Sc and neutron-capture abundances in Galactic low- and high-alpha field halo stars

0.44 MB·

by  C. K. Fishlock

Mon.Not.R.Astron.Soc.000,1–11(2013) Printed11January2017 (MNLATEXstylefilev2.2)
Sc and neutron-capture abundances in Galactic low- and high-α
field halo stars(cid:63)
C. K. Fishlock1, D. Yong1†, A. I. Karakas1,2, A. Alves-Brito3, J. Mele´ndez4,
7 P. E. Nissen5, C. Kobayashi6, and A. R. Casey7
1
0 1ResearchSchoolofAstronomy&Astrophysics,AustralianNationalUniversity,CanberraACT2611,Australia
2 2MonashCentreforAstrophysics,SchoolofPhysicsandAstronomy,MonashUniversity,VIC3800,Australia
3InstitutodeFisica,UniversidadeFederaldoRioGrandedoSul,PortoAlegre,RS,Brazil
n
4DepartamentodeAstronomiadoIAG/USP,UniversidadedeSa˜oPaulo,Brazil
a
5StellarAstrophysicsCentre,DepartmentofPhysicsandAstronomy,AarhusUniversity,NyMunkegade120,DK-8000AarhusC,Denmark
J
6CentreforAstrophysicsResearch,ScienceandTechnologyResearchInstitute,UniversityofHertfordshire,AL109AB,UK
0 7InstituteofAstronomy,UniversityofCambridge,Cambridge,CB30HA,UK
1

]
R Accepted....Received...;inoriginalform...
S
.
h ABSTRACT
p
-
Wedeterminerelativeabundanceratiosfortheneutron-captureelementsZr,La,Ce,Nd,
o
r and Eu for a sample of 27 Galactic dwarf stars with −1.5< [Fe/H] <−0.8. We also mea-
t sure the iron-peak element Sc. These stars separate into three populations (low- and high-α
s
a haloandthick-discstars)basedonthe[α/Fe]abundanceratioandtheirkinematicsasdiscov-
[ ered by Nissen & Schuster. We find differences between the low- and high-α groups in the

abundanceratiosof[Sc/Fe],[Zr/Fe],[La/Zr],[Y/Eu],and[Ba/Eu]whenincludingYandBa
1
v fromNissen&Schuster.Forallratiosexcept[La/Zr],thelow-α starshavealowerabundance
3 compared to the high-α stars. The low-α stars display the same abundance patterns of high
2 [Ba/Y]andlow[Y/Eu]asobservedinpresent-daydwarfspheroidalgalaxies,althoughwith
4 smaller abundance differences, when compared to the high-α stars. These distinct chemical
2 patternshavebeenattributedtodifferencesinthestarformationratebetweenthetwopopula-
0 tionsandthecontributionoflow-metallicity,low-massasymptoticgiantbranch(AGB)stars
1. to the low-α population. By comparing the low-α population with AGB stellar models, we
0 placeconstraintsonthemassrangeoftheAGBstars.
7
Keywords: Stars:abundances–Galaxy:halo
1
:
v
i
X
r 1 INTRODUCTION liest times (e.g., Eggen, Lynden-Bell & Sandage 1962; Searle &
a
Zinn1978;Vennetal.2004;Frebel&Norris2015).Duetotheir
UnderstandingtheearliestphasesoftheassemblyofourGalaxy,
longlifetimes,FGK-typedwarfstarsareidealfortracingtheevo-
the Milky Way, remains a key issue in contemporary astronomy.
lutionoftheGalaxy.Theatmospheresofdwarfstarscloselymir-
Cosmological simulations predict that mergers and accretion of
ror the chemical composition of the gas clouds from which they
satellite dwarf galaxies may play an important role through hi-
formed.Thismeansthatspectroscopicstudiesofthese‘stellarfos-
erarchical structure formation (e.g., Bullock & Johnston 2005;
sils’willenableustoprobethechemicalandkinematichistoryof
Pillepich et al. 2014). Our Galaxy is no exception. Observational
ourGalaxyovercosmictime.
support for the hierarchical nature of Galactic formation, and the
Comparison of chemical abundances between the Galactic
role dwarf galaxies play in that process, comes from the accre-
haloandsatellitedwarfspheroidalgalaxies(dSph)revealsdiffer-
tionoftheSagittariusdwarfgalaxy(Ibata,Gilmore&Irwin1994)
encesintheirchemicalcomposition,mostnotablyinthe[α/Fe]ra-
andnumeroustidalstreamsobservedintheGalactichalo(e.g.,Be-
tio(e.g.,Shetrone,Coˆte´&Sargent2001;Tolstoyetal.2003;Venn
lokurovetal.2006).
etal.2004;Tolstoy,Hill&Tosi2009).Onaverage,dSphstarswith
The chemical abundances and kinematics of Galactic halo
[Fe/H]>−2havelower[α/Fe]thantheGalaxy.The[α/Fe]ratio
starsholdvaluablecluesconcerningGalacticformationattheear- ∼
dependsontherelativenumberofTypeIatoTypeIIsupernovae,
andthisquantityissensitivetothestarformationrate(e.g.,Tins-
(cid:63) Thispaperincludesdatagatheredwiththe6.5meterMagellanTelescopes ley1979;Matteucci&Greggio1986;Wheeler,Sneden&Truran
locatedatLasCampanasObservatory,Chile. 1989; Nomoto, Kobayashi & Tominaga 2013). The lower [α/Fe]
† [email protected] observedindSphwouldthereforesuggestaslowerstarformation
(cid:13)c 2013RAS
2 Fishlocketal.
raterelativetotheGalaxy.Differencesinthe[α/Fe]ratiobetween precision,instarslikelytohavebeenaccretedfromdwarfgalax-
the halo and present-day dwarf galaxies preclude the continuous ies.Starscurrentlyindwarfgalaxiesareveryfaintsuchthathigh
merging of such systems although it may be that the mergers of precision chemical abundance studies cannot be conducted with
massivedwarfgalaxiesoccurredatveryearlytimesintheforma- existing facilities. The aim of this study is to investigate the role
tionoftheGalactichalo(Vennetal.2004). ofAGBstarsinthechemicalenrichmentoftheGalactichaloand
Although a handful of α-poor stars were known to exist in dSph.Specifically,webuildupontheworkofNS10andNS11by
theGalactichalo(e.g.,Carneyetal.1997;King1997;Preston& studying additional s-process elements (Zr, La, Ce, and Nd). We
Sneden 2000; Ivans et al. 2003), the study by Nissen & Schuster also measure, for the first time with this unique sample, a rapid
(2010)(hereafterNS10)providedthefirstcompellingevidencethat neutron-captureprocess(r-process)element(Eu).Finally,wemea-
the halo hosts two populations that are well separated in [α/Fe]. sureScwhichhasbeenobservedtobehavelikeanαelement(e.g.,
NS10 studied 94 halo dwarf stars in the solar neighbourhood us- Nissenetal.2000).
ingacarefuldifferentialanalysisresultinginabundanceerrorsfor
[α/Fe] of only 0.02 dex. They argued that the low-α stars have
kinematicsconsistentwithaccretionfromasatellitegalaxywitha
2 SAMPLESELECTIONANDOBSERVATIONS
slowerstarformationratethanthehigh-α populationwhichmost
likelyformedin-situintheGalaxy.Additionally,NS10foundthat Our observations focused on obtaining high quality blue spectra
thelow-andhigh-αpopulationsseparateinthe[Na/Fe]vs.[Ni/Fe] forstarsintheNS10sample.IntheanalysesofNS10andNS11,
abundanceplane.Noseparationwasseenfor[Cr/Fe]. only lines redder than 4700 A˚ were measured. Our targets were
The study of NS10 was extended by Ram´ırez, Mele´ndez & selectedtoincludebothlow-andhigh-αstarsacrossthemetallicity
Chaname´ (2012). Through an analysis of the 777 nm O I triplet rangeof−1.5<[Fe/H]<−0.8.Thesampleconsistsof13low-
lines, they found that the low-α stars have a lower [O/Fe] abun- α stars,6high-α stars,and8thick-disc(TD)stars,includingthe
dancecomparedtothehigh-αandthick-discstars.Thelow-αstars tworeferenceTDstarsusedinthestudiesbyNS10andNS11,HD
alsoshowacosmicstar-to-starscatterin[O/Fe]atagiven[Fe/H]. 22879andHD76932.
