The Chemistry of the Trailing arm of the Sagittarius DwarfGalaxy

Stefan C. Keller, David Yong and Gary S. Da CostaResearch School of Astronomy and Astrophysics, Australian National University,

Mt. Stromlo Observatory, Cotter Rd. Weston ACT 2611 Australia.

stefan@mso.anu.edu.au

ABSTRACT

We present abundances of C, O, Ti, and Fe for eleven M-giant stars in the trailing tidal arm ofthe Sagittarius dwarf (Sgr). The abundances were derived by comparing synthetic spectra withhigh-resolution infrared spectra obtained with the Phoenix spectrograph on the Gemini Southtelescope. The targeted stars are drawn from two regions of the Sgr trailing arm separated by 66◦

(5 stars) and 132◦ (6 stars) from the main body of Sgr. The trailing arm provides a more directdiagnostic of the chemical evolution of Sgr compared to the extensively phase-mixed leading arm.

Within our restricted sample of ∼2–3 Gyr old stars, we find that the stream material exhibitsa significant metallicity gradient of −(2.4 ± 0.3) × 10−3 dex / degree (−(9.4 ± 1.1) × 10−4 dex/ kpc) away from the main body of Sgr. The tidal disruption of Sgr is a relatively recentlyevent. We therefore interpret the presence of a metallicity gradient in the debris as indicative ofa similar gradient in the progenitor. The fact that such a metallicity gradient survived for almosta Hubble time indicates that the efficiency of radial mixing was very low in the Sgr progenitor.

No significant gradient is seen to exist in the [α/Fe] abundance ratio along the trailing arm.Our results may be accounted for by a radial decrease in star formation efficiency and/or radialincrease in the efficiency of galactic wind-driven metal loss in the chemical evolution of the Sgrprogenitor. The [Ti/Fe] and [O/Fe] abundance ratios observed within the stream are distinctfrom those of the Galactic halo. We conclude that the fraction of the intermediate to metal-richhalo population contributed by the recent dissolution (< 3 Gyr) of Sgr-like dwarf galaxies cannot be substantial.

Subject headings: Galaxy: halo — Galaxy: structure — stars: abundances — galaxies: individual:Sagittarius dSph

1. Introduction

The paradigm of ΛCDM cosmology enshrinesthe importance of hierarchical assembly to the pro-cess of galaxy formation. Galaxies such as theMilky Way are expected to arise from the coa-lescence of numerous smaller systems. The ideathat the stellar halo of the Milky Way was formedfrom the disruption of smaller systems was firstproposed by Searle & Zinn (1978). The studyof Carollo et al. (2007) argues that the relativecontribution of stars originating in external sys-tems to the halo is dependent on Galactocentric

radius (RGC). The outer halo (RGC > 15 kpc) isdominated by lower metallicity material and ex-hibits kinematics indicative of derivation from ac-cretion. On the other hand, at RGC < 10 kpcthe inner halo is dominated by generally highermetallicity and prograde kinematics as would arisefrom an Eggen et al. (1962) in-situ formation sce-nario. Further, observational studies of the spa-tial (Bell et al. 2008) and kinematic (Starkenburget al. 2009) properties of the halo conclude thataccretion has a significant role to play in the con-struction of the Halo. These studies find accretionis responsible for between 10-100% of the extant

1

arX

iv:1

006.

4885

v1 [

astr

o-ph

.GA

] 2

4 Ju

n 20

10

population depending on the assumed merger treefor the Galaxy.

An apparent argument against substantial con-tribution of accreted stars to the Halo is thediscrepancy between the chemistry of the most-likely accreted systems, namely, dwarf spheroidal(dSphs) galaxies, and the chemistry of the Halo(see for example, Unavane et al. 1996, and re-cently, Tolstoy et al. 2009). At [Fe/H]> −1 dex,dSph stars exhibit substantially lower [α/Fe] ra-tios compared to their Galactic halo counterparts(Tolstoy et al. 2003; Shetrone et al. 2003; Vennet al. 2004). Chemical evolution models (Gilmore& Wyse 1991; Lanfranchi & Matteucci 2004; Mat-teucci 2008; Calura & Menci 2009) demonstratethis is due to substantially different star formationhistories, combined with differences in the abilityto retain processed gas, between the Milky Wayand dSph populations. A solution to the chemi-cal mismatch is that the stellar populations of thepresent-day dSphs are not representative of thebulk of the material previously contributed to thehalo via tidal stripping. Indeed the different chem-istry between the present-day dSphs and the halomight simply reflect that these dSphs were able tosurvive for a Hubble time and this has provided anextended timescale for chemical evolution to occurwithin them (Lagos et al. 2009).

As pointed out by Abadi et al. (2006) and Fontet al. (2006), tidal disruption of a dSph in whichthere is a strong metallicity gradient would lead tothe deposition of stars distinct from those surviv-ing in the present-day core. In low mass systems,where orbit swapping processes are expected tobe much less effective (Roskar et al. 2008), anddistinct from low mass early-type galaxies thathave suffered major mergers (Spolaor et al. 2009),the sense of this metallicity gradient is one of de-creasing metallicity with increasing radial distance(Stinson et al. 2009). Furthermore, it has becomeapparent that the chemistry of some of the mostmetal-poor ([Fe/H]< −3 dex) dSph stars is indis-tinguishable from that of the Halo (Frebel et al.2010; Norris et al. 2010). Hence tidal stripping ofthe metal-poor outskirts of a dSph could lead tothe deposition of material that is old, metal-poorand exhibits abundance ratios matching those ofthe Halo (Cohen & Huang 2009; Munoz et al. 2005;Majewski et al. 2003).

In this paper we examine the chemistry of the

stellar population that has been deposited intothe outer halo of the Galaxy from the Sagittar-ius dwarf galaxy (Sgr, Ibata et al. 1994). Thetidal disruption of the Sgr dwarf galaxy is theMilky Way’s most prominent ongoing accretionevent. Tidal debris on leading and trailing armsare traced via RR Lyrae variables (Vivas et al.2004; Keller et al. 2008; Prior et al. 2009), bluehorizontal branch stars (Yanny et al. 2000, 2009;Newberg et al. 2007), M-giants (Yanny et al. 2009;Majewski et al. 2003), subgiants (Keller et al.2009) and main-sequence turn-off stars (Belokurovet al. 2006; Juric et al. 2008). The tidal debrisextends around the sky and represents multipleorbits of Sgr around the Milky Way.

