First Stars VIII -- Enrichment of the neutron-capture elements in the early Galaxy
P. Francois, E. Depagne, V. Hill, M. Spite, F. Spite, B. Plez, T. C. Beers, J. Andersen, G. James, B. Barbuy, R. Cayrel, P. Bonifacio, P. Molaro, B. Nordström, F. Primas
aa r X i v : . [ a s t r o - ph ] S e p Astronomy & Astrophysics manuscript no. 7706 c (cid:13)
ESO 2018October 26, 2018
First Stars VIII – Enrichment of the neutron-capture elements inthe early Galaxy. ⋆ P. Fran¸cois , , E. Depagne , , V. Hill , M. Spite , F. Spite , B. Plez , T. C. Beers , J. Andersen , , G.James , , B. Barbuy , R. Cayrel , P. Bonifacio , , P. Molaro , B. Nordstr¨om and F. Primas GEPI, Observatoire de Paris-Meudon, CNRS, Univ. de Paris Diderot, Place Jules Janssen, F-92190 Meudon, France GRAAL, Universit´e de Montpellier II, F-34095 Montpellier Cedex 05, France Dept. of Physics & Astronomy, CSCE: Center for the Study of Cosmic Evolution, and JINA: Joint Institute forNuclear Astrophysics, Michigan State University, E. Lansing, MI 48824, USA IAG, Universidade de S˜ao Paulo, Departamento de Astronomia, CP 3386, 01060-970 S˜ao Paulo, Brazil The Niels Bohr Institute, Astronomy Group, Juliane Maries Vej 30, DK-2100 Copenhagen, Denmark Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Trieste, Via G.B. Tiepolo 11, I-34131 Trieste, Italy European Southern Observatory (ESO), Karl-Schwarschild-Str. 2, D-85749 Garching b. M¨unchen, Germany European Southern Observatory (ESO), Alonso de Cordova 3107, Vitacura, Casilla 19001, Santiago 19, Chile Nordic Optical Telescope, Apartado 474, ES-38700 Santa Cruz de La Palma, Spain Las Cumbres Observatory, Santa Barbara, California, USAReceived 24/04/2007 / accepted 18/09/2007
ABSTRACT
Context.
Extremely metal-poor (EMP) stars in the halo of the Galaxy are sensitive probes of the production of the firstheavy elements and the efficiency of mixing in the early interstellar medium. The heaviest measurable elements in suchstars are our main guides to understanding the nature and astrophysical site(s) of early neutron-capture nucleosynthe-sis.
Aims.
Our aim is to measure accurate, homogeneous neutron-capture element abundances for the sample of 32 EMPgiant stars studied earlier in this series, including 22 stars with [Fe/H] < − Methods.
Based on high-resolution, high S/N spectra from the ESO VLT/UVES, 1D, LTE model atmospheres, andsynthetic spectrum fits, we determine abundances or upper limits for the 16 elements Sr, Y, Zr, Ba, La, Ce, Pr, Nd,Sm, Eu, Gd, Dy, Ho, Er, Tm, and Yb in all stars.
Results.
As found earlier, [Sr/Fe], [Y/Fe], [Zr/Fe] and [Ba/Fe] are below Solar in the EMP stars, with very large scatter.However, we find a tight anti-correlation of [Sr/Ba], [Y/Ba], and [Zr/Ba] with [Ba/H] for − . < [Ba/H] < − . r -process as measured by [Ba/H]. Spectra of even higher S/Nratio are needed to confirm and extend these results below [Fe/H] ≃ − .
5. The huge, well-characterised scatter of the[n-capture/Fe] ratios in our EMP stars is in stark contrast to the negligible dispersion in the [ α /Fe] and [Fe-peak/Fe]ratios for the same stars found in Paper V. Conclusions.
These results demonstrate that a second (“weak” or LEPP) r -process dominates the production ofthe lighter neutron-capture elements for [Ba/H] < − .
5. The combination of very consistent [ α /Fe] and erratic [n-capture/Fe] ratios indicates that inhomogeneous models for the early evolution of the halo are needed. Our accuratedata provide strong constraints on future models of the production and mixing of the heavy elements in the earlyGalaxy. Key words.
Stars: abundances – Stars: Population II – Galaxy: abundances – Galaxy: halo – Nucleosynthesis
1. Introduction
In cold dark matter models for hierarchical galaxy forma-tion, the very first generation of metal-free (PopulationIII) stars are thought to be born in sub-galactic frag-ments of mass
M > M ⊙ (Fuller & Couchman 2000;Yoshida et al. 2003; Madau et al. 2004). Recent models ofprimordial star formation (Abel et al. 2000; Bromm 2005)suggest that these stars were very massive ( M > M ⊙ ),although substantial uncertainties remain. Send offprint requests to : P. Fran¸cois ⋆ Based on observations made with the ESO Very LargeTelescope at Paranal Observatory, Chile (program ID 165.N-0276(A); P.I: R. Cayrel).
Correspondence to : [email protected]
It is likely that none of these stars survives in the Galaxytoday. However, this first generation of stars left imprintsof its nucleosynthetic history in the elemental abundancepatterns of the most metal-poor lower-mass stars that wecan observe at present. Detailed chemical analyses of themost metal-poor stars can therefore provide insight into thesynthesis of the first heavy elements and how efficiently theywere mixed and incorporated in later stellar generations -i.e. how large spiral galaxies such as our own were firstassembled.In Paper V of this series (Cayrel et al. 2004), we con-firmed the existence of relatively uniform α -element over-abundances in 32 very metal-poor halo giants down to[Fe/H] ≃ − .
2, as expected for material enriched by mas-sive progenitors. The very small dispersion in [ α /Fe] showed Fran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements that previous findings of significant scatter in [ α /Fe] and[Fe-peak/Fe] at low metallicity were due to problems inthe data and/or analyses (low S/N, uncertain stellar atmo-spheric parameters, combinations of data using differentline lists, different analysis techniques, etc.). The results ofPaper V thus suggested that mixing of the ISM in the earlyGalaxy was quite efficient.In contrast, the neutron-capture elements have beenfound to behave very differently (Molaro & Bonifacio1990; Norris et al. 1993; Primas et al. 1994). For ex-ample, the [Ba/Fe] and [Sr/Fe] ratios are found to begenerally below solar for stars with [Fe/H] < − . − < [Fe/H] < − − < [Fe/H] < −
3, and [Fe/H] < −
4, re-spectively (Beers & Christlieb 2005). We will not discussthe Carbon-Enhanced Metal-Poor (CEMP) stars, many ofwhich exhibit peculiar abundances and may be binary sys-tems (Lucatello et al. 2004, and in preparation).
