Free Access
Issue
A&A
Volume 562, February 2014
Article Number A146
Number of page(s) 14
Section Galactic structure, stellar clusters and populations
DOI https://doi.org/10.1051/0004-6361/201322796
Published online 24 February 2014

© ESO, 2014

Table 1

Objects and observations.

1. Introduction

Ultra-faint dwarf spheroidal galaxies (UFDs), recently discovered in the Local Group, provide us with an excellent laboratory for studying the formation of low-mass galaxies and production of heavy elements in their earliest stages in the Universe. The faint Boötes I dwarf spheroidal galaxy (dSph) is one of the most interesting objects of its kind as a candidate for first galaxies and/or as building blocks of the Milky Way (MW). It was first discovered with the help of photometric data obtained by the Sloan Digital Sky Survey (SDSS; York et al. 2000) as one of the faintest galaxies known to date (Belokurov et al. 2006). Subsequent spectroscopy to measure line-of-sight velocities of candidate member stars suggests that Boötes I is a dark-matter-dominated system with a total mass on the order of 107  M (Muñoz et al. 2006; Martin et al. 2007). Koposov et al. (2011) has developed a reduction technique to derive more accurate radial velocities, and they suggest that the velocity dispersion of this system is represented better by two components with 2.4 km s-1 for a “cold” component and ~9 km s-1 for a “hot” one, which could be interpreted as the result of velocity anisotropy in this galaxy.

Detailed analyses of a color-magnitude diagram (CMD) for Boötes I suggests that it consists of a very old stellar population with age >10 Gyr, which is similar to the most metal-poor globular clusters in the MW (Okamoto et al. 2012). Chemical abundance analyses based on lower resolution spectroscopy report very low metallicities for Boötes I stars at [Fe/H] ~ −2.5 on average with a large spread of Δ [Fe/H] ≳ 1.0 dex (Norris et al. 2008; Lai et al. 2011).

Detailed chemical abundances for Boötes I allow us to further address whether possible progenitors of the metal-poor component of the MW are similar to Boötes I and whether this galaxy is indeed an example of the first galaxies formed in the early Universe. However, these questions have not yet been clearly answered because of the discrepancy in abundance results among previous studies. Chemical abundances based on high-resolution spectroscopy for Boötes I stars have previously been obtained by Feltzing et al. (2009, hereafter F09), Norris et al. (2010b), and Gilmore et al. (2013, hereafter G13). Feltzing et al. (2009) analyzed seven stars with −3 ≲ [Fe/H] ≲ −2 based on the high-resolution spectra using an LTE abundance analysis code that takes the sphericity of stellar atmospheres into account. They report the presence of a star with an anomalous [Mg/Ca] ratio, Boo-127, and interpret it as a signature of nucleosynthesis yields of individual supernove (SNe).

Gilmore et al. (2013) carried out the double-blind analysis by two groups of the co-authors, Norris & Yong (NY) and Geisler & Monaco (GM), for six stars in Boötes I with −3.7 < [Fe/H] < −1.9. They find that the chemical abundance pattern for the Boötes I stars are generally similar to those observed in field halo stars with small scatter. However, in contrast to what is generally known for the field halo stars with the similar metallicities, a signature of decreasing [α/Fe] ratios with increasing [Fe/H] is also suggested. They observed Boo-127 in common with F09 but did not confirm the [Mg/Ca] anomaly of this object. Although these data provide important implications for the early chemical enrichment of this galaxy, they provide the contrasting pictures, namely, the former study reports the presence of an anomalous star, which may suggest that the chemical evolution in Boötes I proceeded in an inhomogeneous manner, while the latter does not support the presence of this inhomogeneity.

To obtain a consistent picture of the early chemical evolution of Boötes I, we present an independent analysis of six Boötes I stars with −3.2 < [Fe/H] < −2.0 based on high-resolution spectroscopic data obtained with the Subaru/High Dispersion Spectrograph (HDS). Five stars and three stars in our sample have been analyzed in common with F09 and G13, respectively.

This paper is organized as follows. In Sect. 2, we describe our sample, the method of observation, and equivalent width (EW) measurements. Section 3 presents the method for the stellar parameter and abundance estimates for the sample stars. In Sect. 4 we report the results of our abundance analysis and their comparison with previous results in the literature. Section 5 first presents the abundance comparison of the present Boötes I sample with the field MW halo stars and with other dSphs. Then, possible implications of the chemical evolution of Boötes I are discussed. Finally, Sect. 6 concludes the paper.

2. Observation and measurements

2.1. Sample and observation

The sample of six bright red giant branch stars in Boötes I was selected from Norris et al. (2008) to cover a wide metallicity range (−3 ≲ [Fe/H] ≲ −2). The high-resolution spectra of six bright red giants were obtained with the Subaru Telescope High Dispersion Spectrograph (HDS; Noguchi et al. 2002) on 16–17 May 2009 and 16 May 2010. The objects are listed in Table 1 with observation details. We applied CCD on-chip binning (2 × 2 pixels), resulting in a resolving power of R = λ/δλ = 40   000. Our spectral ranges cover 4030–6800 Å. Spectra of two comparison stars in the Galactic halo (HD 216143 and HD 85773) were obtained with similar setups of the instrument in our previous run (Aoki et al. 2009).

A standard data reduction was made with IRAF echelle package. Cosmic-ray hits were removed by the method described in Aoki et al. (2005). The observations were carried out mostly during dark nights, and the sky background was not significant. One exposure of Boo-121 on 16 May 2010 was exceptionally affected by the Moon. We thus subtracted the sky background estimated from the slit image around the object. The spectra obtained from individual exposures were combined to obtain final spectra.

Radial velocities of the sample stars are measured from 2550 Fe I lines for each object. The results are given in Table 1 (vhelio). The measured values agree with those derived by G13 within 1.5 km s-1 for the three stars observed in common.

thumbnail Fig. 1

Comparison of the measured equivalent widths (EWs) in this work (TW) and in the two analyses, “NY” (closed circles) and “GM” (open circles), in Gilmore et al. (2013).

Open with DEXTER

2.2. Equivalent width measurement

The EWs of absorption lines were measured by Gaussian fitting. The atomic line list used in Ishigaki et al. (2012, 2013) was adopted. The errors in the EW measurements estimated from the formula presented in Cayrel (1988) range from ~5 to 12 m Å, depending on the signal-to-noise ratios (~10 per pixel at ~4500 Å and ~30 at ~6000 Å) of the continuum.