This study utilises high resolution IR spec-troscopy obtained on the Gemini South telescopeto target stars in two regions along the length ofthe Sgr trailing arm. This enables us to determinethe chemistry of material lost on two consecutiveperigalactic passages of Sgr and hence to investi-gate the presence of any chemical abundance gra-dients evident in the stripped material.

2. Sample Selection

The targets for our abundance study are takenfrom the M-giants catalogued in Majewski et al.(2003). We selected two samples of stars seen to-wards the trailing arm of Sgr debris that possessradial velocities and distances appropriate for thetrailing arm. The radial velocities of the targetsample are shown in Figure 1. The two samplesat Λ� = 66◦ and 132◦ are shown in Figure 1 inthe Sgr plane in the context of literature modelsfor the Sgr debris (Law et al. 2005; Fellhauer et al.2006; Helmi & White 2001). As seen in the simula-tions of Law et al. (2005) the two regions representstars lost from the Sgr system approximately 0.5Gyr and 1.3 Gyr ago respectively.

We have chosen to focus on the trailing arm forthe following reasons. Firstly, Law et al. (2005)show that the leading arm debris is much less spa-tially differentiated with respect to the epoch atwhich material was stripped compared with thetrailing arm material (see their Figure 1). Con-sequently, stars lost in successive orbits overlapsignificantly. Furthermore, the radial velocity dis-tribution of leading arm stars is less coherent thanthat seen in the trailing arm (see Law et al. 2005

2

Figure 10). For trailing arm material selection onthe basis of radial velocity is more stringent andcontamination from field star interlopers is lesslikely than that for a corresponding leading armsample.

3. Spectroscopic Observations

Table 1 reports the details of our observations.The spectra of individual stars along the trail-ing arm of Sgr were obtained in the H-band ata resolution of R = 50000. The spectra wereobtained using the 8 m Gemini South telescopeand the Phoenix spectrograph (Hinkle et al. 2003)in queue mode observing (Program GS-2008B-Q-33). The observations were centred at 1.555 µmand utilised the H6420 order separating filter toprovide a wavelength coverage of 75A. The spec-tra were observed using a 4-pixel slit (0.34′′ wideand 14′′ long). Each target star was observed attwo positions, left and right of the midpoint of thespectrograph slit length, separated by 2.5′′. Thesky and dark background were removed by sub-tracting exposures taken at the alternating po-sitions on the detector array. For each night 10dark and flat frames were also acquired. A seriesof spectra of hot stars for telluric line correctionwere also acquired. Examination of the observedwavelength window reveals only a very few weaktelluric features.

The frames were reduced in the IRAF envi-ronment following a procedure similar to that de-scribed in Smith et al. (2002) and Melendez et al.(2003). Dark and flat frames are combined andthe resulting dark is subtracted from the flat. Aresponse image was derived from the flat and thescience frames were then divided through by thisresponse frame. The spectra were then extractedand wavelength calibrated using the stellar ab-sorption lines evident. Finally, the spectra cor-responding to left and right displacements werecombined and continuum normalised.

4. Stellar Parameters and Abundance De-termination

The reddening to each object is derived fromthe 100 µm Galactic map of Schlegel et al. (1998)and is provided in Table 1. Infra-red photome-try is derived from the 2MASS database havingfirst transformed from the 2MASS system to the

Johnson-Glass system1 for the purpose of temper-ature determination. We dereddened the coloursof the target stars using a ratio of E(J−K)/E(B−V ) = 0.53 (Bessell et al. 1998). The effectivetemperature scale for cool giants is well definedfrom occulation measurements. We use the re-lation of Bessell et al. (1983), namely Teff =7070/(J − K + 0.88), to derive the temperaturesgiven in Table 2. Surface gravities were derivedfrom interpolation of appropriate low metallicityisochrones (Marigo et al. 2008). The microturbu-lent velocity was determined using the followingrelation, ξt = 4.2 − (6 × 10−4Teff ), adopted fromthe optical analysis by Melendez et al. (2008) ofthick disk and bulge stars. We estimate that in-ternal uncertainties in the stellar parameters areTeff ±75K, logg g ± 0.4 dex, and ξt±0.4 km s−1.

Abundances for a given line were derived bycomparing synthetic spectra with the observedspectra following the analysis of Yong et al. (2008).The synthetic spectra were generated using thelocal thermodynamic equilibrium (LTE) stellarline analysis program MOOG (Sneden 1973, 2007version) and LTE model atmospheres from theMARCS grid (Gustafsson et al. 2008). The linelist used in the generation of synthetic spectra wastaken from Melendez & Barbuy (1999); Melendezet al. (2001, 2003). First we derived abundancesfor O from the OH molecular lines at 15535.462,15536.705, and 15565.880 A(see Figure 2). Next,abundances for C were obtained from the COmolecular lines near 15576 A. Since the abun-dances of C and O are coupled, we iterated untilself-consistent abundances were obtained, whichalways occurred within one iteration. N measure-ments are possible from CN lines in the H-band.However, examination of the spectra revealed thatreliable N abundances cannot be obtained giventhe weakness of the CN lines and the modest S/Nratios. Fe abundances were obtained from the Fe Ilines at 15534.260A, 15550.450A, and 15551.430A.Ti abundances were determined from the Ti I lineat 15543.780 A. In Figures 3, we show examples ofsynthetic spectra fits to derive abundances in oursample, and in Table 3 we present the final abun-dances. In the discussion to follow we express ourabundances relative to the solar values of Asplundet al. (2006). The abundance dependences on the

1http://www.astro.caltech.edu/∼jmc/2mass/v3/transformations/

3

stellar parameters are given in Table 4.

5. Metallicity Gradient

In Figure 4 we present our [Fe/H] determina-tions for the two sample regions in the upper panelof Figure 4. In the lower panel of Figure 4 wegraphically represent the median [Fe/H] and itsinterquartile range as a function of angular dis-tance (Λ� as defined in Law et al. (2005)) fromthe main body of Sgr. A two-sided K-S test isused to determine the probability that the streamsamples were drawn from the core sample (Monacoet al. 2005). The percentages above each samplein the lower panel of Figure 4 show the proba-bility that this is the case. Low values of thisprobability indicate that the assumption that thestream samples are similar to that of the core isa poor one. More distant material is seen to beprogressively less like the core sample. A gradientof [Fe/H]=−(2.4±0.3)×10−3 dex/degree is deter-mined from a least-squares fit to the core, Λ�=66◦

and Λ�=132◦ samples. At a mean distance of 22kpc this projects to −(9.4±1.1)×10−4 dex / kpc.