2. Observations
The observations were performed during several observingruns in 1999 and 2000 at the VLT-Kueyen telescope withthe high-resolution spectrograph UVES (Dekker et al.2000). Details of these observations and the spectrographsettings were given in Paper V, which also provided abun-dances of the lighter elements for the same sample of starsas studied here.The spectra were reduced using the UVES packagewithin MIDAS, which performs bias and inter-order back-ground subtraction (object and flat-field), optimal extrac-tion of the object (above sky, rejecting cosmic-ray hits),division by a flat-field frame extracted with the sameweighted profile as the object, wavelength calibration, re-binning to a constant wavelength step, and merging of alloverlapping orders. The spectra were then added and nor-malized to unity in the continuum.Because UVES is so efficient in the near UV, we achievetypical S/N ratios per resolution element of 50 or more
Table 1.
The observed sample of stars, with adopted modelparameters (T eff , log g , v t , [Fe/H] m ) and final iron abun-dances [Fe/H] c (from Paper V). Star T eff log g v t [Fe/H] m [Fe/H] c HD 2796 4950 1.5 2.1 − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − at 350 nm. Thus, the weak lines from the heavy elementsbecome measurable even in the EMP stars of our sample;most previous studies were based on spectra of substantiallylower quality.
3. Abundance analysis
As described in Paper V, a classical LTE analysisof our spectra was carried out, using OSMARCSmodel atmospheres (Gustafsson et al. 1975; Plez et al.1992; Edvardsson et al. 1993; Asplund et al. 1997;Gustafsson et al. 2003). Abundances were determinedwith a current version of the turbospectrum code(Alvarez & Plez 1998), which treats scattering in detail.Solar abundances were adopted from Grevesse & Sauval(2000).Line detection and equivalent-width measurement wasfirst carried out with the line list of the appendix of Paper I(Hill et al. 2002) and the automatic code fitline , whichis based on genetic algorithms. As most of the lines areweak and located in crowded spectral regions, this turnedout to be less than optimal, so we decided to determine theabundances by fitting synthetic spectra to all visible lines(and therefore do not list individual measured equivalentwidths here). ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 3
It soon became clear that establishing upper limits forthe abundances of many of the heavy elements could alsobe useful, even when no line from the strongest predictedtransition could be detected. These upper limits were com-puted by comparing the synthetic and observed spectra,and changing the abundance until the computed strengthof the line was of the same order as the noise in the observedspectrum.All the measured abundances and upper limits are givenin Tables 3–5 and shown in detail in Figs. 6–9.
The procedures employed to derive T eff , log g , and micro-turbulent velocity estimates v t for our stars were describedin detail in Paper V (Sect. 3). In summary, T eff is de-rived from broadband photometry, using the Alonso et al.(1999) calibration. The surface gravity is set by requiringthat the Fe and Ti abundances derived from neutral andsingly ionised transitions be identical. Micro-turbulent ve-locities are derived by eliminating the trend in abundanceof the Fe I lines as a function of equivalent width. Table 1lists the atmospheric parameters adopted from Paper V. Table 2.
Estimated errors in the element abundance ratios[X/Fe] and [X/Ba] for BS 17569-049. The other stars yieldsimilar results. [X/Fe] ∆T eff = ∆log g = ∆ v t =+100 K +0.2 +0.2 km s − Sr − − − − − − − − − eff = ∆log g = ∆ v t =+100 K +0.2 +0.2 km s − Sr − − − − − − − − − − − − − − − − − − − For all of the stars in our sample, we adopt the [Fe/H]abundances derived in Paper V, which are based on a largenumber of lines (60–150 Fe I lines and 4–18 Fe II lines). Theline list used to determine the heavy-element abundancesis taken from Paper I, but updated with recent determi-nations of oscillator strengths and hyperfine structure cor-rections (Den Hartog et al. 2003; Lawler et al. 2004) forseveral of the elements.The solar abundances from Grevesse & Sauval (2000)have not been corrected for the changes introduced by thesecorrections, as they are small and only affect some of thetransitions in each element.
Table 2 lists the computed errors in the heavy-elementabundance ratios due to typical uncertainties in the stellarparameters. These errors were estimated by varying T eff ,log g , and v t in the model atmosphere of BS 17569-049 bythe amounts indicated; other stars of the sample yield sim-ilar results. As will be seen, errors in the basic parameterslargely cancel out in the abundance ratios between elementsin similar stages of ionization and with similar excitationpotentials.The global error of an element abundance [A/H], in-cluding errors in fitting of the synthetic line profile to theobserved spectra, is of the order of 0.20 − −
4. Abundances of the neutron-capture elements
In the Solar System, the abundances of Sr, Y, and Zr aredominated by s -process production (Arlandini et al. 1999).A small fraction of these elements can be produced by theweak s -process (Prantzos et al. 1990), but this process isnot expected to be efficient at the low metallicities observedin our sample.Fig. 1 shows the abundance ratios [Sr/Fe], [Y/Fe],and [Zr/Fe] as functions of [Fe/H], as determined hereand by Honda et al. (2004), together with data selectedfrom earlier papers (Ryan et al. 1991; Norris et al. 2001;Gratton et al. 1987; Gilroy et al. 1988; Gratton et al.1988; Gratton & Sneden 1991; Edvardsson et al.1993; Gratton & Sneden 1994; McWilliam et al. 1995;Carney et al. 1997; Nissen & Schuster 1997; McWilliam1998; Stephens 1999; Burris et al. 2000; Fulbright 2000;Carretta et al. 2002; Johnson & Bolte 2002). Only resultsbased on high-resolution, high-S/N spectroscopy are shownhere; thus we do not include the recent lower-S/N data byBarklem et al. (2005).Fig. 1 shows a rather similar behaviour for these threeelements, i.e. [X/Fe] ≃ ≃ − . − . Fran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements
Fig. 1.
The abundance ratios [Sr/Fe], [Y/Fe], and [Zr/Fe] as functions of [Fe/H].