Although Mg is a key element in our work, the gf values could be uncertain because of the lack of laboratory measurements for most of the Mg I lines (Aldenius et al. 2007). We inspected the gf values of Mg I lines adopted in previous studies for Boötes I stars F09 and G13 and summarize them in Table 2 (the values of F09 was provided by Feltzing, private communication). The last column gives the log   gf values adopted in this work. The Mg I lines at 5172.68, 5183.60, and 5711.09 Å were not used in this work because EWs of these lines for our sample stars were too strong or too weak to be useful in the abundance analysis. As can be seen from Table 2, the log   gf values for the 5528.40 Å line, which were common to all of these studies, are different by up to ~0.3 dex.

We confirmed that the Mg abundances derived from different lines using the log   gf values adopted in this work (last column of Table 2) agree within ~0.3 dex for the comparison stars, HD 216143 and HD 85773, for each.

Table 2

The Mg line lists.

2.3. Equivalent width comparisons

Figure 1 presents the comparison between the EWs measured in this work and those in G13 for the three stars analyzed in common. For all of these stars, the EWs measured in this work are systematically lower than the values obtained by G13. The mean differences between the EW in this work and in the analysis of “NY” in G13 are 1626 mÅ, while those between in this work and in the other independent analysis, “GM”, in G13 are 1018 mÅ, which are modestly larger than the errors in the EWs measurement (~12 mÅ).

To test whether the uncertainty in the continuum placement could be the reason for the discrepancy, we examined the change in the measured EWs by adopting different continuum levels for one of the Fe I line as shown in Fig. 2. For this line, the discrepancy in the measured EWs among the three analyses is particularly large (72.0 mÅ, 96.2 mÅ and 102.5 mÅ in this work, GM, and NY, respectively). If the continuum level of the normalized spectrum is changed by ± 0.05, which is slightly more than the scatter in the continuum around this line (widths in the shaded regions), the measured EW varies in the range 52.789.2 mÅ. This result illustrates that the variation in the continuum level partly explains the EW difference among the three analyses. However, as can be seen in Fig. 2, the error in the continuum level that is as large as ± 0.05 is unlikely for this line, so it is difficult to completely illuminate the discrepancy. In this paper, we use the measured EWs without any correction. Instead, we examine the effects of this offset in derived abundances in Sect. 3.3.

thumbnail Fig. 2

EW measurements for one of the Fe I lines by adopting the different continuum levels. The dotted line shows the best-fit Gaussian, where continuum level is estimated from the shaded wavelengths region. The dashed lines show the Gaussian fit by fixing the continuum levels to be ± 0.05 of the best-fit value.

Open with DEXTER

3. Abundance analysis

3.1. Stellar atmospheric parameters

Effective temperatures (Teff) were estimated based on griz color indices using the temperature scale of Castelli1. The griz photometry was adopted from the SDSS DR9 data set. The Galactic extinction in the griz magnitudes toward individual stars was corrected based on the revised extinction measurement by Schlafly & Finkbeiner (2011) via the NASA/IPAC Extragalactic Database2. The extinction in the g-band varies from 0.052 to 0.060 among our six sample stars. These values are translated into E(B − V) in the Schlegel et al. (1998) from 0.016 to 0.018 (Schlafly & Finkbeiner 2011), which are consistent with the previously adopted value of E(B − V)= 0.02 (Norris et al. 2008).

The surface gravity (log g) for the sample stars are then estimated by comparing the adopted temperature with a 12 Gyr Yonsei-Yale (YY) isochrone (Demarque et al. 2004) by assuming that the stars lie on the red giant branch of the system. In the above procedure for obtaining Teff and log   g, the [Fe/H] values are needed as input. We adopt the initial guess of the [Fe/H] values from Norris et al. (2010a).

The microturbulent velocity (ξ) is estimated by requiring that the Fe abundances from individual Fe I lines do not depend on their measured EWs. In this step, we have used the Fe I lines having a reduced EW: log   (EW/λ) < −4.7.

After the standard 1D-LTE abundance analysis is performed, the procedures are iterated until a consistent set of parameters (Teff, log   g, ξ, and [Fe/H]) is obtained. The resulting atmospheric parameters are summarized in Table 3.

The adopted Teff and log   g in this work and in the two independent analyses (“NY” and “GM”) presented by G13 agree within 100 K and 0.2 dex, respectively, for the three stars analyzed in common. The ξ values are different between the two analyses in G13 by 0.51.3 km s-1. The relatively high ξ value for Boo-094 reported by the “NY” analysis is in excellent agreement with ours, while the ξ values for Boo-117 and Boo-127 are more consistent with those obtained by the “GM” analysis.

Table 3

Atmospheric parameters.

Table 4

Equivalent widths and abundances.

3.2. Abundances

Abundances of individual elements are determined by standard 1D-LTE analyses using the measured EWs. We restrict the analysis to the absorption lines with log   (EW/λ) < −4.7 to minimize the errors in derived abundances. For Na I lines, we relax this restriction to log   (EW/λ) ≤ −4.5, since only strong resonance lines are detected for this species. Similarly, only strong resonance Sr II lines are available for the abundance estimate of Sr. A spectral synthesis is applied to estimate Sr abundances as described in detail in Sect. 4.7. The hyperfine splitting effect is included in the analysis of Ba lines (McWilliam 1998), assuming the r-process isotope ratios. The EWs and abundances estimated from individual lines are summarized in Table 4.

Carbon abundances are determined from the comparisons of synthetic spectra to the observed spectrum of the CH molecular band at around 4323 Å. For Boo-094, Boo-121, and Boo-009, only an upper limit of carbon abundance is estimated. Because the signal-to-noise ratios are low for the wavelength range of the G-band, the continuum level is not determined well, which produces the abundance uncertainty of 0.10.3 dex.

3.3. Abundance errors

The errors due to the line-to-line scatter (σline), which are calculated as the standard deviation of the abundances from individual lines divided by the square root of the number of lines, are summarized in Table 5. If only one line is available for the abundance estimate, σline is assumed to be equal to the standard deviation for the Fe I lines, which are typically ~0.2 dex in the present work.

The abundance errors due to uncertainties in Teff, log   g, and ξ are examined by calculating the abundance differences when changing these parameters by ±100 K, ±0.3 dex, and ±0.3 km s-1, respectively. Results of this exercise are summarized in Table 5. The error bars in Figs. 5 and 6, and σtot in Table 5 correspond to a quadratic sum of the contribution from the uncertainty in the three atmospheric parameters and line-to-line scatter in the abundances.

The systematic offset in the EWs from those in Gilmore et al. (2013) mentioned in Sect. 2.3 are an additional source of uncertainty in measuring absolute abundances in this work. Figure 3 shows changes in log ϵ(X) abundances when the EWs are increased by 15 mÅ to reduce the systematic offset from the Gilmore et al. (2013) analyses for the most metal-poor (Boo094) and metal-rich (Boo127) stars in our sample. As can be seen, after the EW offset is applied, the abundances are systematically increased by ~0.150.30 dex. On the other hand, the amount of correction is mostly similar for all elements, and thus difference in the abundance ratios relative to iron ([X/Fe]) partly cancel out. Although the absolute abundances could be affected by the systematic errors in EW measurements, the abundance ratios ([X/Fe]) are consistent with those in Gilmore et al. (2013).