Before continuing, it is important to recognise that any at- ProgrammestarswereobservedusingtheMagellanInamori
temptstoinferthechemicalenrichmenthistoryofastellarpopula- Kyocera Echelle (MIKE) spectrograph (Bernstein et al. 2003) on
tionrequiresadetailedunderstandingofthenucleosyntheticorigin the Magellan (Clay) telescope during February 2011 and June
of individual elements (Nissen 2013). Low- to intermediate-mass 2011.Weusedthe0.35”slitsizewhichprovidesaresolvingpower
starsundergorichnucleosynthesisoncetheyreachtheasymptotic of83,000intheblue.Typicalexposuretimesrangedfrom500to
giantbranch(AGB)phase.Ithasbeenobservationallyconfirmed 5000 seconds per target and the signal-to-noise ratio per pixel of
that the slow neutron-capture process (s-process) takes place in thespectrawasapproximately400at4500A˚.Thewavelengthcov-
AGBstars(e.g.,Wallersteinetal.1997;Busso,Gallino&Wasser- erageprovidedbyMIKEis3350to9000A˚.Datareductionwasac-
burg 1999). The s-process is responsible for the production of complishedusingIRAF1 withtheMIKEMTOOLS2 package.Con-
aroundhalfoftheabundanceoftheheavyelementsbeyondiron. tinuum fitting was performed using the Spectroscopy Made Hard
Throughstrongmassloss,AGBstarsejectenrichedmaterialinto (SMH) package as used in Casey et al. (2014) and described in
the interstellar medium and pollute the next generation of stars Casey(2014).
(Herwig2005;Karakas&Lattanzio2014).
AGBstarsmayplayanimportantroleintheoriginofchem-
icalabundancedifferencesbetweentheGalaxyandnearbydwarf
galaxies.Inadditionto[α/Fe],differencesintheratiosof[Ba/Y] 3 ABUNDANCES
and[Y/Eu]arefoundbetweendSphandtheGalaxy(e.g.,Tolstoy,
FollowingNS10,weperformadifferentialline-by-lineanalysisrel-
Hill&Tosi2009).Thehigh[Ba/Y]observedindSphwasattributed
ativetotheTDstarHD22879.Thisapproachenablesustoobtain
tolow-massAGBstarsenrichingtheinterstellarmediumasare-
high-precision abundance ratios which are necessary in identify-
sult of the slower chemical evolution. To explain the low [Y/Eu]
ingpossibleseparationsbetweenthelow-andhigh-α populations.
observed,Vennetal.(2004)suggestedthatmetal-poorAGBstars
Themodelatmospheresusedintheanalysiswereone-dimensional,
didnotcontributeasmuchY(comparedtomoremetal-richAGB
plane parallel, local thermodynamic equilibrium (LTE) ATLAS9
stars) because the production of second s-process peak elements
models by Castelli & Kurucz (2003). Interpolation between T ,
suchasBaandLawerefavouredoverfirsts-processpeakelements eff
logg, [Fe/H], and [α/Fe] was required to produce model atmo-
suchasZrandY(Bussoetal.2001).
spheres.Thestellarparameters,T ,ξ ,andlogg,aswellasthe
eff turb
Nissen&Schuster(2011)(hereafterNS11)determinedabun-
abundanceratiosof[Fe/H]and[α/Fe]weretakenfromNS10and
dancesfortwoneutron-captureelementsproducedbythes-process
NS11(seeTable1).Foracompletediscussiononthedetermination
in AGB stars, Ba and Y (along with the iron-peak elements Mn,
oftheseparametersseeNS10andNS11.Weusestellarparameters
Cu, and Zn). They discovered that the low- and high-α popula-
from NS10 without the correction for effective temperature of ∼
tionscouldbeseparatedin[Cu/Fe]and[Zn/Fe].Moreimportantly,
100KfromNS11.Giventhedifferentialnatureoftheabundance
NS11foundthatthe[Ba/Y]ratiodiffersbetweenthetwopopula-
analysisandthatmoststarshavefairlysimilarstellarparameters,a
tions with the low-α stars showing a higher [Ba/Y] value. Those
chemicalabundancepatternsaresimilartothatfoundindSphby
Vennetal.(2004)andreinforcesthehypothesisthatthelow-αstars
1 IRAF is distributed by the National Optical Astronomy Observatory,
arelikelytohavebeenaccretedfromdSph.
whichisoperatedbytheAssociationofUniversitiesforResearchinAs-
TheNS10sampleprovidesauniqueopportunitytostudyaddi-
tronomy(AURA)underacooperativeagreementwiththeNationalScience
tionalneutron-captureelementabundancesinthelow-andhigh-α Foundation.
populations.Suchananalysiswillenableustoexaminetheabun- 2 www.lco.cl/telescopes-information/magellan/
dancesforalargerrangeofneutron-captureelements,withhigh- instruments/mike/iraf-tools/iraf-mtools-package
(cid:13)c 2013RAS,MNRAS000,1–11
Neutron-captureandScabundancesoffieldhalostars 3
Table1.Atmosphericparametersandabundanceratiosfortheprogramme 90
stars(takenfromNS10andNS11).
µ= 0.16, σ= 2.18
80
70
ID Teff logg ξturb [Fe/H] [α/Fe] Classa 60
(K) (cgs) (kms−1) )
m 50
CD-453283 5597 4.55 1.0 −0.91 0.12 low-α (S
CD-610282 5759 4.31 1.3 −1.23 0.22 low-α WN40
G16-20 5625 3.64 1.5 −1.42 0.26 (low-α) E 30
G18-28 5372 4.41 1.0 −0.83 0.31 high-α
G18-39 6040 4.21 1.5 −1.39 0.34 high-α 20
G31-55 5638 4.30 1.4 −1.10 0.29 high-α
10
G56-30 5830 4.26 1.3 −0.89 0.11 low-α
G56-36 5933 4.28 1.4 −0.94 0.20 low-α 0
G66-22 5236 4.41 0.9 −0.86 0.12 low-α 0 10 20 30 40 50 60 70 80 90
10
G170-56 5994 4.12 1.5 −0.92 0.17 low-α
)
HD3567 6051 4.02 1.5 −1.16 0.21 low-α 5
m
HD22879 5759 4.25 1.3 −0.85 0.31 TD ( 0
W
HD25704 5868 4.26 1.4 −0.85 0.24 TD E 5
∆
HD76932 5877 4.13 1.4 −0.87 0.29 TD 10
HD103723 5938 4.19 1.2 −0.80 0.14 low-α 0 10 20 30 40 50 60 70 80 90
HD105004 5754 4.30 1.2 −0.82 0.14 low-α
EW (m )
HD111980 5778 3.96 1.5 −1.08 0.34 high-α SMH
HD120559 5412 4.50 1.1 −0.89 0.30 TD
HD126681 5507 4.45 1.2 −1.17 0.35 TD Figure 1. Comparison between EW measurements for HD 22879 from
HD179626 5850 4.13 1.6 −1.00 0.32 high-α NS10 and NS11 (EWNS) and by using SMH (EWSMH, this work). The
HD189558 5617 3.80 1.4 −1.12 0.35 TD dashedlineshowstheone-to-onecorrespondenceline.Thebottompanel
HD193901 5650 4.36 1.2 −1.07 0.17 low-α showsthedifferencebetweenthetwoanalyses.
