Spectroscopic analysis of metal-poor stars from LAMOST: early results

0.61 MB·English

by  Hai-Ning Li

DraftversionJanuary14,2015
PreprinttypesetusingLATEXstyleemulateapjv.05/12/14
SPECTROSCOPICANALYSISOFMETAL-POORSTARSFROMLAMOST:EARLYRESULTS
Hai-NingLi1,GangZhao1,NorbertChristlieb2,LiangWang1,WeiWang1,YongZhang3,YonghuiHou3,andHailongYuan1
(Received;Accepted)
DraftversionJanuary14,2015
ABSTRACT
Wereportonearlyresultsfromapilotprogramsearchingformetal-poorstarswithLAMOSTandfollow-up
5 high-resolutionobservationacquiredwith theMIKE spectrographattachedtothe MagellanIItelescope. We
1 performeddetailed abundanceanalysisfor eightobjects with iron abundances[Fe/H] < −2.0, includingfive
0 extremelymetal-poor(EMP;[Fe/H]<−3.0)starswithtwohaving[Fe/H]<−3.5. Amongtheseobjects,three
2
arenewlydiscoveredEMPstars,oneofwhichisconfirmedforthefirsttimewithhigh-resolutionspectralob-

n servations.Threeprogramstarsareregardedascarbon-enhancedmetal-poor(CEMP)stars,includingtwostars
a withnoenhancementintheirneutron-captureelements,whichthuspossiblybelongtotheclassofCEMP-no
J stars;oneoftheseobjectsalsoexhibitssignificantenhancementinnitrogen,andisthusapotentialcarbonand

3 nitrogen-enhancedmetal-poorstar. The[X/Fe]ratiosofthesamplestarsgenerallyagreewiththosereported
1 intheliteratureforothermetal-poorstarsinthesame[Fe/H]range.Wealsocomparedtheabundancepatterns
ofindividualprogramstarswiththeaverageabundancepatternofmetal-poorstars, andfindonlyonechemi-

] callypeculiarobjectwithabundancesofatleasttwoelements(otherthanCandN)showingdeviationslarger
R
than0.5dex.Thedistributionof[Sr/Ba]versus[Ba/H]agreesthatanadditionalnucleosynthesismechanismis
S neededasidefromasingler-process.TwoprogramstarswithextremelylowabundancesofSrandBasupport
. theprospectthatbothmainandweakr-processmayhaveoperatedduringtheearlyphaseofGalacticchemical
h
evolution.Thedistributionof[C/N]showsthattherearetwogroupsofcarbon-normalgiantswithdifferentde-
p
- greesofmixing.However,itisdifficulttoexplaintheobservedbehaviorofthe[C/N]ofthenitrogen-enhanced
o unevolvedstarsbasedoncurrentdata.
r Subjectheadings:Galaxy:halo—stars: abundances—stars: PopulationII
t
s
a
[ 1. INTRODUCTION spectroscopy have been performed to identify various