For the Sr abundances estimated by the spectral syntheses, uncertainty in the continuum level (1σ ~ 0.1) causes an error as large as ±1.0 dex. as detailed in Sect. 4.7.

thumbnail Fig. 3

Changes in abundances when all of the EWs in this work are increased by 15 mÅ to partly correct for the EWs offset from Gilmore et al. (2013).

Open with DEXTER

4. Abundance results

Our abundance results are summarized in Table 5. Relative abundances ([X/Fe]) were determined adopting the solar abundances from Asplund et al. (2009). Figures 5 and 6 plot the relative abundances of C, Na, Mg, Ca, Sc, Ti I, Ti II, Cr, Ni, Zn, Sr, and Ba as a function of [Fe/H] for the sample of six Boötes I stars (filled circles) and the two comparison field halo stars (open circles). Triangles represent the abundance results from the “GM” analysis in G13, whose systematic offset in the EWs from those in this work is smaller than that in the “NY” analysis.

4.1. Comparison with Feltzing et al. (2009)

Five objects in the present sample have also been studied by F09, who determined Fe, Mg, Ca and Ba abundances for seven objects in Boötes I. The [Fe/H] of the sample stars determined by the present work agree well with the results of F09 within ~0.1 dex, except for the lowest metallicity star Boo-094 in our sample, for which the estimated [Fe/H] is −3.4 dex in our study but −2.9 in F09. This difference can be partly attributed to the higher log   g and larger ξ adopted in our analysis.

The abundances of Mg and Ca also marginally agree with those estimated by F09 within 0.30 dex, except for Boo-127. For this star, F09 report an exceptionally high [Mg/Ca] ratio ([Mg/Ca] > 0.6), while this work has obtained [Mg/Ca] = 0.08. The estimated log   ϵ(Mg) in this work is ~0.5 dex lower than that reported by F09. Several factors seem to be responsible for the abundance difference, including (1) a higher log   gf value for Mg I 5528 Å (see Table 2); (2) higher log   g; and (3) higher ξ in this work than those adopted in F09. While the error in the EWs by up to ~1015 mÅ may also contribute to the discrepancy, it does not totally explain the large difference.

To illuminate the uncertainty in the log   gf values, we also performed a simple differential analysis to obtain [Mg/Ca] for Boo-127 with respect to one of the comparison stars, HD 217143. Since the Teff, log   g, and [Fe/H] values are similar between the two stars, a direct comparison of EWs for the Mg I and Ca I lines provides the [Mg/Ca] estimate. Figure 4 shows the wavelength regions that contain the Mg I and Ca I lines for Boo-127 and HD 216143. While both stars show Ca I line with similar strengths, the Boo-127 shows slightly weaker Mg I line than HD 216143, which does not support the anomalously high [Mg/Ca] in Boo-127 compared to the reference star. More quantitatively, the relative abundance log   (A/Aref) is expressed as (1)where κν and κν,ref are continuum opacity, Δθex (θ = 5040/T) is the difference in an excitation temperature, and χ is the excitation potential of the line (Gray 2005). If we assume that the κν term is negligible and Δθex is constant and equal to the one computed from the effective temperature (Δθex = Δθex,eff) for all relevant optical depths, the relative abundance can be obtained from the ratios of the EWs (Smith & Harmer 1982). This results in the [Mg/Ca] ratio for Boo-127 being ~0.04 dex lower than that of HD 216143, which is, again, inconsistent with the anomalously high [Mg/Ca] ratio. The low, but not extremely low, Ba abundances ([Ba/Fe] ~−0.6) of Boo-121 and Boo-911 found by F09 are confirmed by our measurement. The log   ϵ(Ba) in the two studies agree within 0.3 dex.

thumbnail Fig. 4

Spectra of Boo-127(red) and the comparison star, HD 216143 (black), for the wavelength regions that contain Mg I and Ca I lines.

Open with DEXTER

4.2. Comparison with Gilmore et al. (2013)

Comparisons of the relative abundances in this work and in the “GM” analysis of G13 are presented in Fig. 5 (triangles that are connected with the data points in this work) for the three stars analyzed in common. Within the quoted error bars, our abundance results generally agree with those obtained by G13 with a few exceptions. One relatively large discrepancy is found in the [Fe/H] value for Boo-094, the most metal-poor star in our sample, which is ~0.2 dex lower in this work than obtained by G13. The smaller EWs (see Sect. 2.3) by ~10 mÅ and larger ξ adopted in this work than in G13 are probably responsible for the discrepancy.

thumbnail Fig. 5

[X/Fe] ratios plotted against [Fe/H] for the sample of six Boötes I stars (filled red circles) and the two comparison field halo stars (open blue circles). Black triangles represent the abundance results from the “GM” analysis in G13. Small gray circles present the abundances of field halo stars from Ishigaki et al. (2012, 2013). In the plot for [C/Fe]–[Fe/H], results from Norris et al. (2010a) for Boötes I stars are shown with crosses.

Open with DEXTER

4.3. Carbon

The carbon abundance is estimated for three of the six sample stars, while only an upper limit is obtained for the other three objects. For the six sample stars, extremely carbon-rich objects are not found. These sample stars all show [C/H] < −2.7 and [C/Fe] < 0.2. For the three objects with measured carbon abundances, the average [C/Fe] is −0.7 dex with small scatter. These values are close to the lowest bounds of the [C/Fe] distributions obtained by Lai et al. (2011) and Norris et al. (2010a) for this galaxy. This [C/Fe] value is comparable to what is seen in field halo stars having similar luminosity (Spite et al. 2005; Norris et al. 2010a).

4.4. Sodium

For the present Boötes I sample stars, sodium abundances are only estimated using the strong Na I resonance lines at 5889/5895 Å. These lines are known to suffer from strong non-LTE (NLTE) effects such that LTE abundances tend to be larger than corresponding NLTE values (Takeda et al. 2003). The NLTE correction to Na abundances ranges from −0.1 dex to −0.5 dex, depending on the atmospheric parameters of stars (Takeda et al. 2003). Without any NLTE correction, the average [Na/Fe] value for the six Boötes I stars is ~−0.1 dex with small scatter. This value is similar to the uncorrected [Na/Fe] values obtained for field halo stars with comparable metallicities.