The target stars have been taken from the studyof Majewski et al. (2003) and are selected thereinon the basis of their 2MASS colours as M-giantstars. The judicious selection of Majewski et al.isolates the upper red giant branch (RGB) ofSgr with low contamination from the Milky Way(MW) field. However, it also imposes a bias to-wards metal-rich stars as detailed in Majewskiet al. (2003). Metallicities of [Fe/H]< −1 dex areessentially excluded due to this colour selection.To minimise the effects of this imposed metallic-ity bias on our findings, the above figures compareour results with literature data that impose identi-cal colour selection of the M-giants. Utilisation ofM-giants also imposes an age range to the sampleof stars we study. An M-giant of [Fe/H] = −0.4dex (typical of the Sgr core; Monaco et al. 2005)possesses an age of 2–2.5 Gyr. At lower metal-licities an older age is required to reach the sameJ−K color (for example a 1M� [Fe/H] = -0.4 staris 0.1 dex younger than a [Fe/H] = -0.7 star atJ−K=1.0; Marigo et al. 2008).

Our results may be compared to the metallicitygradient observed in the more extensively studiedleading arm material. As noted above, the lead-ing arm is more extensively phased mixed. That is

to say the material lost in successive orbits is notas spatially differentiated as in the trailing arm.This effect would be expected to reduce the ap-parent metallicity gradient along the leading armcompared to the trailing arm. In their study ofthe leading arm M giants, Chou et al. (2007), re-port the metallicity distribution function (MDF)in two regions; one centred at Λ� ∼ 230◦ of around100◦ in extent, and another region at Λ� ∼ 30◦

(proposed to be old leading arm debris displaced∼ 390◦ from the main body). The mean metallici-ties are found to be −0.7 dex and −1.1 dex respec-tively. Taken together with the mean metallicityof the core, this equates to a metallicity gradient of−2.2 × 10−3 dex/degree. This is compatible withthe present results for the trailing arm.

The [Fe/H] gradient we derive here is alsocompatible with the mean metallicity determined,again from M-giants, in the sample of Monacoet al. (2007) (marked M07 in Figure 4). Further,it is noteworthy that our observed [Fe/H] gradientcontinues to the Chou et al. (2007) ’North Galac-tic Cap positive velocity’ group which is ascribedby Chou et al. to an old wrap of the trailing arm.The North Galactic Cap sample is not used in ourdetermination of the metallicity gradient of thetrailing arm since there is possible confusion withother kinematically distinct substructures in thedirection of the North Galactic Cap sample. TheNorth Galactic Cap sample occupies an area of thesky in which there is both leading and trailing armmaterial as well as material from the Virgo Stel-lar Stream (Duffau et al. 2006; Vivas et al. 2008;Keller 2010; Prior et al. 2009). The dynamicalmodels of Law et al. (2005) predict that while theleading arm material at the position of the NorthGalactic Cap sample will possess negative veloc-ities, the trailing material will possess velocitiesof 100 < VGSR < 200 kms−1. The Virgo StellarStream material possesses a colder velocity pro-file centred at VGSR ∼ 100 kms−1. Prior et al.(2009) shows that considerable overlap in radialvelocity exists between the two systems and con-sequently it is not clear how much of the NorthGalactic Cap sample is due to the Virgo StellarStream. Prior et al. (2009) finds that the metallic-ity of the RR Lyrae members of the Virgo StellarStream are uniformly metal-poor at [Fe/H]∼ −1.7.At such low metallicity few objects would be con-tributed to the 2MASS colour selection discussed

4

above. However, the RR Lyraes of the Virgo Stel-lar Stream represent the old stellar population andnothing is known of the metallicity distribution ofother populations within this halo substructure.

5.1. Implications of the metallicity gradi-ent in Sgr debris

We now discuss the implications of our find-ings of a metallicity gradient in the stars along thetrailing arm of Sgr in the context of models of thechemodynamic evolution of Sgr. Dynamical mod-els of the tidal disruption of Sgr have been pre-sented by numerous authors. Ibata et al. (2001);Helmi (2004); Martınez-Delgado et al. (2004); Lawet al. (2005) present models that seek to accountfor the distance to, and radial velocity of, the M-giants of Majewski et al. (2003). Fellhauer et al.(2006) and Martınez-Delgado et al. (2007) includefurther observational constraints from the SDSSregarding the distance to the Sgr leading arm. Acommon feature of these studies is that while theymanage to qualitatively match the features of thestream, they do not provide a consistent model forthe shape of the Milky Way’s dark matter halo. Inparticular, the leading material is best matchedby an oblate halo, whereas the trailing materialis best matched by a prolate figure (Prior et al.2009; Newberg et al. 2007; Yanny et al. 2009; Lawet al. 2005). Law et al. (2009) demonstrates thatby adopting a triaxial halo model, rather than theaxisymetric models assumed in the studies above,a concordant solution is achievable. However, asLaw et al. (2009) point out the solution is some-what unsatisfactory since such a configuration isexpected to be dynamically unstable.

While the above uncertainties remain in themodelling of the orbit of Sgr, the pertinent fea-tures for the present study are that stars are re-leased preferentially during perigalactic passagewith an orbital period of ∼ 0.85 Gyr (Law et al.2005). Consequently, our sample at Λ�=66◦ waslost from Sgr on the present perigalactic passageapproximately 0.5 Gyr ago, and the Λ�=132◦

sample was lost ∼ 1.3 Gyr ago on the previouspassage of Sgr (see figure 1 of Law et al. 2005).

Studies of the star formation history of the mainbody of Sgr (Layden & Sarajedini 2000; Siegelet al. 2007) have revealed a complex and pro-tracted star formation history with three majorphases. These studies find an old (11 Gyr) metal-

poor population of [Fe/H] ∼ −1.3 dex, a domi-nant intermediate age population (6–4.5 Gyr, withpossible bursts) with [Fe/H] ∼ −0.6 dex and ayoung (2–3 Gyr) population of [Fe/H] ∼ −0.4 to−0.1 dex. If the age-metallicity relation (AMR)of Sgr was spatially uniform then we would expectour samples of M-giants to possess the metallicityappropriate for their relatively young age (∼2–3Gyr), namely −0.4 < [Fe/H] < −0.1 dex (as dom-inates the core sample). Furthermore we wouldexpect a negligible metallicity gradient in the de-bris stream.

Our observations imply that the progenitor ofthe present-day Sgr did not possess a spatially uni-form AMR. Seen another way, the time betweenperigalacticons does not allow sufficient time forchemically homogeneous in-situ elevation of themean [Fe/H] to the levels we observe between suc-cessive orbits. Rather, as recognised by Chouet al. (2007), the abundance gradient observedmust arise due to the stripping of the outer re-gions of the Sgr progenitor over which a metal-licity gradient (and/or concomitant age gradient)was present.

