A&A 480, 379-395 (2008)
DOI: 10.1051/0004-6361:20064854

Chemical abundances in LMC stellar populations

I. The inner disk sample^,,

L. Pompéia^1,2 - V. Hill³ - M. Spite³ - A. Cole^4,5 - F. Primas⁶ - M. Romaniello⁶ - L. Pasquini⁶ - M.-R. Cioni⁷ - T. Smecker Hane⁸

1 - IP&D, Universidade do Vale do Paraíba, Av. Shishima Hifumi, 2911, São, José dos Campos, 12244-000 SP, Brazil
2 -  Instituto Astronômico e Geofísico (USP), Rua do Matão 1226, Cidade Universitária, 05508-900 São Paulo, Brazil
3 - Observatoire de Paris-Meudon, GEPI and CNRS UMR 8111, 92195 Meudon Cedex, France
4 - School of Mathematics and Physics, University of Tasmania, Private Bag 37, Hobart, TAS 7001, Australia
5 - Kapteyn Astronomical Institute, University of Groningen, Postbus 800, 9700 AV Groningen, The Netherlands
6 - European Southern Observatory, Karl Schwarschild Str. 2, 85748 Garching b. München, Germany
7 - Edinburg SUPA, School of Physics, University of Edinburgh, IfA, Blackford Hill, Edinburgh EH9 3HJ, UK
8 - Department of Physics and Astronomy, 4129 Frederick Reines Hall, University of California, Irvine, CA 92697-4575, USA

Received 13 January 2006 / Accepted 13 November 2007

Abstract
Aims. We have used FLAMES (the Fibre Large Array Multi Element Spectrograph) at the VLT-UT2 telescope to obtain spectra of a large sample of red giant stars from the inner disk of the LMC, $\sim$2 kpc from the center of the galaxy. We investigate the chemical abundances of key elements to understand the star formation and evolution of the LMC disk: heavy and light [s-process/Fe] and [$\alpha$/Fe] give constraints on the time scales of formation of the stellar population. Cu, Na, Sc, and the iron-peak elements are also studied aiming to better understand the build up of the elements of this population and the origin of these elements. We aim to provide a more complete picture of the LMC's evolution by compiling a large sample of field star abundances.
Methods. LTE abundances were derived using line spectrum synthesis or equivalent width analysis. We used OSMARCS model atmospheres and an updated line list.
Results. We find that the alpha-elements Ca, Si, and Ti show lower [X/Fe] ratios than Galactic stars at the same [Fe/H], with most [Ca/Fe] being subsolar. The [O/Fe] and [Mg/Fe] ratios are slightly deficient, with Mg showing some overlap with the Galactic distribution, while Sc and Na follow the underabundant behavior of Ca, with subsolar distributions. For the light s-process elements Y and Zr, we find underabundant values compared to their Galactic counterparts. The [La/Fe] ratios are slightly overabundant relative to the galactic pattern showing low scatter, while the [Ba/Fe] are enhanced, with a slight increasing trend for metallicities [Fe/H] > -1 dex. The [heavy-s/light-s] ratios are high, showing a slow, increasing trend with metallicity. We were surprised to find an offset for three of the iron-peak elements. We found an offset for the [iron-peak/Fe] ratios of Ni, Cr, and Co, with an underabundant pattern and subsolar values, while Vanadium ratios track the solar value. Copper shows very low abundances in our sample for all metallicities, compatible with those of the Galaxy only for the most metal-poor stars. The overall chemical distributions of this LMC sample indicates a slower star formation history relative to that of the solar neighborhood, with a higher contribution from type Ia supernovae relative to type II supernovae.

Key words: stars: abundances - galaxies: Magellanic Clouds - Galaxy: abundances - galaxies: evolution

1 Introduction

During the last decade, the operation of the new class of large telescopes, has led to analysis of the elemental abundances of large samples of individual stars in external galaxies for the first time. Thanks to new optical technologies, objects fainter than supergiant stars, planetary nebulae, or HII regions are now possible targets suitable for extragalactic research, allowing the study of older objects and the exploration of earlier phases of galaxy evolution. The abundance patterns of diverse elements in numerous stars in a galaxy give information on different domains such as the kinematic and chemical evolution, nucleosynthesis channels, the star formation history (SFH), and the initial mass function (IMF) of its stellar population(s).

One of the most interesting extragalactic objects in the study of stellar populations is the Large Magellanic Cloud (LMC), our nearest companion after the Sagittarius dwarf galaxy (that is in the process of merging with the Milky Way). The LMC is an irregular galaxy located within 50 kpc of the Sun, with a kinematically-defined disk, a bar, and a thick disk or flattened halo (e.g. Westerlund 1997). The almost face-on position of its disk, with a tilt of $\sim$30$^{\circ}$ relative to the plane of the sky, gives us the precious opportunity to study stars from its different components.

The star formation (SF) and cluster formation histories of this galaxy have been studied for more than three decades (e.g. Butcher 1977; van den Bergh 1979; Olszewski et al. 1996, and references therein; Cioni et al. 2006, and references therein) although a final picture is far from complete (the current status of the research deals with the detailed SF and cluster formation within the different components and regions of this galaxy, e.g. Geha et al. 1998; Smecker-Hane 2002; Subramaniam 2004; Javiel et al. 2005; Cole et al. 2005). The clusters of the LMC show an ancient population with ages >11.5 Gyr, followed by a hiatus when just one single cluster seems to have formed (ESO 121-SC03) (e.g. van den Bergh 1998, and references therein). Some 2-4 Gyr ago, a new formation event was triggered and some other clusters were built (e.g. Da Costa 1991). The SF in the disk field shows a different evolution, with nearly constant rate over most of the history of the LMC (Geha et al. 1998). The SFR appears to have been enhanced some 1-4 Gyr ago, with the timing and amplitude of the ``burst'' seeming to vary between locations (Holtzman et al. 1999; Olsen et al. 1999). The SFH of the bar field appears to more closely track the cluster formation history, with a strong burst $\approx$3-6 Gyr ago (Smecker-Hane et al. 2002; Cole et al. 2005). The lack of a field-star age gap means that field star properties can be used to trace the history of the LMC during the 3-11 Gyr cluster age gap (Da Costa 1999; van den Bergh 1999).

The elemental distributions of the LMC stars are still poorly known, because of the paucity of data, but the present picture is changing fast due to new observational programs (e.g. Evans et al. 2005; Dufton et al. 2006; Johnson et al. 2006). We briefly summarize here the results for elemental abundances in the LMC (for a detailed discussion see Hill 2004). The abundance analysis of B stars and HII regions (Garnett 1999; Korn et al. 2002; Rolleston et al. 2002) show a deficient abundance of O, Mg, and Si compared to their solar neighborhood counterparts, with mean $\log~ ($X/H $) - \log~ ($X/H) $_{\odot} \sim -0.2$ dex for oxygen, -0.2 dex for magnesium, and -0.4 dex for silicon abundances (this last value is only for the B stars, while HII regions show a much lower value of $\sim$-0.8 dex), but compatible with galactic supergiant values. Russell & Dopita (1992), Hill et al. (1995), and Luck et al. (1998) studied samples of supergiants in the field of the LMC, and Hill & Spite (1999) derived abundances for supergiants in clusters. They found a similar behavior for the $\alpha$-elements when compared to the galactc disk values, while for the heavy elements (those with $Z \geq 56$) the abundance ratios are enhanced by a factor of $\sim$2. Huter et al. (2007) derived C, Mg, O, Si and N abundances for three globular clusters from the LMC, and found an average value 0.3 dex lower than that of the Galactic Clusters for all the analyzed elements, except for N. Red giant branch stars (herafter RGB stars) from the field (Smith et al. 2000, hereafter SM02) and from globular clusters (Hill et al. 2000, 2003, hereafter H00 and H03; Johnson et al. 2006, hereafter JIS06) have also been studied. A general tendency toward low [$\alpha$/Fe] ratios compared to the stars of the galactic disk with similar metallicities is detected (with the exception of Si and Mg in JIS06), while the same overabundant pattern found for the LMC supergiants has been derived for the heavy-elements. JIS06 inferred the [Y/Fe] ratios and found abundances compatible to the solar value. Sodium abundances are different in field and clusters stars. While SM02 find low [Na/Fe] ratios and [Sc/Fe] ratios close to zero, JIS06 find that [Sc/Fe] and [Na/Fe] ratios are similar to their galactic counterparts. JIS06 have derived the [iron-peak/Fe] abundances to find that Ni, V, and Cu abundances fall below their corresponding galactic values.

An observational project aiming at a full analysis of the elemental abundances of significant samples ($\sim$70-100) of stars from different locations in the LMC has been developed, taking advantage of the FLAMES multiplex facility at the VLT. We obtained spectra from stars in three different regions of the LMC: the inner disk (characterized by a galactocentric radius of $R_{\rm C}=2$ kpc); the outer disk (with $R_{\rm C}=4$ kpc); and a field near the optical center of the bar. Stars were selected based on kinematics and metallicity data derived from the near-infrared calcium triplet (CaT and CaT metallicities), trying to sample the whole metallicity range of this galaxy as evenly as possible. In the present paper, we focus on a sample of RGB stars on the inner disk region, previosly studied by Smecker-Hane et al. (2002), who derived the ages, metallicities (CaT) and kinematics of this sample. They have identified two kinematical groups in the inner disk field, one with a velocity dispersion of $13\pm4$ km s^-1, characterizing a thin disk, and one with velocity dispersion of $34\pm6$ km s^-1, probably pertaining to the flattened halo. The metallicities of these two groups are different: the low-dispersion velocity group has metallicities ranging $-0.6 \leq [$Fe/H $] \leq -0.3$ dex, while the high-dispersion velocity component has $-2 \leq [$Fe/H $] \leq -0.4$. The ages derived for this inner disk population has shown that stars have continuously formed during the past $\sim$1 to 15 Gyr, with a possible enhancement in the SFR some 3 Gyr ago.

As the prototype galaxy of the Magellanic irregular class, to learn the evolutionary history of the LMC is clearly a vital step towards the global understanding of galaxies near the dwarf-giant boundary. Because the Magellanic Clouds have evolved in such close proximity to the Milky Way, their histories have been intimately tied to that of our own galaxy. The ongoing impact of the LMC on the structure and kinematics of the Milky Way is manifest in the warp of the Galactic disk and possibly in the presence of the central bar (e.g., Weinberg 1999), while Bekki & Chiba (2005) have used N-body simulations to show that the LMC could have made a significant contribution to the build up of the Milky Way halo as a result of tidal stripping.