4.5. α/Fe ratios

The [Mg/Fe] ratios in our sample stars are similar to or slightly lower than those obtained for the field halo stars. A hint of a modest decreasing trend of the [Mg/Fe] with increasing [Fe/H] is seen, as was also reported in G13. The [Ca/Fe] ratios are also similar to or slightly lower than the value for the field halo stars. Excluding the lowest metallicity star, a scatter in the [Ca/Fe] ratios is remarkably small (~0.1 dex). The [Ti I/Fe I] and [Ti II/Fe II] ratios differ by up to ~0.3 dex in our Boötes I sample stars, as reported for the field halo stars (e.g., Lai et al. 2008). The Boötes I stars tend to show slightly lower [Ti I/Fe I] abundance ratios than the halo stars, while the [Ti II/Fe II] ratios for these samples are similar.

4.6. Iron-peak elements

Scandium abundances are estimated from a few to several Sc II lines in our sample stars. The mean abundances are similar to or slightly lower than those reported for the field halo stars in [Fe/H] > −2.5.

Chromium abundances are obtained from a few Cr I lines for the Boötes I stars. The [Cr/Fe] ratios are similar to those reported for the field halo stars for two of the most metal-rich stars, while other stars show lower [Cr/Fe] ratios. However, since only one or two lines with low-signal-to-noise ratio are used in the abundance estimate, the significantly lower [Cr/Fe] ratios for these stars require further confirmation.

In the four higher metallicity stars in our sample, the [Ni/Fe] ratios are similar to those obtained for the field halo stars, except for one object (Boo-121). Two Ni I lines used to obtain Ni abundance in this stars both suggest a subsolar [Ni/Fe] ratio.

Since EWs of Zn I lines are not reliably measured for our sample stars, we used spectral synthesis to estimate upper limits of the Zn abundances. The results are shown by bars with downward arrows in the bottom right hand panel of Fig. 5. The upper limits of the [Zn/Fe] ratios are similar to or lower than the field halo samples in the metallicity range [Fe/H] > −2.7.

thumbnail Fig. 6

Same as Fig. 5 but for Sr and Ba. Abundance data extracted from the SAGA database (Suda et al. 2008) (RGB, R > 30   000 with no flags of “C-rich” nor of “CEMP”) are additionally plotted with gray crosses to supplement the MW halo data.

Open with DEXTER

4.7. Neutron-capture elements

The strontium abundances are obtained by fitting synthetic spectra to the observed ones for the Sr II resonance line at 4215.5 Å. The log   gf value for this line is taken from NIST database. Figure 7 shows the resulting synthetic and the observed spectra around the Sr II line for the Boötes I stars for which the Sr abundances are reasonably obtained. The other three sample stars are excluded in the analysis since the spectral region is too noisy to estimate Sr abundances.

In the fitting procedure, we redefine the continuum level using the wavelength regions 4214.10–4215.25 Å and 4216.80–4217.10 Å. To estimate abundance errors due to the uncertainty in the continuum level, we performed the fitting changing the continuum level of the synthetic spectrum by ±1σSTDV, where σSTDV is a standard deviation in the observed continuum level. The error estimated in this procedure is ~0.71.5 dex, as summarized in Table 5.

thumbnail Fig. 7

Observed (black dots) and best-fit synthetic (solid line) spectra of the three Boötes I stars around the Sr II 4215.5 Å line. The obtained [Fe/H] and [Sr/H] values are indicated in the bottom-left corner of each panel (“Name/[Fe/H]/[Sr/H]”). The dotted lines show the synthetic spectra fitted after changing the continuum levels by ± 1σSTDV (in each panel, the continuum levels have been shifted to match the best-fit value). For comparison, the spectrum of one halo giant, HD 107752, is shown in the bottom panel.

Open with DEXTER

For comparison, the bottom panel of Figure 7 shows the observed spectrum of a field halo giant, HD 107752, which has atmospheric parameters similar to those of the present sample (Teff= 4820 K, log   g= 1.6, ξ = 1.9 km s-1, and [Fe/H] = −2.8) (Ishigaki et al. 2013). While the Sr II line of the HD 107752 appears to be stronger than a nearby Fe I line, those in the three Boötes I stars seem to have comparable strength as the Fe I lines.

Andrievsky et al. (2011) studied the NLTE effects on Sr abundance determination for extremely metal-poor stars and report that the NLTE correction for the Sr II 4215.5 Å line is at most ~0.2 dex for stars having atmospheric parameters similar to the present Boötes I sample. The correction is smaller than the other sources of uncertainty in the present analysis, so that is not included in our abundance results.

The top panel of Fig. 6 shows resulting [Sr/Fe] values plotted against [Fe/H] with the error bars. The crosses and small filled circles represent the abundances from literature extracted from SAGA database and those from Ishigaki et al. (2013), respectively. As can be seen, even though the errors are very large, the [Sr/Fe] ratios of the three Boötes I stars are significantly lower than the solar value, which is not typical for the field halo stars at similar metallicities. Most of the field halo sample in literature show the solar value in [Fe/H] < −2.5 and increasing dispersion toward lower metallicity with [Sr/Fe] ~ −2.0 to ~0.5. On the other hand, the Boötes I stars have [Sr/Fe] ~ −1 in [Fe/H] > −2.5, and the most metal-poor star in our sample show even lower [Sr/Fe] of ~−2.

The bottom panel of Fig. 6 plots the [Ba/Fe] ratios against [Fe/H]. The [Ba/Fe] of Boötes I stars are subsolar, generally confirming the result of G13. The [Ba/Fe] ratios for these stars tend to distribute at the lower bound of the [Ba/Fe] distribution for the field halo stars.

One interesting feature is that Boo-127, which is the most metal-rich star in our sample, shows lower [Ba/Fe] than the field halo stars at similar metallicities ([Fe/H] ~ −2). As shown in Fig. 4, the Ba II line is significantly weaker in Boo-127 than seen in the comparison star with similar [Fe/H].

5. Discussion

5.1. Comparison with the field MW halo

If the MW halo was assembled from smaller satellite galaxies similar to the surviving dSphs, as suggested by the standard cosmological model, we naively expect that chemical abundances in the MW field halo stars are similar to these dSphs. In this section, we address the question of whether the observed abundances in Boötes I stars are similar to those obtained for the field halo stars.

thumbnail Fig. 8

The abundance pattern of the Boötes I stars (filled circles) and the MW halo stars (open circles) with −3.0 < [Fe/H] < −2.0.

Open with DEXTER

thumbnail Fig. 9

log   ϵ(Ba) versus log ϵ(Sr) for the three Boötes I stars. For comparison, the abundance data for MW halo giants are extracted from SAGA database (Suda et al. 2008) with the criteria −3.5 < [Fe/H] < −2.0 and a spectral resolution R > 30   000 with no flags of “C-rich” nor of “CEMP”.