Indeed such population gradients are typicallyobserved in dwarf galaxies. As reviewed by Stin-son et al. (2009), dwarf irregular galaxies univer-sally show extended envelopes dominated by oldRGB stars. Such haloes of old–intermediate agestars are seen across a range of galaxy luminosityand under a range of tidal conditions. Stellar pop-ulation gradients are notably less distinct amongstthe local dSph galaxies. However, gradients in themorphology of the horizontal branch are not un-common and are indicative of metallicity and/orage gradients (Harbeck et al. 2001).

It is plausible therefore, that the original Sgrprogenitor possessed an extended and correspond-ingly metallicity segregated, halo consistent withour observations. As Sgr experienced strong tidalinteraction with the Milky Way, the tidal radius ofSgr decreased with each successive orbit, leadingto the loss of stars from outer regions of the pro-genitor. In this picture, it is the intermediate ageM-stars (having arisen in perhaps the most recentstar formation episode) of this outer Sgr halo pop-ulation that we see in the tidally disrupted debristails. The radial metallicity gradient imprinted byprevious chemodynamical evolution gives rise tothe metallicity gradient we observe here. Since the

5

stars of the present study are amongst the mostrecent to form in Sgr, the fact that they exhibit a[Fe/H] gradient requires that the abundance gra-dient is long-lived, having survived for approxi-mately a Hubble time. Moreover, our findings im-ply that radial mixing of enriched material mustbe of little significance.

6. Abundance Ratio Gradients

In Figures 5 and 6 we show our findings for[Ti/Fe] and [O/Fe] as a function of [Fe/H]. Bothfigures show a characteristic underabundance ofO and Ti at given [Fe/H] relative to the stars ofthe Galactic disk and halo. We recall that the Mgiants of the present study represent stars of ∼2–3 Gyr; Galactic disk stars of this age reside onthe Galactic locus (the solid line in Figs. 5 & 6)at approximately Solar metallicity.

Figure 7 shows the interquartile ranges of theoffset in [O,Ti/Fe] at given [Fe/H] between ourobserved sample and that defined in the Galacticdisk and halo2 as a function of angular distancefrom Sgr. The probability that each sample isdrawn from the core sample is determined by atwo-sided K-S test. In the case of both [O/Fe]and [Ti/Fe] there is no significant gradient alongthe trailing arm. However, a significant spread in[Ti/Fe] is apparent in the four samples in Figure5. The spread is much larger than the abundanceratio errors (the associated abundance ratio errorsof the studies of Monaco et al. (2005) and Chouet al. (2009) are similar in magnitude to those ofthe present study). Such a spread is not apparentin [O/Fe], shown in Figure 6. The origin of thisspread in [Ti/Fe] is not understood. Our studydoes not however, enable us to define the [Fe/H]of the [O/Fe] or [Ti/Fe] ‘knee’. This is due to theinherent metal-rich bias in our use of M-giants (asdiscussed above) and small sample size. Conse-quently, we do not constrain the expected metal-poor ([Fe/H]< −1 dex) [O/Fe] plateau. Chemi-cal evolution models of Lanfranchi & Matteucci(2004) show that the progressively lower SF effi-ciency should lead to a [α/Fe] plateau at progres-sively lower [α/Fe] at increasing Λ�. Abundancesof radial-velocity selected K-giants (representativeof the metal-poor population) in the trailing arm

2 in the sense [X/Fe]Sgr - [X/Fe]Galactic, where X is theabundance of O or Ti.

would enable clarification of the location of the‘knee’ in the [α/Fe] relation, or otherwise, of achemical gradient in the trailing arm material.

6.1. Implications of α-element abundance

Our observations of the α-element O and theα-like element Ti, show that the [α/Fe] ratio ofSgr material is lower at a given [Fe/H] comparedwith that of the MW (at least for [Fe/H] > −1dex material probed to date). As described in alarge literature of chemical evolution models (seefor example, Gilmore & Wyse 1991; Lanfranchi &Matteucci 2004; Lanfranchi et al. 2006; Matteucci2008) and motivated by numerous observationalstudies (for example, Smith et al. 2002; Venn et al.2004; Cohen & Huang 2009; Chou et al. 2009; Hi-dalgo et al. 2009; Lee et al. 2009) the generallylower [α/Fe] is a consequence of slower chemicalenrichment in lower mass systems. The slowerpace of chemical enrichment (due to lower star for-mation efficiency and/or stronger galactic winds)allows Fe-rich SNIa products to be incorporatedat lower metallicities relative to the MW halo.

The [Fe/H] of the [α/Fe] ‘knee’ feature, seen inthe MW halo at a [Fe/H] ∼ −1.0 dex, provides asnapshot of the metallicity at ∼ 1 Gyr, after whichtime SNIa begin to appreciably raise the Fe con-tent and reduce [α/Fe]. As suggested by Monacoet al. (2005), and demonstrated by Chou et al.(2009) in the Sgr leading arm, for [α/Fe] ≤ −1dex there is a general overlap in the [α/Fe] ratioof MW and Sgr stars. This indicates a commondominance of SNII enrichment at these metallici-ties. For [Fe/H] ≥ −1 dex however, the [α/Fe] ra-tio of Sgr material is significantly below that theof the MW. This possibly reflects a lower earlystar formation rate (SFR) in Sgr compared to theMW, or as proposed by Lanfranchi et al. (2006),that Sgr exhibited a moderate star formation (SF)efficiency at first, but one that was quenched bygalactic winds from ensuing SNII.

The implications of the evolution of [α/Fe] inSgr can be contrasted with those in the LMC. Inthe LMC the [Ti/Fe] and [O/Fe] ratio remainslower that the MW at all metallicities probed todate ([Fe/H].-1.3 (Hill et al. 2000; Smith et al.2002). This is ascribed to a SF efficiency lowerthan that of Sgr and/or more efficient galac-tic winds (discussed in the context of [O/Fe] byGilmore & Wyse (1991) & Smith et al. (2002))

6

and the presence of fewer contributing high-massSNII (from [Ti/Fe], Pompeia et al. (2008)).