According to models of galaxy formation within a hierarchical CDM scenario (D'Onghia & Lake 2004; Moore et al. 1999), the history of the Milky Way depends strongly on its interactions with its environment. It now seems that the abundance patterns in dwarf spheroidal stars are different from those in Milky Way halo stars (e.g., Shetrone et al. 2001; Tolstoy et al. 2003; Geisler et al. 2005), ruling them out as analogues to the accreting fragments that built the halo up. Study of the LMC takes on added significance in this light, because of the hypothesis by Robertson et al. (2005) that the accretion of LMC-like fragments circumvents this difficulty with the hierarchical accretion scenario. Deeper knowledge of the abundances in the oldest LMC stars therefore has direct bearing on the evolution of our own Galaxy.

In the present paper, we focus on the inner disk region of the LMC, presenting abundance results for iron-peak, heavy and light s-process elements, and $\alpha$elements for a total of 59 stars. With this detailed information in hand, we aim to shed light on the following questions. (i) What are the chemical abundance patterns of the inner disk of the LMC? (ii) What do these elemental distributions tell us about the formation and evolution of the LMC? (iii) Are they similar to any component of the Milky Way? (iv) Or to the populations of other Local Group galaxies? (v) Based on the elemental distributions, is a merging scenario with LMC debris a likely solution for the Galactic halo formation?

The paper is organized as follows. In Sect. 2 the observations and the reduction procedure are described, the calculation of stellar parameters is presented in Sect. 3, Sect. 4 describes the abundance determination procedures, Sect. 5 reports the results for the abundance ratios and compares them to Milky Way samples, in Sect. 6 we compare our results to those for the dSph galaxies. We discuss the results in Sect. 7, and finally in Sect. 8 give a summary of the work.

2 Sample selection, observations and reductions

To best measure the elemental abundances of the LMC disk and their evolution in time, we selected a field located $1.7^{\circ}$southwest of the LMC bar, in the bar's minor axis direction to ensure a negligible contribution of its stellar populations. An HST color-magnitude diagram study of this field (SMH02) found it to have experienced a rather smooth and continuous history of star formation over the past 13 Gyr, with a possibly increased SFR over the past 2 Gyr. This contrasts with the history of the bar itself, in which significant episodes are seen to have commenced 4-6 Gyr ago (SMH02; see also Holtzman et al. 1999, and references therein). This field has also more recently been the target of a low-resolution spectroscopy campaign (Cole et al. 2000; SMH), using the CaT to derive its metallicity distribution and break the age-metallicity degeneracy inherent to color-magnitude diagram CMD analyses.

We have these infrared CaT metallicities from SMH to select a sample of RGB members of the LMC (based on their radial velocities) distributed uniformly (i.e., with the same number of stars in each metallicity bin) over the whole metallicity range of the LMC disk. In this way, we have been able to sample the lower metallicity bins of the LMC very efficiently. The most metal-poor stars convey essential information on the evolution of the elements of this galaxy, but they are rare, so their number would have been significantly lower if we had selected our sample by picking stars randomly across the RGB. The final sample consists of 67 stars with CaT metallicities ranging from -1.76 to -0.02 dex (including 13 stars with metallicities below -1.0 dex), drawn from the 115-star sample of SMH. In Fig. 1 we show the sample stars overplotted on the color-magnitude diagram of the LMC inner disk region (CTIO photometry from SMH). The sample mean magnitude is V=17.25 mag, bright enough to allow reasonable S/N high-resolution spectra to be acquired.

	Figure 1: V vs. (V-I) color-magnitude diagram of the disk region (following SMH02), with our sample stars overplotted: asterisks are stars with [Fe/H] $_{\rm CaT} \geq - 0.5$ dex, triangles $-1.0 \leq [$Fe/H] $_{\rm CaT} < - 0.5$ dex, squares [Fe/H] $_{\rm CaT} < - 1.0$ dex.
Open with DEXTER

The observations were made at the VLT Kueyen (UT2) telescope at Paranal during the science verification of FLAMES/GIRAFFE (Pasquini et al. 2000) in January, February, and March 2003, complemented by one night of the Paris Observatory Guaranteed Time Observations in January 2004. In its MEDUSA mode, GIRAFFE is a multiobject spectrograph with 131 fibers of which 67 were used for the present project. The remaining fibers were allocated to targets of other science verification projects in the LMC. The detector is a $2048 \times 4096$ EEV CCD with 15 $\mu$m pixels. We used the high-resolution grating of GIRAFFE in three different setups: (i) H14 $\lambda$638.3-$\lambda$662.6 nm with R=28 800, (ii) H13 $\lambda$612.0-$\lambda$640.6 nm with R=22 500, and (iii) H11 $\lambda$559.7-$\lambda$584.0 nm with R=18 529. Exposure times are 6 h for H14 and H13 setups and 7h30 for H11. The setups were chosen to cover the maximum number of key elements such as Fe I and Fe II for spectroscopic calculations of stellar parameters, and $\alpha$, iron-peak, and s-process elements, for the abundance analysis. The average signal-to-noise ratio of the spectra is $S/N\sim80$ per resolution element.

The data reduction was carried out using the BLDRS (GIRAFFE Base-Line Data Reduction Software http://girbldrs.sourceforge.net/) and consists of bias subtraction, localization and extraction of the spectra, wavelength calibration and rebinning. We also used the MIDAS packages for sky subtraction and co-addition of individual exposures.

Table 2: Stellar parameters.

3 Determination of stellar parameters

3.1 Photometric stellar parameters

A first guess of the stellar parameters was made using photometric data of CTIO (V, I from SHM) and 2MASS (J, H, K). Bolometric magnitudes and effective temperatures were derived from calibrations of Bessell et al. (1998, hereafter BCP). The observed CTIO and 2MASS colors were transformed into the corresponding photometric systems using Fernie (1983, V-I  Cousins to Johnson) and Carpenter (2001, K, V-K& J-K 2MASS). Photometric data are given in Table 1, while Table 2 gives the derived effective temperatures ( $T_{\rm phot}$) and surface gravities ( $\log g_{\rm phot}$). The $T_{\rm phot}$ is derived using the BCP calibration of the deredenned V-Iand V-K colors, and the surface gravity is computed using the following relation:

$$\log {g_{\rm phot}=4.44 +\log~(M)+4\log~\left( \frac{T_{\rm phot}}{5790}\right)+0.4(M_{\rm bol}-4.75)}$$

where $M_{\rm bol}$ is computed from the dereddened K magnitude of the star, the bolometric correction BC_K taken from BCP, and the mass of the stars (M) is assumed to be 2 $M_\odot$. A distance modulus based on Hipparcos data and the period-luminosity relations from LMC Cepheids of $18.44 \pm 0.05$ mag is assumed (Westerlund 1997; Madore & Freedman 1998). Uncertainties of this value stem from the specific subsets of the Cepheids chosen for the comparison (Madore & Freedman 1998). For the reddening, two values were checked: E(B-V)=0.03, which was derived by SMH02 for the sample of the inner disk, using Strömgren photometry; and E(B-V)=0.06, a mean value for the whole disk (Bessell 1991). We adopted CaT metallicities from SMH as our initial guesses and reported them in the [Fe/H] $_{\rm CaT}$ column of Table 2.

We derived temperatures from V-I, V-K, and J-K colors. We found some trends when comparing temperatures from different colors: $T_{\rm eff}$(V-I) is 65 K hotter than $T_{\rm eff}$(V-K) in the mean, with $\sigma=59$ K; $T_{\rm eff}$(J-K) is 21 K hotter than $T_{\rm eff}$(V-K) and shows a highly dispersed relation, with $\sigma=118$ K (these numbers vary only slightly when choosing a reddening of E(B -V)=0.06 or 0.03). As initial values of our stellar temperatures we have chosen to use a weighted mean of the estimates from V-I and V-K, omitting the less sensitive J-K color. We assign higher weight to the more temperature-sensitive (V-K), according to the expression

$${T_{\rm eff} = (T_{\rm eff}(V-I)+ 2 \times T_{\rm eff}(V-K))/3.}$$

In Table 2 the inferred temperatures are given for the two values of reddening, $T_{\rm photLow}$ and $T_{\rm phot}$ for E(B-V) = 0.03 and 0.06 respectively.

3.2 Spectroscopic parameters

The final stellar parameters used for the abundance determination of the sample stars were derived spectroscopically using abundances derived from the equivalent widths (EW) of iron lines. Although 67 stars were observed, 8 of them have one or two setups with low S/N, compromising the determination of stellar parameters. These stars have not been included in the abundance analysis. Due to low S/N ratios, the H13 setup has not been used for the following stars: RGB_601, RGB_646, RGB_672, RGB_699, RGB_705, RGB_710, RGB_720, RGB_731, RGB_748, RGB_756, RGB_773 and RGB_775; and for RGB_666 the H11 setup has been discarded. We have estimated the stellar parameters as follows: effective temperatures are calculated by requiring no slope in the A(Fe I) vs. $\chi _{\rm exc}$ (excitation potential) plot ( $\chi _{\rm exc}$ is the excitation potention of the line); microturbulent velocities, $V_{\rm t}$, are derived demanding that lines of different EW give the same iron abundance, also checking for no slope in the [Fe/H] vs. log (W/$\lambda$) plot (iron abundance vs. the reduced EW); and surface gravities are determined by forcing the agreement between Fe I and Fe II iron abundances (within the accuracy of the abundance determination of Fe II). For the temperature and surface gravity ranges covered by our current sample of stars, $T_{\rm eff}$, and $\log g$determinations are correlated well and the calculation of stellar parameters is made iteratively. In Fig. 2 we show an example of the excitation equilibrium calculation for RGB_656, and in Fig. 3, the [Fe/H] vs. $\lambda$with the log (W/$\lambda$) check and the [Fe/H] vs. EW are given for RGB_625. The spectroscopic and photometric parameters of all our stars are reported in Table 2, together with the barycentric radial velocities calculated from the spectra.