Black rectangles: present study; redtriangles ; Honda et al. (2004); blue rectangles : Johnson & Bolte (2002); green crosses : selected results from earlier lit-erature (see references listed in text). CS 31082-001 (Paper I) is shown by a magenta star . Fig. 2. [Y/Sr] as a function of [Fe/H]. Symbols as in Fig. 1.
Strontium is a key element for probing the early chemi-cal evolution of the Galaxy, because its resonance lines arestrong and can be measured even in stars with metallicitiesas low as [Fe/H] = − .
0. For most of our stars, only theresonance lines at 4077.719 ˚A and 4215.519 ˚A are visible inour spectra.We adopt the gf values from Sneden et al. (1996)and confirm the large underabundance of Sr in EMP starsreported e.g. by Honda et al. (2004). It has long beenrealized that the [Sr/Fe] ratio exhibits very high disper-sion for stars with [Fe/H] ≤ − . ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 5 Table 3.
Abundance results. Numbers in parenthesis indicate the number of lines measured.
Object [Fe/H] [Sr/Fe] [Y/Fe] [Zr/Fe] [Ba/Fe] [La/Fe] [Ce/Fe]HD 2796 − − − − − − − − − − − < − − − < − < − − − − < − − < +0.02 < +0.71BS 16467-062 − − < − < − < − < +0.30 < +1.19BS 16477-003 − − − < − < +0.38BS 17569-049 − − − − − − < − < − − − − < − − < − < +0.58CS 22186-025 − − − − − − − − < − < +0.41CS 22873-055 − − − − − − < − − − − − − − − − − − < +0.17CS 22885-096 − − < − − − < − < +0.90CS 22891-209 − − − − < − − − − − < − < +0.33CS 22948-066 − − − − − < − < − − − − − − < − < +0.05CS 22953-003 − − − − − − < − < +0.25CS 22966-057 − − − − − < +0.34CS 22968-014 − − < − < − − − < +0.48CS 29491-053 − − − − < − < − − − − − − − < − − − < − < − − < − < +0.51CS 29516-024 − − − − − − < − − − − − < +0.20CS 30325-094 − − < − < − − < − Fig. 3. [ < Sr,Y,Zr > /Fe] vs. [Fe/H]. Symbols as Fig. 1. The Y lines are somewhat weaker than those of Sr in thistemperature range and are not readily detected in our mostmetal-poor stars. However, nine lines of similar strength(354.9 nm, 360.07 nm, 361.10 nm, 377.43 nm, 378.86 nm,381.83 nm, 383.29 nm, 395.03 nm, and 439.80 nm) can bemeasured in stars with [Fe/H] > − . Fig. 4. [ < Sr,Y,Zr > /Ba] vs. [Ba/H]. Symbols as Fig. 1.Sr, i.e., a solar ratio down to [Fe/H] ≃ − .
0, and lowervalues of increasing dispersion at even lower metallicities.Unlike Sr, which displays a relatively high dispersion atall metallicities, the weaker and sharper lines of Y yield avery small dispersion in its abundance at intermediate orhigher metallicities. The similarity we stress here is that thedispersions in both [Sr/H] and [Y/H] increase by at least afactor of 2 below [Fe/H] ≃ − . Fran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements
Table 4.
Abundance results (continued).
Object [Fe/H] [Pr/Fe] [Nd/Fe] [Sm/Fe] [Eu/Fe] [Gd/Fe] [Dy/Fe]HD 2796 − − − − − − − < +0.05 − − < − − < +0.83 < +0.19 < +1.28 < +0.38 < +0.77 < +0.95BS 16467-062 − < +1.31 < +0.57 < +1.46 < +0.76 < +0.75 < +1.43BS 16477-003 − < +0.40 < +0.36 < +0.65 < +0.25 < +0.44 < +0.72BS 17569-049 − − < +0.18 < − < − < − < − < − − < +0.50 < − < +1.05 < +0.05 +0.54 (2) < − − < +0.59 +0.54 (1) +0.68 (1) +0.53 (2)CS 22189-009 − < +0.23 < − < +0.88 < − < − − − − < +0.18 − − − − − − < − − < − < − − < +0.09 < − < +0.34 − < − − − < +0.62 < +0.58 < +1.27 < +0.47 < +0.86 < +0.24CS 22891-209 − − < +0.28 − − − − − − < +0.50 +0.01 (1) < +0.50 < − < +0.29 < − − < − < − < − < − < − < − − < +0.07 < − < − < − < +0.01 < − − − < +0.37 < − < +0.03 < +0.02 < +0.21 < − − < +0.76 +0.47 (2) < +0.32 +0.41 (1) < +0.30 +0.48 (1)CS 22968-014 − < +0.40 < − < +0.16 < +0.05 < +0.14 < − − < − − < − − < − < − − − < − − − < − − < +0.23 < − < − < − < +0.07 < +0.55CS 29516-024 − < +0.10 − < − − < − − − < − < − < − < +0.26CS 30325-094 − < +0.64 (1) 0.00 (1) < < − < +0.18 < − Zr is similar to Y in line strength, and we can measure5-10 lines in stars with [Fe/H] > − .
50 (see Fig. 1). Wefind a similar pattern for [Zr/Fe] as for Sr and Y, witha slightly lower average underabundance, large dispersionbelow [Fe/H] ≃ − .
0, and somewhat smaller scatter at in-termediate and higher metallicities.
If two elements are formed by the same process, their ra-tio should not vary with metallicity, and the dispersionaround the mean value should yield a good estimate ofthe errors on the abundance determinations. Fig. 2 showsthe ratio [Y/Sr] as a function of [Fe/H] for our data,along with those by Burris et al. (2000), Johnson & Bolte(2002), and Honda et al. (2004). We confirm that [Y/Sr]is constant with rather low scatter around the mean value:[ < Y/Sr > ] = − ± Fig. 5. [Y/Sr] vs. [Sr/H]. Symbols as in Fig. 1.
The first-peak elements are known to have a more complexorigin than the heavier neutron-capture elements like Ba orEu, which are only produced by the “main” components ofthe r - and s -processes. In solar-type material, Sr, Y and Zrare formed in the “main” s -process, but at lower metallicitythe “weak” s -process (Busso et al. 1999) also contributes.In our EMP stars, we expect a pure r -process origin forthe neutron-capture elements, and we wish to explore thenature of those processes in more detail. ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 7 Table 5.