1 Very metal-poor ([Fe/H]4. < −2.0) and extremely metal- abundance patterns in metal-deficient stars (e.g., Ryanetal.
1996;Laietal.2008;Cohenetal.2008;Norrisetal.2013b).
v poor(EMP;[Fe/H] < −3.0)stars arebelievedtopreservein
These explorations also resulted in the discovery of a va-
2 theiratmospheresdetailedinformationconcerningthechem-
6 ical compositionsof the interstellar medium at the time and riety of unusual peculiarities in elemental abundances in
0 place thattheywere born. These objectsare expectedto of- metal-poor stars, such as large enhancements in carbon and
3 ferusadirectviewintotheearliestphasesoftheevolutionof nitrogen relative to iron and the abundance ratios seen in
0 theGalaxy,theformationofthefirstgenerationofstars,and the Sun; enhancement of α elements; and neutron-capture
. elements (Aokietal. 2007; Polsetal. 2012; Aokietal.
1 the earliest nucleosynthesis events (McWilliametal. 1995;
2013; Placcoetal. 2013). The ever-increasing sample size
0 Norrisetal. 1996; Beers&Christlieb 2005; Frebel&Norris
of metal-poor halo stars and the enrichment history of
5 2013). Precise analysis of the chemical composition of
the Galactic halo have gradually been revealed through
1 metal-poor stars enables us to indirectly probe the range
statistical investigations in the low-metallicity region of the
: of supernova nucleosynthesis yields in the early Galaxy
v metallicity distribution function (MDF; Carolloetal. 2007;
(Heger&Woosley 2010; Nomotoetal. 2013, and reference
i Scho¨rcketal.2009;Lietal.2010;Yongetal.2013b).
X therein), and to ultimately constrain the primordial nucle-
Recent studies using an increasing number of metal-poor
osynthesis of cosmological models as well (for details, see
r stars have made it clear that there exists a large fraction
a Bromm&Yoshida2011).
in such stars with significant enhancements of carbon, i.e.,
Ever since the discovery of metal-poor stars by
carbon-enhancedmetal-poor(CEMP)stars. Thefrequencyof
Chamberlain&Aller (1951), numerous efforts have been
CEMP stars is suggested to increase with decreasing metal-
devoted to related fields and remarkable progress has been
licity (Cohenetal. 2005; Lucatelloetal. 2006; Carolloetal.
achieved. To explore the nature of the first generation of
2012;Spiteetal.2013;Yongetal.2013b). Largersamplesof
stars, detailed abundance analyses based on high-resolution
CEMPstarsalsoindicatedifferenttypesofchemicalpatterns,
mainlyincludingfoursub-classesdefinedbasedontheabun-
[email protected],[email protected]
1Key Lab of Optical Astronomy, National Astronomical Observato- dances of the neutron-capture elements (Beers&Christlieb
ries,ChineseAcademyofSciences,A20DatunRoad,Chaoyang,Beijing 2005): CEMP-s (enhanced in s-process elements), CEMP-
100012,China r (enhanced in r-process elements), CEMP-rs (enhanced in
2Zentrum fu¨r Astronomie der Universita¨t Heidelberg, Landesstern-
both s-andr-processelements),andCEMP-no(noenhance-
warte,Ko¨nigstuhl12,D-69117Heidelberg,Germany
3NanjingInstituteofAstronomicalOptics&Technology,NationalAs- ment in neutron-capture elements). This diversity in chem-
tronomicalObservatories,ChineseAcademyofSciences,Nanjing210042, istrysuggestsdifferentsitesofcarbonproductionintheearly
China Galaxy,anddetailedanalysesoftheelementalabundancesof
4Thenotationof[A/B]=log(NA/NB)⋆−log(NA/NB)⊙isusedhere,where CEMPstarsallosustounderstandthenatureoftheirprogen-
NAandNBarethenumberdensitiesofelementsAandBrespectively,and⋆ itorstars.
and⊙refertothestarandtheSunrespectively
2 Lietal.
As a key to all these investigations, the number of metal- thesurveyefficiency. Therefore,weareusingthisfacilityto
poor stars has been tremendously increased by recent wide- furtherincreasethenumberofmetal-poorstars.
angle sky surveys, including the HK survey (Beersetal. Thispaperis organizedasfollows. Observationsand data
1992) and the Hamburg/ESO survey (HES; Christliebetal. reduction are addressed in Section 2. The determination of
2008), and more recently using data from the Sloan Dig- stellar parameters and abundance analysis are described in
ital Sky Survey (Yorketal. 2000) and the Sloan Exten- Section3. Theinterpretationsanddiscussionsofthederived
sion for Galactic Understanding and Exploration (SEGUE; elemental abundances including comparisons with literature
Yannyetal. 2009; Rockosietal. 2009), as well as the valuesarepresentedinSection4. Theconclusionsaregiven
SkyMapper Telescope (Kelleretal. 2007) using the novel inSection5.
photometricfiltersystem.
Subsequently, high-resolution spectroscopic research has 2. TARGETSELECTIONANDOBSERVATION
been performed for the metal-poor stars found in these Sample stars were selected from the low-resolution
surveys (e.g., McWilliametal. 1995; Norrisetal. 2001; database (R = 1800)spectra acquiredwith LAMOST. After
Cohenetal.2004;Barklemetal.2005).Studiesoflargesam- robust estimation of stellar parameters and selection of can-
pleshavebeenconducted,includingthe“FirstStars”project didates using LAMOST data, high-resolution(R ∼ 35,000)
(Cayreletal. 2004; Bonifacioetal. 2009), the “0Z project” spectroscopy was obtained for eight interesting objects with
(Cohenetal. 2011), “The Most Metal-Poor Stars” project Magellan/MIKE to carry out detailed abundance analysis of
(Norrisetal. 2013a; Yongetal. 2013a), and the “Extremely thesestars.
Metal-Poor Stars from SDSS/SEGUE” (Aokietal. 2013)
project. Theseeffortshaveresultedindetailedinvestigations 2.1. Low-resolutionSpectroscopicObservationwith
of the abundances of more than 200 EMP stars, including LAMOST
twohypermetal-poor(HMP)starsHE1327−2326([Fe/H]=
Our study is based on the first data release of the LAM-
−5.4,Frebeletal.2005)andHE0107−5240([Fe/H]=−5.3,
OSTsurveywhichwasinternallyreleasedafterthefirstyear
Christliebetal. 2002), and J0313−6708, a star at [Fe/H] <
of LAMOST survey operations. The wavelength coverage
−7.1whichissuspectedtohavebeenpre-enrichedbyasingle
(3700–9100Å)andresolvingpower(R=1800)oftheLAM-
supernova(Kelleretal.2014).
OST spectraallow a robustestimationof the stellar parame-
These surveys and follow-ups have not only identified
ters, includingmetallicities. Threeindependentmethodsare
chemicallypeculiarlow-metallicityobjectsandmadeitpos-
usedtodeterminethemetallicityofanobject.
sible to explorethe evolutionof the Galactic haloin a much
Thefirstmethodisanapplicationofanupdatedversionof
more detailed and systematic manner, but have also led to
the methods described by Beersetal. (1999), which obtain
a number of debates. Recent discoveries of more EMP
stars with [Fe/H] < −3.5 (Yongetal. 2013a; Placcoetal. [Fe/H] by makinguse of the Ca II K line indexKP measur-
ingthestrengthofthislineandtheHP2indexmeasuringthe
2014)haveenabledustodeterminemoreaccuratelythelow-
strengthoftheH line.Thismethodhasalsobeenadoptedby
metallicitytailoftheMDFoftheGalactichalo. Itwasfound δ
previousinvestigationssuchasScho¨rcketal.(2009)towhich
that the sharp cutoff at [Fe/H] ∼ −3.6 previously observed
wereferthereaderfordetails.
in HES data (Scho¨rcketal. 2009; Lietal. 2010) is proba-
Theothertwomethodsmakeuseofagridofsyntheticspec-
bly an artifact caused by small number statistics, and that
tra.Thesespectrawerecomputedbyusingthesynthesiscode
the MDF in fact smoothly decreases at least down to ap-
SPECTRUM (Gray&Corbally 1994), starting from the AT-
proximately [Fe/H] ∼ −4.1 (Yongetal. 2013b). With ex-
LAS9gridofstellaratmosphericmodelsofCastelli&Kurucz
tendedsamplesofEMPstars,Placcoetal.(2014)alsoraises
(2003)withlinearinterpolationinthreedimensions,i.e.,T ,
doubts about the cutoff of the abundance ratio of [Sr/Ba] eff
logg,and[Fe/H]. Amicroturbulencevelocityof2kms−1was
at about [Fe/H] ∼ −3.6, which was previously pointed out
adopted for all of the spectra. The line list and atomic data
by Aokietal. (2013). All of these unsolved questions ur-
gently require additional EMP stars, especially objects with were taken from SPECTRUM 6. The synthetic spectra have a
[Fe/H]<−3.5,tobetterunderstandthetruthandtherealpic- resolution of 0.01Å, covering 4000K ≤ Teff ≤ 9000K in
tureoftheearliestphasesofGalacticchemicalevolution. steps of 250K, 0.0 ≤ logg ≤ 5.0 in steps of 0.5dex, and