	Figure 2: Example of the temperature calculation for RGB_656: [Fe I/H] vs. $\chi _{\rm exc}$.
Open with DEXTER

Figure 3: Examples of microturbulence velocity calculation for RGB_625: [Fe I/H] vs. $\lambda$ ( upper panel); and [Fe I/H] vs. EW ( lower panel). The different values for the reduced EW in the left panel are given with different symbols: 1) squares: $-5.6\leq \log W/\lambda \leq -5.0$ and 12<W<50; 2) crosses: $-4.8 \leq \log W/\lambda \leq -4.0$ and 80<W<300; and 3) times: $-5.0 \leq \log W/\lambda \leq -4.8$ 50<W<80.

Open with DEXTER

3.2.1 Equivalent widths, line list, and model atmospheres

The EW of the lines and the radial velocities (RV, in km s^-1, reported in Table 2) of the stars are computed using the program DAOSPEC written by Stetson (Stetson & Pancino, in preparation). The line list and the atomic data were assembled from the literature and the oscillator strengths references are given in Table 4. DAOSPEC has already been used to measure the EW of spectra for different types of stars yielding reliable results (e.g. Pasquini et al. 2004; Barbuy et al. 2006; Sousa et al. 2006). We made a study of the DAOSPEC EWestimations using GIRAFFE spectra. In Appendix A we show a comparison of DAOSPEC EW with those made by hand using the Splot-Iraf task for six of our sample stars. We found very good agreement between the two methods for the analysis of the GIRAFFE spectra within the expected uncertainties.

MARCS 1D plane-parallel atmospheres models (Gustafsson et al. 1975; Plez et al. 1992; Gustafsson et al. 2003) were kindly provided by B. Plez (private communication).

3.2.2 Comparison with UVES analysis

In a previous observing run (066.B-0331), we obtained UVES spectra (in slit mode) for one of our sample stars, RGB_666. UVES is an echelle spectrograph also mounted on the VLT Kueyen telescope with a higher resolving power of R=45 000 (with a slit of 1 $\hbox{$^{\prime\prime}$ }$) and a much wider wavelength coverage (in the case of our chosen set-up, 4800-6800 Å), and therefore with a better performance to derive EWs. We used this spectrum to evaluate DAOSPEC performance to derive EW from low-resolution spectra. In Fig. 4, EWs derived with DAOSPEC from UVES spectra from 5800 to 6800 Å are compared to those also measured by this program on the GIRAFFE spectra of the same star using the same line list. The top of the plot, the mean differences between analyses are given, together with the dispersion and the number of lines used. Lines of all elements are plotted in this comparison. We can see from this plot that GIRAFFE EW are only slightly higher than UVES EW. Such a difference is probably due to a better definition of the continuum for the UVES spectra, as well as to the increased blending at the lower resolution of GIRAFFE. Using the EW of this figure, we inferred the stellar parameters for UVES to compare the analysis from both spectrographs. We find that the stellar parameters are almost identical to those given for RGB_666 in Table 2, except for $V_{\rm t}$ for which we found $V_{\rm t} = 1.8$ kms^-1 (a difference of $\Delta V_{\rm t} = +0.1$ km s^-1). Comparing the results from the two spectrographs, we have [FeI/H] $_{\rm UVES}-[$FeI/H] $_{\rm GIRAFFE} = -0.11$ dex and [FeII/H] $_{\rm UVES}-[$FeII/H] $_{\rm GIRAFFE} = 0.00$ dex. Therefore it is possible that a systematic uncertainty of [Fe/H $] \sim 0.1$ dex may be present in the following abundance analysis, although robust statements on this uncertainty would require better statistics. Let us further note that this 0.1 dex difference is within the errorbar that we quote for our GIRAFFE metallicities.

	Figure 4: Comparison between UVES and GIRAFFE spectra analyses for RGB_666.
Open with DEXTER

3.3 Behavior of stellar parameters

We found good agreement between spectroscopic and photometric temperatures. Our spectroscopic temperatures are hotter than photometric temperatures derived using the low reddening value, $T_{\rm photLow}$, by 113 K, with $\sigma=91$ K, and by 54 K than $T_{\rm phot}$ (higher reddening value) with $\sigma=64$ K. An interesting result is depicted in Fig. 5 where we compare the spectroscopic temperatures $T_{\rm eff}$(spec) with those derived from colors, $T_{\rm eff}$(V-I) and $T_{\rm eff}$(V-K), and from the equation given in Sect. 3.1, $T_{\rm eff}$(phot), for both values of reddening ( E(B-V)=0.06 in the upper panels, and E(B-V)=0.03 in the lower panels). This figure shows that photometric temperatures inferred using E(B-V) = 0.06 agree much better with spectroscopic temperatures than those derived with the lower E(B-V). Provided that the photometric temperatures and the excitation temperature scale show good agreement, this could indicate that E(B-V)=0.06is a better reddening value for this region.

Figure 5: Comparison between photometric and spectroscopic temperatures (see text). In the bottom plots, photometric temperatures are derived with E(B-V)=0.03 (SMH02), while in the upper plots, photometric temperatures are derived with a higher reddening value, E(B-V)=0.06(Bessel 1991). Solid lines represent $T_{\rm eff(spec)} = T_{\rm eff(phot)}$ and $T_{\rm eff(spec)} = T_{\rm eff(phot)} \pm 100$ K.

Open with DEXTER

On average, spectroscopic surface gravities are lower than the photometric estimates by $\Delta (\log g_{\rm spec} - \log g_{\rm phot}) = -0.38$ dex, as might be expected if NLTE overionization effects are at work (Korn et al. 2003). This systematic effect in $\log g$ corresponds to a 0.2 dex difference between FeI and FeII.

The metallicities that we derive differ on average from those derived from the CaT by $\rm\Delta ([Fe/H]_{CaT} - [Fe/H]_{spec}) = +0.13$ dex with $\sigma= 0.27$ dex. In fact, most of this effect comes from the high-metallicity end of the sample: for [Fe/H] $_{\rm CaT}>-0.6$ dex, CaT seems to overestimate the metallicity systematically by 0.27 dex ( $\sigma= 0.19$ dex), whereas for the metal-poor end of the sample, there is almost no systematic effect ( $\rm\Delta [Fe/H]_{CaT} - [Fe/H]_{spec}) = -0.04$ dex with $\sigma= 0.24$ dex.

Finally, in Fig. 6, abundance ratios of different species, [Cr/Fe], [Ni/Fe], and [V/Fe], against temperatures are plotted to check the quality of the spectroscopic temperatures. As can be seen from this picture, there is no trend of the abundances of the elements with temperature, which means that our temperatures are well-defined. Our final sample comprises 59 red giant stars within $-1.7 < {\rm [Fe/H]} < -0.30$ dex and temperatures ranging from 3900 K to 4500 K.

	Figure 6: Abundance ratios against temperatures. From top to bottom: [Cr/Fe] vs. $T_{\rm eff}$, [Ni/Fe] vs. $T_{\rm eff}$ and [V/Fe] vs. $T_{\rm eff}$.
Open with DEXTER

4 Abundance determination

We have selected a list of lines covering the chosen setups in order to sample the most important elements as much as possible: iron-peak, neutron-capture, and $\alpha$ elements. Abundances are derived from EW mesurements for eight elements (in parenthesis the average number of lines used in the analysis): Fe (45), Ni (7), Cr (4), V (11), Si (3), Ca (10), Ti (7) and Na (3). We also derived abundances by using line synthesis for nine elements (in parenthesis the lines used in the synthesis): O ([O I] 6300 $\rm \AA$), Mg I (5711 $\rm \AA$), Co I (5647 $\rm \AA$), Cu I (5782 $\rm \AA$), Sc II (5657 $\rm \AA$), La II (6320 $\rm \AA$), Y II (6435 $\rm \AA$), Ba II (6496 $\rm \AA$), and Zr I (6134 $\rm \AA$). The code used for the abundance analysis was developed by Monique Spite (1967) and has been improved over the years. We note that both model atmospheres and the line synthesis program are in spherical geometry, so errors due to geometry inconsistencies are minimized (Heiter & Eriksson 2006). For the synthesis of the [O I] line in 6300.311 $\rm \AA$, we took the blend with Ni I 6300.336 $\rm \AA$ (line data from Allende Prieto 2001) into account, but no differences were detected between results with or without such a blend. Hyperfine structures (HFS) have been taken into account for the following elements (the line sources are given in parenthesis): Ba II (Rutten 1978, and the isotopic solar mix following McWilliam 1998), La II (Lawler et al. 2001, with $\log gf$ from Bord et al. 1996), Cu (Biehl 1976), and Co I and Sc II (Prochaska et al. 2000). In Fig. 7 the fitting procedure is shown for the Y I 6435 $\rm \AA$ line in RGB_752 and the La II line 6320 $\rm \AA$ in RGB_690. Abundances are given relative to solar abundances of Grevesse & Sauval (2000). Atomic lines for the synthesis have been chosen according to the quality of the synthetic fit in the Solar Flux Atlas of Kurucz et al. (1984). The derived abundances are given in Tables 5 to 8.

Figure 7: Example of the line synthesis procedure for the Y I and La II lines: left panel: Y I $\lambda$6435$\rm \AA$ line fitting for RGB_752; right panel: La II $\rm \AA$ 6320 line fitting for RGB_690. The black circles depict the observed spectra and the lines are the synthetic spectra. Abundances of the synthetic spectra are [Y/Fe] = -0.55 (dashed line), -0.45 (continuous line - best fit), -0.25 (dotted line), [La/Fe] = 0.56 (dashed line), 0.66 (continuous line - best fit), and 0.76 (dotted line).

Open with DEXTER

Errors in the derived abundances have three main sources: the uncertainties in the stellar parameters, the uncertainties in the measurements of the EW (or spectrum synthesis fitting), and the uncertainties on the physical data of the lines (mainly $\log gf$). The errors due to stellar parameters uncertainties were chosen as the maximum range each parameter could change not to give unrealistic models atmospheres. The errors $\rm \delta([X/Fe])_{model}$ are given in Table 3, assuming the following uncertainties in each of the stellar parameters: $\Delta$( $T_{\rm eff}) = \pm$100 K, $\Delta(\log g) = \pm$0.4 dex, $\Delta(V_{\rm t}) = \pm$0.2 km s^-1, and $\Delta$([Fe/H]$) = \pm$0.15 dex.

