsusedforthe
∼2GyroldopenclusterNGC2420fromAPOGEEspectrafound abundancedetermination.
similarresults.TheRCstarsinNGC2420showthesame13C/14N The problem of the systematic underestimation of [N/Fe] in
asthelowerRGBstars,suggestingthatnoextra-mixingistaking thesub-giantbranchseemstohavebeenovercomeintherecently
placeaftertheluminositybump. published SDSSDR13 (see Appendix). This, however, happened
Unfortunately, our sample of upper RGB stars (also taking on the expense of a much larger scatter in the overlall [N/Fe],
intoaccountthestarsfromTautvaišieneetal.2000)isverysmall, [C/Fe],and[O/Fe]abundances. Inaddition,thepost-FDU[N/Fe]
sothatnostatisticallysignificantconclusioncanbedrawn.Wear- abundanceisnotconsistentanymorewiththeotherspectroscopic
MNRAS000,1–16(2016)
10 BertelliMottaet al.
studies described in this work and with the evolutionary models KharchenkoN.V.,PiskunovA.E.,SchilbachE.,RöserS.,ScholzR.-D.,
(seeAppendixfordetails). 2013,A&A,558,A53
Finally, we compare the DR12/ASPCAP [C/N] abundances MajewskiS.R.,etal.,2015,preprint, (arXiv:1509.05420)
withstellarevolutionarymodelswhichincludecore-overshooting MartigM.,etal.,2016,MNRAS,456,3655
MasseronT.,GilmoreG.,2015,MNRAS,453,1855
and find that the observational data match the theoretical predic-
MichaudG.,RichardO.,RicherJ.,VandenBergD.A.,2004,ApJ,606,452
tionsverywellwithintheerrorswhentakingintoaccountthefind-
ÖnehagA.,GustafssonB.,KornA.,2014,A&A,562,A102
ingsdescribed intheabove paragraph. Weobtain anew estimate
PaceG.,PasquiniL.,FrançoisP.,2008,A&A,489,403
forthemeanpost-FDU[C/N]inM67of[C/N] =−0.46±0.03
FDU Pichardo B., Moreno E., Allen C., Bedin L. R., Bellini A., Pasquini L.,
which is in good agreement and thus provides a stringent obser- 2012,AJ,143,73
vationaltestofthe[C/N]FDU −t−[M/H]relationofSalarisetal. PietrinferniA.,CassisiS.,SalarisM.,CastelliF.,2004,ApJ,612,168
(2015)(seeFig.9asanupdatedversionofFig.2inSalarisetal. RandichS.,SestitoP.,PrimasF.,PallaviciniR.,PasquiniL.,2006,A&A,
2015). 450,557
RoeserS.,DemleitnerM.,SchilbachE.,2010,AJ,139,2440
SDSSCollaborationetal.,2016,preprint, (arXiv:1608.02013)
Salaris M., Pietrinferni A., Piersimoni A. M., Cassisi S., 2015, A&A,
ACKNOWLEDGEMENTS 583,A87
SandersW.L.,1977,A&AS,27,89
This work was supported by Sonderforschungsbereich SFB 881
SarajediniA.,DotterA.,KirkpatrickA.,2009,ApJ,698,1872
"TheMilkyWaySystem"(subprojectB5)oftheGermanResearch
ShetroneM.D.,SandquistE.L.,2000,AJ,120,1913
Foundation(DFG).Theauthorsthanktheanonymousreferee,Jon SoutoD.,etal.,2016,preprint, (arXiv:1607.06102)
Holtzman,RicardoSchiavon,SabineReffert,andCorradoBoeche TautvaišieneG.,EdvardssonB.,TuominenI.,IlyinI.,2000,A&A,360,499
for helpful discussions. Maurizio Salaris acknowledges support VickersJ.J.,RöserS.,GrebelE.K.,2016,AJ,151,99
fromtheSFBvisitorprogramme.ThisworkhasmadeuseofBaSTI YadavR.K.S.,etal.,2008,A&A,484,609
webtools(http://albione.oa-teramo.inaf.it). YakutK.,etal.,2009,A&A,503,165
Funding for SDSS-III has been provided by the Alfred P. YongD.,CarneyB.W.,TeixeradeAlmeidaM.L.,2005,AJ,130,597
SloanFoundation, theParticipatingInstitutions,theNationalSci-
enceFoundation,andtheU.S.DepartmentofEnergyOfficeofSci-
ence.TheSDSS-IIIwebsiteishttp://www.sdss3.org/.
SDSS-IIIismanagedbytheAstrophysicalResearchConsor-
tium for the Participating Institutions of the SDSS-IIICollabora-
tionincludingtheUniversityofArizona,theBrazilianParticipation
Group,BrookhavenNationalLaboratory,CarnegieMellonUniver-
sity,UniversityofFlorida,theFrenchParticipationGroup,theGer-
manParticipationGroup,HarvardUniversity,theInstitutodeAs-
trofisica de Canarias, the Michigan State/Notre Dame/JINA Par-
ticipation Group, Johns Hopkins University, Lawrence Berkeley
National Laboratory, Max PlanckInstitutefor Astrophysics, Max
Planck Institute for Extraterrestrial Physics, New Mexico State
University,New YorkUniversity,OhioStateUniversity,Pennsyl-
vania State University, University of Portsmouth, Princeton Uni-
versity,theSpanishParticipationGroup,UniversityofTokyo,Uni-
versityofUtah,VanderbiltUniversity,UniversityofVirginia,Uni-
versityofWashington,andYaleUniversity.
REFERENCES
AlamS.,etal.,2015,ApJS,219,12
BelliniA.,etal.,2010a,A&A,513,A50
BelliniA.,BedinL.R.,PichardoB.,MorenoE.,AllenC.,PiottoG.,An-
dersonJ.,2010b,A&A,513,A51
CardelliJ.A.,ClaytonG.C.,MathisJ.S.,1989,ApJ,345,245
Cassisi S., Salaris M., 2013, Old Stellar Populations: How to Study the
FossilRecordofGalaxyFormation.Wiley,NewYork
CharbonnelC.,BalachandranS.C.,2000,A&A,359,563
EisensteinD.J.,etal.,2011,AJ,142,72
FagerholmE.,1906,PhDthesis,,UppsalaUniv.,(1906)
Frolov V. N., Ananyevskaya Y. K., 1986, Astronomicheskij Tsirkulyar,
1432,4
GarcíaPérezA.E.,etal.,2016,AJ,151,144
GellerA.M.,LathamD.W.,MathieuR.D.,2015,AJ,150,97
GilmoreG.,etal.,2012,TheMessenger,147,25
GilroyK.K.,BrownJ.A.,1991,ApJ,371,578
GrattonR.G.,SnedenC.,CarrettaE.,BragagliaA.,2000,A&A,354,169
HoltzmanJ.A.,etal.,2015,AJ,150,148
MNRAS000,1–16(2016)