TheLargeskyAreaMulti-ObjectfiberSpectroscopicTele- −4.0 ≤ [Fe/H] ≤ −0.5 in steps of 0.5dex. Considering the
scope (LAMOST, also known as the Wang-Su Reflecting factthatwearesearchingforcandidatemetal-poorstars, the
SchmidtTelescopeortheGuoshoujingTelescope)5 isanew [α/Fe]was set to +0.4dex, which is the typicalvalue in this
type of wide-field telescope with a large aperture (Cuietal. [Fe/H]region(e.g.,McWilliam1997).
2012). LAMOST started its pilot survey in 2010 (Luoetal. Thesecondmethodusescomparisonsoflineindices.Based
2012), and has conducted a five year regular survey since onthemostup-to-datewavelengthdefinitionsofLickindices
2011. The combination of a large aperture (4m) and high 7,wehavecalculatedcorrespondingindicesforallofthesyn-
multiplex factor (4000 fibers) with a 5◦ field of view makes theticspectra,resultinginalineindexgridcoveringthecor-
it a uniquefacility. Its special designand highspectrum ac- responding space of stellar parameters. The same series of
quiring rate allow LAMOST to efficiently carry out system- indices are then calculated for each observed spectrum, and
atic spectroscopic surveys of the various stellar components compared with this synthetic grid to find the best match of
oftheGalaxy(Zhaoetal.2006,2012),includingmetal-poor parameters.
stars. Furthermore,withthelow-resolution(R = 1800)spec- Thethirdmethodthatweadoptisbasedonadirectcompar-
troscopicdatafromLAMOST,itispossibletoreliablyiden- isonofthenormalizedobservedfluxwiththenormalizedsyn-
tifymetal-poorstarsinsurveymode,thusevidentlyenhancing theticspectrainthewavelengthrange4500Å≤ λ ≤ 5500Å.
5 See http://www.lamost.org/public/ for more detailed informa- 6http://www.appstate.edu/grayro/spectrum/spectrum.html
tion,andtheprogressoftheLAMOSTsurveys. 7http://astro.wsu.edu/worthey/html/index.table.html
3
TABLE1
LogoftheMagellan/MIKEobservations.
ID R.A. Decl. r Date Exptime S/N vr
(mag) (s) (pixel−1) (kms−1)
LAMOSTJ0006+1057 000617.20 +105741.8 12.59 2013Dec 2700 86 −281.6
LAMOSTJ0102+0428 010212.66 +042824.2 10.87 2013Aug 360 72 −201.3
LAMOSTJ0126+0135 012658.58 +013515.4 12.30 2013Dec 1800 75 −197.3
LAMOSTJ0257−0022 025706.93 −002233.7 14.97 2013Aug 1890 48 2.7
LAMOSTJ0326+0202 032653.88 +020228.1 11.55 2013Dec 1200 106 121.2
LAMOSTJ0343−0227 034322.87 −022756.1 13.43 2013Dec 2400 62 −4.3
LAMOSTJ1626+1721 162614.78 +172112.1 14.29 2013Aug 1300 56 −177.1
LAMOSTJ1709+1616 170959.79 +161613.3 13.13 2013Aug 900 43 −324.2
Note.—TheS/Nratioperpixelwasmeasuredatλ∼4500Å.
TABLE2
EquivalentwidthMeasurementsandLine-by-lineAbundances.
J0006+1057 J0102+0428 J0126+0135 J0257−0022
Ion λ χ loggf W logǫ(X) σ W logǫ(X) σ W logǫ(X) σ W logǫ(X) σ
(Å) (eV) (mÅ) (mÅ) (mÅ) (mÅ)
NaI 5889.95 0.00 0.108 134.7 3.05 0.09 186.1 3.69 0.09 146.3 2.86 0.11 ... ... ...
NaI 5895.92 0.00 −0.194 ... ... ... 155.8 3.50 −0.09 112.7 2.65 −0.11 ... ... ...
MgI 3829.35 2.71 −0.208 ... ... ... 198.4 5.36 0.12 134.0 4.38 −0.07 119.7 5.30 −0.17
MgI 4167.27 4.35 −0.710 ... ... ... 52.8 5.15 −0.09 ... ... ... ... ... ...
MgI 4702.99 4.33 −0.380 61.5 4.95 −0.05 69.8 5.03 −0.22 26.7 4.26 −0.19 45.1 5.49 0.02
MgI 5172.68 2.71 −0.450 ... ... ... 211.2 5.36 0.12 161.7 4.49 0.04 117.1 5.40 −0.07
MgI 5183.60 2.72 −0.239 187.8 5.10 0.09 230.4 5.37 0.13 178.9 4.56 0.11 syn 5.55 0.08
MgI 5528.40 4.34 −0.498 ... ... ... 80.4 5.25 0.01 ... ... ... 43.5 5.59 0.12
AlI 3943.99 0.00 −0.640 ... ... ... 154.4 3.56 −0.08 syn 2.27 0.08 60.6 3.43 −0.02
AlI 3961.52 0.01 −0.340 112.5 2.61 0.05 182.4 3.72 0.08 syn 2.12 −0.08 77.3 3.46 0.02
SiI 3905.52 1.91 −1.092 ... ... ... 203.1 5.22 0.09 ... ... ... 127.2 5.80 0.07
SiI 4102.94 1.91 −3.140 61.9 4.72 0.03 85.3 5.05 −0.09 61.9 4.47 0.03 ... ... ...
Note. —Thethree columns foreach objects refertothemeasuredequivalent width, corresponding abundance inlogǫ(X), andtheuncertainty causedbythe
equivalentwidthmeasurement.The“syn”indicatesthattheabundancehasbeenderivedusingspectralsynthesis.
(Thistableisavailableinitsentiretyinamachine-readableform.)
Theχ2 minimizationtechniqueofLeeetal.(2008)isusedto ForJ0326+0202,high-resolutionobservationandabundance
findthebest-matchingsetofparameters. analysis have been performed as described in Holleketal.
Taking into account the typical uncertainty of ∼ 0.1 – (2011), and we include this object for comparison purpose.
0.3dexfor metallicitiesderivedfromlow-resolutionspectra, For the latter object, J0126+0135, since there has been no
an objectis regardedas an EMPcandidateif it is within the high-resolution observation to confirm this object, we have
temperaturerange4000K<T <7000K,andifatleasttwo alsoincludeditintoourtargetlist.
eff
of the three methods described above yield [Fe/H] ≤ −2.7. The setup of the observationsincluded a 0.7 slit with 2×
¨
The constraint on T is adopted to exclude low-luminosity, 2 and 1 × 2 binning CCD readout modes (used for the two
eff
foregroundlate-typestarsbelongingtothediskpopulation(s), runsrespectively),yieldingresolvingpowersofR∼35,000in
and to ensure that only stars redward of the main-sequence the blue spectral range (3300 — 4900Å) and R ∼28,000 in
turnoffforstarswithagesof∼ 13Gyr(aspredictedbytheo-
theredspectralrange(4900—9400Å).Thespectrahavean
reticalisochrones)enterthesample.Thespectraoftheabove
averagesignal-to-noiseratioperpixelofS/N ∼75at4500Å.
automatically selected candidates were visually inspected to
Moreinformationabouttheobservations,aswellasthetarget
removefalse positives such as coolwhite dwarfs, or objects
coordinatesandmagnitudes,arelistedinTable1.
thatwereselectedbecausetheirspectraweredisturbedbyre-
The raw data were reducedwith the standard IRAF 8 pro-
ductionartifacts.
cedurestoobtain1Dspectrathatareflat-fielded,wavelength-
2.2. High-resolutionSpectroscopywithMagellan/MIKE calibrated, co-added,and continuum-normalized. The radial
velocities were obtained using the IRAF procedure fxcor,
ForeightofcandidatesselectedasdescribedinSection2.1,
employingasyntheticspectrumwithlow-metallicityasatem-
high-resolution spectra were acquired during two runs in
plate.
2013Augustand2013December,withtheMagellanInamori
Equivalentwidthswere measuredby fitting Gaussian pro-
KyoceraEchellespectrograph(MIKE;Bernsteinetal.2003)
at the Magellan-Clay Telescope at Las Campanas Observa-
8 IRAF is distributed by the National Optical Astronomy Observatory,
tory. Among these objects, J0126+0135 and J0326+0202
whichisoperatedbytheAssociationofUniversitiesforResearchinAstron-
werepreviouslyselectedfromHES(Frebeletal.2006),also omy,Inc.,undercooperative agreementwiththeNationalScienceFounda-
known as HE 0124−0119and HE 0324+0152, respectively. tion.
4 Lietal.
TABLE3
AdoptedStellarParametersoftheProgramStars.
MIKEMeasurements LAMOSTMeasurement
ID Teff logg [Fe/H] ξ Teff logg [Fe/H]
(K) (km−1) (K)
LAMOSTJ0006+1057 4520 0.6 −3.26 2.2 4800 1.6 −3.03
LAMOSTJ0102+0428 4540 0.6 −2.75 2.5 4940 1.2 −2.78
LAMOSTJ0126+0135 4330 0.1 −3.57 2.8 4810 1.2 −3.15
LAMOSTJ0257−0022 6330 4.2 −2.24 1.5 6260 3.6 −2.65
LAMOSTJ0326+0202 4700 1.1 −3.36 1.8 4830 1.8 −3.15
LAMOSTJ0343−0227 4790 1.5 −2.42 2.0 4795 1.8 −2.57
LAMOSTJ1626+1721 5930 3.6 −3.20 1.6 5830 3.4 −3.00
LAMOSTJ1709+1616 5780 3.5 −3.71 1.9 6070 3.0 −3.33
filestoisolatedatomicabsorptionlinesbasedonthelinelist
of Frebeletal. (2013), supplemented with K I, Mn I, and
Y II linesfromPlaccoetal. (2014). Table1 showsthemea-
surementsofthe adoptedatomiclinesforallofthe program
stars. Figure 1 compares our measured equivalent widths 40
)
ofJ0326+0202with thosemeasuredbyHolleketal. (2011). ek
Thereisameandifferenceof−0.6mÅwithσ = 3.4mÅbe- ll 20
o
tweenourmeasurementsandthoseofHolleketal.(2011)for H
linesincludedinouranalysis. - 0
s
i
3. STELLARPARAMETERSANDABUNDANCE th -20
(
ANALYSIS
W
Forourabundanceanalysis,we use1Dplane-parallel,hy- E -40
drostatic model atmospheres of the ATLAS NEWODF grid
4000 5000 6000
of Castelli&Kurucz (2003), assuming a mixing-length pa-
Wavelength (Angstrom)
rameter of α = 1.25, no convective overshooting, and
MLT
local thermodynamic equilibrium. We use an updated ver-
sion of the abundanceanalysis code MOOG (Sneden 1973), 200
whichtreatscontinuousscatteringasasourcefunctionwhich
)
sums both absorptionand scattering rather than true absorp- 11
0
tion(Sobecketal.2011). 2 150