Errors in the EW measurement we computed by DAOSPEC during the fitting procedure, then propagated into an abundance uncertainty for each line, and finally combined into an abundance error on the mean abundance for each element ( $\delta_{\rm DAOSPEC}$). Errors due to the combined uncertainties on the line data and line measurement are reflected in the abundance dispersion observed for each element, provided that the number of lines is large enough to measure this dispersion in a robust way ($N\geq3$). We therefore combined these error estimates conservatively as given below:

$$N_{\rm X} 3\!: \delta\rm ([X/H]) = \delta_{DAOSPEC},$$ <

$$N_{\rm X} \geq 3\!: \delta([{\rm X/H}]) = {\rm Max}\left(\delta_{\rm DAOSPEC}, \frac{\sigma({\rm X})}{\sqrt{N_{\rm X}}}\right)$$ (1)

where $N_{\rm X}$ is the number of lines of the element X and $\sigma$(X) the dispersion among lines.

These errors were calculated for each element and given in Tables 5 to 7, together with the abundances derived from the EW. For elements measured by synthesis spectrum fitting, an error estimate was carried out of the typical abundance change for which two different synthetic spectra (i.e. computed with two slightly different abundances) still satisfactorily fit the same line. On average, these values are the following for each element: ${\delta}$[Zr/H] = 0.15 dex, ${\delta}$[Y/H] = 0.15 dex, ${\delta}$[La/H] = 0.20 dex, ${\delta}$[Ba/H] = 0.25 dex, ${\delta}$[Co/H] = 0.10 dex, ${\delta}$[Cu/H] = 0.20 dex, ${\delta}$[Sc/H] = 0.10 dex, ${\delta}$[Mg/H] = 0.15 dex, and ${\delta}$[O/H] = 0.20 dex. For the error bars reported in our abundance plots (always shown in the lower left corner of Figs. 8-12), we adopted two error sources. The first, caused by stellar parameter uncertainties (leftmost side of the plots), comes directly from Table 3, whereas the second (more to the right side) represents the error associated with the abundance analysis. For those abundances derived from the EW, this is the mean error of Tables 5-7, and for those elements with abundances derived from spectrum synthesis, it is the value described earlier on in this section.

5 Abundance distributions and comparison to galactic samples

In Figs. 8 to 12 we depict the elemental distributions for the $\alpha$-elements, the iron-peak group, Na, Sc, Cu, and s-elements for our stars compared to different samples of the Galaxy and the LMC. Our data are represented as dots. The references of the disk are Fulbright (2000, crosses), Reddy et al. (2003, open squares), Allende Prieto et al. (2004, open stars), Prochaska et al. (2000, open triangles), Burris et al. (2000, stars - only for the heavy-elements plots), Johnson & Bolte (2002, open triangles - only for the heavy-elements plots), Simmerer et al. (2004, open hexagons), Nissen & Shuster (1997, asterisks, only stars with low [$\alpha$/Fe] ratios), Nissen et al. (2000, asterisks - Sc abundances for the low-$\alpha$ stars), and Bensby et al. (2004, open squares - only for the oxygen plot). LMC globular clusters (GC) stars from Hill et al. (2000, hereafter HI00) for O, and Hill (2004, hereafter HI04) for Na, Mg, Ca, and Si are plotted as downward-pointing, open triangles. The LMC GC stars from JIS06 are represented as open diamonds and field LMC red giants of SM02 as open pentagons. Error bars as described in Sect. 4. are shown in the lower left side of the plots.

5.1 Ca, Si, and Ti

In Fig. 8, the elemental distributions for Ca I, Si I, and Ti I are depicted. We find that [Si/Fe] follows roughly the solar ratio with some scatter. The [Ca/Fe] shows a slight decrease with metallicity. Compared to the distribution of the Galactic halo, both Si and Ca mean abundances are deficient by a factor of 3. The Ti I ratios are also underabundant relative to Galactic disk and Galactic halo samples, and agree very well with the results of SM02, who derived Ti abundances from neutral lines for a sample of red giants from LMC disk. There is a hint of a decreasing trend of Ti abundances for higher metallicity stars, especially when SM02 datapoints are taken into account with our sample. Compared to the LMC GC of H04, we find that the star of our sample with metallicity similar to those of those of Hill et al. (2004) also has a similar [Ca/Fe] ratio. The JIS06 sample of LMC GC stars seems to overlap our [Ca/Fe] and [Ti/Fe] distributions, while their [Si/Fe] ratios are enhanced.

Figure 8:  Abundance distributions for the inner disk LMC stars: [$\alpha$/Fe] vs. [Fe/H] (dots). LMC samples are depicted with polygons (downward-pointing triangles - Hill et al. 2000; pentagons - Smith et al. 2002; diamonds - Johnson et al. 2006); and the remaining symbols are data for the galactic stars (crosses - Fulbright 2000; open squares - Reddy et al. 2003; asterisks - Nissen & Schuster 1997). Error bars depict a) leftmost side of the plots - errors due to stellar parameter uncertainties (Table 3), and b) more to the right side - errors associated with the abundance analysis - for those derived from the EW, is the mean error of Tables 5-7; for those elements with abundances derived from spectrum synthesis, the value is described in Sect. 4.

Open with DEXTER

A very interesting result emerges when comparing our data with those of Nissen & Shuster (1997, hereafter NS97, asterisks). NS97 discovered a sample of stars from the Galactic halo with abnormal abundances: low [$\alpha$/Fe], [Na/Fe], and [Ni/Fe] ratios compared to ``standard'' halo stars. Such chemically peculiar or ``low-$\alpha$'' halo stars play an important role in elucidating the possible merging history of the Galactic halo. Because of their chemical properties, they indicate that this group has formed in another stellar system that evolve separately and has been captured or ejected to the halo. Comparing our LMC distribution to the low-$\alpha$ stars, we have found that NS97 stars show a slightly enhanced mean $\alpha$ abundance compared to our LMC stars.

Silicom, Ca and Ti are predicted to be produced in intermediate mass type II SNe (SNe II) with a smaller contribution from type Ia SNe (SNe Ia) (e.g. Tsujiomoto et al. 1995; Thielemann et al. 2002), while Fe is mostly produced by SNe Ia (e.g. Thielemann et al. 2001; Iwamoto et al. 1999). The low [$\alpha$/Fe] ratios observed indicate that SNe Ia has contributed more to the ISM content in the past than the SNe II.

5.2 Mg, O, Na and Sc

In Fig. 9, abundance ratios are given for O, Mg, Sc, and Na. Nucleosynthetic predictions attribute the main source of O, Mg, and Na to high-mass stars, with $M > 25~M_\odot$, which explode as SNe II (Woosley & Weaver 1995, hereafter WW95), with Na production controlled by the neutron excess. Although WW95 attribute the origin of Sc to SNe II, the main source of Sc production is still unclear (e.g. McWilliam 1997; Nissen et al. 2000).

	Figure 9:  Abundance distributions for the inner disk LMC stars: [O, Mg, Na, Sc/Fe] vs. [Fe/H] (symbols are the same as in Fig. 8, and we added: open stars - Allende Prieto et al. 2004; crosses - Bensby et al. 2004, only for oxygen; asterisks - Nissen et al. 2000, for Sc; Nissen & Schuster 1997, for the other elements).
Open with DEXTER

As can be seen in the upper panel of Fig. 9, oxygen ratios fall in the lower envelope of the Galactic halo and disk distributions. For higher metallicities, it shows a faster decline with metallicity compared to stars from the Galactic disk. In the second plot we see that the [Mg/Fe] ratios for the LMC Inner Disk overlap those of the Galaxy, but with lower mean values. In contrast, Na and Sc behaviors are similar to those of the $\alpha$-elements Ti and Si. Both elements are deficient and show lower values for higher metallicities, while for the metal-poor tail, a match to the Galactic samples is observed. From this figure we see that the different LMC samples agree very well for all elements, Mg, O, Na, and Sc, even the LMC globular clusters of H00, H04 and JIS06. A few stars in the NS00 sample of low-$\alpha$ stars show low [Sc/Fe] ratios and overlap our sample, but most of them show solar [Sc/Fe] values, higher than in our LMC sample. Sodium abundances in NS97 sample are similar to our values, although with a higher mean abundance. It is important to notice that sodium abundances in giants are still uncertain. Pasquini et al. (2004) find that [Na/Fe] ratios in giant stars are slightly higher than those from dwarf stars in the same cluster. High [Na/Fe] ratios are also inferred from giants in M 67 (Tautvaisiene et al. 2000). But such results have not been confirmed in the reanalysis of [Na/Fe] in giants and dwarfs of M 67 (Randich et al. 2006).

Nissen et al. (2000) also found that Sc behaves similarly to Na, showing lower [Sc/Fe] ratios in their low-$\alpha$ stars, suggesting a correlation among those elements. To test the hypothesis of a correlation among Na and $\alpha$-elements and Sc and $\alpha$-elements we applied a statistical test to check for the existence and significance of such correlation, calculating the linear correlation coeficient, which varies from 1 or -1 (maximum correlation or anti-correlation) to 0 (no correlation). We have found that the correlations are weak: for Na-Ca, a correlation coefficient $\phi = -0.06$is found, and for Sc-Ca, $\phi = 0.39$.

5.3 Iron-peak elements

Abundance distributions for the iron-peak elements are shown in Fig. 10. The iron-peak elements Co, Ni, and Cr display a very distinct pattern in the LMC inner disk stars, with underabundant values compared to the Galactic distributions and many subsolar ratios. [Co/Fe], [Cr/Fe] and [Ni/Fe] show a flat trend for most of the metallicity range, with mean abundances of $\sim$-0.18 dex for Cr, $\sim$-0.24 for Ni, and $\sim$-0.14 dex for Co. The [V/Fe] ratios are similar to the galactic halo and disk patterns and track the solar value, with a group of stars showing lower values. Results from the LMC GC of JIS06 seem to agree with our samples for Co, Ni, and Cr. Vanadium in their sample shows an offset, with abundance ratios corresponding to the stars with lower values in our sample. NS97 low-$\alpha$ stars overlap our sample for Ni and Cr, but lie in the high abundance envelope of the distributions.

	Figure 10:  Abundance distributions for the inner disk LMC stars: [iron-peak/Fe] vs. [Fe/H] (symbols are the same as in Figs. 8 and 9, and the solid downtriangles depict upper limits for our sample stars).
Open with DEXTER

According to nucleosynthetic predictions, iron-peak elements are mainly produced in SNe Ia (Iwamoto 1999; Travaglio et al. 2005): while each SN Ia produces $\approx$0.8 $M_\odot$ of the solar iron-peak elements, SN II produce $\approx$0.1 $M_\odot$ each (Timmes et al. 2003). The difference in the distributions from one environment to the other are evidence that the production factors for each iron-peak element are not the same in the different types of SNe and depend on the SFH of the parent population. This will be further discussed in Sect. 7.