Monaco et al. (2005) therefore point out thatthe early chemical evolution of Sgr was more akinto that of the MW than that of the LMC andthe local dSphs. They propose that the progeni-tor of Sgr was most likely a gas-rich, star form-ing galaxy that was substantially more massivethat the extant remnant. The Lanfranchi et al.(2006) results amplify this argument. The studyof Lanfranchi et al. defines a sequence of relativeSFR in early galaxy evolution that ranges fromthe dSphs which exhibit the lowest SFR, to theLMC with a moderate SFR, and finally Sgr withthe highest SFR evident amongst the MW satel-lites. Thus the advanced chemical evolution of Sgrmaterial argues strongly that the progenitor wasmore massive that the LMC. However, at presentthe dynamical models of Law et al. (2005) assumea progenitor mass of 2–5×108M� compared to theLMC mass of 1–2×109M�. An increase in theprogenitor mass of Sgr would imply a more ex-tended history of mass-loss to the MW and anenhanced impact on the chemistry of the Halo.We also note that a large mass progenitor wouldbe in line with the observed extended SFH of Sgrthat indicates that Sgr was actively forming starsuntil 1–2 Gyr ago (Siegel et al. 2007, indeed if ithad not, there would be no M giants with whichto trace the tidal streams) although it has beentidally interacting with the MW for ∼ 3 Gyr (seefor example Law et al. 2005). The ability to re-tain the gas required for such star formation in theface of Galactic tidal interaction would argue fora large progenitor mass.

6.2. Sgr and the formation of the GalacticHalo

An important role is expected for accretedsatellites in the formation of the Galactic Halowithin Lambda-CDM cosmology. Here hierarchi-cal mergers lead to massive galaxies surrounded bya flotilla of lower mass satellite halos (at least astraced by dark matter-only simulations). Hence,the stellar content of the Halo must presumablycontain some contribution from the disruption ofsuch satellites. A number of observational chal-lenges to such a scenario have arisen.

Firstly, the number of lower mass halos ob-served about the MW was found to be low com-

pared to predictions of cosmological simulations(the so-called ‘missing satellite problem’ Klypinet al. 1999). However, recent searches have re-vealed increased numbers of extremely faint dSphs(Walsh et al. 2009; Belokurov et al. 2006). Whenconsidered together with survey selection effects,the shortfall of local dSphs is substantially ame-liorated (Koposov et al. 2009).

Secondly, amongst metal-poor stars ([Fe/H <−2.5 dex) the MDF and chemistry of the Galactichalo were seen to be very different to those ex-hibited by dSphs (Geisler et al. 2007). For exam-ple, the metal-poor tail of the MDF in dSphs wasseen to be deficient compared to that of the Halo(Helmi et al. 2006). However, subsequent studiesof Starkenburg et al. (2010) do not uphold thisdeficiency. Schorck et al. (2009), Cohen & Huang(2009) and Norris et al. (2008) conclude that theMDF of dSphs and the Halo are in good agree-ment. Furthermore, evidence now shows that theextremely metal-poor stars in dSphs show similar[α/Fe] to the Halo (Frebel et al. 2010; Norris et al.2010). It is therefore possible that present-daydSphs could have contributed to the metal-poorHalo.

However, divergent chemistry remains at highermetallicity. For [Fe/H]> −1, the [α/Fe] of thedSph sample is uniformly, and significantly, lowerthan that of the Halo. Only Sgr, thought to be(in the form of its progenitor) the most massiveof the local dwarf galaxies, has some stars with[α/Fe] similar to the Halo at intermediate metallic-ities. Signature of slow chemical enrichment, the[α/Fe] of an average low-mass dSph implies thatsuch systems can not be responsible for a substan-tial fraction of the metal-poor halo.

In order to explain the Halo chemistry, it is nec-essary to capture [Fe/H] & −2 and α-enhancedmaterial. One mechanism to introduce such ma-terial is from a number of massive satellites thatexperienced rapid star formation and were thenaccreted rapidly, before the onset of SNIa contri-butions (Robertson et al. 2005; Font et al. 2006;De Lucia & Helmi 2008). Lagos et al. (2009) makea clear distinction between ‘building block’ satel-lites (i.e. those that are accreted) and survivingsatellites. From cosmological simulations, theyfind that ’building blocks’ collapse and form starsearlier than surviving satellites that instead formstars in a quiescent manner. Cooper et al. (2009)

7

finds that the merger of less than five massive (108

M�) ‘building blocks’ at early times (1 < z < 7)can account for the Galactic halo.

Our findings show that the chemistry of thestars contributed by Sgr to the Galactic halo in thetrailing arm debris are significantly different fromthat of the halo at similar [Fe/H]. Therefore Sgr isnot a ‘building block’. The debris are increasinglymetal-poor further from the mainbody of Sgr andso, in terms of the [Fe/H] the material is increas-ingly ‘halo-like’. However, we see no signs of con-vergence between the [α/Fe] ratio in stars of thetrailing arm and stars of similar [Fe/H] from theGalactic Halo. Consequently, our findings showthat the fraction of the intermediate – metal-richhalo population that can have been contributedby the dissolution of Sgr-like objects over the lastseveral Gyr can not be substantial.

7. Summary

In this study we have presented the abundanceanalysis of eleven M-giant stars in two regionsalong the trailing arm of the Sgr dwarf. Thetwo populations, together with existing data fromthe literature, enable us to explore the metal-licity and chemistry of stars deposited into theMilky Way halo from Sgr over the past ∼ 1.3 Gyr.Within our limited sample of ∼2–3 Gyr old stars,we find a significant gradient in metallicity alongthe debris stream. The metallicity gradient is−(2.4±0.3)×10−3 dex / degree (−(9.4±1.1)×10−4

dex / kpc) decreasing away from the main bodyof Sgr. The [α/Fe] ratio at a given metallicity isseen to be approximately constant along the ex-tent of the trailing arm. The change in metallicityalong the stream can be understood to be due tothe tidal disruption of a Sgr progenitor that pos-sessed a radial gradient in [Fe/H]. Such a gradientcould have arisen due to a radial decrease in thestar formation efficiency and/or a radial increasein the efficiency of galactic wind-driven metal loss.The physical mechanism that produced a radial[Fe/H] gradient that persisted over a Hubble timedid not, however, produce a detectable gradient in[α/Fe].

There is no significant [α/Fe] gradient along thetrailing arm from our sample of M-giants. The[α/Fe] ratios of our targets is distinct from thehalo. It can therefore be concluded that while

some portion of the metal-poor halo might be con-tributed via the accretion of Sgr-like objects, themetal-rich component of the Halo can not featurea substantial contribution from such objects overthe last several Gyr.

We thank Jorge Melendez for providing the linelist. This research has been supported in part bythe Australian Research Council through Discov-ery Project Grants DP0343962 and DP0878137.Based on observations obtained at the Gemini Ob-servatory for Program GS-2008B-Q-33, which isoperated by the Association of Universities for Re-search in Astronomy, Inc., under a cooperativeagreement with the NSF on behalf of the Gem-ini partnership: the National Science Foundation(United States), the Science and Technology Facil-ities Council (United Kingdom), the National Re-search Council (Canada), CONICYT (Chile), theAustralian Research Council (Australia), Ministe-rio da Ciencia e Tecnologia (Brazil) and Ministe-rio de Ciencia, Tecnologia e Innovacion Productiva(Argentina).