HD194598 5942 4.33 1.4 −1.09 0.20 low-α
HD199289 5810 4.28 1.3 −1.04 0.30 TD
HD205650 5698 4.32 1.3 −1.17 0.30 TD
HD219617 5862 4.28 1.5 −1.45 0.28 (low-α) Table2.EWmeasurements(inmA˚)ofZrII,CeII,andNdIIforeachstar.
HD230409 5318 4.54 1.1 −0.85 0.27 high-α
aForstarswith[Fe/H]<−1.4,theclassificationisuncertain(NS10). ID ZrII CeII CeII NdII
4208.98A˚ 4562.36A˚ 4628.16A˚ 4706.54A˚
CD-453283 21.3 7.6 5.3 7.0
systematicshiftof100Kislikelytohaveaminimaleffectonthe CD-610282 17.7 5.2 2.7 4.0
finaldifferentialabundances. G16-20 22.1 6.6 4.1 4.0
ElementalabundancesforZr,Ce,andNdwerederivedfrom G18-28 29.1 10.3 ... 9.4
theanalysisofequivalentwidths(EWs)usingSMH.Forlineswith G18-39 17.3 ... 2.0 3.0
EW<60mA˚,aGaussianprofilewasfittedtotheline.Forlines G31-55 23.8 4.9 3.6 4.6
G56-30 23.1 6.6 7.2 2.9
with EW larger than 60 mA˚, a Voigt profile was used. Figure 1
G56-36 26.1 6.7 4.4 5.4
presentsacomparisonof85EWmeasurements(excludingFeIand
G66-22 26.2 ... 8.9 10.6
FeII)ofNS10andNS11totheEWsmeasuredusingSMHwiththe G170-56 23.4 6.8 3.4 4.2
samelinelistforHD22879.ThemeanEWdifferencebetweenthe HD3567 21.3 4.9 5.3 4.0
twomethodsis0.16±0.24mA˚ (σ =2.18mA˚). HD22879 31.4 8.2 6.5 7.9
WeusetheatomicdatacompiledbyYongetal.(2005)forthe HD25704 22.4 6.5 6.1 7.1
measured absorption lines. The abundance of Zr is derived from HD76932 34.3 8.4 5.1 7.9
the EW of the Zr II line at 4208.98 A˚ which is unblended with HD103723 27.5 9.4 7.4 7.7
otherlines,allowingforareliablemeasurement.TheZrIIlineis HD105004 24.6 ... 9.4 5.0
HD111980 37.4 10.6 6.5 6.6
detectedinallstarswithallEWmeasurementsbeinggreaterthan
HD120559 23.8 ... 4.1 7.5
10mA˚ (seeTable2).Ceabundanceshavebeendeterminedusing
HD126681 28.9 7.4 5.4 5.0
the EWs of two Ce II lines at 4562.36 A˚ and 4628.16 A˚. Seven
HD179626 27.6 5.9 4.1 4.1
starsonlyhaveoneCeIIlineavailableformeasurement:fivestars
HD189558 39.3 10.9 8.9 7.1
donothaveameasurable4562.36A˚ linewhereastwostarsdonot HD193901 17.4 5.2 ... 5.3
haveameasurable4628.16A˚ line(seeTable2).Forallbutthree HD194598 15.8 3.6 3.2 2.9
Celines,theEWislessthan10mA˚. HD199289 19.1 3.3 2.5 3.6
NdabundanceshavebeendeterminedusingtheNdIIlineat HD205650 19.1 4.6 3.4 3.0
4706.54A˚.TheEWmeasurementsarelessthan10mA˚ exceptin HD219617 10.7 ... 3.3 2.0a
twostars(G66-22andHD230409).Wearbitrarilysetaminimum HD230409 26.2 8.2 5.8 10.5
reliableEWmeasurementlimitof2mA˚.Onestar,themostmetal- aUpperlimit
poorinthesample(HD219617),fallsbelowour2mA˚ limitand
thereforeweadoptanupperlimitof2mA˚ fortheNdEW.
(cid:13)c 2013RAS,MNRAS000,1–11
4 Fishlocketal.
Table3.AdoptedsolarelementalabundancesfromAsplundetal.(2009).
1.0 Sc
Element Z log(X/H)+12
0.9 Sca 21 3.05
x
u Zr 40 2.58
Fl
0.8 La 57 1.10
Ce 58 1.58
Nd 60 1.42
0.7
Eu 63 0.52
aMeteoriticvalue.
5526.4 5526.6 5526.8 5527.0 5527.2
3.1 Zr
1.0 Eu
The neutron-capture element Zr is predominately produced by
AGBstarsandisafirsts-processpeakelement.Thes-processcon-
0.9
x tributiontoZrintheSolarSystemis∼80%(e.g.,Simmereretal.
u
Fl 2004; Sneden, Cowan & Gallino 2008; Bisterzo et al. 2014). By
0.8
determiningtheabundanceofZr,weexpandonthemeasurements
ofY,anotherfirsts-processpeakelementwhoseabundanceswere
0.7 previouslyobtainedbyNS11.
Figure3(a)shows[Zr/Fe]asafunctionof[Fe/H].Thedisper-
4129.0 4129.2 4129.4 4129.6 4129.8 4130.0 sionin[Zr/Fe]is∼0.12dexwhichislargerthantheerrorof0.04
dex.Whenconsideringthetwopopulations,thescatterin[Zr/Fe]is
Wavelength ( )
smallerforthelow-αstarswithadispersionof0.06dexcompared
to a dispersion of 0.13 dex when combining the high-α and TD
Figure 2. Synthetic spectrum fit (solid line) to the observations (black stars.Inadditiontothedifferenceinthedispersionof[Zr/Fe],the
points)oftheSclineat5526A˚ (top)andtheEulineat4129A˚ (bottom)
low-α stars have a systematically lower [Zr/Fe] abundance com-
forHD22879.Thedashedlinesshowsyntheticspectrawith[Sc/Fe]and
paredtothehigh-αandTDstars.
[Eu/Fe]of±0.1.
This possible separation in the high- and low-α populations
is in agreement with the slight separation in abundance measure-
mentsof[Y/Fe]forthetwopopulationsseeninNS11.AsZrand
Yarebothfirst-peaks-processelements,theyareexpectedtofol-
lowthesametrend.Figure4demonstratesthedifferenceintheZr
abundance for a low-α star (HD 194598) and a high-α star (HD
111980) at nearly the same metallicity (see Table 1). The low-α
ForSc,La,andEu,wefitsyntheticspectrageneratedbythe
starhasaweakerZrIIlinethanthehigh-α star.Whiletheeffec-
LTEspectrumsynthesisprogramMOOG(v.2013,Sneden1973)to
tivetemperaturediffersbetweenthesetwostarsbyapproximately
determineabundances.Theseelementsrequiretheconsiderationof
150K,allotherlinesinthisregionexhibitverysimilarstrengths.
hyperfinesplitting(HFS)and/orisotopicshifts.Figure2illustrates
A comparison between [Y/Fe] as measured by NS11 and
thefitofthesyntheticspectrumtotheobservedspectrumforthe
[Zr/Fe]ispresentedinFigure5.TheabundanceofZrfollowsthe
ScandEulinesofHD22879.Thebestfittingsyntheticspectrum
wasdeterminedusing χ2 minimisation.ForSc,theloggf values abundanceofYwhichisexpectedifYandZrhaveacommonori-
gin.Whilethereisasystematicoffsetbetween[Y/Fe]and[Zr/Fe]
ofNissenetal.(2000)wereutilisedforthe5526.81A˚ absorption
whichmaybeduetoerrorsinthegf valuesofthelines,theimpor-
line.LawasdeterminedusingoneLaIIlineat4086.71A˚ withlog
tantpointisthattheslopeofthedataiscloseto1.Thehigh-αstars
gf valuesfromLawler,Bonvallet&Sneden(2001).FortheEuII
haveatightercorrelationbetween[Y/Fe]and[Zr/Fe]comparedto
spectrallineat4129.72A˚ weusetheloggf valuesfromLawler
thelow-αstars.
etal.(2001)andsolarisotoperatiosfromLodders(2003).