5.4 Copper

	Figure 11: Abundance distributions for inner disk LMC stars: Copper. The symbols are the data from: our sample stars (dots); Mishenina et al. (2002, stars), Prochaska et al. (2002, open triangles), Reddy et al. (2003, open squares), Johnson et al. (2006, diamonds).
Open with DEXTER

In Fig. 11 we show the plot for Cu. We have found that in the inner disk LMC stars, the copper distribution is flat, with a mean value of [Cu/Fe] = -0.68 dex. Comparing it to the Galaxy, there is an overlap between the LMC and halo stars at the metal-poor end ([Fe/H] < -1.3 dex); for the higher metallicity range, the distributions diverge, with LMC stars showing a clear underabundance with respect to the Galactic disk. JIS06 also found an offset in their [Cu/Fe], compatible to our abundance ratios.

Although originally associated with the iron-peak elements, the origin of copper is still much-debated (e.g. Bisterzo et al. 2004; Cunha et al. 2004; Mishenina et al. 2002). Sometimes its main source is attributed to SNe Ia (Matteucci et al. 1993; Cunha et al. 2002; Mishenina et al. 2002) and sometimes to SNe II, particularly to a metallicity-dependent mechanism (Bisterzo et al. 2004; McWilliam & Smecker-Hane 2005). If the elemental behavior of the present sample, with low [$\alpha$/Fe], low [iron-peak/Fe] ratios, is due to a higher contribution from SNe Ia, the overall low [Cu/Fe] pattern indicates that thermonuclear supernovae cannot be the main source of Cu production.

5.5 s-process elements

We find interesting elemental distributions for the s-process elements for our sample stars (Fig. 12). While the light s-process elements (hereafter ls: elements made by the s-process with atomic number lower than $\sim$45) Zr and Y, show subsolar ratios with mean abundances the heavy s-process elements (hereafter hs: elements made by the s-process with atomic number higher than $\sim$50), La and Ba show supersolar values with enhanced pattern compared to those of the Galaxy. The underabundance of ls elements is quite strong, [Y/Fe] = -0.33 dex and [Zr/Fe] = -0.48 dex, and Zr shows a hint of decreasing with increasing metallicities. Of the hs elements, Ba has a peculiar behavior with a high value for one metal-poor star ([Fe/H] < -1.4 dex), mild enhancements until [Fe/H$] \sim -$1.15 dex, increasing again towards higher metallicities. Lanthanium shows no trend with metallicity, with mild enhancements everywhere. One star, RGB_1118, has particularly high La and Ba abundances ([Ba/Fe] and [La/Fe$]\geq+1$.0 dex) and could be a star enriched in s-process elements (via mass-transfer from a former AGB companion), although it is not possible from our present data to distinguish between enhancements of s-process or r-process elements. The s-process elements in JIS06 sample are different when compared to our results. While they find no offset for the lselements compared to the galactic distribution, therefore showing a higher abundance compared to our stars, their hs elements (Ba and La) are less enhanced than ours. Comparing NS97 low-$\alpha$ stars with our sample, we find that these stars show abundances nearer those of normal disk stars for Ba and Y than the LMC stars.

	Figure 12:  Abundance distributions for inner disk LMC stars: [s-elements/Fe] vs. [Fe/H]. The large dots (or downtriangles for upper limits) depict our sample stars, while open symbols represent galactic samples: symbols as in Fig. 8, plus Burris et al. (2000, stars), Johnson & Bolte (2002, open trianlges), Simmerer et al. (2004, open hexagons).
Open with DEXTER

The hs/ls ratios are high, showing large scatter, with a mean value of [hs/ls] = +0.77 dex, as can be seen in Fig. 13. This is very different from what is observed for the Galactic halo and disk stars, which fall around -0.2 to +0.2 dex (e.g. Pagel & Tautvaisiene 1997; Travaglio et al. 2004). A slowly, increasing trend with metallicity is observed.

High abundances of elements heavier than Zr were also derived for LMC and SMC supergiants (Russell & Bessell 1989; Spite et al. 1993; Hill et al. 1995). Hill et al. (1995) for example, found that the light s-elements Zr and Y show solar composition in LMC supergiants, while heavier s-elements (Ba, La, Nd), as well as the r-process element Eu, are enhanced by +0.30 dex. As discussed by these authors, the overabundance of the heavier s-process and r-process elements seems to be a characteristic of the Magellanic Clouds and indicates a particular evolution of that Galactic system, although no satisfactory explanation is proposed for it.

	Figure 13: Observed abundance ratios [hs/ls] = [Ba+La/Y+Zr].
Open with DEXTER

	Figure 14: The NaNi and NaMg abundance relations. Our sample stars are depicted as dots and NS97 low-$\alpha$ stars as starred symbols.
Open with DEXTER

To evaluate the r-process and s-process contributions within our sample, we analyzed the r-process content of one of our sample stars for which we have UVES spectra that cover the Eu $\lambda$ 6645 $\rm \AA$ line. The Eu and Ba abundances were derived from these spectra in the same way as for GIRAFFE spectra. For RGB_666 we find [Ba/Fe] = +0.52, and [Eu/Fe] = +0.40 dex. The corresponding [Ba/Eu] ratio of 0.12 (to be compared with the solar r-process [Ba/Eu]=-0.55 and the solar s-process [Ba/Fe]=+1.55, following Arlandini et al. 1999) indicate that this star contains a significant r-process contribution at a value close to the solar s/r mix at intermediate metallicities (RGB_666: [Fe/H]=-1.10).

A high content of r-process elements seems to contradict the observed low [$\alpha$/Fe] ratios (both produced in massive stars). More data on Eu abundances are needed to confirm this high content of r-process elements and, in particular, the trend of the s/r fraction (traced by [Ba/Eu]) as a function of metallicity will help to constrain the source of the high content of heavy s-process elements in the LMC disk. We intend to tackle this issue in the two other fields (bar and outer disk) of our LMC program, since one of the MEDUSA wavelength ranges covers the Eu line for these fields.

5.5.1 The NaMg, NaNi relations

In the paper by NS97, the authors find a correlation between Na and Ni for their halo stars (both ``normal'' and ``low-$\alpha$'' stars). Such a correlation has been confirmed for a group of stars in the dwarf spheroidal galaxies (Shetrone et al. 2003, SH03; Tolstoy et al. 2003, TO03; Venn et al. 2004). To evaluate this trend, we plot the [Ni/Fe] vs. [Na/Fe] relation for our sample stars (dots) in Fig. 14, together with NS97 low-$\alpha$ stars. We see that the LMC stars also show a correlation between Na and Ni, although with a flatter pattern than the increasing trend observed for the NS97 sample. According to Tsujimoto et al. (1995), Ni can be produced in SNe Ia without Na production; therefore, a higher contribution from SNe Ia would flatten the NaNi relation (Venn et al. 2004) and could explain the behavior of the LMC stars. In Fig. 14 we also analyze the correlation between Na and Mg and find decreasing [Na/Mg] ratios for increasing [Mg/H] ratios. The NS97 low-$\alpha$ stars seem a continuation of the observed trend.

6 Comparison to the dwarf spheroidal galaxies

In Figs. 15 and 16 we compare the chemical distributions of our LMC sample to those of the dSph galaxies of Shetrone et al. (2003) and Tolstoy et al. (2003), and the Sagittarius dwarf galaxy (Sgr) of Bonifacio et al. (2004) and Sbordone et al. (2007). The elemental distributions of most dSph galaxies are more concentrated in the metallicity range for which we have the lowest number of stars, [Fe/H] < -1.2, so the present analysis is not ideal. In Fig. 15 the distributions for the $\alpha$-elements Ca, Ti, and Si and for Cu are depicted (the description of the different symbols are given in the figure captions). As can be seen from these figures and also observed for O, Mg, Na, and Sc, there is an overlap among the LMC abundance ratios and those of the dSph galaxies. The same occurs for the iron-peak elements Cr, V, and Ni and for Cu (with the exception of Fornax, which shows higher values for Cu). In particular, the agreement among our data and those of the Sagittarius dwarf galaxy is very good, except that this galaxy shows [Ti/Fe] and [Mg/Fe] ratios to be slightly underabundant relative to our values.

Figure 15:  Comparison of the inner disk LMC stars with stars from the dwarf spheroidal galaxies and the Sgr galaxy. 1. Alpha elements. The symbols are our sample (dots), Leo I (open pentagons), Sculptor (open dots), Fornax (open triangles), Carina (open squares), and Sgr (Bonifacio et al. 2005 - stars; Sbordone et al. 2007 - asteriks). Solid lines depict mean values of the Galactic distributions for each element.

Open with DEXTER

For the s-process elements, depicted in Fig. 16, the dSph galaxies show enhanced hs and deficient ls compared to the Galactic behavior, although the general pattern is less discrepant than what is shown by the LMC inner disk stars, except for Sgr, which shows striking similar ratios when compared to our data. Fornax has a more metal-rich star (Fnx21) with high s content, which may be an s-enriched star. The [Ba/Y] ratios show a large offset relative to Gsalactic samples, of the same order magnitude we found. Venn et al. (2004) attribute such an offset to primary s-process production by low-metallicity AGB stars.

The very similar elemental distributions of the Sgr galaxy indicate that this galaxy must have been very similar to LMC, i.e., with a higher mass content, which nowadays may be hidden in streams and/or dynamically mixed to the Galaxy.

7 Discussion

It is an amazing opportunity to have so much data on the amount of various elements of stars in an external galaxy. With this unique dataset, we can now explore the SFH in more detail and understand the evolution of the LMC disk better. The overall low [X/Fe] ratios indicate that such stars have undergone a global process which is different from what is experienced by the average halo and disk stars in the Galaxy. In this section we discuss the possible explanations for this behavior.

We have found an overall low abundance pattern for the $\alpha$-elements, in agreement with many previous works on stars in this galaxy (Sect. 1). The heavy s-elements show an enhancement relative to the Galactic disk distributions, as inferred before for supergiants and red giants in the LMC. New results from the present work include low light-sabundance ratios ([Y/Fe] and [Zr/Fe]), with most of the stars showing subsolar values and an unexpected offset for the iron-peak elements Ni, Cr, and Co, and in some stars, also for V. Both Na and Sc are deficient with many subsolar ratios relative to iron, and Cu shows a very low abundance in all stars from the present sample, with mean [Cu/Fe $] \sim -0.7$ dex, and no trend with metallicity.