Abundance results (continued).
Object [Fe/H] [Ho/Fe] [Er/Fe] [Tm/Fe] [Yb/Fe]HD 2796 − < +1.26 − − < +0.43 (1) +0.47 (3) +0.75 (1) +0.52 (1)BD +17:3248 − − − − − < +0.53 < +0.86 ... ...BS 16467-062 − < +1.71 < +1.14 ... ...BS 16477-003 − < +0.80 < +0.53 ... ...BS 17569-049 − − < − < − − < +0.70 < +0.53 ... ...CS 22186-025 − − < +0.83 < +0.56 ... ...CS 22873-055 − < − − − < − − − < +0.19 < +0.02 ... ...CS 22885-096 − < +0.82 < +0.55 ... ...CS 22891-209 − < − − − − − − < +0.45 < +0.18 ... ...CS 22948-066 − < − < − − < − < +0.00 ... ...CS 22953-003 − − < +0.87 < +0.20 ... ...CS 22966-057 − < +0.46 +0.64 (2) ... ...CS 22968-014 − < +0.10 < +0.43 ... ...CS 29491-053 − < − < − − < − − < +0.46 +1.83 (1) ...CS 29516-024 − < − − − < − < − − < +0.34 < +0.07 ... ... Fig. 3 shows the average [ < Sr+Y+Zr > /Fe] ratio for ourstars and from recent literature as a function of [Fe/H].Only stars with data for all three elements are included,which limits the sample towards the lowest metallicities).We find a clear increase in the dispersion of this ratio withdecreasing metallicity. Note also that the two most metal-poor stars ([Fe / H] ≃ − .
5) in Fig. 3 are nearly one dexbelow the solar value, reflecting the strong deficiency of allthree elements in the most metal-poor stars.Because Fe and the neutron-capture elements form un-der quite different conditions, it may be more informativeto study their abundances as functions of another heavyelement. The strong resonance lines of Ba can be mea-sured in stars down to almost [Fe/H] = − .
0, so we selectBa as our alternative reference element. Fig. 4 shows themean [ < Sr+Y+Zr > /Ba] ratio as a function of [Ba/H]. Wefind a striking, tight anti-correlation, especially for starsbelow [Ba/H] ≃ − . ≤ Z ≤ ) This range in atomic mass includes the well-studied ele-ments Ba, Eu, and La. Ba and Eu played a key role inunderstanding early nucleosynthesis, when Truran (1981)first suggested that the [Ba/Eu] vs. [Fe/H] observations ofSpite & Spite (1978) could be naturally understood if both of these neutron-capture elements were synthesised by the r -process in massive stars during early Galactic evolution(85% of the Ba in the Solar System is due to the s -process).Due to the high UV efficiency of UVES, we have beenable to determine abundances or upper limits in many ofour stars for several other heavy neutron-capture elements(Ce, Pr, Nd, Sm, Gd, Dy, Ho, Er and Tm). The results areshown in Figs. 10 - 13 as functions of [Fe/H], together withthose by Johnson & Bolte (2002), Honda et al. (2004),and data selected from earlier literature. These data enableus to discuss the nature of the early r -process nucleosyn-thesis in considerable detail.As noted above, Ba is a particularly interesting element,in part because the resonance lines are strong enough tobe measured in all but two of our stars and permit usto explore mean trends and scatter amongst the neutron-capture elements down to [Fe/H] = − .
2; see Fig. 10. Allour abundance results for Ba have been derived assum-ing the isotopic composition corresponding to the r -process(McWilliam 1998).In the metallicity range − − − .
0. Although the number of stars in thisrange remains small, Fig. 10 suggests that [Ba/Fe] contin-ues to decline to a mean value of [Ba/Fe] = − . − − . − . Fran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements
Fig. 6.
Abundance patterns for the neutron-capture elements in our sample. The full line shows the Solar-system r -processabundance pattern from Arlandini et al. (1999), scaled to match the observed abundance of Ba in each star.The greatest scatter in [Ba/Fe] (a factor of 1000) occursin the metallicity range − . ≤ [Fe / H] ≤ − .
8. Thus, if theBa and Fe in these stars was created by the same classof progenitor objects, their yields would have to vary bya similarly large factor, whatever model of chemical evolu-tion for the early Galaxy one adopts. The yield of Ba couldbe extremely metallicity-dependent or, perhaps more likely,the early production of Ba and Fe was not correlated with each other, and Ba and Fe were produced in different astro-physical sites, as suggested by Wanajo et al. (2001, andreferences therein).As Fig. 10 shows, we do not observe a single star witha [Ba/Fe] ratio above solar below [Fe/H] ≃ − .
2; however,we do note that Barklem et al. (2005) do detect at least afew stars with [Ba/Fe] above solar at metallicities down to[Fe/H] ≃ − . ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 9 Fig. 7.
As Fig. 6, for the next 8 stars.The metallicity interval showing the largest scatter in[Ba/Fe] ( − . ≤ [Fe / H] ≤ − .
8) is also where the extremely r -process-enhanced metal-poor stars are found; i.e. thosewith [r-element/Fe] > +1.0, referred to as r-II stars byBeers & Christlieb (2005). CS 22892-052, CS 31082-001,the eight new r-II stars found by Barklem et al. (2005),and the most recent discovery HE 1523-0909 (Frebel et al.2007), all fall in this range. It is interesting that bothCS 22892-052 and CS 31082-001 fit into the same region of Fig. 10 as the “normal” (non-r-II) stars (albeit at thevery upper limit), so these extreme r-II stars are not excep-tional as far as the [Ba/Fe] ratio is concerned.Like Ba, both La and Ce are primarily due to the s -process at solar metallicity. For La and Ce, we can deter-mine abundances for stars with [Fe/H] > − .
0, but onlyupper limits for the more metal-poor stars. It is interest-ing, however, that we find the same increase of the scat-ter with declining metallicity in the range − . < [Fe/H] Fig. 8.
As Fig. 6, for the next 8 stars. < − . s - and r -process in roughly equal proportions. For Pr, theonly earlier data are from Honda et al. (2004). We confirmthe high [Pr/Fe] ratios found by these authors down to[Fe/H] ≃ − .