We adopt solar elemental abundances from Asplund et al.
(2009)asdetailedinTable3.Sinceweperformadifferentialanal-
3.2 La,Ce,andNd
ysis to measure abundances, the choice of solar abundance only
shiftsthestellarabundancesupordownanddoesnotaffecttherel- We extend upon the Ba measurements of NS11 by determining
ativeabundances.Toreiterate,weareprimarilyinterestedinrela- abundances for other elements belonging to the second s-process
tiveabundances,andthereforeabundancedifferences,betweenthe peak,namelyLa,Ce,andNd.
low-andhigh-αpopulations.TheabundancesaregiveninTable4. Thes-processcontributiontoLaintheSolarSystemis75%
Weestimatederrorsusingthestellarparameteruncertainties andthereareanumberofadvantageswithusingLaasans-process
from NS10 where the differential errors for σ(logg) and σ(T ) indicator compared to Ba (e.g., Simmerer et al. 2004; Winckler
eff
are±0.05and±30K,respectively.The1-σ errorsfromNS10for etal.2006).WhileBaconsistsoffivestableisotopesandrequires
[Fe/H]and[α/Fe]were0.04and0.02dex,respectively.Weesti- theconsiderationofisotopeshifts,99.91%ofLaisintheisotope
matedthe1-σerrorsin[Zr/Fe]and[Ce/Fe]tobe0.04.For[Sc/Fe], 139La (Lodders 2003; Asplund et al. 2009). Figure 3(b) presents
[La/Fe],and[Eu/Fe],theerrorsare0.05.Theerrorfor[Nd/Fe]is [La/Fe]asafunctionof[Fe/H].Thereisnocleardifferenceinthe
0.07. Laabundancebetweenthelow-α,high-α,andTDstars.Thedis-
(cid:13)c 2013RAS,MNRAS000,1–11
Neutron-captureandScabundancesoffieldhalostars 5
Table4.Abundancesforeachofthemeasuredelements.Thevaluesfor[Fe/H]aretakenfromNS10.
Star [Fe/H] [Sc/Fe] [Zr/Fe] [La/Fe] [Ce/Fe] [Nd/Fe] [Eu/Fe]
CD-453283 −0.91 −0.11 −0.07 0.00 −0.01 0.36 0.31
CD-610282 −1.23 −0.15 0.00 −0.01 −0.05 0.32 0.26
G16-20 −1.42 −0.11 −0.00 0.02 −0.04 0.17 0.32
G18-28 −0.83 0.11 0.03 −0.02 0.04 0.34 0.13
G18-39 −1.39 −0.00 0.21 0.07 −0.03 0.47 0.26
G31-55 −1.10 −0.03 0.06 −0.08 −0.11 0.26 0.05
G56-30 −0.89 −0.22 −0.12 −0.06 −0.04 −0.09 0.09
G56-36 −0.94 −0.02 0.04 −0.04 −0.05 0.31 0.13
G66-22 −0.86 −0.16 −0.08 −0.04 0.02 0.34 0.30
G170-56 −0.92 −0.14 −0.10 −0.02 −0.16 0.14 0.16
HD3567 −1.16 −0.04 0.05 0.16 0.07 0.33 0.39
HD22879 −0.85 0.12 0.06 −0.09 −0.04 0.34 0.16
HD25704 −0.85 0.09 −0.16 −0.02 −0.09 0.33 0.15
HD76932 −0.87 0.15 0.13 0.03 −0.08 0.36 0.21
HD103723 −0.80 −0.03 −0.05 −0.03 0.01 0.33 0.17
HD105004 −0.82 −0.04 −0.13 −0.12 0.10 0.10 0.04
HD111980 −1.08 0.00 0.25 0.13 0.09 0.33 0.14
HD120559 −0.89 0.17 −0.05 −0.05 −0.22 0.32 0.20
HD126681 −1.17 0.00 0.28 0.14 0.13 0.34 0.20
HD179626 −1.00 0.12 0.03 0.02 −0.14 0.14 0.03
HD189558 −1.12 0.09 0.24 0.07 0.08 0.24 0.17
HD193901 −1.07 −0.19 −0.14 −0.00 −0.08 0.31 0.24
HD194598 −1.09 −0.07 −0.11 −0.03 −0.11 0.17 0.27
HD199289 −1.04 0.08 −0.09 −0.12 −0.30 0.15 0.09
HD205650 −1.17 −0.01 0.01 −0.03 −0.07 0.14 0.19
HD219617 −1.45 −0.11 −0.05 −0.11 0.18 0.03a 0.21
HD230409 −0.85 0.05 −0.03 −0.00 −0.07 0.43 0.27
aUpperlimit
persion in the abundance of [La/Fe] is ∼0.07 dex, the lowest of investigatethecontributionofr-processnucleosynthesistothelow-
all the s-process abundances measured here. The [La/Fe] ratio is andhigh-αstars.
almostconstantwith[Fe/H]andthemeanabundanceclosetothe Figure6(a)presentsthe[Eu/Fe]abundanceswherethereisa
solarvalue. potentialseparationbetweenthetwopopulations:thelow-α stars
Thes-processcontributiontoCeintheSolarSystemis81% appeartoshowahigher[Eu/Fe]abundancecomparedtothehigh-
where140Ceistheonlystableisotopeaccessiblebythes-process αstars.However,duetothelackofstarsatlowermetallicities,the
(Sneden,Cowan&Gallino2008).Figure3(c)presentstheresults separationisnotasclearaswhatisfoundforotherelementssuch
for [Ce/Fe]. The majority of stars have a subsolar [Ce/Fe] abun- asZr.Thelow-α groupalsoshowsapotentialdecreasein[Eu/Fe]
dance with a mean value of around −0.04 dex and no apparent with increasing metallicity. A larger sample of stars is needed to
separationbetweenthelow-andhigh-α stars.Thedispersionsfor confirmtheseresults.
thelow-αandhigh-αstarsareessentiallythesame(0.09dexcom-
paredto0.10dex,respectively).
ForNd,58%oftheSolarSystemabundanceisduetothes-
process(Sneden,Cowan&Gallino2008).Figure3(d)presentsthe
abundanceof[Nd/Fe]against[Fe/H].DuetotheweakNdIIline
forHD219617,weprovideanupperlimitof[Nd/Fe]=0.03for
thisobjectbasedonanEWlimitof2mA˚.Thereisnoobviousdif- 3.4 Sc
ference in [Nd/Fe] between the two populations except for a few
low-α starshavingalowerNdabundance.Thiscausesthedisper- ThereisonlyonestableisotopeofSc(45Sc)anditismainlysyn-
sion to be higher for the low-α stars relative to the high-α stars, thesisedduringcarbonandneonburninginmassivestars(Woosley,
0.14 dex compared 0.10 dex. The mean abundance of [Nd/Fe] is Heger&Weaver2002).ItisalsoproducedtoalesserextentinAGB
0.26dexandallbutonestarhasasupersolar[Nd/Fe]abundance. stars(e.g.,Fishlocketal.2014).WemeasureScinordertoverify
the conclusions of Nissen et al. (2000) in which the low-α stars
havealower[Sc/Fe]abundancecomparedtothehigh-αstars.