Open with DEXTER

Figure 8 compares mean abundances of the Boötes I stars and field halo stars in the metallicity range of −3.0 < [Fe/H] < −2.0. To minimize systematic errors in the abundance comparison, eight halo giants (Teff< 5000 K and log   g< 3.0, including one of the comparison stars HD 85773) from Ishigaki et al. (2012, 2013) are only considered for the field halo sample. In their work, the adopted 1D-LTE abundance analysis technique and the line list are the same as in this work. Error bars in Fig. 8 represent standard deviations around the mean values.

In comparison to the field halo stars, the Boötes I stars show slightly lower [Mg/Fe] and [Ca/Fe] ratios in the metallicity range of −3.0 <  [Fe/H]  < −2.0. Kolmogorov-Smirnov (KS) tests suggest that the probability at which [X/Fe] ratios of the halo stars and the Boötes I stars can be drawn from the same distribution is less than 5% for Mg and Ca, which marginally supports the abundance difference. On the other hand, the [Na/Fe] ratios for the Boötes I stars are indistinguishable from those of the halo stars.

As mentioned in the previous sections, the Boötes I and the MW halo stars show a hint of an abundance difference in Sr. Figure 9 shows log   ϵ(Sr) plotted against log   ϵ(Ba) for the Boötes I stars and MW halo giants. The data for the MW halo giants were extracted from the SAGA database (Suda et al. 2008) with criteria, −3.5 < [Fe/H] < −2.0 and a spectral resolution R > 30   000 with no flags of “C-rich” or “CEMP”. As can be seen, most of the halo stars show Sr/Ba > 1, while the Boötes I stars show Sr/Ba ≲ 1. Aoki et al. (2005) found that some of the halo stars with [Fe/H] < −3.0 show Sr/Ba > 1, while the excess of Sr is not seen for stars with higher Ba abundances. They interpret the result as evidence that the two astrophysical sites are responsible for producing these neutron-capture elements, one of which produces both Sr and Ba and the other produces more Sr than Ba. Since the signature of enhanced Sr relative to Ba is not evident in our Boötes I sample, the latter astrophysical site may be lacking or less frequent in this galaxy. We discuss the origin of Sr and Ba in Bootes I in Sect. 5.3.5.

To summarize, the present sample of Boötes I stars show a hint of difference in abundances of some of the α-elements and neutron-capture elements from the bulk of MW halo stars. This result suggests that the chemical abundances of Boötes I stars alone are not enough to fully explain the observed chemical abundances for field halo stars.

Corlies et al. (2013) have investigated conditions that yield a difference in chemical abundances between surviving satellite galaxies and progenitors of the MW halo using an N-body simulation. They suggest that the progenitors of the stellar halo usually experienced more cross pollution by neighboring subhalos, while the surviving satellites tend to have evolved chemically in isolation. Whether the degree of the cross pollution explains the observed abundance difference may have interesting implications for the hierarchical formation of the MW halo.

Since the present result is based on a small sample, a larger sample is clearly desirable to extract a conclusion for the range in elemental abundances spanned by the whole stellar population in Boötes I. Moreover, abundances of field halo stars are often determined only for relatively nearby objects (within a several kpc), while the estimated distance to Boötes I is 64    ±    3 kpc (Okamoto et al. 2012). A systematic abundance survey for more distant halo stars would be needed to examine possible connection between dSphs similar to Boötes I and the MW field halo.

thumbnail Fig. 10

Top panel: abundance comparisons with classical (Draco: triangles, Sextans: three-pointed asterisks, Ursa Minor: squares) and ultra-faint (Coma Berenices: pentagons, Hercules: stars, Ursa Major II: asterisks) dwarf galaxies from the literature (Aoki et al. 2009; Cohen & Huang 2009, 2010; Frebel et al. 2010; Koch et al. 2008). For comparison, the abundance data for MW halo giants extracted from the SAGA database (Suda et al. 2008) (R > 30   000 with no flags of “C-rich” nor of “CEMP”) and from Ishigaki et al. (2012) are plotted with crosses and dots, respectively. Bottom panels: mean [Mg/Fe] ratios for each galaxy plotted against velocity dispersion (σlos, left) and luminosity (log   L(L), right) for that galaxy. The σlos and L data are taken from Walker et al. (2009).

Open with DEXTER

5.2. Comparison with other dSphs

Figure 10 compares the [Mg/Fe] ratios of the Boötes I stars in this work and those of other MW dSphs that include very metal-poor ([Fe/H] < −2) stars from the literature (Aoki et al. 2009; Cohen & Huang 2009, 2010; Frebel et al. 2010; Koch et al. 2008). Relatively luminous (LV, ⊙ ≳ 105  L), so-called classical dSphs (Draco, Sextans, Ursa Minor) are shown with green symbols, while less luminous dSphs (UFDs; Boötes I, Coma Berenices, Hercules, Ursa Major II) are shown in red symbols. According to these available data, the more luminous dSphs appear to show relatively low [Mg/Fe] ratios than field halo stars at [Fe/H] ~ −2 and/or a decreasing [Mg/Fe] trend with increasing [Fe/H]. On the other hand, the fainter dSphs such as Coma Berenices or Ursa Major II appear to show comparable or higher abundance ratios with large scatter. The Boötes I stars generally show the [Mg/Fe] ratios more similar to some of the classical dSphs, such as Ursa Minor rather than the fainter UFDs, such as Coma Berenices or Ursa Major II, although there seems to be large variation in the abundance ratios among the classical dSphs.

thumbnail Fig. 11

Same as Fig. 10 but for [Ca/Fe].

Open with DEXTER

To examine whether the abundance ratios can be related to the global properties of dSphs (e.g., dynamical masses or luminosity), the lower panels in Fig. 10 plot average [Mg/Fe] ratios of these dSphs against their central line-of-sight velocity dispersions (σlos; left) and logarithms of total luminosity (L; right). The data for σlos and L are taken from Walker et al. (2009) except for the σlos of Boötes I, for which an updated value from Koposov et al. (2011) ( km s-1, assuming a single component velocity distribution) is available. These figures suggest that the dSphs with larger velocity dispersions or higher luminosity tend to show lower [Mg/Fe] ratios. Again, Boötes I, which has σlos as low as that of Coma Berenices and L midway between these two classes of dSphs, show the mean [Mg/Fe] ratio to be more similar to the classical dSphs, such as Ursa Minor or Draco. A similar trend can be seen for [Ca/Fe] ratios as shown in Fig. 11. The exceptionally high [Mg/Fe] ratios are reported for the two Hercules stars, which may suggest an anomalous star formation history for this galaxy (see also François et al. 2012). According to the available data, differences in other elemental abundances lighter than zinc, such as [Na/Fe] ratios, are not very clear among all dSphs including Boötes I for [Fe/H] < −2.

The MW dSphs have been reported to follow a correlation between mean [Fe/H] and L (Kirby et al. 2011b) extending for a wide range of luminosities (3.5 < log    (L/L) < 7.5). Although precision and sample size may not be sufficient to examine the [Mg,Ca/Fe]–σlos, and [Mg,Ca/Fe]–L relations, if any, from currently available data, the locations of Boötes I in these diagrams may have implications for the chemical evolution of this galaxy, which we discuss in Sect. 5.3.