0. Our upper limits show that a rather large scatter in [Pr/Fe] exists down to [Fe/H] ≃ − .
0; for Nd andSm, the scatter clearly increases with declining metallicityuntil its maximum at [Fe/H] ≃ − .
0. Note that we have Ndmeasurements for three stars with [Fe/H] < − . r -process, also in Solar-system material (93%,84%, and 87%, respectively, according to Arlandini et al.(1999)). Figure 12 shows that they behave similarly to theother elements of the second neutron-capture peak and dis- ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 11 Fig. 9.
As Fig. 6, for the last 8 stars.play increasing overabundances with declining metallicity,accompanied by increasing scatter. Once again, it appearsthat the scatter is at maximum at [Fe/H] ≃ − .
0, as foundby Honda et al. (2004).Note that for [Eu/Fe], low values ( ≤ ≃ − .
0. Barklem et al. (2005) did find starswith high [Eu/Fe] ( > − .
0, but from a much larger sample of stars than ours.This indicates that stars with high [Eu/Fe] ratios are quite rare at very low metallicity, so that dedicated surveys areneeded to uncover additional examples. For Gd, we do mea-sure high [Gd/Fe] values in two stars (CS 22172-002 andCS 22189-009) below [Fe/H] = − r -process. Veryfew previous results exist for these three elements, whichwe find to be generally overabundant, as also reported byHonda et al. (2004) for Er and Tm. Once more, the large Fig. 10. [Ba/Fe], [La/Fe], and [Ce/Fe] as functions of[Fe/H]. Symbols as in Fig. 1. Blue stars in the upper panel:Data from Barklem et al. (2005).scatter in the element ratios appears maximal at [Fe/H] = − .
0. Our few results for Yb (not plotted) follow the samegeneral trend.
5. Discussion
Our accurate, detailed, and homogeneous abundance datafor the neutron-capture elements in a large sample of VMPand EMP stars enables us to address two important ques-tions regarding the first stages of heavy-element enrichmentin the Galaxy: (i):
The nucleosynthesis process(es) thatformed the first heavy elements, and (ii): the efficiency withwhich the newly synthesised elements were incorporated inthe next generation(s) of stars, including those that havesurvived until today. We discuss each of these in turn inthe following. r -process(es) in EMP stars We begin by repeating the classical Truran (1981) test ofthe relative weight of the r - and s -process as a function ofmetallicity. Ba and La are produced mostly by the “main” s -process in Solar-metallicity stars (92% and 83%, respec-tively, according to Arlandini et al. (1999), but in EMPstars they should be due to the r -process. Fig. 14 showsthe [Eu/Ba] and [Eu/La] ratios as a function of [Fe/H] forour stars, along with earlier data. The dashed lines in both Fig. 11. [Pr/Fe], [Nd/Fe], and [Sm/Fe] as functions of[Fe/H]. Symbols as in Fig. 1.panels indicate the Solar-system r -process abundance ratios(Arlandini et al. 1999).Our [Eu/Ba] ratios do cluster around the Solar-system r -process value at low metallicity, but a substantial scatterremains. Some of this may be due to the Ba data because ofthe broad hyperfine structure of the Ba lines: If the mix ofBa isotopes in the star is different from that assumed in thesynthetic spectrum, the fit to the observed spectrum maybe less stable than for single-component lines. Indeed, the[Eu/La] ratios exhibit substantially smaller dispersion at allmetallicities, demonstrating that the scatter in [Eu/Ba] isessentially due to the Ba, not the Eu abundances. Together,the two panels of Fig. 14 confirm that the neutron-captureelements in EMP stars were produced predominantly orexclusively by the r -process.Given the large scatter of the [n-capture element/Fe]ratios as functions of [Fe/H] (Figs. 1 and 10 – 13), we pro-ceed to compare elemental abundances within the neutron-capture group itself in the following. As noted earlier, wechoose Ba as the reference element because data are avail-able for nearly all our stars.Fig. 15 shows the [Sr/Ba], [Y/Ba], and [Zr/Ba] ratiosvs. [Ba/H] as determined by us and previous authors. Wefind a tight anti-correlation of [X/Ba] with [Ba/H] for allthree elements, at least down to [Ba/H] = -4.5. We em-phasize that most stars in our sample are not enrichedin r -process elements, but note that the two extreme r-II stars CS 22892-052 and CS 31082-001 do in fact follow ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 13 Fig. 12. [Eu/Fe], [Gd/Fe], and [Dy/Fe] as functions of[Fe/H]. Symbols as in Fig. 1. Blue stars in the upper panel:Data from Barklem et al. (2005).the same relation as the “normal” stars. In particular, wefind no stars that are both Sr-poor and Ba-rich, as sus-pected already by Honda et al. (2004); however, such cases are found among the C-enhanced metal-poor (CEMP) stars(Sivarani et al. 2006).Our most Ba-poor stars, below [Ba/H] ≃ − .
5, seem todepart from the correlation and show roughly Solar valuesfor [Sr/Ba] and [Y/Ba], although we note that Honda et al.(2004) do find a couple of high [Sr/Ba] ratios in this region.This might indicate that the additional production channelfor Sr, Y, and Ba discussed below may not operate in thevery first stellar generations. However, the sample is verysmall (these are among our most metal-poor stars, with[Fe/H] < − . The diagrams discussed above amply demonstrate thatnot all the neutron-capture elements in metal-poor starswere produced by a single r -process, as discussed byTravaglio et al. (2004, and references therein); an addi-tional process must contribute preferentially to the produc-tion of the first-peak elements in VMP/EMP stars, previ-ously called the “weak” r -process; we will discuss below theaptness of this term. Fig. 13. [Ho/Fe], [Er/Fe], and [Tm/Fe] as functions of[Fe/H]. Symbols as in Fig. 1.Travaglio et al. (2004) explored the issue by followingthe Galactic enrichment of Sr, Y, and Zr using homoge-neous chemical evolution models. They confirmed that aprocess of primary nature ( r -process) is required to explainthe observed abundance trends, argued that massive starswere the likely sites as these elements occur at very lowmetallicity, and coined the term “Lighter-Element PrimaryProcess” (LEPP) for it. However, regardless of nomencla-ture, the actual process, site, or progenitor stars have notbeen identified.Cescutti et al. (2005) came to similar conclusions,based on the behavior of Ba and Eu. They confirmed theneed for a primary source to explain the behaviour of[Ba/Fe] vs. [Fe/H] and suggested that the primary produc-tion of Eu and Ba is associated with stars in the mass range10-30 M ⊙ . Ishimaru et al. (2004) computed the evolutionof [Eu/Fe], using inhomogeneous chemical evolution modelswith induced star formation, and concluded that the obser-vations implied that the low-mass range of supernovae werethe dominant source of Eu.The observations shown in Fig. 15 clearly cannot beexplained by single r -process. The trends suggest the exis-tence of three different regimes: (i): [Ba/H] ≥ − .