Figure 6(b) reveals a clear separation in [Sc/Fe]. The low-α
starshaveasubsolar[Sc/Fe]abundancewithameanvalueof−0.11
3.3 Eu
dexwhereasthehigh-α starshaveamostlysupersolarabundance
Euispredominatelyanr-processelementwithlessthan2%ofthe withameanvalueof0.07dex.Theseresultsconfirmtheseparation
SolarSystemabundancearesultofs-processnucleosynthesis(Sne- in[Sc/Fe]foundintheearlierworkbyNissenetal.(2000)between
den,Cowan&Gallino2008).WehaveincludedEuinthisstudyto thelow-andhigh-αpopulations.
(cid:13)c 2013RAS,MNRAS000,1–11
6 Fishlocketal.
0.5
0.4 (a) 1.0
0.3 0.9
0.2
Fe] 0.1 0.8
Zr/ 0.0 ux 0.7
[ 0.1 Fl
I
0.2 0.6 r I
Z
0.3 0.5
0.4
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.4
HD 111980 HD 194598
0.5
0.4 (b) 4207 4208 4209 4210
0.3 Wavelength ( )
0.2
]
Fe 0.1 Figure4.MIKEspectraaroundtheZrlineat4208.98A˚ foralow-α star
a/ 0.0 (HD194598)andahigh-αstar(HD111980).
L
[
0.1
0.2
0.3
0.3
0.4
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
0.2
0.5
0.4 (c)
0.1
0.3
]
Fe] 00..21 [Y/Fe 0.0
Ce/ 0.0 0.1
[
0.1
0.2 0.2
0.3
0.4 0.3
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.3 0.2 0.1 0.0 0.1 0.2 0.3
0.5 [Zr/Fe]
0.4 (d)
0.3 Figure5.Comparisonof[Y/Fe](takenfromNS11)and[Zr/Fe].Thered
0.2 circlesindicatelow-αstars,theblueopencirclesindicatehigh-αstars,and
e] theplussignsindicateTDstars.Thedashedlineshowstheone-to-onecor-
F 0.1
Nd/ 0.0 respondenceline.The1-σerrorisshowninthetop-left.
[ 0.1
0.2
now seek to quantify this potential limitation by introducing our
0.3
methodologyonhowwedecidewhetheragivenabundanceratio
0.4
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 exhibitsadifference(i.e.,separation)betweenthelow-andhigh-α
groups.
[Fe/H]
We first consider the NS10 measurements of Cr, a repre-
sentative element for which NS10 found no significant separa-
Figure3.Fours-processelements(Zr,La,Ce,andNd)versus[Fe/H].The
tion between the low- and high-α groups. Using a two sam-
redcirclesindicatelow-αstars,theblueopencirclesindicatehigh-αstars,
pleKolmogorov-Smirnov(KS)testbetweenthelow-andhigh-α
andtheplussignsindicateTDstars.The1-σerrorisshowninthebottom-
groups(withoutdistinguishingTDstars),wefindap-valueof0.07
leftofeachplot.
when considering [Cr/Fe] for all stars in the NS10 sample with
−1.45(cid:54)[Fe/H](cid:54)−0.8.
Inordertoinvestigatehowoursmallersamplesizeaffectsour
4 ANALYSIS
abilitytoidentifyabundanceseparations,weperformedthefollow-
AswiththeNS10andNS11studies,wecombinetheTDandhigh- ingtests.Werandomlyselect13low-α and14high-α starsfrom
α halostarsintoonehigh-α group.Duetoasmallersamplesize thefullNS10samplewith−1.45(cid:54)[Fe/H](cid:54)−0.8,applytheKS
comparedtoNS11,wedonotremovetheabundancedependence test and obtain the p-value. We repeat this exercise 10,000 times
on [Fe/H] for the high-α stars as it cannot be definitively deter- andfindthatforapproximately25%oftherealisations,thep-value
mined. Given that our sample size is smaller than in NS10, we isbelow0.07.Ifwearbitrarilydefinethethresholdtobea p-value
(cid:13)c 2013RAS,MNRAS000,1–11
Neutron-captureandScabundancesoffieldhalostars 7
Table5.Meanvalueandstandarddeviationofeachabundanceratioforthe
0.4 (a) low-α groupandthehigh-α group.YandBaaretakenfromNS11.The
0.3 p-valuebetweenthetwogroupsisalsogiven.
] 0.2
e
F
u/ 0.1 low-α high-α
E x˜ σ x˜ σ p-value
[ 0.0
0.1 [Sc/Fe] −0.107 0.062 0.067 0.063 0.000
[Zr/Fe] −0.058 0.061 0.069 0.131 0.017
0.2
[La/Fe] −0.022 0.065 0.004 0.076 0.589
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 [Ce/Fe] −0.012 0.087 −0.058 0.114 0.340
[Nd/Fe] 0.217 0.137 0.299 0.098 0.614
0.4 (b) [Eu/Fe] 0.222 0.097 0.161 0.068 0.173
0.3 [La/Zr] 0.037 0.061 −0.066 0.084 0.005
[ls/Fe] −0.109 0.057 0.049 0.117 0.004
] 0.2
e [hs/Fe] 0.005 0.056 0.038 0.084 0.303
F
c/ 0.1 [hs/ls] 0.114 0.046 −0.010 0.065 0.000
S
[ 0.0 [Y/Eu] −0.382 0.101 −0.133 0.122 0.000
[ls/Eu] −0.332 0.092 −0.112 0.126 0.001
0.1
[Ba/Eu] −0.387 0.122 −0.253 0.122 0.024
0.2 [La/Eu] −0.244 0.064 −0.157 0.082 0.173
[hs/Eu] −0.218 0.082 −0.122 0.081 0.022
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
[Y/Fe] −0.160 0.064 0.028 0.107 0.000
[Fe/H]
[Ba/Fe] −0.165 0.070 −0.092 0.115 0.303
[Ba/Y] −0.005 0.055 −0.120 0.030 0.000
Figure6.(a)[Sc/Fe]and(b)[Eu/Fe]versus[Fe/H].Theredcirclesindicate
low-αstars,theblueopencirclesindicatehigh-αstars,andtheplussigns
indicateTDstars.The1-σerrorisshowninthebottom-leftofeachplot.
of0.03,roughly10%oftherandomrealisationsresultinap-value
below0.03. Zr,theabundanceratiosforelementsbelongingtothesecondpeak
Wenowexamine[Ba/Y],anabundanceratiothatwasfoundto [La/Fe], [Ce/Fe], and [Nd/Fe] show no significant difference be-
exhibitasignificantseparationbetweenthelow-andhigh-α pop- tweenthelow-andhigh-α groups.Theseconds-processpeakele-
ulations.ForthefullNS10samplewith−1.45(cid:54)[Fe/H](cid:54)−0.8, mentBa,measuredbyNS11,alsoshowsnoseparationin[Ba/Fe]
wecomputea p-valueof2.9×10−9.Wethenseektoquantifythe betweenthetwogroups.