Facilities: Gemini:South (Phoenix)

REFERENCES

Abadi M. G., Navarro J. F., Steinmetz M., 2006,MNRAS, 365, 747

Asplund M., Grevesse N., Sauval A. J., 2006, Nuc-PhysA, 777, 1

Battinelli P., Demers S., Mannucci F., 2007, A&A,474, 35

Bell E. F., Zucker D. B., Belokurov V., et al., 2008,ApJ, 680, 295

Belokurov V., Zucker D. B., Evans N. W., et al.,2006, ApJL, 642, L137

Bessell M. S., Castelli F., Plez B., 1998, A&A, 333,231

Bessell M. S., Wood P. R., Evans T. L., 1983, MN-RAS, 202, 59

Calura F., Menci N., 2009, MNRAS, accepted

Carollo D., Beers T. C., Lee Y. S., et al., 2007,Nature, 450, 1020

8

Chou M., Cunha K., Majewski S. R., Smith V. V.,Patterson R. J., Martinez-Delgado D., GeislerD., 2009, ArXiv e-prints

Chou M., Majewski S. R., Cunha K., et al., 2007,ApJ, 670, 346

Cioni M., Habing H. J., 2005, A&A, 442, 165

Cohen J. G., Huang W., 2009, ApJ, 701, 1053

Cooper A. P., Cole S., Frenk C. S., et al., 2009,preprint (astro-ph/0910.3211)

De Lucia G., Helmi A., 2008, MNRAS, 391, 14

D’Ercole A., Brighenti F., 1999, MNRAS, 309, 941

Duffau S., Zinn R., Vivas A. K., Carraro G.,Mendez R. A., Winnick R., Gallart C., 2006,ApJL, 636, L97

Eggen O. J., Lynden-Bell D., Sandage A. R., 1962,ApJ, 136, 748

Fellhauer M., Belokurov V., Evans N. W., et al.,2006, ApJ, 651, 167

Font A. S., Johnston K. V., Bullock J. S., Robert-son B. E., 2006, ApJ, 646, 886

Frebel A., Kirby E., Simon J. D., 2009, preprint(astro-ph/0912.4734)

Frebel A., Simon J. D., Geha M., Willman B.,2010, ApJ, 708, 560

Geisler D., Wallerstein G., Smith V. V., Casetti-Dinescu D. I., 2007, PASP, 119, 939

Gilmore G., Wyse R. F. G., 1991, ApJL, 367, L55

Gustafsson B., Edvardsson B., Eriksson K.,Jørgensen U. G., Nordlund A., Plez B., 2008,A&A, 486, 951

Harbeck D., Grebel E. K., Holtzman J.,Guhathakurta P., Brandner W., 2001, AJ, 122,309

Helmi A., 2004, ApJL, 610, L97

Helmi A., Irwin M. J., Tolstoy E., et al., 2006,ApJ, 651, L121

Helmi A., White S. D. M., 2001, MNRAS, 323,529

Hidalgo S. L., Aparicio A., Martınez-Delgado D.,Gallart C., 2009, ApJ, 705, 704

Hill V., Francois P., Spite M., Primas F., 2000,A&A, 364, L19

Hinkle K. H., Joyce R. R., Sharp N., et al., 2003,Proc. SPIE, 4834, 353

Ibata R., Lewis G. F., Irwin M., Totten E., QuinnT., 2001, ApJ, 551, 294

Ibata R. A., Gilmore G., Irwin M. J., 1994, Na-ture, 370, 194

Ishigaki M., Chiba M., Aoki W., 2009, preprint(astro-ph/0912.0329)

Juric M., Ivezic Z., Brooks A., et al., 2008, ApJ,673, 864

Keller S. C., 2010, PASA, 27, 45

Keller S. C., Bessell M. S., Cook K. H., Geha M.,Syphers D., 2002, AJ, 124, 2039

Keller S. C., da Costa G. S., Prior S. L., 2009,MNRAS, 394, 1045

Keller S. C., Murphy S., Prior S., da Costa G.,Schmidt B., 2008, ApJ, 678, 851

Keller S. C., Wood P. R., 2006, ApJ, 642, 834

Kirby E. N., Simon J. D., Geha M., GuhathakurtaP., Frebel A., 2008, ApJ, 685, L43

Klypin A., Kravtsov A. V., Valenzuela O., PradaF., 1999, ApJ, 522, 82

Koposov S. E., Yoo J., Rix H., Weinberg D. H.,Maccio A. V., Escude J. M., 2009, ApJ, 696,2179

Lagos C. D. P., Padilla N. D., Cora S. A., 2009,MNRAS, 397, L31

Lanfranchi G. A., Matteucci F., 2004, MNRAS,351, 1338

Lanfranchi G. A., Matteucci F., Cescutti G., 2006,A&A, 453, 67

Law D. R., Johnston K. V., Majewski S. R., 2005,ApJ, 619, 807

Law D. R., Majewski S. R., Johnston K. V., 2009,ApJL, 703, L67

9

Layden A. C., Sarajedini A., 2000, AJ, 119, 1760

Lee M. G., Yuk I., Park H. S., Harris J., ZaritskyD., 2009, ApJ, 703, 692

Letarte B., Demers S., Battinelli P., Kunkel W. E.,2002, AJ, 123, 832

Mac Low M., Ferrara A., 1999, ApJ, 513, 142

Majewski S. R., Skrutskie M. F., Weinberg M. D.,Ostheimer J. C., 2003, ApJ, 599, 1082

Marcolini A., D’Ercole A., Battaglia G., GibsonB. K., 2008, MNRAS, 386, 2173

Marigo P., Girardi L., Bressan A., GroenewegenM. A. T., Silva L., Granato G. L., 2008, A&A,482, 883

Martınez-Delgado D., Gomez-Flechoso M. A.,Aparicio A., Carrera R., 2004, ApJ, 601, 242

Martınez-Delgado D., Penarrubia J., Juric M., Al-faro E. J., Ivezic Z., 2007, ApJ, 660, 1264

Matteucci F., 2008

Melendez J., Asplund M., Alves-Brito A., CunhaK., Barbuy B., Bessell M. S., Chiappini C.,Freeman K. C., Ramırez I., Smith V. V., YongD., 2008, A&A, 484, L21