5, whereall ratios are close to Solar; (ii): − . ≤ [Ba/H] ≤ − . (iii): [Ba/H] ≤ − . ≃ −
3, i.e. the
Fig. 14. [Eu/Ba] and [Eu/La] as function of [Fe/H]; sym-bols as in previous figures. The dashed lines indicate theSolar-system r -process abundance ratios (Arlandini et al.1999)metallicity range in which all the highly r -process enrichedmetal-poor stars are have been found so far – the r-II starsas defined by Beers & Christlieb (2005).It appears from these plots, and from the great uni-formity of the r -process element patterns in the r-II starsobserved so far, that the main r -process dominates the to-tal abundance pattern of the heavy elements once they havebeen enriched beyond the level of [Ba/H] ≥ − .
5. At lev-els up to 2 dex below this threshold, another process con-tributes increasingly to the production of the first-peak el-ements Sr, Y, and Zr. We want to clarify the propertiesof this process as independently of the main r -process aspossible.To do so, we have computed the mean residuals of Sr, Y,and Zr in each of our stars from the Solar-system r -processabundance pattern of Arlandini et al. (1999) as shown inFigs. 6–9. Thus, these abundance residuals should repre-sent the pure production of the unknown process, free ofinterference from the main r -process.The result is shown in Fig. 16 and shows that, far frombeing “weak”, the LEPP is responsible for 90-95% of thetotal abundance of these elements at [Ba/H] ≃ − .
3, wherethe [ < Sr,Y,Zr > /Ba] ratio may split into two branches, assuggested on theoretical grounds by Ishimaru & Wanajo(1999) and Ishimaru & Wanajo (2000). Fig. 15. [Sr/Ba], [Y/Ba], and [Zr/Ba] vs. [Ba/H]. Symbolsas in Fig. 1.
Fig. 16.
Average abundance residuals of Sr, Y, and Zrfrom the Solar-system curves in Figs. 6–9 vs. overall heavy-element content as measured by [Ba/H]. Symbols as Fig. 1.One would surmise that qualitative differences in neu-tron exposure or the nature of the available seed nu-clei in the most extreme metal-poor stars could causesuch differences. E.g., Qian & Wasserburg (2007) pro-pose that the first-peak elements (Sr, Y, Zr) are formedby charged-particle reactions in the so-called α -process(Woosley & Hoffmann 1992) in all supernovae, while heavy r -process elements would form only in low-mass SNe withO-Ne-Mg cores and iron only in high-mass SNe. The corre- ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 15 lation shown in Fig. 16 would appear difficult to reconcilewith this otherwise interesting scenario.As an alternative, a new nucleosynthesis process(the νp -process) has been proposed very recently byFroehlich et al. (2006). This process should occur in core-collapse supernovae and would allow for the nucleosynthesisof nuclei with mass number A > The scatter in the observed abundance ratios are an indica-tion of the efficiency of mixing in the ISM in the era beforethe formation of the oldest stars we can observe today. Theresults so far are contradictory.In Paper V, we demonstrated that the [ α /Fe] ratios inthe EMP giant stars of our sample exhibit very little scatterbeyond the observational uncertainty. The great uniformityin the [ α /Fe] ratios of metal-poor stars has recently beendemonstrated in more limited samples of turnoff (dwarf)stars also by Cohen et al. (2004), Arnone et al. (2005),and Spite et al. (2005) and will be further discussed in thenext paper of this series (Bonifacio et al., in preparation).These results are clearly inconsistent with current inhomo-geneous chemical evolution models, which predict a scatterof order 1 dex for such elements (Argast et al. 2002).As emphasized by Argast et al. (2002), the initial scat-ter of a given element ratio, [X/Fe], is determined by theadopted nucleosynthesis yields. The details of the chemicalevolution model will then determine how fast a homoge-neous ISM is achieved through mixing of the enriched re-gions. The results of Paper V indicated that, in order toreproduce the observed low scatter in [ α /Fe], the galac-tic chemical evolution model must employ yields of [ α /Fe]with little or no dependence on the mass of the progenitor.In homogeneous chemical evolution models (Fran¸cois et al.2004), instantaneous mixing is assumed and more variationin the yield can be allowed, because it is integrated overthe different stellar masses as the galaxy evolves.As the number of EMP stars with high-resolution, highS/N spectroscopy has increased, our ability to quantify thetrends and scatter about such trends for individual elementshas improved dramatically as well. In such studies, it isparticularly important to use data sets that are reducedand analysed in as homogeneous a manner as possible, soas to minimise the influence of spurious “observer” scatteron the behaviours that one seeks to understand. It was akey goal of our project to produce such data sets.Thus, Table 6 presents estimates of the observed scat-ter of the elemental ratios reported here, following the or-der of the figures presenting the information, but based ex-clusively on the stars analysed by ourselves. The first twocolumns in the table present the ordinates and abscissaecorresponding to each of the figures. The number of starsconsidered in each range of abscissae listed in the table ap-pears in the third column, while the range in the parameterunder discussion is listed in the fourth column.In order to obtain robust estimates of scatter, we mustfirst de-trend the distributions of the observed ratios. Thisis accomplished by determination of robust locally weightedregression lines ( loess lines), as described by Cleveland(1979, 1994). Such lines have been used before in similarscatter analyses (see, e.g., Ryan et al. 1996; Carretta etal. 2002). The scatter about these lines is then estimated by application of the biweight estimator of scale, S BI , de-scribed by Beers, Flynn, & Gebhardt (1990) .The first entry in the last column of Table 6 lists thisestimate. The quantities in parentheses in this column arethe 1 − σ confidence intervals on this estimate of scatter,obtained from analysis of 1000 bootstrap resamplings ofthe data in each of the given ranges. In this listing, CLrepresents the lower interval on the value of the scatter,while UL represents the upper interval. These errors areuseful for assessing the significance of the difference betweenthe scales of the data from one range to another.[Sr/Fe], [Y/Fe], and [Zr/Fe] shows a similar increase ofthe scatter as the metallicity decreases, with a more pro-nounced effect for Sr. The mean ratio [ < Sr+Y+Zr > /Fe]shows the same behaviour with a lower amplitude.Large scatter is also seen in Ba and La, but its variationas a function of metallicity differs from the lighter elements.The dispersion found for Ba seems independent of metallic-ity, whereas the scatter of La appears much smaller for themost metal-poor stars. Ce, Pr and Nd show much smallerscatter again, in particular Ce for which we measure a bi-weight estimator of only 0.038 dex for the whole sample.Pr and Nd behave like La with smaller scatter for the mostmetal-poor stars.Eu shows a rather high scatter, decreasing as the metal-licity decreases. In contrast, Gd, Dy and Er follow the samebehaviour as Sr, i.e. an increase in scatter as the metallicitydecreases.If we now consider the ratios [Eu/Ba] and [Eu/La] as afunction of [Fe/H], the scatter is smaller by almost an orderof magnitude, confirming the common origin of these ele-ments. It is also noteworthy that the scatter is even smallerfor the most metal poor metallicity bin. The apparently contradictory abundance results for the α -and various neutron-capture elements in VMP and EMPstars might be reconciled if the sites of significant r -processproduction were diverse and (some of them) rare. And wecaution that r-II stars are rare: Barklem et al. (2005) es-timate that they constitute roughly 5% of the giants with[Fe/H] < − .