effectofhavingasmallersamplesize.Asabove,werandomlyse- We confirm the separation in [Ba/Y] reported by NS11 (see
lect13low-α and14high-α starsfromthefullNS10samplewith Figure 7(a)). In Figure 7(b) we plot [La/Zr] and are able to re-
−1.45(cid:54)[Fe/H](cid:54)−0.8,applytheKStestandobtainthep-value. produce the separation between a light-s (ls) element at the first
We repeat the exercise 10,000 times. For a threshold p-value of s-processpeakandaheavy-s(hs)elementatthesecondpeak.The
0.03,about99.7%ofrandomrealisationshaveap-valuebelowthis p-valueof0.005islowerthantheindividual p-valuesfor[Zr/Fe]
level,indicatingthatweareabletoidentifytheseparationin[Ba/Y] (0.017)and[La/Fe](0.589)demonstratingthat aclearseparation
inallbut0.3%ofcases. betweenthelow-andhigh-α populationscanbeseenwhencon-
Basedonthesetests,wechoosea p-valueof0.03whichen- sideringcertaincombinationsofelements.
suresthatthereisatleasta90%chancethattheabundancesepara- Assumingthatthehigher[La/Zr]ratiosinthelow-α starsare
tioninoursampleof27starsisgenuinedespitehavingasmaller duetoenrichmentfromAGBstars,weusethemodelpredictions
number of stars than NS10 and NS11. We also confirm using a ofFishlocketal.(2014)with[Fe/H]=−1.2toestimatethemass
KStestthatthehigh-α populationandTDstarshaveasignificant rangeofthoseAGBstars.Startingatlowermasses(1to3M(cid:12)),cal-
probability of coming from the same population. In Table 5 we culationspredict[La/Zr]ratiosabovesolar;forexample,the3M(cid:12)
present the mean of each abundance ratio and its standard devia- modelpredictsafinal[La/Zr]ratioof0.32dex.Movingtotheinter-
tionforbothlow-andhigh-α groups.Wealsopresentthe p-value mediatemasses(3.25to7M(cid:12)),calculationspredict[La/Zr]ratios
ofthetwosampleKStestbetweenthelow-andhigh-αgroups.The belowsolar;forexample,the6M(cid:12) modelpredictsafinal[La/Zr]
analysesof[Y/Fe],[Ba/Fe],and[Ba/Y]areincludedinTable5for ratio of −0.68 dex. Therefore, based on the [La/Zr] ratio, AGBs
comparisonusingtheabundancesmeasuredbyNS11foroursam- starsintherange1to3M(cid:12)couldberesponsiblefortheabundance
ple of 27 stars. For comparison, the mean p-value calculated for differencebetweenthelow-andhigh-αgroups.
[Mg/Fe]whenrandomlyselecting13low-αand14high-αstarsis For stars in the range −1.5< [Fe/H] <−1, there is a pos-
4.9×10−7. sible separation in [Eu/Fe] where the low-α group has a higher
ForabundancesrelativetoFe(e.g.,[Zr/Fe]),ScandZrshowa [Eu/Fe]abundancecomparedtothehigh-αgroup.However,higher
separationbetweenthelow-andhigh-αgroups.Forbothelements, metallicity stars show no difference between the two groups and
thehigh-αgrouphasahigher[X/Fe]ratiothanthelow-αgroup. thehintofaseparationisnotstatisticallysignificantwithap-value
We find that the separation between the low- and high-α of0.173.Wereturntothisintriguingpossibilityofaseparationin
groupsisdetectedfor[Y/Fe]despitehavingasmallersamplesize Section5.
thanNS11.Incontrasttothefirsts-processpeakelementsYand Thes-processindicatorsaretheaverageabundanceoftheele-
(cid:13)c 2013RAS,MNRAS000,1–11
8 Fishlocketal.
[ls/Fe]=([Y/Fe]+[Zr/Fe])/2 (1)
0.3
(a) and
0.2
0.1 [hs/Fe]=([Ba/Fe]+[La/Fe]+[Ce/Fe]+[Nd/Fe])/4. (2)
]
Y
a/ 0.0 The[ls/Fe]and[hs/Fe]ratiosagainst[Fe/H]arepresentedin
B
[ 0.1 Figure 7(c) and 7(d). The low-α stars have a lower [ls/Fe] ratio
thanthehigh-α stars.For[hs/Fe],thereisnoapparentdifference
0.2
betweenthetwopopulations.
0.3
Figure7(e)showsthereisaclearseparationinthes-process
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
0.3 indicator [hs/ls] between the low- and high-α groups. The low-
(b) α populationhasahigher[hs/ls]ratiothanthehigh-α population
0.2
withamean[hs/ls]valueof0.11dexcomparedto−0.01dexfor
] 0.1 thehigh-αpopulation.Allthelow-αstarshaveasupersolar[hs/ls]
r
a/Z 0.0 ratioandthiswouldbeexpectedifthestarswereenrichedbyprevi-
L ousgenerationsoflow-massAGBstars.Forexample,thelow-mass
[ 0.1
modelsofFishlocketal.(2014)endtheAGBphasewith[hs/ls]ra-
0.2
tiosofbetween0.45and0.59dex,excludingthe1M(cid:12)modelwhich
0.3 experiencesminimalthirddredge-up.Incontrast,theintermediate-
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 mass models have final [hs/ls] ratios between −0.71 and −0.41
0.3
dex.Thedifferencesbetweenthelow-massandintermediate-mass
(c)
0.2 modelsarearesultofthedifferentneutronsources(Lugaroetal.
0.1 2012).
]
e The abundance ratio of s-process elements to r-process ele-
F 0.0
/s mentsoffersapowerfuldiagnosticforGalacticchemicalevolution
[l 0.1 (Wheeler,Sneden&Truran1989;McWilliam1997).InFigure8
0.2 we present the abundance ratios of [ls/Eu], [Y/Eu], [hs/Eu], and
[La/Eu]against[Fe/H].Thelow-α starshavealower[ls/Eu]ratio
0.3
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 comparedtothehigh-α starswiththeseparationmoreevidentat
0.3 [Fe/H]<−1.Thereisalsoanapparentincreasein[ls/Eu]within-
∼
(d)
0.2 creasing[Fe/H]forthelow-αgroupwhichisnotseeninthehigh-α
group.
0.1
e] Figure8(b)presentstheabundanceratioofthelselement,Y,
/Fs 0.0 relativetoEu.Thelow-α grouphasalowerabundanceof[Y/Eu]
[h 0.1 comparedtothehigh-α groupwitha p-valueof2.2×10−4.This
p-valueismoresignificantthanthe[ls/Eu]separation.Wealsofind
0.2
thatthe[Y/Eu]abundanceforthelow-αstarswith[Fe/H]<−1is
0.3
almost constant at approximately −0.4 dex with minimal scatter.
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
0.3 Similarbehaviourisalsoseenfor[ls/Fe].
(e) Theseparationin[hs/Eu]isnotasclearasthatof[ls/Eu],al-
0.2
thoughithasap-valueof0.022,justbelowthechosensignificance
0.1 level.Thelow-α grouphasa[hs/Eu]ratiothatisclosertother-
]
s
/sl 0.0 processcontributionofthehselementsandEuintheSolarSystem
[h 0.1 thanthehigh-αgroup.Aswith[ls/Fe],thereisanapparentincrease
in[hs/Eu]withincreasing[Fe/H]forthelow-αstars.