We adopt the photospheric Solar abundances of .
l
Asplundetal. (2009) when calculating the [X/H] and a
[X/Fe]abundanceratios. et 100
k
3.1. AtmosphericParameters le
l
TheeffectivetemperaturesTeff ofthestarsweredetermined (Ho 50
by minimizing the trend of the relationship between the de-
W
rived abundancesand the excitation potentials of Fe I lines. E
Previous investigations and experience have shown that this
0
procedure yields effective temperatures with systematic off-
0 50 100 150 200
setscomparedtothosedeterminedfrome.g.,broadbandopti- EW (this work)
calandinfraredphotometry. Based onaliteraturesampleof
metal-poorstars,Frebeletal.(2013)derivesalinearrelation
betweenthespectroscopicandphotometriceffectivetempera- Fig.1.—ComparisonoftheequivalentwidthsofJ0326+0202,respectively,
tures,andthisrelationcanbeusedtocorrectsuchsystematic measuredbythisworkandHolleketal.(2011).Top:residualsoftheequiv-
alentwidths(thiswork-Hollek)vs.thewavelength.Bottom:directcompar-
deviations. Therefore,wehaveadoptedthiscorrectiontoob-
isonofthetwomeasurementsoftheequivalentwidths. Theone-to-oneline
tainthefinaleffectivetemperaturesofoursample. isplottedforreference.
Microturbulentvelocitiesξweredeterminedfromtheanal-
ysisofFe I lines, byforcingtheironabundancesofindivid-
uallinestoexhibitnodependenceonthereducedequivalent no Fe II lines could be detected in their spectra. Therefore,
widths.ForobjectsthathaveasufficientnumberofFeIIlines we resorted to the theoretical isochrones of Demarqueetal.
detected, the surface gravity logg was determined by min- (2004)toestimatethesurfacegravitiesfortheseobjectsgiven
imizing the difference between the average abundances de- their temperaturesand metallicities. Isochroneswith an age
rivedfromtheFeIandFeIIlines. of 13Gyr were adopted, and all three parameters (i.e., Teff,
However,therearetwoobjectsatlowmetallicities,namely logg, ξ) were iterated until convergence was reached, i.e.,
J1626+1721 and J1709+1616, for which only spectra of with the derivedsurfacegravity, the “zero”slope canbe ob-
lower-than-average S/N are available; therefore only one or tained for derived abundances versus excitation potentials,
5
and abundances versus EWs, for corresponding T and ξ. presented in the left panels of Figure 3. The molecular line
eff
Forthesetwoobjects, twosolutionsofsurfacegravitycould data for the spectrum synthesis were taken from the Kurucz
be deduced from the theoretical isochrones for each set of database. 9 Thelinebroadeningwasdeterminedusingasin-
temperature and metallicity, e.g., logg of 4.57 and 3.62 for gle Gaussian profile representing the combined effects from
J1626+1721, and 4.63 and 3.52 for J1709+1616. However, the instrumental profile, atmospheric turbulence, and stellar
onlythesub-giantloggvaluecouldresultinconvergencedur- rotation. The width of the Gaussian was determined by fit-
ing the iterations, and therefore3.62 and 3.52 were adopted tinganumberofisolatedandclearFe IandTiIIlineswhose
forJ1626+1721andJ1709+1616,respectively. abundanceshavebeenderivedfromEWmeasurements.
The nitrogen abundance was measured from the CN-BX
electronic transition at 3883Å, by comparing the observed
spectrawithsyntheticspectrageneratedwithMOOG(seethe
0 [Fe/H]=-2.5 rightpanelsofFigure3forexamples). Wecouldonlydeter-
[Fe/H]=-3.0 mine the nitrogen abundancesof seven of the program stars
[Fe/H]=-3.5
whose CN-BX features are strong enough for reliable mea-
1
surements.WedidnotusetheNHbandat3360Å,becauseof
anS/N(typicallylowerthanfive)ofourspectraatthatspec-
2 tralregionthatistoolowtoperformspectralsynthesis.
g
The carbon and nitrogen abundancesof the program stars
g
o arelistedinTable1.
l
3
3.3. AbundancesDeterminedfromAtomicLines
For 19 elements between Na and Ba, the abundances
4
werecomputedusingthemeasuredequivalentwidthsandthe
atomicdatapresentedinTable1,andthestellarparametersin
Table1.
5
Weperformedsanitychecksforcaseswheretheabundance
6500 6000 5500 5000 4500
Teff ofanelementwasderivedfroma singleline, orwhenabun-
dancesdeviatedbymorethan3σfromtheaveragecomputed
Fig.2.—Teff vs. loggofthesample. Filledcirclescorrespondtocarbon- foranatomicspeciesfrommultiplelines. Inthesecases,the
enhancedobjects,andtheopencirclesrefertothecarbon-normalones,whose
abundanceswereverifiedwithspectralsynthesisandmodified
definitionisdescribedinSection3.2. Forreference, the13Gyrtheoretical
isochrones with [α/Fe] = +0.4and [Fe/H] = −2.5, −3.0, and −3.5 from ifnecessary. Forclearlydetectedandisolatedlines,synthetic
Demarqueetal.(2004)areplottedwithdash-dotted,solidanddashedlines spectrawithcorrespondingabundancesderivedfromthemea-
respectively. suredequivalentwidthscouldwellmatchtheobservedspec-
tra. However, for lines that suffer from blending or prob-
Duringtheiterationsfordeterminingthestellarparameters, lems in setting the continuum level, the abundance derived
any Fe line with an offset of more than 3σ from the mean fromequivalentwidthscouldnotproperlymatchtheobserved
abundancewasrejected.Theadoptedstellarparametersofall spectralline, andhencetheabundancederivedfromspectral
theprogramstarsarelistedinTable1. Thederivedstellarpa- synthesiswasadopted.Table1liststheusedatomiclinedata,
rametersofJ0326+0202areveryclosetothevaluesgivenby themeasuredequivalentwidths,aswellasthederivedabun-
Holleketal. (2011), with a differenceof−75K in Teff, −0.1 dances,wherelinesmarkedas“syn”refertoabundancesde-
inlogg,and−0.1in[Fe/H],respectively.Theparametersde- terminedwithspectralsynthesis.
rivedfromLAMOST low-resolutionspectra arealso shown, Foralloftheprogramstars,theabundancesofαelements
usingtheaveragevaluesfortheparametersderivedfromthe (Mg,Si,Ca,andTi)havebeendetermined. Amongthelight
threemethodsasdescribedinSection2.1. Arelativelylarge odd-Zelements(Na, Al, K, and Sc), abundancesof Sc have
discrepancy is found in logg for the cool giants, which is been determined for all of the program stars, while for the
partlydueto thelargesteps(0.5dex)inlogg oftheadopted warmerstarsnotallthelinesaredetectable,e.g.,thereddou-
synthetic spectra template. The general agreement between blet K lines are too weak in J0257−0022, J1626+1721, and
thetwosetsofparametersshowsthattheLAMOSTselection J1709+1616to derive an abundance. For Fe-peak elements
pipeline is able to derive reliable parameters, especially Teff (Cr, Mn, Co, Ni, and Zn), the abundances of Cr and Mn
and [Fe/H]. In addition, the [Fe/H] derived from LAMOST couldbe determinedforall oftheprogramstars; meanwhile
spectra for J0126+0135(−3.15)is quite consistentwith that the abundances of Co, Ni, and Zn could not be derived for
fromHESfollow-upobservations(−3.1,Frebeletal.2006). the most metal-deficient object, J1709+1616 due to its low
The distribution of Teff versus logg of all of the program Fe abundance and possibly insufficient S/N. In the cases of
starsisplottedinFigure2,withfilledandopencirclesrefer- theheavyelements(Sr,Y,Ba,andEu),theabundancesofSr
ring to the carbon-enhanced and carbon-normal stars in the and Ba were determined for all of the program stars except
sample (as will be described in Section 3.2); the theoretical J1709+1616(themostmetal-poor)andJ1626+1721,respec-
isochrones from Demarqueetal. (2004) are also shown for tively;meanwhileabundancesofYcouldonlybedetermined
reference. forfourofthesestars,andtheEulineshavebeendetectedand
measuredinonlytwoobjects,J0102+0428,andJ0343−0227.
3.2. CarbonandNitrogen
All of the elemental abundances of the programs stars are
The carbonabundancesof the programstars were derived showninTable1. Wehavealsocomparedtheabundancesof
bymatchingtheobservedCH-AXbandheadat4310Å (i.e.,
theGband)tothesyntheticspectra.Examplesofsuchfitsare 9http://kurucz.harvard.edu/molecules.html
6 Lietal.
Fe
1.0 Co Ti CN
1.0
x
u
Fl 0.8 0.9