Melendez J., Barbuy B., 1999, ApJS, 124, 527

Melendez J., Barbuy B., Bica E., Zoccali M., Or-tolani S., Renzini A., Hill V., 2003, A&A, 411,417

Melendez J., Barbuy B., Spite F., 2001, ApJ, 556,858

Monaco L., Bellazzini M., Bonifacio P., BuzzoniA., Ferraro F. R., Marconi G., Sbordone L.,Zaggia S., 2007, A&A, 464, 201

Monaco L., Bellazzini M., Bonifacio P., FerraroF. R., Marconi G., Pancino E., Sbordone L.,Zaggia S., 2005, A&A, 441, 141

Munoz R. R., Frinchaboy P. M., Majewski S. R.,et al., 2005, ApJ, 631, L137

Newberg H. J., Yanny B., Cole N., Beers T. C.,Re Fiorentin P., Schneider D. P., Wilhelm R.,2007, ApJ, 668, 221

Norris J. E., Gilmore G., Wyse R. F. G., Wilkin-son M. I., Belokurov V., Evans N. W., ZuckerD. B., 2008, ApJ, 689, L113

Norris J. E., Yong, D., Gilmore G., Wyse R. F. G.,2010, ApJ, 711, 350

Pompeia L., Hill V., Spite M., et al., 2008, A&A,480, 379

Prior S. L., Da Costa G. S., Keller S. C., 2009,ApJ, 704, 1327

Prior S. L., Da Costa G. S., Keller S. C., MurphyS. J., 2009, ApJ, 691, 306

Recchi S., Spitoni E., Matteucci F., LanfranchiG. A., 2008, A&A, 489, 555

Robertson B., Bullock J. S., Font A. S., JohnstonK. V., Hernquist L., 2005, ApJ, 632, 872

Roskar R., Debattista V. P., Quinn T. R., StinsonG. S., Wadsley J., 2008, ApJL, 684, L79

Sbordone L., Bonifacio P., Buonanno R., MarconiG., Monaco L., Zaggia S., 2007, A&A, 465, 815

Schlegel D. J., Finkbeiner D. P., Davis M., 1998,ApJ, 500, 525

Schorck T., Christlieb N., Cohen J. G., et al., 2009,A&A, 507, 817

Searle L., Zinn R., 1978, ApJ, 225, 357

Shetrone M., Venn K. A., Tolstoy E., Primas F.,Hill V., Kaufer A., 2003, AJ, 125, 684

Siegel M. H., Dotter A., Majewski S. R., et al.,2007, ApJL, 667, L57

Smith G. H., Sneden C., Kraft R. P., 2002, AJ,123, 1502

Sneden C., 1973, ApJ, 184, 839

Spolaor M., Proctor R. N., Forbes D. A., CouchW. J., 2009, ApJ, 691, L138

Starkenburg E., Helmi A., Morrison H. L., et al.,2009, ApJ, 698, 567

Starkenburg E., Hill V., Tolstoy E., et al., 2010,astro-ph:1002.2963

10

Stinson G. S., Dalcanton J. J., Quinn T., Gog-arten S. M., Kaufmann T., Wadsley J., 2009,MNRAS, 395, 1455

Tolstoy E., Venn K. A., Shetrone M., Primas F.,Hill V., Kaufer A., Szeifert T., 2003, AJ, 125,707

Tolstoy E., Hill V., Tosi M., 2009, ARAA, 47, 371

Unavane M., Wyse R. F. G., Gilmore G., 1996,MNRAS, 278, 727

Venn K. A., Irwin M., Shetrone M. D., Tout C. A.,Hill V., Tolstoy E., 2004, AJ, 128, 1177

Vivas A. K., et al., 2004, AJ, 127, 1158

Vivas A. K., Jaffe Y. L., Zinn R., Winnick R.,Duffau S., Mateu C., 2008, AJ, 136, 1645

Walsh S., Willman B., Jerjen H., 2009, AJ, 137,450

Whitelock P. A., Menzies J. W., Feast M. W.,Matsunaga N., Tanabe T., Ita Y., 2009, MN-RAS, 394, 795

Yanny B., Newberg H. J., Johnson J. A., et al.,2009, ApJ, 700, 1282

Yanny B., Newberg H. J., Kent S., et al., 2000,ApJ, 540, 825

Yong, D., Melendez, J., Cunha, K., Karakas, A. I.,Norris, J. E., & Smith, V. V. 2008, ApJ, 689,1020

This 2-column preprint was prepared with the AAS LATEXmacros v5.2.

11

leading a

rm

trailing arm

Sgr

20kpc

40kpc

60kpc

132o Sample

66o Sample

RA

Fig. 1.— Top: The spatial distribution of targetstars in the debris stream of Sgr (red: Λ = 66◦

and blue: Λ = 132◦) projected in RA. Concen-tric circles show heliocentric distances of 20-80kpc. M-giants from Majewski et al. (2003) (boldblack: radial velocities present and appropriate forSgr material) and detections from Belokurov et al.(2006) (boxes)) are shown. Overlaid are modelsfrom Law et al. (2005, solid line), Fellhauer et al.(2006, dashed), and Helmi & White (2001, shortdashed). Bottom: The galactocentric radial ve-locities of the selected targets. Radial velocitiesare as determined by Majewski et al. (2003).

12

15520 15540 15560 155800

0.2

0.4

0.6

0.8

1

CO Bandhead

Fe

Fe

Ti OHFe

OH

Fig. 2.— Observed spectrum of 2350361m200216. Lines used in the abundance analysis are indicated.

13

15576 15577 15578 15579 155800.7

0.8

0.9

1

A(C)=6.81, 6.96, 7.110348437p065236

A(O)=7.64, 7.79, 7.940348437p065236

15532 15534 15536 155380.4

0.6

0.8

1

15538 15540 15542 15544 15546 15548 155500.2

0.4

0.6

0.8

1

A(Ti)=3.94, 4.09, 4.240342225p054745

Fig. 3.— Observed spectra (squares) and syn-thetic spectra for C (top); O (middle); and Ti(bottom). The synthetic spectra show the best fit(black line) and unsatisfactory fits (red and bluelines), A(C,O,Ti) ± 0.15 dex.

