Free Access

This article has an erratum: [erratum]

Issue
A&A
Volume 586, February 2016
Article Number A49
Number of page(s) 24
Section Galactic structure, stellar clusters and populations
DOI https://doi.org/10.1051/0004-6361/201527385
Published online 25 January 2016

© ESO, 2016

1. Introduction

Elements with atomic numbers up to iron can be synthesized via nuclear fusion in the interiors of stars. Heavier elements are formed by the addition of neutrons in the stellar interiors and are called neutron-capture elements. There are two main ways to add neutrons, through slow neutron-capture (s-process) or through rapid neutron-capture (r-process), depending on whether the neutron-capture is slow or rapid compared to the timescale for β decay (Burbidge et al. 1957). The s-process requires a low neutron flux, and the creation of the new elements moves along the valley of β stability. In the case of the r-process, the neutron flux is intense, permitting the creation of elements outside the valley of stability.

An important step in the study of neutron-capture elements was the discovery that metal-poor stars show high relative abundances of certain neutron-capture elements compared to Fe, meaning that the r- and s-processes were already active at early times (e.g., Frebel & Norris 2013). In particular for the r-process, this was by the significant number of stars with high levels of the r-process element Eu at very low metallicities (e.g., the neutron-capture review by Sneden et al. 2008). However, the production sites for r-process elements are poorly understood, and currently there are at least three possible scenarios. First, the classic scenario for r-process production is neutrino-induced winds from type-II supernovae (SN II; Woosley et al. 1994). Extremely energetic neutrinos are produced during the collapse of the SN II, and they are potentially able to interact with the dense material that is falling onto the core of the star. This interaction can heat the material, giving to it the additional energy needed to recreate the energy output observed of 1051 erg. Other scenarios are the merging of neutron stars (Freiburghaus et al. 1999) or the merging of a neutron star with a black hole (Surman et al. 2008). Rosswog et al. (2014) show that that dynamic ejecta from this merging produce r-process elements with A> 130 with a pattern independent of the kind of merging, while neutrino-driven winds are responsible for nucleosynthesis of elements from A = 50 to A = 130, but in this case the exact output depends on the merging parameters. Lastly, polar jets from SN II with the use of a pure magnetohydrodynamic explosion seems to lead to the right conditions to have r-process nucleosynthesis (Nishimura et al. 2006). Unfortunately, there are still theoretical problems in the modeling of SN II explosions with neutrino wind that prevent a definitive confirmation of the latter production site. In addition to this, theoretical predictions for r-process production have not been entirely successful in synthesizing the observed total abundance distribution of r-process nuclei (Cescutti et al. 2006).

The production of s-process elements, on the other hand, can occur in massive stars, in the He-burning core, and in the convective C-burning shell (Pignatari et al. 2010), as well as in asymptotic giant branch (AGB) of lower mass stars at solar and lower metallicities (Bisterzo et al. 2011). Considering that the s-process is responsible for producing approximately half of the nuclides from Fe to Bi – in particular feeding the groups Sr-Y-Zr, Ba-La-Ce-Pr-Nd, and Pb – understanding how and where the s-process elements are produced is very important. However, our understanding is poor, especially with respect to the where, when, and how many of these elements are produced. Abundance surveys of metal-poor dwarf stars in the halo have revealed the presence of very old stars that are extremely rich in s-process elements (Sneden et al. 2008). These high abundances are difficult to explain with stellar evolution and nucleosynthesis theories because dwarf stars on the main-sequence cannot produce these elements. However, if a dwarf star experienced accretion from a more massive giant companion, the peculiar atmospheric abundance of the dwarf star can be explained. In fact, if the giant companion has already experienced dredge-up, bringing s-process elements produced during its AGB phase to the surface, these elements can be transferred onto the dwarf companion (Aoki et al. 2001).

Most neutron-capture elements can be produced by both s- and r-processes, and it is not always easy to constrain the processes that are involved in the creation path(s) of each element. An important work on the production rates for neutron-capture elements was the one from Arlandini et al. (1999). They claim that the s-process production is responsible for 85% of Sr abundance, 83% of Zr, 62% of La, 77% of Ce, 30% of Sm, and only 7% of Eu. Thanks to its very limited s-process component, Eu is considered a “pure” r-process element, and it is well suited to determining the corresponding r-process contribution for other elements (Winckler et al. 2006). For Nd, the production is equally divided between the s- and r-processes (56% produced by s-process). Ever since Arlandini et al. (1999), several works have derived new production rates and yields for the neutron-capture elements, including better cross-section measurements for the reactions and evolutionary model for stars of different metallicities and masses. However, large uncertainties are still present in the s- and r-process calculations (Karakas 2014). At the same time, observations usually show spread in abundances at a given metallicity, and this can be due to the dependence of neutron-capture process on metallicity (Travaglio et al. 2004).

The relative contribution from each of the processes to the abundance of an element and to the various abundance ratios change during the evolution of the Galaxy. This means that a complete overview of the stellar populations at different metallicities is needed for a better understanding of the evolution of these elements. Considering the possible production sites listed above, and considering that the first low-mass stars in the Universe reached their AGB phase about 500 million years after the Big Bang (see, for example, Sneden et al. 2008), the s-process enrichment occurs with some delay with respect to SN II, and they start to explode after a few million years after the onset of star formation. The study of this delay and the way the different elemental abundances relate to each other at different metallicities can help us to understand the evolution of the Galactic disk. For this purpose, dwarf stars are very suitable because they have very long lifetimes, comparable to the age of the Universe (Sackmann et al. 1993), and their surface abundance can be considered to be representative of the chemical composition of the gas cloud they were born in (Lambert 1989; Freeman & Bland-Hawthorn 2002). This means that the information that can be derived from these types of stars is directly related to the enrichment processes that occurred in the previous stellar generations, all the way back to the first billion years of Milky Way history.

In this paper we derive abundances of some neutron-capture elements (Sr, Zr, La, Ce, Nd, Sm, and Eu) in the solar neighbourhood, and in addition to this, in our investigation, we made use of the abundances of Ba derived by Bensby et al. (2014). Even if the method used is not the same (for Ba equivalent width measurements were used), one of the goals of this paper is to analyze elements that were not studied in Bensby et al. (2014).

The paper is organized as follows. In Sect. 2 the stellar sample and the abundance analysis are described. Section 3 give the abundance results for Sr, Zr, La, Ce, Nd, Sm, and Eu. Section 4 discusses possible origins of these neutron-capture elements and their evolution in the Milky Way. Finally, Sect. 6 summarizes our findings.