e
v
i
t 0.8
a
l 0.6
e
R
J0257-0022 J0257-0022
0.7
[C/Fe]=+0.56 [N/Fe]=+2.26
[C/Fe]=+0.76 [N/Fe]=+2.41
[C/Fe]=+0.96 [N/Fe]=+2.56
0.4
0.6

Fe
1.0 Co Ti CN
1.0
x 0.9
u 0.8
l
F
0.8
e
v
ti 0.6 0.7
a
l
e
R 0.6
J0006+1057 J0006+1057
0.4
[C/Fe]=-0.42 [N/Fe]=+1.33
[C/Fe]=-0.32 0.5 [N/Fe]=+1.48
[C/Fe]=-0.22 [N/Fe]=+1.63
0.2
4307 4308 4309 4310 4311 4312 3880 3881 3882 3883
Wavelength Wavelength
Fig.3.—Left: examplesoffittingoftheCHbandusedtodeterminetheCabundance. Right: examplesoffittingoftheCNbandusedtodeterminetheN
abundance.Thedottedlinefilledcirclerepresentstheobservedspectra;thesolidlinereferstothebestfit,andtheupperandlowerdash-dottedlinescorrespond
tosyntheticspectrawithchangesof1σfittinguncertaintyin[C/Fe]and[N/Fe].TheupperandlowerpanelsrefertothefittingofthespectraofJ0257−0022and
J0006+1057,respectively.
J0326+0202with those from Holleketal. (2011), as shown
inFigure4. Thedifferencebetweentheabundancefromthis 1.0
work and Hollek is −0.14 ± 0.18, indicating a rather good
)
agreement. TheonlyspecieswithsignificantdifferenceisY. k
e
SinceourmeasurementshaveusedthreeY linesresultingin ll
o 0.5
astandarddeviationof0.08inthederivedabundance,further H

informationshouldberequiredtoexplainthedifference. -
k
r
o
3.4. AbundanceErrorEstimation w 0.0
s
Theerrorofthederivedabundancesmainlyarisesfromtwo hi
t
aspects, the uncertainties in the equivalent width measure- (

ments and those caused by the uncertaintiesfrom the stellar )
X
parameters. A( -0.5