0.2
Figure 8(d) presents the abundance ratios of the hs element,
0.3 La,relativetoEuwhereallthestarshaveasubsolar[La/Eu]ratio.
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
Unliketheseparationfoundfor[hs/Eu],wedonotfindaseparation
[Fe/H] in [La/Eu] between the two groups. We also show the r-process
contributiontotheSolarSystemabundancesofLaandEu(andits
Figure7.[Ba/Y](fromNS11),[La/Zr],[ls/Fe],[hs/Fe],and[hs/ls]versus uncertainty)inFigure8(d)asdeterminedbyWinckleretal.(2006).
[Fe/H]wherelsistheaverageabundanceofYandZrandhsistheaverage ThetheoreticalpredictionofanAGBstellarmodelwithaninitial
abundanceofBa,La,Nd,andCe.Theredcirclesindicatelow-αstars,the massof3.25M(cid:12)atametallicityof[Fe/H]=−1.2,fromFishlock
blueopencirclesindicatehigh-αstars,andtheplussignsindicateTDstars. etal.(2014)isalsopresentedinFigure8.Ofallthemodelscalcu-
latedbyFishlocketal.(2014),thelowest[La/Eu]ratioof0.08dex
occursforthe3.25M(cid:12)model.
mentsateachofthetwos-processpeaks.WeincludetheYandBa
resultsfromNS11anddefine[ls/Fe]and[hs/Fe]3tobe:
3 Asnoted,theSolarSystemabundanceofNdhasa∼40%contribution
from the r-process. Had we excluded Nd from hs, as did Lugaro et al.
(2012),ourconclusionswouldbeunchanged.
(cid:13)c 2013RAS,MNRAS000,1–11
Neutron-captureandScabundancesoffieldhalostars 9
ement. The simplest explanation is that Sc is preferentially made
0.1
in higher mass stars along with the α elements and that Type Ia
(a)
0.0 SNe have a negligible contribution to Sc (Iwamoto et al. 1999;
0.1 Woosley & Weaver 1995; Romano et al. 2010). Indeed, Nissen
] (2016)findthatthe[Sc/Fe]ratioexhibitsasimilarbehaviourwith
u 0.2
E ageas[Mg/Fe]intheirsampleofsolartwins.Thelow-αstarshave
[/ls 0.3 low[Sc/Fe]ratiosduetotheextracontributionofFefromTypeIa
0.4 SNe.
Asnotedintheintroduction,thereisanotherchemicalsigna-
0.5
turebesides[α/Fe]4 thatdistinguisheshalostarsfromtheirdSph
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 counterparts, namely, the neutron-capture element abundance ra-
0.1
tios.IndSph,roughlytwo-thirdsofthestarsshowahigher[Ba/Y]
(b)
0.0 ratio and a lower [Y/Eu] ratio compared to Galactic halo stars
0.1 (Vennetal.2004;Tolstoy,Hill&Tosi2009).McWilliam,Waller-
Eu] 0.2 stein & Mottini (2013) analysed three Sagittarius dSph stars and
Y/ foundlow[α/Fe],[Na/Fe],[Al/Fe],and[Cu/Fe]ratiosalongwitha
[ 0.3 high[La/Y]ratiorelativetotheGalactichalo.ThestudyoftheCa-
0.4 rinadSphbyVennetal.(2012)alsofoundthatthelow-metallicity
starshaveslightlylower[Y/Eu]ratios(andslightlyhigher[Ba/Y]
0.5
ratios) when compared to Galactic stars. The dSph of today are
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 chemically different to the galaxies that merged early during the
0.1
formationoftheGalaxy.
(c)
0.0 The NS11 study found that the low-α stars show a higher
0.1 [Ba/Y] ratio than the high-α stars. We also find differences for
u] [Zr/Fe], [La/Zr], [ls/Fe], [hs/ls], [Y/Eu], [Ba/Eu], [ls/Eu], and
E 0.2
/s [hs/Eu]betweenthetwopopulations.Thesefindingsrevealthatthe
[h 0.3 low-α populationdisplaysmanychemicalsignaturesalsoseenin
0.4 dSphtherebysupportingtheaccretionhypothesisfromNS10.
Wenowturnourattentiontothepossiblesourcesthatcould
0.5
produce these abundance differences among the neutron-capture
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
elements. Due to the slower rate of chemical evolution, it has
been suggested that the dSph and low-α population have been
(d)
0.2 3.25 M enriched from low-metallicity, low-mass AGB stars (NS11, Venn
fl etal.2012).Intermediate-massAGBstars(>3M(cid:12))producesub-
u] 0.0 solar [Ba/Y] ratios (e.g., Karakas & Lattanzio 2014), i.e., these
E
a/ objectscannotexplainthebehaviourof[Ba/Y]inthelow-α pop-
L 0.2
[ ulation. On the other hand, low-metallicity, low-mass AGB stars
0.4 (∼<3M(cid:12))produce[Ba/Y]ratiosabovesolar(e.g.,Cristalloetal.
2011; Fishlock et al. 2014), and this is consistent with the low-
α population. The high [La/Zr] and [hs/ls] ratios for the low-α
1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8
populationfurthersupportthehypothesisofenrichmentfromlow-
[Fe/H] metallicity, low-mass AGBs in the low-α populations as a result
oftheslowerrateofchemicalevolution(e.g.,inlow-massAGBs,
Figure8.[ls/Eu],[Y/Eu],[hs/Eu],and[La/Eu]versus[Fe/H]wherelsis thefirsts-processpeakelementsarebypassedinfavourofsecond
theaverageabundanceofYandZrandhsistheaverageabundanceofBa, s-processpeakelements).
La,Nd,andCe.Theredcirclesindicatelow-αstars,theblueopencircles Oneissuewiththisscenarioisthatwewouldthereforeexpect
indicatehigh-αstars,andtheplussignsindicateTDstars.Theredsolidline theabundanceofYtobegreaterthan(orequalto)Zr,i.e.,[Y/Fe](cid:62)
representsther-processratiofor[La/Eu]andthereddashedlineitsupper [Zr/Fe](e.g.,Bisterzoetal.2014;Trippellaetal.2016).Theobser-
uncertainty(Winckleretal.2006).Thefinalsurfaceabundanceof[La/Eu]
vationsindicateotherwise,andthisissuehasalsobeenidentifiedin
forthe3.25M(cid:12),Z = 0.001AGBmodelfromFishlocketal.(2014)isalso CEMP-sstars(Abateetal.2015).Thereasonforthediscrepancy
indicated(dash-dottedline).
betweenobservationsandtheoryisunknown.AsnotedinSection
3.1,whiletheabundancesofYandZrfolloweachother,theremay
beazero-pointoffsetduetoerrorsintheatomicdata(e.g.,gf val-
5 DISCUSSION
ues).
Wehavestudiedchemicalabundancesinasubsampleoflow-and If low-metallicity, low-mass stars have contributed to chem-
high-α starsfromNS10inwhichthelow-α starswerelikelyac-
creted from satellite dwarf galaxies whereas the high-α stars are
representativeofGalactichaloobjects.Wenowseektounderstand
4 We note that there is some overlap in [α/Fe] between dSph stars and
theabundancedifferencesbetweenthesepopulations.
Galactichalostarswithextremeretrogradeorbits(Fulbright2002;Stephens
ForSc,weconfirmandextenduponresultsfromNissenetal. &Boesgaard2002).Additionally,themostmetal-poorstarsindSphappear
(2000)inwhichthelow-αstarshavealower[Sc/Fe]ratiothanthe tohavehalo-like[α/Fe]ratios(Frebeletal.2010;Norrisetal.2010;Venn
high-α stars. That is, Sc appears to behave like an α-capture el- etal.2012).