2. Abundance analysis

2.1. Stellar sample and stellar parameters

The stars in this study are a subset of the 714 F and G dwarf star sample of Bensby et al. (2014), namely those 593 stars observed with the FEROS and MIKE high-resolution spectrographs that have complete wavelength coverage from about 3500 to above 9000 Å. This is important because many of the spectral lines that we use are located in the blue spectral region. The spectra have spectral resolutions between R = 48 000 to 65 000 and the signal-to-noise ratios are generally greater than S/N> 200. Further details are given in Bensby et al. (2014).

Stellar parameters, ages, and elemental abundances for O, Na, Mg, Al, Si, Ca, Ti, Cr, Fe, Ni, Zn, Y, and Ba were determined for all 714 stars in Bensby et al. (2014) where the reader is directed for full details. Briefly, the stellar parameters were determined by requiring excitation balance of abundances from Fe i lines for the effective temperature (Teff), ionization balance between Fe i and Fe ii lines for surface gravity (log g), and abundances from Fe i lines that are independent of reduced line strength to get the microturbulence parameter (ξt). Stellar ages were determined using a grid of α-enhanced Yonsei-Yale isochrones by Demarque et al. (2004) using probability distribution functions as explained in Bensby et al. (2014). In short, the age probability distribution of each star is constructed by considering the errors in effective temperature, surface gravity, and metallicity of that specific star, which permit us to derive the most likely age for the stars (and these ages are the ones used in the paper), together with a lower and higher estimation (used to calculate the errors on the ages). In addition, Sc, V, Mn, and Co abundances were determined in Battistini & Bensby (2015) for part of the sample.

Table 1

Analyzed elements and spectral lines.

Table 2

Elemental abundances for individual lines in the solar spectrum.

2.2. Spectral line synthesis

The spectral lines from the heavy neutron-capture elements are usually located in the blue regions of the visible spectrum. The blue region is very crowded with spectral lines and is especially so for metal-rich disk stars, meaning that blends from other species are frequent. The crowdedness of the blue spectral regions also makes the placement of the continuum difficult. In addition, heavy elements usually consist of several isotopes, and their spectral lines can be affected by hyperfine splitting (hfs). Therefore, to determine abundances from these lines, one has to use spectral line synthesis in order to accurately account for wavelength shifts of the different isotopes, hyperfine components, and other blending features.

Table 1 lists the spectral lines that we use in this study. It shows the analyzed lines with the isotopic ratios of the elements, ionisation stages, wavelengths, and excitation potential, along with reference work where these details were gathered. For Sm and Nd, which suffer in large part from isotopic shifts and also hfs, we use data from Roederer et al. (2008), for La we used hfs-only components from Ivans et al. (2006) since La has only one naturally-occurring isotope, and for Eu we used shift correction for isotopic and hfs substructures from Lawler et al. (2001). However, some of the lines of the elements affected by hfs are treated as single lines because the wavelength splitting is small enough to be ignored. In our linelists the lines treated as single lines are La ii at 4748 Å, Nd ii at 5130 Å, and at 5319 Å, and Sm ii at 4523 Å, 4577 Å, and 4669 Å. All lines from Sr ii, Zr ii, and Ce ii lines are treated as single lines since there is no information available on their hyperfine structure and isotopic shifts or because they do not suffer from any of these problems. Atomic data for blending lines and other nearby lines were taken from the compilation available in the Vienna Atomic Line Database (VALD, Piskunov et al. 1995; Ryabchikova et al. 1997; Kupka et al. 1999, 2000). The complete linelists with hyperfine components can be found in Appendix A.

2.3. Abundance determination

The methodology for abundance determination is the same as in Battistini & Bensby (2015) where we analyzed Sc, V, Mn, and Co for the same sample of stars. It is based on comparisons between observed spectra and synthetic spectra to find the best fitting abundance. The synthetic spectra were calculated with the MARCS2012 code (Gustafsson et al. 2008), under the assumption of local thermodynamic equilibrium (LTE), and one-dimensional plane-parallel model atmospheres.

The synthetic spectra were created using Spectroscopy Made Easy (SME, Valenti & Piskunov 1996; Valenti & Fischer 2005) with stellar parameters from Bensby et al. (2014) as input. The abundance analysis is done through a minimization routine of an unnormalised χ2 function based on the difference between the observed and the synthetic spectra. An example of the comparison between observed and synthetic spectra can be seen in Fig. 1 for the Zr ii at 4208 Å in one of the available solar spectra. In Table 2 the abundances from individual lines are listed for the different solar spectra that we analyzed. The fewer abundances usually available for La occur because the lines are generally weak so more similar to the spectral noise, while for Sr the line is not well synthesized in most cases.

thumbnail Fig. 1

Synthesis of the Zr ii line at 4208 Å in the solar spectrum (in this case Vesta, observed with the MIKE spectrograph in January 2006). The upper panel shows synthetic spectra with different Zr abundances in steps of 0.04 dex (blue lines). The difference between the observed spectrum and the synthetic spectra is plotted in the lower panel with the best abundance spectrum plotted as a red line in both the upper and lower panels. The inset shows the χ2 fit to determine the best abundance value expressed as log ϵ(Zr) (shown as a red dot).

Open with DEXTER

The analysis was done strictly differentially with respect to the Sun. The solar spectra for each observation run is listed in Table 1 in Bensby et al. (2014) and were used to determine abundances for the different lines using the same methodology as for stars in the sample. Bedell et al. (2014) note that the derived abundances from different solar spectra observed with different instruments can differ by up to 0.04 dex. We therefore decided to normalize the abundances on a line-by-line basis, using the correspondent solar spectrum from the different runs, when available. For the observation runs that do not have a solar observation, an average abundance based on abundances from all solar spectra was used. Table 2 gives the abundances we derived for individual lines in the different solar spectra, Figs. B.1B.5 show all line fits in one of the solar spectra, and an example is shown in Fig. 1.