Figure 12 shows [Sr/H] and [Ba/H] ratios plotted against [Fe/H] for Boötes I and other dSphs. Classical dSphs generally show increasing [Sr/H] and [Ba/H] ratios with increasing [Fe/H], similar to the trend seen for field halo stars. On the other hand, the abundances for fainter dSphs tend to distribute at lower bounds than for the field halo stars. Lower abundances of neutron-capture elements for very-metal-poor stars in dSphs have already been noticed in previous studies (e.g., Fulbright et al. 2004; Koch et al. 2008, 2013; Frebel et al. 2010; Honda et al. 2011). Our sample of Boötes I stars show [Ba/H] ratios that are more similar to the classical dSphs, while [Sr/H] ratios are lower, similar to those observed in fainter dSphs.

Whether the neutron-capture elements are more deficient in dSphs than in the MW field halo stars needs to be investigated with a larger sample of high-quality spectra for distant objects, which may be challenging with the current instrumentation. Nevertheless, the detection of Sr and Ba in Boötes I suggests that at least one event that produces neutron-capture elements must have occurred at the earliest epoch of star formation in this galaxy (cf., Roederer 2013).

thumbnail Fig. 12

Comparison of the [Sr/H] and [Ba/H] ratios in Boötes I stars obtained in this work and those in very metal-poor stars in other dSphs. Symbols are the same as in Fig. 10.

Open with DEXTER

5.3. Chemical evolution of Boötes I

In Sect. 5.1, we reported a possible offset in the abundances of α and neutron capture elements between the Boötes I stars and the field halo stars in the metallicity range −3 < [Fe/H] < −2. These differences, if confirmed with larger samples, may suggest the differences in chemical evolution history between the Boötes I and dominant progenitors of the MW stellar halo. As discussed in Sect. 5.2, the [Mg/Fe] and [Ca/Fe] ratios in the observed Boötes I stars are generally similar to the values reported for brighter (classical) dSphs in the metallicity range −3 < [Fe/H] < −2, while their velocity dispersion and luminosities are more similar to those of UFDs (Figs. 10 and 11).

In this section, we discuss possible implications of the results of these comparisons on the chemical evolution of Boötes I in terms of 1) contribution from Type Ia SNe; 2) galactic winds, 3) tidal interaction with the MW; 4) stochastic effects; and 5) individual supernova yields.

5.3.1. Contribution from Type Ia SNe

In chemical evolution models, trends in abundances of α-elements with [Fe/H] are diagnostics of the Type Ia SNe contribution relative to Type II SNe to chemical enrichments in galaxies (e.g., Matteucci & Greggio 1986). For many of known dSphs, decreasing trends of [α/Fe] with increasing [Fe/H] are reported (Tolstoy et al. 2009; Kirby et al. 2011a, 2013; Vargas et al. 2013), which could be indicative of chemical enrichment proceeded partly through Type Ia SNe.

Whether Boötes I has experienced a significant contribution from Type Ia SNe to its chemical enrichments is an important question for constraining the star formation history in this galaxy. Photometric observations suggest that Boötes I is dominated by an old stellar population with little age spread (Okamoto et al. 2012), which suggests that the duration of star formation was shorter than the uncertainty of the photometric age estimates (a few Gyr). Gilmore et al. (2013) suggest a contribution of Type Ia SNe in order to interpret the apparent decreasing [α/Fe] trend with [Fe/H]. On the other hand, the low-resolution spectroscopic study of Lai et al. (2011) does not find a signature of decreasing [α/Fe] ratios with increasing [Fe/H], which does not support the Type Ia contribution.

The observed [Mg/Fe] and [Ca/Fe] ratios for the Boötes I stars in this work appear to show a modest decreasing trend with [Fe/H]. The Pearson test for linear correlation indicates that the probability for a null correlation is 30% for [Mg/Fe][Fe/H] and 7% for [Ca/Fe][Fe/H], thus only marginally supporting the correlation. Although a firm conclusion about the trend could not be obtained for the present sample, the hint of the decreasing trend suggests that the star formation in Boötes I lasted at least until Type Ia SNe began to contribute to the chemical enrichment.

The decreasing [Ca/Fe] trend with [Fe/H] has also been reported for a sample of red giant stars in Hercules dSph, which spans the [Fe/H] range similar to the present Boötes I sample. This may imply that the contribution of Type Ia SNe to the chemical enrichment is common among galaxies as faint as Hercules or Boötes I, if the Type Ia SNe is indeed responsible for lowering [Ca/Fe]. Analyses of the delay-time distribution of Type Ia SNe suggest that a certain fraction of Type Ia SNe occur within <1 Gyr of star formation (Maoz et al. 2010), which may support their contribution to the chemical enrichments in Boötes I.

An alternative interpretation of the lower [Mg, Ca/Fe] ratios in Boötes I is a different initial mass function (IMF) in this galaxy in comparison to the dominant progenitors of the MW halo. Since [Mg/Fe] yields in Type II SNe could depend on masses of the progenitor stars (Kobayashi et al. 2006), a difference in IMF may lead to systematic difference in these abundance ratios.

5.3.2. Galactic wind

Galactic winds driven by SNe are thought to significantly affect subsequent chemical evolutions in dSphs (Arimoto & Yoshii 1987). Chemical evolution models of Lanfranchi & Matteucci (2004) particularly suggest that the strength and time of occurance of the galactic winds play an important role in determining the [α/Fe] ratios in the dSphs.

The winds are expected to occur when part of the energy injected by SNe heat up interstellar gas so that thermal energy of gas exceed the binding energy of the galaxy (Arimoto & Yoshii 1987). In this scenario, the strength of the wind may depend on both the initial gas content and gravitational potential of the galaxy.

As can be seen in the lower left panels of Figs. 10 and 11, the dSphs with smaller velocity dispersion, and thus lower dynamical masses, such as Coma Berenices or Ursa Major II, show higher [Mg/Fe] and [Ca/Fe] ratios than the dSphs with larger velocity dispersions. The likely correlation between the velocity dispersion and the abundance ratios suggests that the chemical abundance depends on the gravitational potential of galaxies, if the velocity dispersion traces the dynamical mass of these systems.

Using the measured velocity dispersion profile, Walker et al. (2009) estimate the dynamical mass of Boötes I within the half-light radius of this galaxy (rhalf = 242    ±    21 pc) to be ~106  M. If we assume that the mass has not changed significantly over the age of this galaxy and if duration of star formation is indeed longer than the timescale of Type Ia SNe, as suggested in the previous section, then the above scenario implies that a dark matter halo with mass as low as that of Boötes I could host enough initial gas to retain a certain duration of star formation, or that galactic winds were not strong enough to truncate all subsequent star formation in such a halo. Whether this scenario could be possible depends on the possible tidal effect on the initial gas and dark matter content of this galaxy as discussed below.