(cid:13)c 2013RAS,MNRAS000,1–11
10 Fishlocketal.
ical enrichment of the low-α population, then we can place con- The low abundance of [Y/Eu] found in the low-α popula-
straintsontheenrichmenttimescalesinvolvedintheearliestdwarf tioncomparedtothehigh-α populationmatchesoneofthechem-
galaxies. Models from Fishlock et al. (2014) predict that low- icalsignaturesofpresent-daydSph.Vennetal.(2004)discardthe
metallicity,low-mass(∼<3M(cid:12))starshavelifetimesbetween0.29× possibilitythattheGalactichalocouldconsistoflow-massdwarf
109yrto6.8×109yr. galaxies that have continuously merged with the Galaxy as high
WealsopresentmeasurementsofEu;ther-processisrespon- [Ba/Y] and low [Y/Eu] ratios are not observed in Galactic halo
sibleforproducingthemajorityofEuintheSolarSystem(Sneden, stars.TheideaofmoremassivemergerswiththeGalaxywasnot
Cowan&Gallino2008).Theevolutionof[Eu/Fe]forbothGalac- ruledoutbyVennetal.(2004).Inaddition,modelsofAGBstars
ticstarsanddSphstarsgenerallyfollows[α/Fe](Vennetal.2004; areunabletoexplainthelower[hs/Eu]ratiosobservedinthelow-α
Tolstoy,Hill&Tosi2009)indicatingacommonoriginbetweenα starscomparedtothehigh-αstars.
elementsandEu,mostlikelytobemassivestars.Iftrue,thenwe Theresultspresentedhereoffernewandimportantobserva-
would expect the low-α stars to have a lower [Eu/Fe] abundance tionalevidenceregardingthenatureoftheGalactichalo.Itisap-
thanthehigh-αstarsatagivenmetallicity.However,thereissome parentfromthestudyofNS10andNS11thatthereareatleasttwo
evidencethatthelow-α starshaveahigher[Eu/Fe]ratiothanthe populationsintheGalactichalo,onethatformedin-situaswellas
high-αstars,incontrasttoourexpectations.Thisseparationisonly oneaccretedfromdwarfgalaxies.Thelow[Y/Eu]abundancesin
observedat[Fe/H]<−1andincludeshalfthesample.Whenusing dSphhavebeenattributedtothecontributionofmetal-poorAGB
thewholemetallicityrange,theKStestdoesnotfindastatistically stars(Vennetal.2004).Low-metallicityAGBstarscouldalsobe
significantseparationbetweenthetwopopulations(seeTable5). responsible for the high [Ba/Y] abundances. Only with detailed
If the low-α stars do have a higher Eu abundance than the chemicalevolutionmodels,alongwithmeasurementsofadditional
high-α stars at lower metallicities, it would suggest that the r- r-process elements beyond Eu, will there be a better understand-
process contributed more to the chemical evolution of the low-α ingofthechemicalenrichmentoftheearliestdwarfgalaxiesthat
starsatagivenmetallicity.Thisisdifficulttoreconcilewithchem- mergedwiththeGalactic halo.Inparticular,throughabetter un-
icalevolutionmodelswhichsupporttheideathatintensegalactic derstandingofthes-process,wewillbeabletoprovideconstraints
winds remove Eu from a system (Lanfranchi, Matteucci & Ces- ontheformationsite(s)ofther-process.
cutti 2008). Another possibility is that the α and Eu abundances
areduetodifferentinitialmassfunctionsratherthantheaddition
ofSNeIamaterial(Ishimaru&Wanajo1999;Travaglioetal.1999;
Kobayashietal.2006).Todetermineifthereisthepossibilityofa 7 ACKNOWLEDGEMENTS
statisticallysignificantseparation,additionalobservationsofEufor
We thank the referee, Maurizio Busso, for helpful comments.
theNS10samplearerequired,particularlyforstarswith[Fe/H]<
ThisworkhasbeensupportedbytheAustralianResearchCouncil
−1.
(grantsFT110100475andFT140100554)andtheEuropeanUnion
Weobserveconstant[Y/Eu]abundances(∼−0.4dex)inthe
FP7 programme through ERC grant number 320360. JM thanks
lower metallicity low-α stars (−1.5(cid:54) [Fe/H] (cid:54)−1) suggesting
support by FAPESP (2012/24392-2 and 2014/18100-4). Funding
thatYandEuhaveacommonoriginortimescaleforenrichment,
fortheStellarAstrophysicsCentreisprovidedbyTheDanishNa-
most likely to be the r-process, at these metallicities. The dSph
tional Research Foundation (Grant DNRF106). Australian access
chemical evolution models of Lanfranchi, Matteucci & Cescutti
to the Magellan Telescopes was supported through the National
(2008)showconstant[Y/Eu]ratiosuptoa[Fe/H]ratioofapproxi-
Collaborative Research Infrastructure Strategy of the Australian
mately−1.5.Theratioof[Y/Eu]thenincreasesathighermetallic-
FederalGovernment.
itiesduetotheproductionofYbyAGBstars.
We observe lower [Y/Eu], [ls/Eu], and [hs/Eu] ratios in the
low-α starscomparedtothehigh-α starswiththeseparationbe-
comingmorenoticeableatlowermetallicities.ThestudyoftheCa-
REFERENCES
rinadSphbyVennetal.(2012)alsofoundthatthelow-metallicity
starshaveslightlylower[Y/Eu]ratios(andslightlyhigher[Ba/Y] Abate C., Pols O. R., Izzard R. G., Karakas A. I., 2015, A&A,
ratios) when compared to Galactic stars. Venn et al. (2012) con- 581,A22
cludedthatthesedifferencesarearesultofmetal-poorAGBstars AsplundM.,GrevesseN.,SauvalA.J.,ScottP.,2009,ARA&A,
contributing to the chemical enrichment of Carina, and this may 47,481
alsoapplytothelow-αpopulation. BelokurovV.etal.,2006,ApJ,642,L137
Bernstein R., Shectman S. A., Gunnels S. M., Mochnacki S.,
AtheyA.E.,2003,inProc.SPIE,Vol.4841,InstrumentDesign
andPerformanceforOptical/InfraredGround-basedTelescopes,
6 CONCLUSIONS
IyeM.,MoorwoodA.F.M.,eds.,pp.1694–1704
We found that the low-α and high-α populations of NS10 sepa- BisterzoS.,TravaglioC.,GallinoR.,WiescherM.,Ka¨ppelerF.,
ratewhenusingtheabundanceratiosof[Sc/Fe],[Zr/Fe],[La/Zr], 2014,ApJ,787,10
[ls/Fe],[hs/ls],[Y/Eu],[Ba/Eu],[ls/Eu],and[hs/Eu].Thesediffer- BullockJ.S.,JohnstonK.V.,2005,ApJ,635,931
encesinchemicalabundanceratioshaveonlybeendetectedusing BussoM.,GallinoR.,LambertD.L.,TravaglioC.,SmithV.V.,
adifferentialanalysisrelativetoareferencestarwithcomparable 2001,ApJ,557,802
stellar parameters to minimise the errors in abundance measure- BussoM.,GallinoR.,WasserburgG.J.,1999,ARA&A,37,239
ments. The separation observed in [La/Zr] between the low- and CarneyB.W.,WrightJ.S.,SnedenC.,LairdJ.B.,AguilarL.A.,
high-α groupsconfirmstheresultsofNS11whofindaseparation LathamD.W.,1997,AJ,114,363
in[Ba/Y],whereZrandYarefirsts-processpeakelementsandLa CaseyA.R.,2014,PhDthesis,AustralianNationalUniversity
andBaareseconds-processpeakelements. CaseyA.R.etal.,2014,MNRAS,443,828
(cid:13)c 2013RAS,MNRAS000,1–11