A difficulty in the analysis of these elements is that most of the lines are weak, meaning that even if the signal-to-noise of the spectrum is high, in some cases the noise is comparable to the strength of the lines. In the solar spectrum, these lines show equivalent widths ranging from a few mÅ (Ce at 5187 Å, 4 mÅ, and La at 4748 Å, 5.3 mÅ) to several tens of mÅ (Eu at 4129 Å, 35.6 mÅ, Sr at 4607 Å, 44.4 mÅ, and Zr at 4208Å, 78 mÅ). At the end of the fitting procedure of each spectral line, a visual inspection was therefore performed to evaluate whether the value for the best abundance from each line actually accurately reproduced the shape of the observed line. On a total of 597 stars from the FEROS and MIKE spectra, we have Sr abundances for 156 stars, Zr for 311 stars, La for 242 stars, Ce for 365 stars, Nd for 395 stars, Sm for 280 stars, and Eu for 378 stars. All abundances from individual lines are given in Table 3, while the solar-normalized averaged values are given in Table 4. When results from more than two lines are available, we use median value because it is less affected by outliers.

2.4. Systematic and random error estimation

Performing a differential analysis relative to the Sun means that systematic errors arising from uncertainties in atomic data and the analysis methods largely cancel out. To check for systematic uncertainties further, we search the literature for other studies that have analyzed the same stars to investigate possible offset in stellar parameters and derived abundances. Table 5 lists the differences in stellar parameters and abundances for the stars in common with Reddy et al. (2003, 2006), Mashonkina et al. (2004), Mashonkina et al. (2007) and Mishenina et al. (2013). Among the different works, we share the highest number of stars with neutron-capture abundances determination with Reddy et al. (2006), while for the other works such as Reddy et al. (2003) and Mishenina et al. (2013), we only have one or two stars in common, giving no real information on possible offsets. Only the comparison with Reddy et al. (2006) contains enough stars (more than ten) to show that there is good agreement in this case.

Table 3

Abundance from single lines.

Table 4

Stellar parameters and abundance results for the entire sample.

Table 5

Comparisons of stellar parameters and abundances for stars in common with other studies.

thumbnail Fig. 2

Elemental abundance trends with Fe as the reference element. The number of stars for which the abundance has been derived is indicated in each plot.

Open with DEXTER

The estimation of the random errors for our abundances was done by deriving how much the errors in the stellar parameters would affect the final abundances. We followed the same procedure as in Battistini & Bensby (2015), selecting a random subsample of stars that probes different part of the stellar parameters space and calculates the new abundances when the random errors from Bensby et al. (2014) are applied to each star. First we calculated the difference between the abundances without and with the errors on stellar parameters applied, then all the differences for each element were used to calculate the final square mean error. We considered an error on the abundance determination of 0.05 dex as wrong continuum placement and as imperfectly fit lines. The abundance errors, together with the mean standard random errors on stellar parameters, are listed in Table 6.

As is visible in Table 6, the error on Eu is the smallest between the elements we analyzed and is comparable to the errors on [Fe/H]. On average, our errors for the elements are around 0.1 dex but are higher for Sr, 0.15 dex to be precise, probably because Sr presents high spread (as explained in Sect. 3).

3. Results

In this section we present the abundance results for the neutron-capture elements that we analyzed, and Fig. 2 shows the [X/Fe] versus [Fe/H] abundance trends.

3.1. Strontium

Strontium in Fig. 2a shows increasing [Sr/Fe] abundance ratios with decreasing metallicity: [Sr/Fe] ≈ −0.2 for solar metallicity stars reaching [Sr/Fe] ≈ 0 at [Fe/H] ≈ −1. There are a few stars that show high [Sr/Fe] ratios, which could be because only one Sr line was used. A similar spread is found in giant stars from Burris et al. (2000) and dwarf stars from Mashonkina & Gehren (2001), Reddy et al. (2003), and (Brewer & Carney 2006, even though they derived abundances from Sr ii lines) for stars in a metallicity range similar to ours (−2.0 ≲ [Fe/H] ≲ 0.2), where the trend is on average solar. [Sr/Fe] for very metal-poor stars ([Fe/H] < −2.5), as for giants in Burris et al. (2000) and Andrievsky et al. (2011), presents a downturn. We notice that Andrievsky et al. (2011) used non-LTE (NLTE) values for Sr, but the same trend is still present even when compared to the LTE case. In addition to literature results, more recently, Ishigaki et al. (2013) observed stars in thick disk and halo and found solar [Sr/Fe] ratios for [Fe/H] ≲ −1. In addition to the literature results, more recently, Ishigaki et al. (2013) have observed stars in thick disk and halo and found solar [Sr/Fe] ratios for [Fe/H] ≲ −1.

Table 6

Random errors in stellar parameters and abundances.

3.2. Zirconium

Zirconium in Fig. 2b shows an increasing trend with decreasing metallicity, from [Zr/Fe] ≈ −0.1 at [Fe/H] ≈ 0.3 to [Zr/Fe] ≈ 0.3 for the most metal-poor stars in the sample at [Fe/H] ≈ −1. Burris et al. (2000) found a [Zr/Fe] trend that seems to be only slightly super-solar with a wide spread for stars with [Fe/H] < −1.5, and Brewer & Carney (2006) found basically solar [Zr/Fe] values for all their stars. The same was seen by Reddy et al. (2006) even if there seems to be a small trend from slightly super-solar [Zr/Fe] at solar metallicity to slightly subsolar [Zr/Fe] for [Fe/H] < −0.6. On the other hand, the trend found by Mashonkina et al. (2007) is similar to what we find and has an increasing trend from a solar [Zr/Fe] value for solar metallicity stars to [Zr/Fe] ≈ 0.3 for [Fe/H] ≈ −1.5. Recently, Mishenina et al. (2013) have observed F, G, and K dwarf stars in the Galactic disk and find a similar trend to ours, even though the stars at super-solar metallicities show solar or slightly enhanced [Zr/Fe] values. Also, Ishigaki et al. (2013) studied Zr, finding a similar [Zr/Fe] trend to ours for their thick disk stars, while for halo stars there is a decreasing trend toward negative [Zr/Fe] values at low metallicities.

3.3. Lanthanum

As shown in Fig. 2c, La presents mostly solar [La/Fe] values, with a possible trend from super-solar metallicity with [La/Fe] ≈ −0.3 to roughly solar [La/Fe] for [Fe/H] ≈ −0.6. Burris et al. (2000) observed average solar [La/Fe] for [Fe/H] ≲ −1 with increasing spread as metallicity decreases. Simmerer et al. (2004) derived abundances of La for giants and dwarfs in the metallicity range 3 < [Fe/H] < 0.3: for stars with the metallicity as in our sample, the agreement is good even if an offset of 0.2 dex seems to be present. Brewer & Carney (2006) found a trend similar to what we found, with subsolar [La/Fe] for solar metallicity stars that then increases to reach around solar [La/Fe] for [Fe/H] ≈ −0.6. More recently, Mishenina et al. (2013) have results in agreement with our trend, as well as with the thick disk sample of Ishigaki et al. (2013).