0. The lower probability of finding them atmetallicities below [Fe/H] ≃ − .
20 may introduce an arti-ficial decrease of the observed scatter.The highly r -enriched (r-II) stars have all been found ina very narrow range around [Fe/H] = − . r -process signature. Did the stars with [Fe/H] < − . < −
4; infact, only three are currently known (Christlieb et al.2002; Frebel et al. 2005; Norris et al. 2007). In a standardclosed-box model (Fran¸cois et al. 1990), we would expect The scale matches the dispersion for a normal distribution.6 Fran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements
Table 6.
Robust estimates of scatter for the observed abundance ratios (see text). “[l/Fe]” represents [ < Sr , Y , Zr > / Fe].
Ord. Abs. N Range S BI (CL, CU) Ord. Abs. N Range S BI (CL, CU)[Sr / Fe] [Fe / H] 31 ≤ − / Fe] [Fe / H] 8 ≤ − − − ≤ − / Fe] [Fe / H] 18 ≤ − − − / Fe] [Fe / H] 26 ≤ − ≤ − − − ≤ − / Fe] [Fe / H] 14 ≤ − − − / Fe] [Fe / H] 26 ≤ − ≤ − − − ≤ − / Fe] [Fe / H] 13 ≤ − − − / Sr] [Fe / H] 25 ≤ − ≤ − − − ≤ − / Fe] [Fe / H] 9 ≤ − / Sr] [Sr / H] 25 ≤ − / Fe] [Fe / H] 16 ≤ − − − − − ≤ − ≤ − / Fe] [Fe / H] 24 ≤ − / Fe] [Fe / H] 6 ≤ − − − ≤ − / Fe] [Fe / H] 6 ≤ − / Ba] [Ba / H] 24 ≤ − / Ba] [Fe / H] 18 ≤ − − − − ≤ − ≤ − / Fe] [Fe / H] 30 ≤ − / La] [Fe / H] 15 ≤ − − − − − ≤ − ≤ − / Fe] [Fe / H] 18 ≤ − / Ba] [Ba / H] 29 ≤ − − − − ≤ − ≤ − / Fe] [Fe / H] 13 ≤ − / Ba] [Ba / H] 26 ≤ − − − − ≤ − ≤ − / Fe] [Fe / H] 14 ≤ − / Ba] [Ba / H] 25 ≤ − − − − ≤ − ≤ − / Fe] [Fe / H] 18 ≤ − − − ≤ − to have found several more, if the IMF did not change sub-stantially over time; however, the preferred scheme for thehalo formation is an open model, where infall is invoked toexplain this “UMP desert” (Chiappini et al. 1997).In the context of an inhomogeneous model of chemi-cal evolution of the Galaxy (Argast et al. 2002), simula-tions show that the density of stars at [Fe/H] = − . − . − .
6, that of the outer haloaround [Fe/H] = − . ran¸cois et al.: First Stars VIII – Abundances of the neutron-capture elements 17 thus have been dominated by inner-halo objects. If so, sim-ple models of Galactic chemical evolution that match theMDFs derived from such surveys may not provide adequateexplanations for the formation of the Milky Way halo, norfor the detailed chemical composition of its most primitivestars.
6. Conclusions
This paper has presented accurate, homogeneous abun-dance determinations for 16 neutron-capture elements ina sample of 32 VMP and EMP giant stars, for which abun-dances of the lighter elements have been determined earlier(Paper V). Our data confirm and refine the general resultsof earlier studies of the neutron-capture elements in EMPstars, and extend them to lower metallicities. In particu-lar, the sample of stars below [Fe/H] = − . ≃ − .
0. Below [Fe/H] ≃ − .
2, we do not find starswith large overabundances of neutron-capture elements rel-ative to the solar ratio. We note, however, that the large“snapshot” sample of Barklem et al. (2005) does identifyat least a few stars below [Fe/H] = − . ≃ − . ≃ − . r -process (Busso et al. 1999; Qian & Wasserburg2000; Wanajo et al. 2001), LEPP process (Travaglio et al.2004), or CPR process (Qian & Wasserburg 2007), or evenan entirely new nucleosynthesis mechanism in massive,metal-poor stars ( ν p-process, Froehlich et al. (2006). Bysubtracting the contributions of the main r -process, weshow that this mechanism is responsible for 90-95% of theamounts of Sr, Y, and Zr in stars with [Ba/H] > − . α/F e ] and[Fe-peak/Fe] ratios as functions of [Fe/H] is very small. Wediscuss the implications of these apparently contradictoryresults on the efficiency of mixing of the primitive ISM interms of homogeneous vs. inhomogeneous models of galac-tic chemical evolution. Acknowledgements.