14

Core M07 NGCC09

-1.4

-1.0

-0.6

-0.2 17.1% 8.8% 5.6% 0.1%

Fig. 4.— [Fe/H] as a function of angular distancefrom the main body of Sgr along the trailing arm.The upper panel shows the individual points. Inthe lower panel the distribution of [Fe/H] in eachsample is displayed as a box plot in which the solidline represents the median, the boxed region spansthe first to third quartiles (i.e. the interquartilerange), and the bars represent ±1.5 × the in-terquartile range. The core sample is taken fromMonaco et al. (2005). The Monaco et al. (2007)study, centred at Λ�=100◦, is derived from a 27.5◦

wide region of the trailing arm. The ‘North Galac-tic Cap’ sample is that from Chou et al. (2009).The percentage above the various groups along thetrailing arm records the probability that each sam-ple is drawn from the Sgr core sample (see textfor details). The dotted line shows the result ofa least squares linear fit to the core, Λ�=66◦ andΛ�=132◦ samples.

15

Galactic Trend

knee

Fig. 5.— [Ti/Fe] vs. [Fe/H]. Sgr core (open cir-cles: Monaco et al. (2005)), trailing arm sampleat Λ�=66◦ (red circles), trailing arm sample atΛ�=132◦ (blue squares), and North Galactic Capsample (crosses: Chou et al. (2009)) are shown.For reference the Galactic locus is shown (solidline) as described in Venn et al. (2004). A repre-sentative error bar is shown.

16

Galactic Trend

Fig. 6.— [O/Fe] vs. [Fe/H]. Sgr core (open cir-cles: Sbordone et al. (2007)), trailing arm sampleat Λ�=66◦ (red circles), and trailing arm sampleat Λ�=132◦ (blue squares) are shown. For ref-erence the Galactic locus is shown (solid line) asdescribed in Venn et al. (2004). A representativeerror bar is shown.

17

Del

ta [T

i/Fe]

259Λ0

CoreNGCC09

(degrees)

Del

ta [O

/Fe]

−0.4

0.0

0.4

0 66 132

−0.8

−0.4

0.0

0.4

Fig. 7.— Top: The offset in [Ti/Fe] at a given[Fe/H] relative to the Galactic halo+disk locus(Venn et al. 2004) as a function of angular dis-tance from the main body of Sgr. The dotted lineshows the least squares linear fit to the samples.Other details of the figure are as described in Fig.4. No spatial gradient in [Ti/Fe] is apparent, how-ever, the spread of [Ti/Fe] at given [Fe/H] is large.Bottom: Same as above for [O/Fe].

18

Table1

Log

ofPhoenix

obse

rvations.

Thefirst

fivestarscorresp

ondtothe‘66degree’group,and

theremaining

sixcorresp

ond

to

the‘132degree’groupdiscussed

inthetext.

Sta

rR

A(J

2000)

Dec

(J2000)

Λ�

(degre

es)

JK

E(B−V

)(J−K

) 0a

UT

Date

Exp

osu

re(s

econds)

S/N

2329301m

245810

23:2

9:3

0.1

-24:5

8:1

060.0

011.5

90

10.4

99

0.0

21.1

45

2008-0

8-2

16×

562

94

2345417m

264456

23:4

5:4

1.7

-26:4

4:5

662.6

311.4

13

10.3

93

0.0

21.0

72

2008-0

8-2

16×

562

98

2350361m

200216

23:5

0:3

6.1

-20:0

2:1

666.4

111.6

64

10.5

76

0.0

21.1

42

2008-0

9-1

76×

562

68

2353194m

205041

23:5

3:1

9.4

-20:5

0:4

166.7

212.5

71

11.6

21

0.0

21.0

00

2008-0

9-1

712×

562

65

0003528m

194047

00:0

3:5

2.8

-19:4

0:4

769.3

512.0

24

11.0

07

0.0

31.0

69

2008-0

9-1

96×

562

52

0334210p051809

03:3

4:2

1.0

+05:1

8:0

9126.9

312.5

05

11.3

45

0.2

61.3

55

2008-0

9-1

912×

562

110

0340164p090338

03:4

0:1

6.4

+09:0

3:3

8130.0

312.9

71

11.7

57

0.3

51.4

62

2009-0

1-1

112×

562

30

0342225p054745

03:4

2:2

2.5

+05:4

7:4

5128.9

311.9

76

10.8

19

0.2

21.3

28

2008-0

8-2

26×

562

78

0348437p065236

03:4

8:4

3.7

+06:5

2:3

6130.9

711.9

76

10.8

36

0.2

01.2

99

2009-0

2-0

26×

562

97

0357262p053258

03:5

7:2

6.3

+05:3

2:4

8132.1

211.3

75

10.1

30

0.2

91.4

59

2008-1

2-1

96×

562

79

0408285p044043

04:0

8:2

8.5

+04:4

0:0

4134.1

912.6

81

11.4

57

0.4

01.5

01

2008-0

9-1

96×

562

65

aD

ere

ddened

colo

rin

the

Johnso

n-G

lass

syst

em

.

19

Table 2: Derived stellar parameters for targetstars.

Star Teff(K) log g ξt (km s−1)2329301m245810 3531 0.03 2.082345417m264456 3665 0.22 2.012350361m200216 3536 0.09 2.082353194m205041 3807 1.16 1.920003528m194047 3670 0.22 2.000334210p051809 3654 0.22 2.000340164p090338 3646 0.16 2.010342225p054745 3616 0.11 2.030348437p065236 3627 0.13 2.020357262p053258 3526 0.02 2.080408285p044043 3683 0.23 1.99

20

Table 3: Derived stellar abundances for targetstars.

Star A(C) A(O) A(Ti) A(Fe)2329301m245810 7.27 8.03 4.02 6.972345417m264456 7.19 7.93 4.14 6.772350361m200216 7.34 8.13 4.29 7.022353194m205041 7.31 8.03 4.30 6.930003528m194047 7.01 7.95 4.49 6.870334210p051809 7.21 7.93 4.79 7.000340164p090338 7.21 7.93 4.29 6.920342225p054745 6.91 8.05 4.09 6.820348437p065236 6.96 7.79 3.99 6.720357262p053258 7.24 8.03 4.09 6.520408285p044043 7.09 7.84 3.81 6.62

Note.—A(X) = log[n(X)/n(H)] + 12. For reference the Asplund et al. (2006) Solar values of the above elemental abundancesare: A(C)=8.39, A(O)=8.66, A(Ti)=4.90, and A(Fe)=7.45.

21

Table 4: Stellar abundance dependencies on modelparameters for 0348437p065236

Species Teff ± 75K log g ± 0.4dex ξt ± 0.4km s−1 Totala

∆A(C) . . . 0.08 0.09 0.06 0.13∆A(O) . . . 0.12 0.10 0.06 0.15∆A(Ti) . . . 0.08 0.08 0.09 0.15∆A(Fe) . . . -0.08 0.04 -0.04 0.10

aThe total value is the sum in quadrature of the individual abundance dependencies.

22