As seen in previous sections, low [$\alpha$/Fe] ratios have already been observed in other stellar systems such as the chemically peculiar halo stars (NS97, NS00), the dSph galaxies of the Local Group (Shetrone 2003; Tolstoy 2003), the Sagittarius galaxy (Smecker-Hane & McWilliam 2002; Bonifacio et al. 2004; Monaco et al. 2005; Sobordone et al. 2007), as well as in samples in the LMC (e.g. Hill et al. 2000, 2003; SM02; Garnett 2000; Korn et al. 2002). It is interesting to notice that the s-process trends in the dSph galaxies (enhanced hs and deficient ls ratios) are the same as for our stars. Correlations between abundances of iron-peak elements and $\alpha$-elements were also observed in other stellar systems. A pattern of slightly deficient Ni and Cr has been observed for the low-$\alpha$ stars of NS97. Bensby et al. (2003) find a correlation among the [iron-peak/Fe] and [Na/Fe] vs. [$\alpha$-elements/Fe] abundance ratios, i.e., sligtly higher [Cr/Fe], [Ni/Fe] and [Na/Fe] ratios in thick disk stars with enhanced [$\alpha$-element/Fe] ratios (see their Fig. 13). Sbodorne et al. (2007) found subsolar ratios for Na, Sc, Co, Ni, and V in their analysis of the Sagittarius dwarf galaxy stars, which also has low [$\alpha$/Fe] ratios. Such behavior may tell us interesting details about the formation of these elements and give clues to low-mass galaxy formation.

Many interpretations have been given for the low [$\alpha$/Fe] ratios observed. One hypothesis is that the SF developed slowly, in short bursts, followed by long quiescent periods without SF, during which the SNe Ia contaminated the ISM and increased the Fe content (e.g. Gilmore & Wyse 1991). Lower SNe II/SNe Ia ratios, therefore a higher frequency of SNe Ia relative to SNe II, have also been invoked, within a bursty or continuous regime and with or without galactic winds (e.g. Pagel & Tautvaisiene 1997; Smith et al. 2002). A steepened IMF relative to that of the solar neighborhood has been proposed by Tsujimoto et al. (1995) and de Freitas Pacheco (1998), whereas alpha-enriched galactic winds, which would lower the [alpha/Fe] content, have been suggested by Pilyugin (1996). Finally, a small (low-mass) SF event that would effectively truncate the IMF, yieding fewer high-mass SNe II than produced by normal SF events has been suggested (Tolstoy et al. 2003). To find explanations for the behavior of the iron-peak elements is more puzzling, since they are predicted to be produced basically in SNe Ia (e.g. Travaglio et al. 2005). A possible explanation is that the yields of the SNe Ia are dependent on metallicity (Timmes et al. 2003).

The abundance distributions observed for the hs and the ls elements, with hs/ls=[Ba+La/Y+Zr], agree with the hypothesis that the s-process in AGB stars is metallicity-dependent (Busso et al. 1999, and references therein; Busso et al. 2001; Abia et al. 2003; Travaglio et al. 2004). It has been noticed that, due to details of the nucleosynthesis of the s-process, hs-elements (e.g. Ba, La and Nd) tend to be produced by metal-poor AGB stars compared to ls elements (e.g. Y, Zr, and Sr), which are most efficiently produced at [Fe/H $] \approx -0.1$ (e.g. Fig. 1 of Travaglio et al. 2004). If the SF is slow, low-metallicity AGB stars have enough time to contaminate the ISM, leaving noticeable chemical signatures for the next generations.

Nevertheless, Venn et al. (2004) discuss the possibility that the abundances of these elements (including Y) in dSph cannot be accounted for solely by the s-process, requiring a strong contribution from the r-process. Also, according to Richtler et al. (1989) and Russell & Dopita (1992), the most probable explanation for the high Ba and La abundances observed in the Magellanic Clouds is an additional r-process component. This would mean that hs and ls elements are produced at different rates by the r-process nucleosynthesis, probably in different sites. Therefore, the analysis of the behavior of the s-elements in the given metallicity range is complex and must take both the r and the scontributions into account.

	Figure 16:  Comparison of the Inner Disk LMC stars with stars from the dwarf spheroidal galaxies. 2. s-process elements (symbols are the same as in Fig. 15).
Open with DEXTER

7.1 Galaxy formation and evolution

One of the most debated themes about galaxy formation in the Universe under a $\Lambda$CDM hierarchical scenario concerns the problem of overprediction of galaxy counts at low-z and underprediction at high-z (Cimatti et al. 2002). One of the consequences for the Local Group is a larger number of small galaxies than is actually observed, although the number of dwarf galaxies observed around the Milky Way and M 31 has lately grown significantly (e.g. Belokurov et al. 2007). According to these models, numerous merging and accretion events play an important role in the formation process of massive galaxies (e.g. Moore et al. 1999), although not all dark matter clumps are predicted to host SF and thereby become visible galaxies (e.g. Bullock & Johnston 2005). The quest for signatures of possible accreted stars from nearby galaxies in the Galactic halo and disk have been carried out, without definite conclusions (NS97; NS00; Ivans et al. 2003; Venn et al. 2004). A careful inspection of the elemental distributions of the different Galactic components reveals a low dispersion in the abundance ratios at each metallicity bin and smooth transitions between them (see e.g. plots from Venn et al. 2004). This seems to indicate a different process: the Galaxy, including the halo, has grown in a holistic way, rather than by many independent accreting events, even for the Galactic halo (see Gilmore & Wyse 2004). Another possibility is that the merging events occurred very early in the building process of our Galaxy, involving mostly dark matter and primordial gas. Such observational features hint at a common history within the same environment rather than a mix of SFHs. The results from the present work strongly support this idea, showing that an LMC-like SFH results in quite a distinctive elemental pattern not seen in any galactic stellar population.

We have found that the elemental compositions of the LMC inner disk stars show a different pattern when compared to their galactic counterparts (if we exclude the low-alpha stars of NS97). This indicates that possible accreting events of LMC and LMC-like fragments (Bekki & Chiba 2005; Robertson et al. 2005), from which our Galactic halo could have been built, are unlikely, but strong conclusions are still not possible because more representative samples are needed from both halo and LMC stars. However, we stress here that the stellar populations probed in the LMC mostly have an intermediate age and would not have been merged into a Milky Way halo or disk, if the accretion of an LMC-like galaxy occurred early on (z>1). Strong conclusions concerning the possible early accretion of LMC-type systems therefore still await detailed analysis of the elemental abundances of representative samples of the oldest populations in the LMC. The elemental distributions of the LMC inner disk also hint at a different process of galaxy formation, showing that the galactic local environment is fundamental for the the amount of various elements of its components.

8 Summary

In the present paper we report abundance ratios for a series of elements, including $\alpha$, s-, and iron-peak elements, Na, Sc, and Cu for a sample of 59 RGB stars of the inner disk of the LMC. We found a very different behavior for most of the elements relative to stars from the Galaxy with similar metallicity, hinting at a very different evolutionary history. On the other hand, there is good overall agreement between the the elemental distributions of our sample stars and previous results of the LMC GC and field stars of Hill et al. (2000, 2003), Smith et al. (2002), and Johnson et al. (2006). The main results are summarized as follows:

• [$\alpha$/Fe] ratios show an overall deficient pattern relative to Galactic distributions, in agreement with a slower SFH in the LMC, leading to a stronger type Ia supernovae influence. However, all $\alpha$-elements do not show the same degree of deficiency: while O/Fe and Mg/Fe are hardly different in the LMC and Milky-Way disks, Si, Ca, and Ti are strongly underabundant. This illustrates that all $\alpha$-elements are not alike from the nucleosynthesis point of view.

• Cu is strongly depleted with respect to iron, [Cu/Fe $] \simeq -0.70$ dex, with no apparent trend with metallicity. This also hints at a strong contribution of type Ia supernovae to the creation of copper.

• The [X/Fe] deficiency of the $\alpha$-elements is also displayed by Na, Sc, and, in an unexpected behavior, by the iron-peak elements Ni, Cr, and Co. The iron peak elements underabundances are not expected in any standard chemical evolution model (i.e. currently not predicted by SNe yields).

• We have found relationships between Na-Ni and Na-Mg, in agreement with those derived by Nissen & Schuster (1997) for a sample of low-$\alpha$halo stars. As Na is predicted to be mainly produced by SNe II, together with O and Mg, a relationship Na-Mg is expected, althoug Na production is also controlled by the neutron excess during carbon burning in massive stars (Umeda et al. 2000). The Na-Ni relationship is also expected if Ni is also produced in SN II, with yields dependent on the neutron excess (Thielemann et al. 1990).

• Heavy neutron capture elements fall into two well-defined groups: while high-mass s-process elements (Ba and La) present an enhanced pattern, low-mass s-process elements (Y and Zr) are deficient compared to the galactic samples. Such behavior has been observed before in LMC and SMC F supergiants and in dSph galaxy RGB stars. It could reflect a strong contribution by metal-poor AGB stars to the metal-enrichment of these systems, as low-metallicity AGB stars tend to produce the heavier s-process elements over the lighter ones (see Travaglio et al. 2004, for the theoretical side; and de Laverny et al. 2006, for the observation of low metallicity AGBs).

• We have derived Eu abundances for one of our intermediate-metallicity stars (RGB_666: [Fe/H]=-1.10), and combined with the measured Ba abundance for this star, this enabled us to disantangle the respective r- and s-process contributions to heavy neutron-capture elements. This star contains a solar mix of r- and s-process elements. Although a single measurement is obviously not enough to lead to a conclusion, we thereby confirm that the high abundances of ls elements observed at intermediate metallicity should be attributed to the s-process.

  For the next two fields of our program (see Introduction), the wavelength range of the spectra covers a Eu line and a better evaluation of such contributions will be possible.

• Compared to the dSph galaxies, similar abundance ratios for almost all the elements have been derived, with slight enhancements of La, Ba, Na, and Y, although the match in metallicity among our sample and the dSph samples is not ideal. The LMC inner disk abundances of Ca, Si, Ti, and Cu are also similar to those of the Sagittarius dwarf galaxy. The commonalities between the LMC inner disk population and the samples in dSph galaxies indicate that all these galaxies may have undergone similar SFH.