3.4. Cerium

In Fig. 2d Ce presents a basically flat and slightly subsolar [Ce/Fe] abundance trend. This behavior with [Ce/Fe] ≈ 0 is observed in Reddy et al. (2006), while a flat trend that is slightly super-solar is found in Brewer & Carney (2006). However, in Mashonkina et al. (2007), the [Ce/Fe] trend is clearly increasing as metallicity decreases. Similar to our result is that of Mishenina et al. (2013), where for solar metallicity and sub-olar metallicities, [Ce/Fe] shows a wide spread around the solar [Ce/Fe] value.

3.5. Neodymium

Neodymium shows a tighter trend compared to the previous elements, as is visible in Fig. 2e. The trend presents similarities to Zr, with subsolar [Nd/Fe] for super-solar metallicity stars that then rises to [Nd/Fe] ≈ 0.2 at [Fe/H] ≈ −0.6, and for even lower metallicity, the trend seems to be flat. Burris et al. (2000) present a significant spread in [Nd/Fe], while Mashonkina et al. (2004) find a tight trend similar to what we found, with increasing [Nd/Fe] for decreasing metallicity up to [Nd/Fe] ≈ 0.3 for [Fe/H] ≈ −1 and then a plateau down to lower metallicity. The same general trend was found in Brewer & Carney (2006), while on the other hand, the trend found in Reddy et al. (2006) is basically flat with [Nd/Fe] ≈ 0. The result from Mishenina et al. (2013) is similar to ours, while Ishigaki et al. (2013) find a flat trend with [Nd/Fe] = 0 for thick disk stars and [Nd/Fe] = 1 at [Fe/H] = −0.6, which decrease to reach solar [Nd/Fe] for [Fe/H] −3 for halo stars.

3.6. Samarium

In Fig. 2f it is possible to distinguish a rising trend from solar [Sm/Fe] at solar metallicity up to [Sm/Fe] ≈ 0.5 at [Fe/H] ≈ −1. Some of the previous listed works studied Sm as well; for example, Mishenina et al. (2013) find a similar trend but with an offset of 0.2 dex, meaning that solar and super-solar metallicity stars have [Sm/Fe] ≈ −0.2. Also Ishigaki et al. (2013) studied [Sm/Fe] , finding a basically solar value.

3.7. Europium

Europium (Fig. 2g) clearly shows a typical α-element trend as is expected for an almost pure r-process element, since rapid neutron capture is believed to occur in SN II. Prochaska et al. (2000) derived Eu abundances for four stars in the metallicity range −0.7 < [Fe/H] < −0.4, showing high [Eu/Fe]. Burris et al. (2000) shows rising [Eu/Fe] as in Mashonkina & Gehren (2001) and Koch & Edvardsson (2002), where Eu shows a rising trend as we observed, with a steady increase from [Eu/Fe] ≈ −0.2 at [Fe/H] ≈ 0 up to [Eu/Fe] ≈ 0.4 for [Fe/H] ≈ −1.0. Good agreement can be found in Simmerer et al. (2004), Bensby et al. (2005), Brewer & Carney (2006), and in Reddy et al. (2006) even if in this case the spread is higher. The same trend with good agreement in values and shape is found in Mishenina et al. (2013), while in Ishigaki et al. (2013) the same trend is shifted to lower metallicity, as [Eu/Fe] ≈ 0 for [Fe/H] ≈ −1, and then [Eu/Fe] rises to + 0.2 at [Fe/H] ≈ −1.8.

4. Origin of s- and r-elements

Most of the neutron-capture elements are produced from a mixture of r- and s-processes, and only in a few cases is only one of the two processes is mainly responsible for the production. The comparison between such “prototype” elements of either the r- or the s-process and other neutron-capture elements can help to constrain the production sites for the neutron-capture elements with uncertain origins. Barium, for instance, is often used in comparisons between neutron-capture elements owing to its high s-process component. For example, Arlandini et al. (1999) derived the contributions of the s- and r-process to the solar Ba abundance, which were 81% and 19%, respectively. However, since the work by Arlandini et al. (1999), several studies have been published with updated r- and s-process rates. For our comparisons, we use the values from Bisterzo et al. (2014).

The Ba abundances that we use in our study are derived with equivalent width from Bensby et al. (2014). Since Ba is known to suffer from NLTE for Teff> 6100 (i.e., Korotin et al. 2011), and for this reason, we discarded the stars with high temperature in the comparisons with Ba. In Fig. 16 in Bensby et al. (2014), the trend of [Ba/Fe] is visible as a function of [Fe/H], together with the typical error on [Ba/Fe] of about 0.1 dex. Once the stars with high temperature are removed, the [Ba/Fe] trend is basically flat.

Several studies of the abundance structure in the Galactic disk have revealed the presence of two different stellar components that differ in age, kinematics, and α-element abundances. The thin disk contains mostly young and kinematically cold stars with low α-abundance for a given metallicity while, on the other hand, the thick disk contains mostly old and kinematically hot stars with high α-abundance for any given metallicity (Fuhrmann 1998; Bensby et al. 2003, 2005; Reddy et al. 2003, 2006; Adibekyan et al. 2012; Bensby et al. 2014). As suggested by Haywood et al. (2013) and Bensby et al. (2014), stellar ages seem to be a better separator between thin and thick disk stellar populations than kinematic properties, that largely overlap between the two. Following Bensby et al. (2014), we considered stars younger than 7 Gyr as thin disk stars and stars older than 9 Gyr as thick disk stars. In all the following figures, if not specified otherwise, black dots are thin disk stars, while white dots are thick disk stars as divided according to our criterion.

thumbnail Fig. 3

[Eu/Ba] as a function of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents a pure r-process ratio derived from Bisterzo et al. (2014). The average error is also presented.

Open with DEXTER

4.1. Eu

The comparison between Ba and Eu can be used as a diagnostic of the neutron-capture process (Mashonkina & Gehren 2001) thanks to the different production of the two elements: Eu almost completely r-process, Ba mostly s-process. In Fig. 3 the [Eu/Ba] ratio is close to the pure r-process line for metal-poor stars, meaning that the r-process was the only neutron-capture process active at the beginning of the formation of the Milky Way. As mentioned before, Eu is almost completely produced by the r-process, but Fig. 3 shows that Ba was also initially produced in this way, as already found, for example, by Burris et al. (2000). As soon as AGB stars start to be present and enrich the ISM with s-process elements, the [Eu/Ba] ratio decreases until it reaches the solar value at solar metallicity. Comparing [Zr/Ba] in Fig. 4 with [Eu/Ba], it is possible to see that the rise in [Eu/Ba] is steeper owing to the almost complete production of Eu by r-process, meaning that as soon as enrichment from AGB stars becomes predominant, the [Eu/Ba] ratio decreases quickly.