WhenN ≥2linesofindividualspeciesofanelementwere g
o
observed,thedispersioninthemeasurementsofmultiplelines l
around the average abundances, i.e., σlogǫ(X), was used to -1.0
presenttheerrorcausedbyrandomuncertaintiesintheequiv- 0 10 20 30 40 50 60
alentwidths,asgiveninTable1. Iftheelementalabundance Atomic Number
was determined from a single line, the statistic error of the
Fig.4.—ComparisonoftheabundancesofJ0326+0202fromthisworkwith
equivalent widths was estimated based on the classical for-
theresultfromHolleketal.(2011),intheformoftheabundancedifference
mulaofCayrel(1988): vs. theatomicnumbers. Thedashedlinesrefertothestandarddeviationof
theabundancedifference (±0.18). Themostdeviated pointcorresponds to
σ =h∆W2i1/2 ≃1.6(w∆x)1/2ǫ (1) theelementY,andisdescribedinthetext.
EW
7
TABLE4
AbundancesofIndividualElementsfortheProgramStars.
LAMOSTJ0006+1057 LAMOSTJ0102+0428 LAMOSTJ0126+0135 LAMOSTJ0257−0022 Sun
logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X)
CH(C) 4.85 −0.32 0.10 1 4.75 −0.93 0.15 1 4.35 −0.51 0.15 1 6.95 0.76 0.20 1 8.43
CN(N) 6.05 1.48 0.15 1 6.45 1.37 0.15 1 5.70 1.44 0.20 1 8.00 2.41 0.15 1 7.83
NaI 3.05 0.07 0.09 1 3.59 0.10 0.13 2 2.75 0.08 0.15 2 ... ... ... ... 6.24
MgI 5.01 0.66 0.08 3 5.24 0.39 0.13 7 4.45 0.42 0.13 5 5.45 0.09 0.12 5 7.60
AlI 2.61 −0.58 0.05 1 3.64 −0.06 0.11 2 2.20 −0.68 0.11 2 3.44 −0.77 0.03 2 6.45
SiI 4.72 0.47 0.03 1 5.14 0.38 0.12 2 4.47 0.53 0.03 1 5.80 0.53 0.07 1 7.51
KI 2.51 0.74 0.01 2 2.86 0.58 0.08 1 2.01 0.55 0.17 1 ... ... ... ... 5.03
CaI 3.46 0.38 0.07 8 3.84 0.25 0.09 18 3.21 0.44 0.09 4 4.38 0.28 0.11 6 6.34
ScII −0.09 0.02 0.07 5 0.23 −0.17 0.11 7 −0.63 −0.22 0.12 5 1.04 0.13 0.05 2 3.15
TiI 1.84 0.15 0.11 6 2.19 −0.01 0.12 21 1.74 0.36 0.14 3 3.32 0.61 0.19 2 4.95
TiII 1.87 0.18 0.12 12 2.27 0.07 0.14 45 1.35 −0.03 0.18 16 3.18 0.47 0.10 15 4.95
VII 0.96 0.29 0.04 1 1.22 0.04 0.12 3 0.60 0.24 0.04 1 ... ... ... ... 3.93
CrI 1.91 −0.47 0.16 5 2.65 −0.24 0.14 15 1.43 −0.64 0.14 6 3.18 −0.22 0.08 3 5.64
CrII ... ... ... ... 3.09 0.20 0.07 1 ... ... ... ... ... ... ... ... 5.64
MnI 1.67 −0.50 0.11 4 2.12 −0.56 0.12 5 0.86 −1.00 0.06 3 2.65 −0.54 0.09 3 5.43
FeI 4.24 0.00 0.14 94 4.75 0.00 0.15 164 3.93 0.00 0.19 61 5.26 0.00 0.14 60 7.50
FeII 4.24 0.00 0.13 11 4.76 0.01 0.10 19 3.93 0.00 0.15 4 5.25 −0.01 0.14 3 7.50
CoI 1.87 0.14 0.04 1 2.13 −0.11 0.16 7 1.59 0.17 0.01 2 3.03 0.28 0.12 1 4.99
NiI 2.90 −0.06 0.06 3 3.25 −0.22 0.18 6 2.57 −0.08 0.07 1 4.07 0.09 0.12 2 6.22
ZnI 1.72 0.42 0.04 2 1.80 −0.01 0.10 1 1.43 0.44 0.07 2 2.71 0.39 0.14 1 4.56
SrII −2.46 −2.07 0.06 2 −1.06 −1.18 0.14 2 −2.31 −1.61 0.04 1 1.05 0.42 0.12 2 2.87
YII ... ... ... ... −1.62 −1.08 0.03 3 ... ... ... ... −0.08 −0.05 0.11 1 2.21
BaII −2.83 −1.75 0.03 1 −1.26 −0.69 0.13 5 −2.53 −1.14 0.16 3 −0.17 −0.11 0.00 2 2.18
EuII ... ... ... ... −2.13 0.10 0.08 1 ... ... ... ... ... ... ... ... 0.52
TABLE4
Continued.
LAMOSTJ0326+0202 LAMOSTJ0343−0227 LAMOSTJ1626+1721 LAMOSTJ1709+1616 Sun
logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X) [X/Fe] σ N logǫ(X)
CH(C) 5.15 0.08 0.10 1 5.90 −0.11 0.15 1 6.30 1.07 0.20 1 6.30 1.58 0.20 1 8.43
CN(N) 5.70 1.23 0.15 1 6.35 0.94 0.15 1 ... ... ... ... ... ... ... ... 7.83
NaI 3.46 0.58 0.01 2 3.26 −0.02 0.14 2 3.02 −0.02 0.03 2 ... ... ... ... 6.24
MgI 4.99 0.75 0.17 8 5.58 0.40 0.03 4 4.90 0.50 0.16 3 4.17 0.28 0.12 3 7.60
AlI 2.36 −0.73 0.04 1 3.50 −0.53 0.06 1 2.92 −0.33 0.08 1 ... ... ... ... 6.45
SiI 4.93 0.78 0.07 2 5.52 0.43 0.04 1 4.77 0.46 0.09 1 3.98 0.18 0.10 1 7.51
KI 2.49 0.82 0.03 2 3.12 0.51 0.08 1 ... ... ... ... ... ... ... ... 5.03
CaI 3.40 0.42 0.08 12 4.23 0.31 0.11 14 3.41 0.27 0.08 2 3.05 0.42 0.11 1 6.34
ScII −0.23 −0.02 0.04 5 0.65 −0.08 0.07 5 0.02 0.07 0.11 1 −0.37 0.19 0.14 1 3.15
TiI 1.81 0.22 0.06 7 2.70 0.17 0.11 15 ... ... ... ... ... ... ... ... 4.95
TiII 1.75 0.16 0.10 32 2.79 0.26 0.11 27 1.92 0.17 0.11 11 1.90 0.66 0.13 10 4.95
VII 0.60 0.03 0.03 1 1.52 0.01 0.05 1 ... ... ... ... ... ... ... ... 3.93
CrI 1.92 −0.36 0.16 6 3.00 −0.22 0.08 9 2.19 −0.25 0.06 3 1.64 −0.29 0.15 1 5.64
CrII ... ... ... ... 3.31 0.09 0.06 1 ... ... ... ... ... ... ... ... 5.64
MnI 1.42 −0.65 0.09 3 2.80 −0.21 0.09 6 2.04 −0.19 0.10 1 1.65 −0.07 0.11 1 5.43
FeI 4.14 0.00 0.10 127 5.08 0.00 0.13 111 4.30 0.00 0.13 46 3.79 0.00 0.14 24 7.50
FeII 4.13 −0.01 0.11 7 5.08 0.00 0.15 14 ... ... ... ... ... ... ... ... 7.50
CoI 1.82 0.19 0.10 5 2.60 0.03 0.07 2 2.42 0.63 0.10 1 ... ... ... ... 4.99
NiI 2.89 0.03 0.09 5 3.90 0.10 0.09 7 3.19 0.17 0.14 2 ... ... ... ... 6.22
ZnI 1.42 0.22 0.04 2 2.35 0.21 0.06 1 ... ... ... ... ... ... ... ... 4.56
SrII −1.18 −0.69 0.01 2 0.29 −0.16 0.07 2 −0.84 −0.51 0.03 2 ... ... ... ... 2.87
YII −1.98 −0.83 0.08 3 −0.58 −0.37 0.09 4 ... ... ... ... ... ... ... ... 2.21
BaII −2.55 −1.37 0.03 2 −0.58 −0.34 0.18 4 ... ... ... ... −1.54 −0.01 0.18 1 2.18
EuII ... ... ... ... −1.99 −0.09 0.04 2 ... ... ... ... ... ... ... ... 0.52
Note.—Thecolumnof“N”referstothenumberoflinesadoptedfordeterminationoftheelementalabundances.ThephotosphericsolarabundancesfromAsplundetal.
(2009)arealsoshownforreference.
8 Lietal.
sumingatypicalmassofM =0.8M forhalostars.
⊙
TABLE5 Figure 5a shows that there are four objects in our sample
Uncertaintiesoflogǫ(X)Propagatedfromthe
that are above the limit, among which there is a red giant
StellarParameters(asDescribedinSection3.4),
ComputedforJ0326+0202asanExample. J0126+0135thatislocatedslightlyoffthelimitandhasvery
low [C/Fe] = −0.51). Therefore, we have decided to clas-
Ion ∆Teff ∆logg ∆ξ σtot sify onlythe threeobjectswith [C/Fe] > 0.7as CEMP stars
+150K +0.3dex +0.3kms−1 (filled circles in Figure 5) and the rest as carbon-normalob-
CH(C) 0.30 −0.15 −0.05 0.34 jects(opencircles).
CN(N) 0.25 −0.05 0.00 0.25 It is known that in addition to [C/Fe], [N/Fe] is notably
NaI 0.17 −0.03 −0.18 0.25 enhanced in many CEMP stars as well, and although this
MgI 0.13 −0.06 −0.08 0.16
phenomenon appears to be rare, a population of nitrogen-
AlI 0.15 −0.05 −0.18 0.24
SiI 0.16 −0.02 −0.03 0.16 enhanced metal-poor (NEMP) stars with [N/C] > 0.5 has
KI 0.12 −0.02 −0.02 0.12 been discovered (Johnsonetal. 2007; Polsetal. 2012). We
CaI 0.11 −0.03 −0.04 0.12 have adoptedthe criteria of Polsetal. (2012) for classifying
ScII 0.10 0.08 −0.08 0.15
stars as NEMP, i.e., [N/Fe] ≥ 1.0 and [N/C] ≥ 0.5. In Fig-
TiI 0.19 −0.03 −0.02 0.19
TiII 0.08 0.08 −0.08 0.14 ure5(b),itcanbeseenthatamongthesixprogramstarswith
VII 0.08 0.09 −0.01 0.12 determinednitrogenabundances,therearefivethatareonthe
CrII 0.18 −0.04 −0.09 0.21 rightsideofthedottedline,andhencearenitrogen-enhanced.
MnI 0.21 −0.05 −0.14 0.26
However,ascanbeseeninFigure2andofFigure5(a),four
FeI 0.17 −0.04 −0.10 0.20
FeII 0.02 0.09 −0.06 0.11 of them have [C/Fe] . 0.0 and [N/Fe] between 1.2 through
CoI 0.21 −0.03 −0.07 0.22 1.5, and are on the upper part of the red giant branch, and
NiI 0.20 −0.05 −0.20 0.29 thustheymaynotbegenuineNEMP,aswillbediscussedin
ZnI 0.09 0.04 0.01 0.10
SrII 0.11 0.07 −0.27 0.30 Section4.5.
YII 0.13 0.08 −0.01 0.15 If we use the separation of so-called “mixed” from “un-
BaII 0.14 0.08 −0.03 0.16 mixed”starsdefinedasinSpiteetal.(2005)(i.e.,thedashed
linein Figure5(b),correspondingto [N/Fe] = 0.5), the four
objectsbelong to the group of “mixed” stars. In these stars,
wherew istheFWHM oftheline, ∆x refersto thesampling nitrogenhasbeenproducedfromthe burningofcarbon,and
stepoftheMIKEspectra,andǫ referstothereciprocalspec- hence they are usually carbon-poor and nitrogen-rich. The
tral S/N in the case of normalized spectra. The uncertainty same holds true when the alternative criterion of Polsetal.
inthederivedabundancecorrespondingto2σ wasadopted (2012) of [(C+N)/Fe] > 0.9 is considered (the dash-dotted
EW
asa conservativeestimateoftherandomerrorcausedbythe line in Figure5(b)). The CEMPturn-offstar J0257−0022is
equivalentwidthmeasurement. Theerrorsrelevanttotheun- well located in the NEMP region, which makes it a poten-
certaintiesofthe equivalentwidth measurementare listed in tial carbon and nitrogen-enhancedmetalpoor (CNEMP) star