The overall pattern of the elemental distributions for the LMC inner disk population can be explained by a higher contribution of type Ia SNe, indicating that the build up of this population has been slower than that of the solar neighborhood stars. A bigger contribution from metal-poor ABG stars is also proposed. The present results support the hypothesis that the elemental distributions of the stars are directly related to the galaxy they belong to.

Acknowledgements

L.P. acknowledges CAPES and FAPESP fellowships #0606-03-0 and #01/14594-2. We thank Peter Stetson for the availability of the DAOSPEC program.

Appendix A: Comparison between equivalent width from DAOSPEC and from Splot-Iraf

In order to evaluate the quality of the DAOSPEC estimates, we have derived by eye inspection the EW of six stars, using the Splot Iraf task. We have chosen stars in a range of S/N ratio typical of our total sample in order to better evaluate the errors: S/N = 54 for RGB_522, 59 for RGB_546, 47 for RGB_664, 42 for RGB_666, 26 for RGB_720, and 47 for RGB_1055. The detailed values are given in Table A.1. We have found two problems relative to the DAOSPEC results from GIRAFFE spectra: a) not all blends have been identified, nevertheless all lines with weak blends (which are most of the lines) have been correctly analyzed by the iterative process of the program; b) cosmic rays also have not been identified, and those lines too near CR features must be discarded. Therefore the use of DAOSPEC requires spectra and line lists as clear as possible from blends and spectra as clean from cosmic hits as possible. As can be seen in Table A.1, we have found a very good agreement between the program results and those from the Splot manual measurements. The average difference EW(Dao) - EW(Splot) is 0.46 mÅ for the six stars, with no strong systematic trend in one direction or the other (the mean difference for each star ranges from -3.7 to +5 mÅ). However, the dispersion of the measurements around the mean are higher, between 9.4 mÅ for RGB_1055 and 22.2 mÅ for RGB_666. These dispersion seem to anticorrelate with S/N as expected (the two stars with the highest dispersion are the two lowest quality spectra), and may also correlate with temperature, although our sample of 6 stars is not quite high enough to investigate these dependancies any further.

	Figure A.1: Difference between the EW derived using the DAOSPEC program and the Iraf task Splot: trends with respect to the EW(Dao) and to the wavelenght of the lines for RGB_522 and RGB_546 ( upper plot) and for RGB_720 and RGB_664 ( lower plot).
Open with DEXTER

Table A.2: Absolute differences Ab(Dao) - Ab(Splot) and the average value (see text).

We have checked for systematic trends on the EW with wavelength and with the EW strength by plotting the differences EW(Dao) - EW(Splot) vs. wavelength and EW(Dao) - EW(Splot) vs. EW(Dao) for four stars, as shown in Fig. A.1. No trends have been found and confort us in the validity of using DAOSPEC EW mesurement for atmospheric parameter determinations (effective temperature and microturbulence velocities).

In Table A.2 we give the differences for the abundances derived from DAOSPEC and Splot, Ab(Dao) - Ab(Splot). As can be seen in this table, the agreement between the two methods is good for most of the elements, always better than the typical errorbars given for our measurement of the corresponding elements, and usually below 0.10 dex. The differences are higher for those elements with fewer lines (e.g. Na I and Ti I), which increases the weight of the scatter among lines. Let us note in particular that elements such as Ca I or Ti I do not seem to be affected by the method used for EW measurement in a systematic direction, so that the strong underabundances found for these elements (with respect to the galactic trends) are robust against EW systematics.