4.2. Sr and Zr

Barium can also be compared to Sr and Zr abundances to diagnose the processes that formed the Sr-Y-Zr peak elements (Travaglio et al. 2004). Figures 4a, b show the trends for [Sr/Ba] and [Zr/Ba] as a function of [Fe/H]. Even if the number of stars in the Sr case is lower than for Zr, the increases in [Sr/Ba] and [Zr/Ba] with decreasing metallicity are quite similar. Compared to [Eu/Ba], [Zr/Ba] shows less steep a rise with decreasing metallicity, probably because Zr has a higher s-process component than Eu. In addition to this, flat [Zr/Ba] and [Sr/Ba] trends are seen for stars around solar metallicity. Moreover, the abundances for thick disk stars at low metallicity is particularly high, similar to what is visible in Fig. 3 for Eu, and is difficult to explain only with the 15% and 20% of r-production of Sr and Zr, respectively, from Arlandini et al. (1999). On the other hand, a shift of 0.2 dex is visible between [Sr/Ba] and [Zr/Ba].

thumbnail Fig. 4

[Sr/Ba] and [Zr/Ba] (panels a) and b), respectively) and [Sr/Eu] and [Zr/Eu] (panels c) and d), respectively) as a function of [Fe/H]. The full sample is divided in thin and thick disk according to our age-selection criterion. The dotted line represent pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER

thumbnail Fig. 5

[La/Ba] and [Ce/Ba] (panels a) and b), respectively) and [La/Eu] and [Ce/Eu] (panels c) and d), respectively) as functions of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents the pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER

thumbnail Fig. 6

[Nd/Ba] and [Sm/Ba] (panels a) and b), respectively) and [Nd/Eu] and [Sm/Eu] (panels c) and d), respectively) as functions of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents the pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER

thumbnail Fig. 7

[X/Fe] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disk. Blue dots represent young thin disk stars, while red dots are for old thick disk stars. The blue and red lines indicate the best fit for thin and thick disk stars. The errors on the ages are from Bensby et al. (2014). The average error on the abundance ratio is indicated in black.

Open with DEXTER

Travaglio et al. (2004) compared their galactic chemical evolution model with observations down to [Fe/H] ≈ −4. The model consists of Sr produced of 71% from low-intermediate mass AGB stars (18 M) defined as main s-process and for 9% from advanced evolution phases of massive stars defined as weak s-process. For Zr they derive a 65% main s-process contribution and an almost negligible contribution from the weak s-process of about 2%. From their results, Travaglio et al. (2004) claim r-process contributions of 20% and 30% that is higher than derived by Arlandini et al. (1999). However, in their comparison with r-process rich and very metal-poor stars, they derive that pure r-process production for Sr and Zr is 10% (more precisely 12% for Sr and 15% for Zr). Summing up the last derived r-contribution with the s-contribution mentioned above, it can be noticed that 8% of Sr and 18% of Zr is missing, and it has a “primary” origin from massive stars at low metallicity. Unfortunately, the real process for this LEPP (lighter element primary process) is still not clear because detailed supernovae model calculations for massive stars at low metallicity are still not available (Travaglio et al. 2004). More recently, Bisterzo et al. (2014) have recalculated the contribution for Sr and Zr and found similar results as in Travaglio et al. (2004), still requiring LEPP as well. The different LEPP contribution for Sr and Zr could be the explanation of the 0.2 dex between Sr and Zr, where the higher plateau of [Zr/Ba] could be explained by the higher contribution from LEPP, i.e., more enrichment of Zr from high massive stars at low metallicity.

The comparison with the Travaglio et al. (2004) model, however, is unable to match our data since it predicts a solar [Sr/Ba] and [Zr/Ba] down to [Fe/H] ≈ −1. A possible explanation for this discrepancy can come from the large uncertainties that still exist for the yields of heavy elements from AGB stars owing to missing models for stars in the range 18 M at different metallicities (Karakas 2014).

thumbnail Fig. 8

[Sr/Zr], [La/Ce], [Nd/Sm], and [Eu/Ba] compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The black line in each plot represents the best fit. The errors on the ages are taken from Bensby et al. (2014). The average error on the abundance ratio is indicated in the lower left part of each plot.

Open with DEXTER

4.3. La, Ce, Nd, and Sm

Figures 5a-d show [La/Ba], [Ce/Ba], [La/Eu], and [Ce/Eu] versus [Fe/H]. La and Ce are part of the second peak of the magic neutron number 82, together with Sm and Eu. From Figs. 2c and 2d, La and Ce do not show any particular trend with metallicity, and this is also reflected in the upper panels of Figs. 5. In the lower panels of Fig. 5, [La/Eu] and [Ce/Eu] are plotted over [Fe/H] to evaluate their small r-process enrichment branch, and the trends look similar. Compared to Sr and Zr, for example, it is possible to see that the increases in [La/Eu] and [Ce/Eu] are fast and occur when AGB stars start to contribute. From Fig. 5d a change in slope is visible at [Fe/H] 0.5. This could be the moment where the enrichment of ISM from AGB stars starts to dominate, represented as a steady increase in [Ce/Eu] as [Fe/H] increases toward solar values.

A similar investigation was performed for Nd, which is produced in almost equal parts by s- and r-processes. In both panels of Fig. 6, Nd shows smooth trends, with high Nd abundance for metal-poor stars that decreases as metallicity increases. Interestingly, the most metal-poor stars in both panels are close to the theoretical abundance ratio for the r-process derived using Bisterzo et al. (2014) values. Very similar results are derived in the Nd work by Mashonkina et al. (2004). The decrease that is present in [Nd/Ba] is due to the higher rate of s-production of Ba compared to Nd (81% for Ba and 56% for Nd), while the decrease in [Nd/Eu] is produced by the higher rate (93%) of r-production for Eu compared to Nd. Similar results can be found for Sm thanks to its high percentage of r-process production. In Fig. 6a, thin disk stars have on average smaller [Nd/Ba] compared to thick disk stars, and are more concentrated along solar [Nd/Ba] values. This means that during thick disk formation and evolution, the r-process contribution was higher than in the thin disk, probably because the thick disk formed in a rapid way with a high star formation rate that produced more massive stars responsible for r-process enrichment.