Table1 foreachindividuallinethathasbeenusedforabun- (Polsetal. 2012). We will further discuss this point in Sec-
danceestimation. tion4.5.
Theabundanceuncertaintiesassociatedwiththeuncertain-
4.2. AbundanceTrendsoftheLightElements
tiesofthestellarparameterswereestimatedbyvaryingT by
eff
+150K, logg by +0.3dex, and ξ by +0.3kms−1 in the stel- The abundance ratios [X/Fe] of the program stars are
lar atmosphericmodel. Table 1 summarizesthe correspond- plotted against [Fe/H] in the right panels of Figure 6 for
ing abundanceuncertaintiesand the total uncertaintyfor the the elements from C through Zn. The observed chemi-
three errors, which was computed as the quadratic sum, for cal abundances of metal-poor stars from the “First Stars”
J0326+0202asanexample.Wenotedthattheabundanceun- program (Cayreletal. 2004), and from Yongetal. (2013a)
certaintiescaused bythe equivalentwidthmeasurementsare and Placcoetal. (2014), are also plotted for comparison.
generallysmallerthanthosepropagatingfromthestellarpa- Note that the “First Stars” sample has been re-analyzed by
rameters,andtherefore,wehaveadoptedthelatteruncertain- Yongetal. (2013a), but since there are a few elements not
tiesasreferenceerrorbarsforfurtherdiscussionsconcerning included in their analysis, we have adopted the abundances
theabundancesofthesample. of the sample of “First Stars” from Cayreletal. (2004),
Spiteetal.(2005),andFranc¸oisetal.(2007).
4. RESULTSANDINTERPRETATION In general, the abundance ratios seen in our sample agree
wellwiththeabundanceratiotrendsdefinedbytheliterature
4.1. CEMPandNEMPObjects
samples. However, for the most iron-deficientobject in our
Despitethelimitedsamplesize,thereareseveralobjectsin sample(J1709+1616),andthecarbonandnitrogen-enhanced
oursampleexhibitingrelativelyhigh[C/Fe]ratios,asshown objectJ0257−0022,afewelements(inparticular,Ti,Mn,and
in Figure 5. To check the fraction of CEMP stars, we have Ba)showslightdeviationsfromthesetrends.
adoptedtheclassificationofAokietal.(2007),whosuggesta Comparedwiththoseofotherelements,dispersionsinthe
schemethattakesintoconsiderationthenucleosynthesisand distributionsofC,N,Na,andAlarenotablylarger.Unlikethe
mixing effects in giants. This is important for our sample, behavior found by Yongetal. (2013a), the abundance ratios
because five of our nine stars are giants (see Figure 2). We of the α elementsMg and Si (as wellas Ca) of the program
therefore consider a star to be CEMP when [C/Fe] ≥ +0.7 starsdistributearoundthetypicalhalovalueof[α/Fe] = 0.4
in the case of stars with luminosities of log(L/L ) ≤ 2.3, (e.g., McWilliam1997,thedottedlinesinFigure6).
⊙
or when [C/Fe] ≥ +3.0−log(L/L ) in the case of stars with Among the iron-peak elements, a small scatter and clear
⊙
log(L/L ) > 2.3. The luminositiesof oursample stars were dependenceon Fe are found for Cr and Zn. The abundance
⊙
computedbasedontheprescriptionofAokietal.(2007),as- ratioof [Cr/Fe] exhibitsthe lowestobservedscatter together
9
3 (a) (b)
2
]
e
F 1
/
C
[ J0257-0022 J0257-0022
0
-1
0 1 2 3 -1 0 1 2 3
log L/L [N/Fe]
Sun
Fig.5.— (a)[C/Fe]ratioasafunctionoftheluminosityestimatedfromTeff andloggofoursample. Thedottedlineindicates thedividinglinebetween
carbon-enhancedandcarbon-normalstarsasdefinedinAokietal.(2007).Thedashedlinecorrespondsto[C/Fe]=1.0.(b)[C/Fe]vs.[N/Fe].Thetwocriteria
forNEMPstarssuggestedbyPolsetal.(2012)arerespectivelyshownindashed([N/Fe]≥1.0and[N/C]≥0.5)anddash-dottedlines([(C+N)/Fe]>0.9).The
filledandopencirclesrespectivelyrefertocarbon-enhancedandcarbon-normalstars.Metal-poorstarsfrom“FirstStars”(Spiteetal.2005),Yongetal.(2013a),
andPlaccoetal.(2014)arealsoplottedforcomparison,infilledtriangles,crosses,andfilledpentacles,respectively. ThepotentialCNEMPobjectJ0257−0022
ismarked.
with a positive slope, which agrees well with the results of tributionofobjectswith[Sr/Ba]largerthantheSolarvalueat
Cayreletal. (2004). These authors interpreted this trend as evenlowermetallicities,butthiswouldrequirealargerbody
anindicationthatCrandFeareproducedinthesamenucle- ofdataforEMPandUMPstarsforfurtherexploration.
osynthesisprocess,sothatmixingintheinterstellarmedium Figure 8(b) shows that if we adopt Ba as a reference ele-
would not result in any significant scatter around the abun- ment, a major trend of anti-correlation between [Sr/Ba] and
dance trend. Furthermore, a clear anti-correlation between [Ba/H] can be found all the way down to [Ba/H] ∼ −4.5,
[Zn/Fe]and[Fe/H]isseen.Thissuggeststhatneutron-capture whiletheabundanceratiorevertsbacktoanearlySolarvalue
processesmaynothavecontributedmuchtotheproductionof when it approaches to lower [Ba/H]. This is quite agree-
ZninprogenitorsoftheseEMPstars,sinceifthatisthecase, able with the results of Franc¸oisetal. (2007), and we sus-
ananti-correlationversusmetallicitieswouldbeexpectedfor pectthattheanti-correlationmayextendto[Ba/H]higherthan
theabundanceratio. −1.5, e.g., to around [Ba/H] ∼ 0.0. The observed distribu-
tiontrendof[Sr/Ba]against[Ba/H]indicatesthatbesidesthe
4.3. Neutron-captureElements main rapid neutron-captureprocess, it is likely that an addi-
tionalnucleosynthesismechanismsuchasasecondneutron-
Theabundanceratiosoftheneutron-captureelementsSr,Y,
captureprocesshascontributedtotheobservedamountsofSr
Ba,andEuversus[Fe/H]areshowninFigure7.Inagreement
(andotherneutron-captureelements)at[Ba/H] > −4.5(e.g.,
withpreviousinvestigations,verylargedispersionsareseen,
Wanajoetal.2001;Travaglioetal.2004;Qian&Wasserburg
especially at [Fe/H] < −2.8 (Ryanetal. 1996; Cayreletal.
2007).
2004). For all of the program stars with measurable Ba
abundances, the [Ba/Fe] ratios are not higher than the So-
4.4. AbundancePatternsandChemicalPeculiarities
larvalue,whichmeansthatthesetwocarbon-enhancedstars
(J0257−0022andJ1709+1616)belongtotheclassofCEMP- Despite the observed diversity in chemistry among low-
no stars. Eu abundances are only measured in the two gi- metallicitystars, basedonhomogeneouslyanalyzedsamples
ants with relatively higher metallicities. Both giants show of metal-poor stars (Cayreletal. 2004; Yongetal. 2013a) it
a near-Solarabundanceratio of [Eu/Fe], and combinedwith isbelievedthatthereexistsapopulationthatconformstothe
the“FirstStars”samplesupporttheindicationthatmetal-poor averageabundancepattern,andthusisregardedasthechem-
starswithhigh[Eu/Fe]arerare(Franc¸oisetal.2007). ically“normal”population. Consideringthefactthatmostof
The abundanceratiosof neutron-captureelements such as thecarbon-normalobjectsin oursample aregiants, we have
[Sr/Ba]canbeusedtoinvestigatethepossiblenucleosynthe- adopted the coefficients of the regression fit of Cayreletal.
sis processes that have occurred in the progenitors of EMP (2004), which were determined by means of a sample of
stars. Aokietal. (2013) has found several remarkable fea- metal-poor giants for which very high-quality spectra were
tures in the distribution of [Sr/Ba] versus [Fe/H], including obtained. The results of our comparison of the abundance
a cutoffatapproximately[Fe/H] ∼ −3.6, anda lower bound patternsare shown in Figure 9. For an element between Na
at about [Sr/Ba] ∼ −0.5. Figure 8(a) shows the distribution and Zn, if the [X/Fe] ratio deviates more than 0.5dex from
of[Sr/Ba]forourprogramstarsandthemetal-poorstarsam- the reference abundance pattern, it is regarded as abnormal
plesfromtheliteratureasafunctionof[Fe/H]. Ascanbeen and marked by filled squares. For five out of the eight pro-
seen, the number of observed objects is indeed very limited gram stars, there is at least one element showing such an
around [Sr/Ba] ∼ −0.5 and below. However, from the cur- abnormal abundance. However, if only carbon-normal ob-
rentlyavailabledatasetofmetal-poorstars,itdoesnotseem jectswithmorethanoneabnormalelementarecounted,only
clear whether the cutoff at [Fe/H] < −3.6 actually exists. It J0326+0202 exhibits a moderate over-abundance in Na and
ispossiblethattheremaybeacutofforsharpdropofthedis- Mg.
10 Lietal.
C N Na Mg
2
] 1
e
F
/
X 0
[
-1