References

• Abia, C., Domínguez, I., Gallino, R., et al. 2002, ApJ 579, 817
• Allende Prieto, C., Barklem, P. S., Lambert, D. L., & Cunha, K. 2004, A&A, 420, 183 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Allende Prieto, C., Lambert, D. L., & Asplund, M. 2001, ApJ, 566, L63 [NASA ADS] [CrossRef]
• Barbuy, B., Zoccali, M., Ortolani, S., et al. 2006, A&A, 449, 349 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Belokurov, V., Zucker, D. B., Evans, N. W., et al. 2007, ApJ, 654, 897 [NASA ADS] [CrossRef] (In the text)
• Bensby, T., Feltzing, S., & Lundstrom, I. 2003, A&A, 410, 527 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Bensby, T., Feltzing, S., & Lundstrom, I. 2004, A&A, 415, 155 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Bessell, M. S. 1991, A&A, 242L, 17 [NASA ADS] (In the text)
• Bessell, M. S., Castelli, F., & Plez, B. 1998, A&A, 333, 231 [NASA ADS]
• Biehl, D. 1976, Ph.D. Thesis, Kiel (In the text)
• Biemont, E., Baudoux, M., Kurucz, R. L., Ansbacher, W., & Pinnington, E. H. 1991, A&A, 249, 539 [NASA ADS]
• Bisterzo, S., Gallino, R., Pignatari, M., et al. 2004, MmSAI, 75, 741 [NASA ADS] (In the text)
• Bonifacio, P., Sbordone, L., Marconi, L., Pasquini, L., & Hill, V. 2004, A&A, 414, 503 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Bord, D. J., barisciano, L. P., & Cowley, C. R. 1996, MNRAS, 278, 997 [NASA ADS] (In the text)
• Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931 [NASA ADS] [CrossRef] (In the text)
• Burris, D. L., Pilachowski, C. A., Armandroff, T. E., et al. 2000, ApJ, 544, 302 [NASA ADS] [CrossRef] (In the text)
• Busso, M., Gallino, R., & Wasserburg, G. J. 1999, ARA&A, 37, 239 [NASA ADS] [CrossRef] (In the text)
• Busso, M., Gallino, R., Lambert, D. L., Travaglio, C., & Smith, V. V. 2001, ApJ, 557, 802 [NASA ADS] [CrossRef] (In the text)
• Butcher, H. 1977, ApJ, 216, 372 [NASA ADS] [CrossRef] (In the text)
• Carpenter, J. M. 2001, AJ, 121, 2851 [NASA ADS] [CrossRef] (In the text)
• Cayrel, R. 1988, in The Impact of Very High S/N Spectroscopy on Stellar Physics, ed. G. Cayrel de Strobel, & M. Spite (Dordrecht: Kluwer), IAU Symp., 132, 345
• Cayrel, R., Depagne, E., Spite, M., et al. 2004, A&A, 416, 1117 [NASA ADS] [CrossRef] [EDP Sciences]
• Cimatti, A., Pozzetti, L., Mignoli, M., et al. 2002, A&A, 391, L1 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Cioni, M.-R., Girardi, L., Marigo, P., & Habing, H. J. 2006, A&A, 448, 77 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Cunha, K., Smith, V. V., Suntzeff, N. B., et al. 2004, AJ, 124, 379 [NASA ADS] [CrossRef] (In the text)
• Da Costa, G. S. 1991, in The Magellanic Clouds, ed. R. Haynes, & D. Milne (Dordrecht: Kluwer), IAU Symp., 148, 183 (In the text)
• Da Costa, G. S. 1999, in New Views on the Magellanic Clouds, ed. Y.-H. Chu, N. Suntzeff, J. Hesser, & D. Bohlender (San Francisco: ASP), IAU Symp., 190, 397 (In the text)
• de Laverny, P., Abia, C., Dominguez, I., et al. 2006, A&A, 446, 1107 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• D'Onghia, E., & Lake, G. 2004, ApJ, 612, 628 [NASA ADS] [CrossRef] (In the text)
• Dufton, P. L., Smartt, S. J., Lee, J. K., et al. 2006, A&A, 457, 265 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• de Freitas Pacheco, J. A. 1998, AJ, 116, 1701 [NASA ADS] [CrossRef] (In the text)
• de Vaucouleurs, G. 1980, PASP, 92, 576 [NASA ADS] [CrossRef]
• Edvardsson, B., Andersen, J., Gustafsson, B., et al. 1993, A&A, 275, 101 [NASA ADS]
• Evans, C. J., Smartt, S. J., Lee, J.-K., et al. 2005, A&A, 437, 467 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Fernie, J. D. 1983, PASP, 95, 782 [NASA ADS] [CrossRef] (In the text)
• Fulbright, J. P. 2000, AJ, 120, 1841 [NASA ADS] [CrossRef] (In the text)
• Garnett, D. R. 1999, in New Views of the Magellanic Clouds, ed. Y.-H. Chu, N. Suntzeff, J. Hesser, & D. Bohlender, IAU Symp., 190, 266 (In the text)
• Geha, M. C., Holtzman, J. A., Mould, J. R., et al. 1998, AJ, 115, 1045 [NASA ADS] [CrossRef] (In the text)
• Gilmore, G., & Wyse, R. F. G. 1991, ApJ, 367, L55 [NASA ADS] [CrossRef] (In the text)
• Gilmore, G., & Wyse, R. F. G. 2004 [arXiv:astro-ph/0411714] (In the text)
• Grevesse, N., & Sauval, A. J. 1998, Space Sci. Rev., 85, 161 [NASA ADS] [CrossRef]
• Grevesse, N., & Sauval, A. J. 2000, in Origin of Elements in the Solar System, Implications of Post-1957 Observations, Proc. of the International Symposium, ed. O. Manuel (Boston/Dordrecht: Kluwer Academic/Plenum Publishers), 261 (In the text)
• Gustafsson, B., Bell, R. A., Eriksson, K., & Nordlund, A. 1975, A&A, 42, 407 [NASA ADS] (In the text)
• Gustafsson, B., Edvardsson, B., Eriksson, K., et al. 2003, in Stellar Atmosphere Modeling, ed. I. Hubeny, D. Mihalas, & K. Werner, ASP Conf. Proc., 288 (ASP: San Francisco), 331 (In the text)
• Hill, V. 2004, in Carnegie Observatories Astrophysics Series, Origin and Evolution of the Elements, ed. A. McWilliam, & M. Rauch (Cambridge: Cambridge Univ. Press), 4, 205 (In the text)
• Hill, V., & Spite, M. 1999, in Galaxy Evolution: Connecting the Distant Universe with the Local Fossil Record, Proc. Coll. Obs. Paris-Meudon, ed. M. Spite (Dordrecht: Kluwer Academic Publishers), 469 (In the text)
• Hill, V., Andrievsky, S., & Spite, M. 1995, A&A, 293, 347 [NASA ADS] (In the text)
• Hill, V., Francois, P., Spite, M., Primas, F., & Spite, F. 2000, A&A, 364, L19 [NASA ADS] (In the text)
• Holtzman, J. A., Gallagher, J. S., III, Cole, A. A., et al. 1999, AJ, 118, 2262 [NASA ADS] [CrossRef] (In the text)
• Hunter, I., Dufton, P. L., Smartt, S. J., et al. 2007, A&A, 466, 277 [NASA ADS] [CrossRef] [EDP Sciences]
• Ivans, I. I., Sneden, C., James, C. R., et al. 2003, ApJ, 592, 906 [NASA ADS] [CrossRef] (In the text)
• Iwamoto, K., Brachwitz, F., Nomoto, K., et al. 1999, ApJS, 125, 439 [NASA ADS] [CrossRef] (In the text)
• Javiel, S. C., Santiago, B. X., & Kerber, L. O. 2005, A&A, 431, 73 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Johnson, J., & Bolte, M. 2002, ApJ, 579, 616 [NASA ADS] [CrossRef] (In the text)
• Johnson, J., Bolte, M., Hesser, J. E., & Ivans, I. I. 2004, in Carnegie Observatories Astrophysics Series, Origin and Evolution of the Elements, ed. A. McWilliam, & M. Rauch, 4, 29
• Korn, A. J., Keller, S. C., Kaufer, A., et al. 2002, A&A, 385, 143 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Korn, A. J., Shi J., & Gehren, T. 2003, A&A, 407, 691 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Kraft, R. P. Sneden, C., Langer, G. E., & Prosser, C. F. 1992, AJ, 104, 645 [NASA ADS] [CrossRef]
• Kurucz, R. L., Furenlid, I., & Brault, J. 1984, National Solar Observatory Atlas, Sunspot, New Mexico, National Solar Observatory, 1, 984 (In the text)
• Lawler, J. E., Bonvallet, G., & Sneden, C. 2001, ApJ, 556, 452 [NASA ADS] [CrossRef] (In the text)
• Luck, R. E., & Lambert, D. L. 1992, ApJS, 79, 303 [NASA ADS] [CrossRef]
• Madore, B. F., & Freedman, W. L. 1998, ApJ, 492, 110 [NASA ADS] [CrossRef] (In the text)
• Matteucci, F., Raiteri, C. M., Busson, M., Gallino, R., & Gratton, R. 1993, A&A, 272, 421 [NASA ADS] (In the text)
• McWilliam, A. 1997, ARA&A, 35, 503 [NASA ADS] [CrossRef] (In the text)
• McWilliam, A. 1998, AJ, 115, 1640 [NASA ADS] [CrossRef] (In the text)
• McWilliam, A., & Rich, M. 1994, ApJS, 91, 949 [NASA ADS] [CrossRef]
• McWilliam, A., & Smecker-Hane, T. 2005, ApJ, 622, 29 [NASA ADS] [CrossRef] (In the text)
• Mishenina, T. V., Kovtyukh, V. V., Soubiran, C., Travaglio, C., & Busso, M. 2002, A&A, 396, 189 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Moore, B., Ghigna, S., Lake, G., et al. 1999, ApJ, 524, L19 [NASA ADS] [CrossRef] (In the text)
• Nave, G., Johansson, S., Learner, R. C. M., Thorne, A. P., & Brault, J. W. 1994, ApJS, 94, 221 [NASA ADS] [CrossRef]
• Nissen, P. E., & Schuster, W. J. 1997, A&A, 326, 752 [NASA ADS] (In the text)
• Nissen, P. E., Chen, Y. Q., Schuster, W. J., & Zhao, G. 2000, A&A, 353, 722 [NASA ADS] (In the text)
• Olszewski, E. W., Suntzeff, N. B., & Mateo, M. 1996, ARA&A, 34, 511 [NASA ADS] [CrossRef] (In the text)
• Pagel, B. E. J., & Tautvaisiene, G. 1997, MNRAS, 288, 108 [NASA ADS] (In the text)
• Pasquini, L., Avila, G., Blecha, A., et al. 2002, The Messenger, 110, 1 [NASA ADS]
• Pasquini, L., Randich, S., Zoccali, M., et al. 2004, A&A, 424, 951 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Pilyugin, L. S. 1996, A&A, 310, 751 [NASA ADS] (In the text)
• Plez, B. 2000, The Carbon Star Phenomenon, ed. R. F. Wing (Dordrecht: Kluwer Academic Publishers), Proc. IAU Symp., 177, 71
• Plez, B., Brett, J. M., & Nordlund, A. 1992, A&A, 256, 551 [NASA ADS] (In the text)
• Prochaska, J. X., Naumov, S. O., Carney, B. W., McWilliam, A., & Wolfe, A. M. 2000, AJ, 120, 2513 [NASA ADS] [CrossRef] (In the text)
• Randich, S., Sestito, P., Primas, F., Pallavicini, R., & Pasquini, L. 2006, A&A, 450, 557 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Reddy, B. E., Tomkin, J., Lambert, D. L., & Allende Prieto, C. 2003, MNRAS, 340, 304 [NASA ADS] [CrossRef] (In the text)
• Richtler, T., Spite, M., & Spite, F. 1989, A&A, 225, 351 [NASA ADS] (In the text)
• Rolleston, W. R. J., Trundle, C., & Dufton, P. L. 2002, A&A, 396, 53 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Russell, S. C., & Bessell, M. S. 1989, ApJS, 70, 865 [NASA ADS] [CrossRef] (In the text)
• Russell, S. C., & Dopita, M. A. 1992, ApJ, 384, 508 [NASA ADS] [CrossRef] (In the text)
• Rutten, R. J. 1978, Sol. Phys., 56, 237 [NASA ADS] [CrossRef] (In the text)
• Sarajedini, A. 1998, AJ, 116, 38 [NASA ADS] [CrossRef]
• Sbordone, L., Bonifacio1, P., Buonanno, R., et al. 2007, 465, 815 (In the text)
• Shetrone, M. D., Venn, K. A., Tolstoy, E., et al. 2003, AJ, 125, 684 [NASA ADS] [CrossRef] (In the text)
• Smecker-Hane, T., & McWilliam, A. 2002, Proc. Symp. Cosmic Abundances as Records of Stellar Evolution and Nucleosynthesis in honor of David L. Lambert, ASP Conf. Ser., ed. T. G. barnes III, & F. N. Bash, San Francisco, 336, 221 (In the text)
• Smecker-Hane, T. A., Cole, A. A., Gallagher, J. S. III, & Stetson, P. B. 2002, SMH02, ApJ, 566, 239 (In the text)
• Smecker-Hane, T., Cole, A., & Mandushev, G. I. 2007, in preparation
• Smith, V. V., Suntzeff, N. B., Cunha, K., et al. 2000, AJ, 119, 1239 [NASA ADS] [CrossRef] (In the text)
• Smith, V. V., Hinkle, K. H., Cunha, K., et al. 2002, AJ, 124, 3241 [NASA ADS] [CrossRef] (In the text)
• Sneden, C., Gratton, R. G., & Crocker, D. A. 1991, A&A, 246, 354 [NASA ADS]
• Sneden, C., McWilliam, A., Preston, G. W., et al. 1996, ApJ, 467, 819 [NASA ADS] [CrossRef]
• Sousa, S. G., Santos, N. C., Israelian, G., Mayor, M., & Monteiro, M. J. P. F. G. 2006, A&A, 458, 873 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Spite, M. 1967, Ann. Astrophys., 30, 21
• Spite, F., Barbuy, B., & Spite, M. 1993, A&A, 272, 116 [NASA ADS] (In the text)
• Subramaniam, A. 2004, A&A, 425, 837 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
• Tautvaisiene, G., Edvardsson, B., Tuominen, I., & Ilyin, I. 2000, A&A, 360, 499 [NASA ADS] (In the text)
• Thielemann, F.-K., Hashimoto, M., & Nomoto, K. 1990, ApJ, 349, 222 [NASA ADS] [CrossRef] (In the text)
• Thielemann, F.-K., Brachwitz, F., Freiburghaus, C., et al. 2001, PPNP, 46, 5 [NASA ADS] (In the text)
• Thielemann, F.-K., Argast, D., Brachwitz, F., et al. 2002, Ap&SS, 281, 25 [NASA ADS] [CrossRef] (In the text)
• Timmes, F. X., Brown, E. F., & Truran, J. W. 2003, ApJ, 590, L83 [NASA ADS] [CrossRef] (In the text)
• Tolstoy, E., Venn, K. A., Shetrone, M. D., et al. 2003, AJ, 125, 707 [NASA ADS] [CrossRef] (In the text)
• Travaglio, C., Gallino, R., Arnone, E., et al. 2004, ApJ, 601, 864 [NASA ADS] [CrossRef] (In the text)
• Travaglio, C., Hillebrandt, W., Reinecke, M., & Thielemann, F.-K. 2005, ApJ, 425, 1029 (In the text)
• Tsujimoto, T., Nomoto, K., Yoshii, Y., et al. 1995, MNRAS, 277, 945 [NASA ADS] (In the text)
• Umeda, H., Nomoto, K., & Nakamura, T. 2000, in The First Stars, ed. A. Weiss, T. Abel, & V. Hill (Heidelberg: Springer), 150 (In the text)
• van den Bergh, S. 1979, ApJ, 230, 95 [NASA ADS] [CrossRef] (In the text)
• van den Bergh, S. 1998, ApJ, 507, L39 [NASA ADS] [CrossRef] (In the text)
• van den Bergh, S. 1999, New Views of the Magellanic Clouds, ed. Y.-H. Chu, N. Suntzeff, J. Hesser, & D. Bohlender, IAU Simp., 190 (In the text)
• van der Marel, R. P., & Cioni, M.-R. L. 2001, AJ, 122, 1807 [NASA ADS] [CrossRef]
• Venn, K. A., Irwin, M., Shetrone, M. D., et al. 2004, AJ, 128, 1177 [NASA ADS] [CrossRef] (In the text)
• Westerlund, B. E. 1997, The Magellanic Clouds (Cambridge: Cambridge Univ. Press) (In the text)
• Woosley, S. E., & Weaver, T. A. 1995, ApJS, 101, 181 [NASA ADS] [CrossRef] (In the text)

 

  
Online Material

Table A.1: Comparison of the equivalent widths derived from DAOSPEC and Iraf-Splot task for six of our sample stars.