We also investigated [Sm/Ba] and [Sm/Eu] versus [Fe/H], and the results can be seen in Figs. 6b and d. From Fig. 6b it is clear that the most metal-poor stars in our sample are close to the pure solar r-process, meaning that r-process was the main active channel for the production of Sm. This is expected, considering that Sm is produced for almost 70% via rapid neutron capture. Since Eu is an r-process, it is interesting to investigate [Sm/Eu] (Fig. 6d). The high percentage of production via rapid neutron capture for both Sm and Eu is responsible for the [Sm/Eu] ≈ 0 until solar metallicity. For super-solar metallicity, [Sm/Eu] seems to be on average higher than solar value, probably because in this metallicity regime, the s-process production of Sm is more important.

The formation of the elements La, Ce, Nd, and Eu was studied by Travaglio et al. (1999), and it is interesting since these elements are part of the magic neutron number N = 82. Travaglio et al. (1999) found that the main s-process contribution to these elements comes from AGB stars in the range 24 M while the r-process contribution is mainly due to stars in the rage 810 M. In their calculation, AGB contribution starts to be important for [Fe/H] ≳ −1.5, meaning that after this point, the Eu abundance decreases while the abundances for the other elements increase, especially for Ce and La because of their larger s-process contribution.

The models of Travaglio et al. (1999) agree with our data, as can be seen by comparing our Figs. 2cf with their Figs. 7, 8, 10, and 11. Their model of Eu enrichment agrees with what we found in Fig. 2g, and it can be explained by r-process derived from SN II from 810 M stars.

thumbnail Fig. 9

[X/H] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disks. Blue dots represent young thin disk stars, while red dots indicate old thick disk stars. The errors on the ages are from Bensby et al. (2014). The average error on the abundance ratio is indicated in black.

Open with DEXTER

thumbnail Fig. 10

[X/H] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disks. Blue dots represent thin disk stars with [Ti/Fe] < 0.2, while red dots indicate thick disk stars identified with [Ti/Fe] > 0.2. The average error on the abundance ratio is indicated in black.

Open with DEXTER

5. Abundance evolution with age

Since the r- and s-processes happen on a different timescale in Galactic evolution, we now investigate how the [X/Fe] abundance ratios vary with time. Only stars with stellar ages with uncertainties less than 3 Gyr are used (derived as the maximum age estimation minus the minimum one). As before, stars with ages less than 7 Gyr are considered to be part of the thin disk, while stars older than 9 Gyr are part of the thick disk. We stress here that the stellar ages are really accurate in relative terms, meaning that the relative age differences between stars is good.

The results are presented in Fig. 7. The first thing that can be noticed is that in most of the cases, there is a change in the slopes at an age of about 8 Gyr, which happens to coincide with the interface between the thin and thick disks. For the thick disk (stars older than 8 Gyr), the indication that we get from Fig. 7 is that enrichment of neutron-capture elements was higher at the early times of the Milky Way, so the r-process could produce large quantities of neutron-capture elements, even if only thanks to stars with masses in the 810 M range. This could be explained assuming an intense and rapid star formation in the process of building up of the thick disk that ended around 8 Gyr ago. The decrease is then due to a reduction of r-process production with, at the same time, the beginnings of s-process production. The basically flat [X/Fe] in thin disk for La, Ce, Nd, Sm, and Eu can be explained as a combination by the contribution from low- and intermediate mass stars in AGB phase becoming more important, but this contribution is similar to Fe production from SN Ia, meaning a flattening in the ratio. In addition to this, since the s-process in the AGB phase requires seed nuclei of iron-group elements, the increase in metallicity of younger stars produces an increase in the yields of neutron-capture elements.

Interestingly, for Sr and Zr we see in Figs. 7a and b that abundances increase for younger thin disk stars. This phenomenon can be explained, as before, by the fact that the s-process is what is mainly responsible for enrichment in the thin disk, but with the addition that these two elements present the highest s-process rate among the other neutron-capture elements in this study (70% for Sr and 65% for Zr, as derived by Bisterzo et al. 2014). Unfortunately, at this moment we cannot explain the high spread in [Zr/Fe] for the very old stars.

Considering the results in Figs. 46, neutron-capture elements can be paired because they seem to share similar properties. For this reason we investigated how the abundance ratios of these element pairs change at different ages. The results are shown in Fig. 8 with the addition of the [Eu/Ba] ratio.

As expected for elements that share the same production sites, the trends are almost flat within the uncertainties. For [Sr/Zr] in Fig. 8a, there are indications of higher Zr production at an early stage than Sr. A possible explanation could be that Zr is said to have higher r-process production (15% from classical r-process plus 18% from LEPP, Travaglio et al. 2004) that will produce more Zr in the early stages of the evolution of the Galaxy. The [La/Ce] ratio in Fig. 8b presents a flat trend, indicating shared sites and mechanisms of production. For [Nd/Sm] in Fig. 8c, the trend is basically flat, which in this case also indicates shared production sites.

The trend of [Eu/Ba] can be explained in a different way. Eu and Ba have clearly different histories of productions and are paired here because they are typical representatives of the r- and s-processes, respectively. In Fig. 8d the decrease in [Eu/Ba] toward younger ages can be explained as an increase in production of Ba and a simultaneous decrease in Eu, because AGB stars become more important than SN II after about 1 Gyr from the bulk of star formation.

We also investigate the trends in [X/H] with respect to age. This kind of investigation is not present in previous works since the age determination was not good enough. Results are presented in Fig. 9, using the same color coding as in Fig. 7, where stars with an age greater than 9 Gyr are considered as thick disk and stars younger than 7 Gyr as thin disk. It is possible to see that in all the elements, apart from Sr (probably due to the big scatter and the lower number of stars with a successful analysis), the trend is very similar. We checked for possible correlations with stellar parameters that could explain the trend in Fig. 9 but found none.

From older ages there is a steep increase that ends at 11 Gyr, which is probably due to the production of these elements via SN II. This rise is then followed by a flatter part that in some cases ends in an increase in [X/H] for younger stars. This final increase can be explained by considering the contribution from low-mass stars that polluted the gas from where the youngest stars were formed. However, in the case of [Eu/H], this is more of a problem for explaining the results because Eu is almost completely produced via the r-process, meaning that the contribution from low-mass stars should be almost negligible.