5.3.3. Tidal Interaction with the MW

We discussed in Sect. 5.2 a possible connection of chemical abundance ratios with global properties, such as σlos and L. The correlation is not currently clear, and Boötes I in particular appears to deviate from a possible relation. Thus, we may ask what makes Boötes I smaller σlos and lower L, while experiencing chemical evolution that is more like the brighter classical dSphs.

One possible mechanism that may affect σlos or L is tidal interaction with the MW potential. Tidal interaction of Boötes I with the MW may have stripped the original content of baryons and dark matter in this galaxy. N-body simulations for dynamical evolution of Boötes I suggest that this galaxy has lost a significant amount of luminous and dark matter contents by the tidal forces as it orbits around the MW potential (Fellhauer et al. 2008). It is possible that Boötes I was initially as luminous as the classical dSphs, while it has lost a significant amount of materials.

Since the full orbital motion of Boötes I is not known, it is difficult to estimate how strong the tidal effects were in this galaxy. The tidal radius of this galaxy at its present location (62 kpc from the Galactic center) is estimated to be 1.2 kpc (Fellhauer et al. 2008), which is larger than the luminous component in this galaxy. Therefore, it is unlikely that the tidal effect is currently significant in stripping material from this galaxy. On the other hand, Collins et al. (2013) report that the estimated maximum circular velocity (Vmax) for Boötes I is relatively low compared to other local group dSphs with wide ranges of luminosities, which could be explained by tidal stripping of dark matter. Therefore, one explanation of the observed similarity may be that Boötes I once experienced similar chemical evolution to, say, Ursa Minor, then within a few Gyr, it followed an orbit that brought it closer to the Galactic center, experiencing strong tidal force that were able to strip luminous and dark matter contents in this galaxy. All subsequent star formation stopped, which results in the current low luminosity and very low metallicity ([Fe/H] < −1.5) in this galaxy (Norris et al. 2010a). The information about the orbital motion of Boötes I would be necessary for examining such a scenario to interpret the observed chemical abundances in this galaxy.

5.3.4. Stochastic effect

In very low-metallicity systems like Boötes I, it is possible that only a small number of SNe of a prior generation of stars have contributed to the chemical enrichment. For example, Frebel & Bromm (2012) suggest that first galaxies that formed exclusively from ejecta of the first generation of stars (Pop III stars) are expected to host relatively few progenitor SNe, which could be around ten, compared to systems with higher metallicity. In this case, the effects of stochastic sampling of an IMF may become important.

The present sample of Boötes I stars are rather homogeneous in abundance ratios for elements lighter than zinc, which does not strongly support the existence of any stochastic effect. Nevertheless, the observed homogeneous [X/Fe] ratios do not completely rule out the stochastic enrichment scenario, since little scatter in [X/Fe] is expected, while producing large scatter in [Fe/H] in such scenarios as proposed by Frebel & Bromm (2012). This scenario thus implies that Boötes I could be a candidate for the first galaxies formed in the Universe.

Lee et al. (2013) have recently investigated the effect of the stochastic IMF sampling as a possible interpretation of the abundance difference and similarity between the UFDs and MW halo stars. In their model, a system enriched by a smaller number of a prior generation of stars is expected to show a lower mean value of [X/Fe] with larger spread, when the supernova yield of an element X has a positive mass dependence (i.e., higher ejected mass for a higher supernova progenitor mass). Therefore, if a UFD has been enriched by fewer SNe (~17 on average) than the progenitors of the MW halo (>7) and if a Sr yield, which is currently uncertain, has a strong positive mass dependence, we would expect lower [Sr/Fe] for the UFDs than the MW halo progenitor. They conclude that the stochastic sampling of IMF for UFDs can simultaneously explain the observed difference in [Sr, Ba/Fe] ratios and the similarity in [Ti/Fe] ratios between the UFDs and MW halo stars.

The result of the present work, a modest difference in [Mg/Fe] and a large difference in [Sr/Fe] from field halo stars appear to both be reproduced with the above model if the Sr yield depends more strongly on progenitor masses than the Mg yields do. Thus, according to this model, the difference in number of prior-generation stars can explain the observed abundance difference between the Boötes I and progenitors of the MW halo.

5.3.5. Individual yields

We finally discuss whether the abundances of the present sample stars are consistent with yields of a single SN of Pop III stars. Pop III stars may have been very massive. Very massive stars with M > 140  M are predicted to be exploded as pair-instability SNe whose yields are expected to show a peculiar abundance pattern, including very large Ca abundances. As can be seen, none of our sample stars show this peculiar pattern and seem to be more consistent with yields of core-collapse SNe (e.g., Tominaga et al. 2007).

An intriguing issue is whether nucleosynthetic yields of SNe are compatible with the observed abundances of Sr and Ba in Boötes I. Core-collapse SNe of massive (>10  M) stars have been considered as one of the promising production sites of heavy neutron-capture elements such as Ba (e.g., Thielemann et al. 2011). However, more recent studies report difficulty in reproducing the required physical condition for the r-process to occur in the environment of the core-collapse SNe.

Qian & Wasserburg (2008) used the phenomenological model to study nucleosynthetic yields of neutron-capture elements in neutrino-driven winds, which are associated with SNe. They pointed out that the low [Sr/Fe] ratios are not reproduced with the predicted yields of Type II SNe with progenitor masses of neither 810 M (H source) nor 1225 M (L sources). Instead, they suggest that another source that produces a large amount of Fe but little Sr is required to explain the abundance patterns of stars with [Sr/Fe] ≪ −0.32. They speculate that an energetic SN, a so-called hypernova, could be a possible astrophysical site that yields the low [Sr/Fe] ratios. In this scenario, the observed Boötes I stars may have formed out of interstellar medium predominantly enriched by a single supernovae with large explosion energy.

A merging of neutron stars in a close binary system is proposed as an alternative site of the r-process. Recent calculations of nucleosynthetic yields for the neutron-star merger successfully explain the abundance patterns of known r-process-rich stars in the Galactic field halo (e.g., Wanajo & Janka 2012; Korobkin et al. 2012). It is interesting for future studies to investigate whether the predicted nucleosynthetic yields of neutron star mergers can explain the observed Sr/Fe and/or Ba/Fe abundances in Boötes I.

The supernova yields depend on various unknown parameters, such as mass cut or explosion energies, and thus properties of the progenitor SN are not fully constraind at this moment. Nucleosynthetic yields of light and heavy neutron capture elements for different SN progenitor masses at extremely low metallicities are required to examine the reason for the Sr deficiency in Boötes I.