Al Si K Ca
2
] 1
e
F
/
X 0
[
-1

Sc II Ti Ti II Cr
2
] 1
e
F
/
X 0
[
-1

Mn Co Ni Zn
2
] 1
e
F
/
X 0
[
C-normal
C-enhanced
-1 Cayrel et al (2004)
Yong et al. (2013)
Placco et al (2014)
-4 -3 -2 -4 -3 -2 -4 -3 -2 -4 -3 -2
[Fe/H] [Fe/H] [Fe/H] [Fe/H]
Fig.6.—[X/Fe]vs. [Fe/H]of15elements(16species)lighterthanZn. Forallαelements(Mg,Si,Ca,andTi),thecanonicalvalueof[α/Fe]∼+0.4forthe
halostars(McWilliam1997)isplottedforreference.ThemeaningsofthedifferentsymbolsarethesameasinFigure5.
Regardingneutron-captureelements, noneof ourstars are great interest, in the sense that they raise the prospect that
strongly enhanced in Sr, Y, Ba, or Eu relative to the Solar at the early phase of Galactic chemical evolution there is
abundance ratios. However, we note that two of our stars, at least one kind of neutron-capture processes operating as
J0006+1057andJ0126+0135,havequitelowSrandBaabun- frequently as the nucleosynthesis mechanisms that produce
dances, respectively with [Sr/H] of −5.33 and −5.18, and lighter elements such as α and iron-peak elements. Inter-
[Ba/H] of −5.01 and −4.71. Using the stellar abundance estingly,J0006-1057andJ0126+0135have[Sr/Ba]ratiosof
data of over 1,000 stars from the Galactic field and dwarf −0.3 and −0.5, respectively (see Table 1 and the right panel
galaxies, Roederer (2013) has noted that strontium and bar- of Figure 8). These values very much resemble those seen
iumhavebeendetectedinallenvironments.Whencomparing in strongly r-process-enhanced stars, such as CS22892-052
J0006+1057andJ0126+0135withRoederer(2013)’sFigure ([Sr/Ba] = −0.4; Snedenetal. 2003), rather than the values
1 which presents the distribution of [Ba/H] versus [Sr/H] in typicalforstarsshowingtheabundancesignatureoftheweak
field stars and dwarf galaxies, both of our programstars are r-process, e.g., HD122563 ([Sr/Ba] = +0.8; Hondaetal.
located in the lower left corner of the plot, and close to the 2006). This suggests that not only the weak r-process, but
detection thresholds. As pointed out by Roederer (2013), alsothemainr-processcontributedsignificantlytotheinven-
such objects with unusually low [Sr/H] and [Ba/H] are of toryofneutron-captureelementsduringtheearliestphasesof