In Fig. 15 in Bensby et al. (2014), it is possible to see the different abundance trends for different α-elements. In particular, Ti shows a clearer distinction between thin and thick disk compared to the other α-elements, so we decided to use Ti to distinguish thin and thick disk stars in the [X/H] versus age comparison. We considered stars to be thin disk stars when [Ti/Fe] < 0.2, while others are thick disk stars when [Ti/Fe] > 0.2. The results are visible in Fig. 10 for all the elements under investigation. Basically, all thick disk stars are located in the old age region and are responsible for the increase that is visible in Fig. 9 up to 11 Gyr. All thin disk stars are instead responsible for the flat and the rising part at a young age.

6. Summary

We performed a detailed chemical abundance analysis on a sample of F and G dwarf stars for several neutron-capture elements Sr, Zr, La, Ce, Nd, Sm, and Eu in order to investigate their formation sites and their evolution in the Galactic disk. There are several works that focus on metal-poor stars because of their interesting highly enriched spectra, while for stars with [Fe/H] > −1, the samples are usually smaller, and not all the elements can be investigated because of the spectra that become extremely rich. In total we determined the Sr abundance for 156 stars, Zr abundance for 311 stars, La abundance for 242 stars, Ce abundance for 365 stars, Nd abundance for 395 stars, Sm abundance for 280 stars, and Eu abundance for 378 stars. Our findings and conclusions are summarized as follows.

  • Strontium and zirconium are part of the same s-process peak production, and they are intensively studied in theoretical works because of their position in the valley of stability owing to the magic neutron number A = 50. They show similar abundance trends with a common flat part around solar metallicity. For Sr, the spread in thin and thick disks is too high to draw any conclusion. For Zr instead, the thin disk stars are grouped close to solar abundance values, indicating a constant production in time, thanks to production in AGB stars in the mass range 18 M. Models from Travaglio et al. (2004) cannot explain our Sr trend, since it is expected to be basically flat around solar [Sr/Fe] down to [Fe/H] ≈ −2. On the other hand, the same models can fit our Zr data, even if we observe subsolar [Zr/Fe] at solar metallicity. The discrepancies could be due to uncertainties in the model on yields for stars in the 18 M range at different metallicities.

  • Lanthanum and cerium are basically flat with solar [La,Ce/Fe]. This flat trend at solar value is basically conserved also when La and Ce are compared to Ba, a typical s-process element. When compared to Eu, some thick disk stars show pure r-process abundance, and it is possible to see a turn in [Ce/Eu] when the s-process become the more important enrichment process at [Fe/H] ≈ −0.5. In Travaglio et al. (1999), r-process La and Ce productions come from SN II from stars of 810 M, while s-process production comes from stars in the range 24 M. This creates a gap in the production of La and Ce and then a fast decrease as soon as AGB stars are actively involved in chemical enrichment, since Eu is basically not produced via this channel.

  • Neodymium and samarium are produced via both the s- and r-processes. For Nd the two processes are almost equally responsible for its enrichment, while for Sm, the r-process is the main productive channel (70%). Even if they are produced in different ways, they share a similar trend as an α-element derived by production from SN II. Considering the results from Travaglio et al. (1999) on different AGB sites for Nd, Sm, and Ba, the smooth decrease in [Nd/Ba] occurs because Nd has a very high r-process production rate compared to Ba, but at solar metallicities, the flat [Nd/Ba] is due to the common production from stars in the mass range 24 M. Samarium, on the other hand, has a higher r-process production rate that makes it more similar to Eu, so [Sm/Eu] is almost flat with values around [Sm/Eu] = 0.

  • Europium is the prototype of the r-process element since it is almost completely produced by rapid neutron capture. It shows a very clean α-trend when compared to metallicity, as do Nd and Sm. This is expected since its high r-process production rate and both thin and thick disk stars present the same [Eu/Ba] decrease. Our data agree with the model of Travaglio et al. (1999) that considers an enrichment from SN II of mass 810 M.

  • The study of abundances as a function of age shows different trends for thin and thick disks in almost all the elements. This is especially true for Sr, Zr, Nd, and Eu. For a thick disk, there is a decrease in the abundance as age decreases that can be explained by the decrease in SN II events, while Fe production from SN Ia increases. In the thin disk, most of the trends are basically flat because of the high production of neutron-capture elements via s-process from AGB stars. This is not the case for Sr and Zr, which show increasing abundances for younger stars owing to the high s-production rate, as derived by Bisterzo et al. (2014). [Sr/Zr] shows an increasing trend with decreasing ages, probably due to higher Zr r-process production. The same production sites produce flat [La/Ce] and [Nd/Sm]. When [Eu/Ba] is related to age, the trend shows a decrease with lower ages, due to the decrease in Eu production by SN II counterbalance by the increase in production by AGB stars of Ba. When the [X/H] ratio is plotted as a function of the age, it is clear that around 11 Gyr, the trend changes clearly in almost all the elements, showing a flatter trend after a rise from older ages up to 6 Gyr, followed in some cases by a clear increase in [X/H] at recent ages. This is particularly relevant in the Eu case, and the reason is not completely clear. In theory younger stars should not present such high abundance in Eu, since Eu is almost completely produced in SN II at the beginning of star formation. In theory if these stars are coming from an inner region of the Galaxy where the metallicity is higher compared to the solar neighborhood, the higher [Eu/H] could potentially be explained even at younger stellar ages. This possible explanation needs to be studied, however, in more detail, also taking the orbit of the stars and possible membership with dynamical streams into account (for example, Hercules streams, since it was studied in Bensby et al. 2014).

Acknowledgments

We would like to thank the referee Prof. Chris Sneden for the useful comments and suggestions for the improvement of this paper. T.B. was supported by the project grant “The New Milky Way” from the Knut and Alice Wallenberg Foundation.

References

Appendix A: Complete linelists

In this section we listed the hfs components for the lines that suffer from hyperfine splitting.

Table A.1

Linelists for the synthesis of Sr.

Table A.2

Linelists for the synthesis of Zr.

Table A.3

Linelists for the synthesis of La.

Table A.4

Linelists for the synthesis of Ce.

Table A.5

Linelists for the synthesis of Nd.

Table A.6

Linelists for the synthesis of Sm.

Table A.7

Linelists for the synthesis of Eu ii.