6. Conclusions

The chemical abundances of C, Na, α, iron-peak, and neutron capture elements in six giant stars in Boötes I ultra-faint dwarf galaxy were estimated based on the high-resolution spectra obtained with Subaru/HDS. Our main results are summarized as follows:

  • 1.

    The very low metallicity of Boo-094 ([Fe/H] = −3.4) previously reported in Feltzing et al. (2009) and Gilmore et al. (2013) is confirmed. For the six sample stars, extremely carbon-rich objects are not found.

  • 2.

    For α and iron-peak elements, the scatter in the abundance ratios is small over the metallicity range of −2.7 < [Fe/H] < −1.8.

  • 3.

    The [Mg/Fe] and [Ca/Fe] ratios for these stars are slightly lower than the field halo stars on average in the metallicity range of −3.0 < [Fe/H] < −2.0. On the other hand, the abundance ratios for Na and Sc in the Boötes I stars are generally similar to those of the field halo stars.

  • 4.

    The [Sr/Fe] abundances of the three Boötes I stars are lower than the field halo values at similar metallicities. These Boötes I stars deviate from the correlation between log   ϵ(Ba) and log   ϵ(Sr) seen in field MW halo stars and generally show lower Sr/Ba ratios.

Highly inhomogeneous chemical abundances that are reported in fainter dSphs are not seen in our Boötes I sample. Instead, our results suggest that the star formation in Boötes I had lasted until Type Ia SNe started to contribute to the chemical enrichment, which has also been reported in other more luminous dSphs. Larger samples of chemical abundances and the kinematics of individual stars, as well as a determination of the orbital motion around the MW, are desirable to conclude whether Boötes I is once more similar to brighter dSphs or is a surviving example of first galaxies formed out of ejecta from Pop III stars.


Acknowledgments

We thank the referee for constructive comments and suggestions that have improved our paper. We are grateful to S. Feltzing for providing us with their Mg linelist. M.N.I. thanks K. Nomoto, L. Vargas, and A. Frebel for helpful discussions on the abundance analysis and chemical evolution of ultra-faint dwarf galaxies. M.N.I. acknowledges financial support from Grant-in-Aid for JSPS (Japan Society for the Promotion of Science) fellow. W.A. was supported by the JSPS Grants-in-Aid for Scientific Research (23224004).

References

Online material

Table 5

Elemental abundances and uncertainties.

All Tables

Table 1

Objects and observations.

Table 2

The Mg line lists.

Table 3

Atmospheric parameters.

Table 4

Equivalent widths and abundances.

Table 5

Elemental abundances and uncertainties.

All Figures

thumbnail Fig. 1

Comparison of the measured equivalent widths (EWs) in this work (TW) and in the two analyses, “NY” (closed circles) and “GM” (open circles), in Gilmore et al. (2013).

Open with DEXTER
In the text
thumbnail Fig. 2

EW measurements for one of the Fe I lines by adopting the different continuum levels. The dotted line shows the best-fit Gaussian, where continuum level is estimated from the shaded wavelengths region. The dashed lines show the Gaussian fit by fixing the continuum levels to be ± 0.05 of the best-fit value.

Open with DEXTER
In the text
thumbnail Fig. 3

Changes in abundances when all of the EWs in this work are increased by 15 mÅ to partly correct for the EWs offset from Gilmore et al. (2013).

Open with DEXTER
In the text
thumbnail Fig. 4

Spectra of Boo-127(red) and the comparison star, HD 216143 (black), for the wavelength regions that contain Mg I and Ca I lines.

Open with DEXTER
In the text
thumbnail Fig. 5

[X/Fe] ratios plotted against [Fe/H] for the sample of six Boötes I stars (filled red circles) and the two comparison field halo stars (open blue circles). Black triangles represent the abundance results from the “GM” analysis in G13. Small gray circles present the abundances of field halo stars from Ishigaki et al. (2012, 2013). In the plot for [C/Fe]–[Fe/H], results from Norris et al. (2010a) for Boötes I stars are shown with crosses.

Open with DEXTER
In the text
thumbnail Fig. 6

Same as Fig. 5 but for Sr and Ba. Abundance data extracted from the SAGA database (Suda et al. 2008) (RGB, R > 30   000 with no flags of “C-rich” nor of “CEMP”) are additionally plotted with gray crosses to supplement the MW halo data.

Open with DEXTER
In the text
thumbnail Fig. 7

Observed (black dots) and best-fit synthetic (solid line) spectra of the three Boötes I stars around the Sr II 4215.5 Å line. The obtained [Fe/H] and [Sr/H] values are indicated in the bottom-left corner of each panel (“Name/[Fe/H]/[Sr/H]”). The dotted lines show the synthetic spectra fitted after changing the continuum levels by ± 1σSTDV (in each panel, the continuum levels have been shifted to match the best-fit value). For comparison, the spectrum of one halo giant, HD 107752, is shown in the bottom panel.

Open with DEXTER
In the text
thumbnail Fig. 8

The abundance pattern of the Boötes I stars (filled circles) and the MW halo stars (open circles) with −3.0 < [Fe/H] < −2.0.

Open with DEXTER
In the text
thumbnail Fig. 9

log   ϵ(Ba) versus log ϵ(Sr) for the three Boötes I stars. For comparison, the abundance data for MW halo giants are extracted from SAGA database (Suda et al. 2008) with the criteria −3.5 < [Fe/H] < −2.0 and a spectral resolution R > 30   000 with no flags of “C-rich” nor of “CEMP”.

Open with DEXTER
In the text
thumbnail Fig. 10

Top panel: abundance comparisons with classical (Draco: triangles, Sextans: three-pointed asterisks, Ursa Minor: squares) and ultra-faint (Coma Berenices: pentagons, Hercules: stars, Ursa Major II: asterisks) dwarf galaxies from the literature (Aoki et al. 2009; Cohen & Huang 2009, 2010; Frebel et al. 2010; Koch et al. 2008). For comparison, the abundance data for MW halo giants extracted from the SAGA database (Suda et al. 2008) (R > 30   000 with no flags of “C-rich” nor of “CEMP”) and from Ishigaki et al. (2012) are plotted with crosses and dots, respectively. Bottom panels: mean [Mg/Fe] ratios for each galaxy plotted against velocity dispersion (σlos, left) and luminosity (log   L(L), right) for that galaxy. The σlos and L data are taken from Walker et al. (2009).

Open with DEXTER
In the text
thumbnail Fig. 11

Same as Fig. 10 but for [Ca/Fe].

Open with DEXTER
In the text
thumbnail Fig. 12

Comparison of the [Sr/H] and [Ba/H] ratios in Boötes I stars obtained in this work and those in very metal-poor stars in other dSphs. Symbols are the same as in Fig. 10.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.