Appendix B: Solar abundance plots

Figures B.1B.5 show the synthesis for all the lines analyzed in this work for the solar spectrum from Vesta observed at Magellan in January 2006. Each plot shows (in the bigger panel) the line fitting with the best fit to the observed one and the different abundances in steps of 0.04 dex. In the lower panel, the differences between the observed spectrum and the synthetic one are shown, with the difference with the best fit highlighted. In the small square panel, the χ2 values for the different abundances are represented, with the red dot representing the minimum, hence the best fit value for the elemental abundance.

thumbnail Fig. B.1

Solar spectrum taken using asteroid Vesta during the run at Magellan in January 2006. Spectral lines are listed in order of element and wavelength. Chemical elements with all the hfs lines are indicated inside the lines. The different colored lines represent the different synthetic spectra with different abundances in steps of 0.04 dex, while the dots are the observed spectra. The lower panels show the values of differences between the real and synthetic spectra for the synthetic spectra plotted above. The red lines represent the best fit derived from un-normalized χ2, visible in the small plot as a red dot. Here the fits for the Sr line and the five Zr lines are shown.

Open with DEXTER

thumbnail Fig. B.2

As in Fig. B.1 but for the four La lines.

Open with DEXTER

thumbnail Fig. B.3

As in Fig. B.1 but for the four Ce lines.

Open with DEXTER

thumbnail Fig. B.4

As in Fig. B.1 but for the five Nd lines.

Open with DEXTER

thumbnail Fig. B.5

As in Fig. B.1 but for the four Sm lines and the two Eu lines.

Open with DEXTER

All Tables

Table 1

Analyzed elements and spectral lines.

Table 2

Elemental abundances for individual lines in the solar spectrum.

Table 3

Abundance from single lines.

Table 4

Stellar parameters and abundance results for the entire sample.

Table 5

Comparisons of stellar parameters and abundances for stars in common with other studies.

Table 6

Random errors in stellar parameters and abundances.

Table A.1

Linelists for the synthesis of Sr.

Table A.2

Linelists for the synthesis of Zr.

Table A.3

Linelists for the synthesis of La.

Table A.4

Linelists for the synthesis of Ce.

Table A.5

Linelists for the synthesis of Nd.

Table A.6

Linelists for the synthesis of Sm.

Table A.7

Linelists for the synthesis of Eu ii.

All Figures

thumbnail Fig. 1

Synthesis of the Zr ii line at 4208 Å in the solar spectrum (in this case Vesta, observed with the MIKE spectrograph in January 2006). The upper panel shows synthetic spectra with different Zr abundances in steps of 0.04 dex (blue lines). The difference between the observed spectrum and the synthetic spectra is plotted in the lower panel with the best abundance spectrum plotted as a red line in both the upper and lower panels. The inset shows the χ2 fit to determine the best abundance value expressed as log ϵ(Zr) (shown as a red dot).

Open with DEXTER
In the text
thumbnail Fig. 2

Elemental abundance trends with Fe as the reference element. The number of stars for which the abundance has been derived is indicated in each plot.

Open with DEXTER
In the text
thumbnail Fig. 3

[Eu/Ba] as a function of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents a pure r-process ratio derived from Bisterzo et al. (2014). The average error is also presented.

Open with DEXTER
In the text
thumbnail Fig. 4

[Sr/Ba] and [Zr/Ba] (panels a) and b), respectively) and [Sr/Eu] and [Zr/Eu] (panels c) and d), respectively) as a function of [Fe/H]. The full sample is divided in thin and thick disk according to our age-selection criterion. The dotted line represent pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER
In the text
thumbnail Fig. 5

[La/Ba] and [Ce/Ba] (panels a) and b), respectively) and [La/Eu] and [Ce/Eu] (panels c) and d), respectively) as functions of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents the pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER
In the text
thumbnail Fig. 6

[Nd/Ba] and [Sm/Ba] (panels a) and b), respectively) and [Nd/Eu] and [Sm/Eu] (panels c) and d), respectively) as functions of [Fe/H]. The full sample is divided into thin and thick disks according to our age-selection criterion. The dotted line represents the pure r-process ratio derived from Bisterzo et al. (2014). The average error is also indicated.

Open with DEXTER
In the text
thumbnail Fig. 7

[X/Fe] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disk. Blue dots represent young thin disk stars, while red dots are for old thick disk stars. The blue and red lines indicate the best fit for thin and thick disk stars. The errors on the ages are from Bensby et al. (2014). The average error on the abundance ratio is indicated in black.

Open with DEXTER
In the text
thumbnail Fig. 8

[Sr/Zr], [La/Ce], [Nd/Sm], and [Eu/Ba] compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The black line in each plot represents the best fit. The errors on the ages are taken from Bensby et al. (2014). The average error on the abundance ratio is indicated in the lower left part of each plot.

Open with DEXTER
In the text
thumbnail Fig. 9

[X/H] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disks. Blue dots represent young thin disk stars, while red dots indicate old thick disk stars. The errors on the ages are from Bensby et al. (2014). The average error on the abundance ratio is indicated in black.

Open with DEXTER
In the text
thumbnail Fig. 10

[X/H] for neutron-capture elements compared to age. Only stars with age uncertainties less than 3 Gyr are plotted. The vertical dotted line at 8 Gyr indicates an approximate age separation between thin and thick disks. Blue dots represent thin disk stars with [Ti/Fe] < 0.2, while red dots indicate thick disk stars identified with [Ti/Fe] > 0.2. The average error on the abundance ratio is indicated in black.

Open with DEXTER
In the text
thumbnail Fig. B.1

Solar spectrum taken using asteroid Vesta during the run at Magellan in January 2006. Spectral lines are listed in order of element and wavelength. Chemical elements with all the hfs lines are indicated inside the lines. The different colored lines represent the different synthetic spectra with different abundances in steps of 0.04 dex, while the dots are the observed spectra. The lower panels show the values of differences between the real and synthetic spectra for the synthetic spectra plotted above. The red lines represent the best fit derived from un-normalized χ2, visible in the small plot as a red dot. Here the fits for the Sr line and the five Zr lines are shown.

Open with DEXTER
In the text
thumbnail Fig. B.2

As in Fig. B.1 but for the four La lines.

Open with DEXTER
In the text
thumbnail Fig. B.3

As in Fig. B.1 but for the four Ce lines.

Open with DEXTER
In the text
thumbnail Fig. B.4

As in Fig. B.1 but for the five Nd lines.

Open with DEXTER
In the text
thumbnail Fig. B.5

As in Fig. B.1 but for the four Sm lines and the two Eu lines.

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.