We thank the ESO staff for assistance during allthe runs of our Large Programme. R.C., P.F., V.H., B.P., F.S. & M.S.thank the PNPS and the PNG for their support. PB and PM acknowl-edge support from the MIUR/PRIN 2004025729 002 and PB fromEU contract MEXT-CT-2004-014265 (CIFIST). T.C.B. acknowledgespartial funding for this work from grants AST 00-98508, AST 00-98549, and AST 04-06784 as well as from grant PHY 02-16783: PhysicsFrontiers Center/Joint Institute for Nuclear Astrophysics (JINA),all awarded by the U.S. National Science Foundation. BN and JAthank the Carlsberg Foundation and the Swedish and Danish NaturalScience Research Councils for partial financial support of this re-search.
References
Abel T., Bryan, G.L., Norman, M.L., 2000, ApJ 540, 39Alonso, A., Arribas, S., Mart´ınez-Roger, C., 1998 A&AS 131, 209Alonso, A., Arribas, S., Martinez-Roger, C. 1999 A&AS 140,261Alvarez R., Plez B., 1998 A&A 330, 1109Arlandini, C. K¨appeler, F., Wisshak, K., Gallino, R., Lugaro, M.,Busso, M., Straniero, O. 1999 ApJ 525, 886Argast, D., Samland, M., Thielemann, F.-K., Gerhard, O. E. 2002A&A 388, 842Arnone, E., Ryan, S.G., Argast, D., Norris, J. E., Beers, T. C. 2005,A&A 430, 507Asplund, M., Gustafsson, B., Kiselman, D., Eriksson, K. 1997A&A 318, 521Barklem, P.S., Christlieb, N., Beers, T.C., Hill, V., Bessell, M.S.,Holmberg, J., Marsteller, B., Rossi, S., Zickgraf, F.-J., Reimers,D. 2005 A&A 439, 129Beers, T.C., Preston, G.W., & Schectman, S.A. 1985, AJ 90, 2089Beers, T.C., Flynn, C., Gebhardt, K. 1990, AJ 100, 32Beers, T.C., Christlieb, N. 2005 ARA&A 43, 531Bromm V. 2005 in
From Lithium to Uranium , IAU Symp. 228 V. Hill,P. Fran¸cois, F. Primas eds., p. 121Burstein, D., Heiles, C. 1982, AJ 87, 1165Burris, D.L., Pilachowski, C.A., Armandroff, T.E., Sneden, C.,Cowan, J.J., Roe, H. 2000, ApJ 544, 302Busso, M., Gallino, R., Wasserburg, G. J. 1999 ARA&A 37,239Carney, B.W., Wright, J.S., Sneden, C., Laird, J.B., Aguilar, L.A., &Latham, D.W. 1997, AJ 114, 363Carollo, D., Beers, T.C., Lee, Y.S., et al., 2007, Nature (submitted)astro-ph/0706.3005v2Carretta, E., Gratton, R., Cohen, J. G., Beers, T. C., Christlieb, N.2002 AJ 124, 481Cayrel, R., Depagne, E., Spite, M., at al. 2004, A&A 416, 1117(
Paper V )Cescutti, G., Fran¸cois, P., Matteucci, F., Cayrel, R., Spite, M. 2005A&A 448, 557Charbonneau, P. 1995 ApJS 101, 309Chiappini C., Matteucci F., Gratton R. 1997 ApJ 77, 765Christlieb, N., Bessell, M., Beers, T.C., et al., 2002, Nature 419, 904Christlieb, N., Beers, T.C., Barklem, P.S., et al. 2004 A&A 428, 1027Cleveland, W.S. 1979, J. Am. Stat. Assoc., 74, 829Cleveland, W.S. 1984 The Elements of Graphing Data (rev. ed.,Summit, NJ: Hobart)Cohen, J.G., Christlieb, N., McWilliam, A., et al. 2004, ApJ 612, 1107Dekker, H., D’Odorico, S., Kaufer, A., Delabre, B., Kotzlowski, 2000in “Optical and IR Telescope Instrumentation and Detectors”Masanori Iye and Alan F. Moorwood (Eds.), Proc SPIE Vol 4008,p534Den Hartog, E.A., Lawler, J.E., Sneden, C., Cowan, J.J., 2003ApJS 148, 543Edvardsson, B., Andersen, J., Gustafsson, B., Lambert, D.L., Nissen,P.E., Tomkin, J. 1993 A&A 275, 101Fran¸cois, P., Vangioni-Flam, E., Audouze, J. 1990 ApJ 361, 487Fran¸cois, P., Matteucci, F., Cayrel, R., Spite, M., Spite, F., Chiappini,C. 2004 A&A 421, 613Frebel, A., Aoki, W., Christlieb, N., et al., 2005, Nature 434, 871Frebel, A., Christlieb, N., Norris, J.E., Thom, C., Beers, T.C., Rhee,J. 2007, ApJ 660, L117Fr¨ohlich, C., Martinez-Pinedo, G., Liebend¨orfer, M., Thielemann,F.-K., Bravo, E., Hix, W.R., Langanke, K., Zinner, N.T. 2006PhysRevLett 96, 142502Fulbright, J.P. 2000, AJ 120, 1841Fuller, T. M., Couchman, H. M. P.2000, ApJ 544, 6Gilroy, K.K., Sneden, C., Pilachowski, C.A., & Cowan, J.J. 1988,ApJ 327, 298Goriely, S., Arnould, M. 1997 A&A 322, 29Gratton, R.G. 1989, A&A 208, 171Gratton, R.G. & Sneden, C. 1987, A&A 178, 179Gratton, R.G. & Sneden, C. 1988, A&A 204, 193Gratton, R.G. & Sneden, C. 1991, A&A 241, 501Gratton, R.G. & Sneden, C. 1994, A&A 287, 927Grevesse, N. & Sauval, A.J. 2000,
Origin of Elements in the SolarSystem. Edited by O. Manuel. p.261Gustafsson, B., Bell, R.A., Eriksson, K., Nordlund ˚A., 1975, A&A 42,407Gustafsson, B., Edvardsson, B., Eriksson, K., Graae-Jørgensen,U., Mizuno-Wiedner, M., Plez, B., 2003, in Stellar AtmosphereModeling, ed. I. Hubeny, D. Mihalas, K. Werner, ASP Conf. Series
Paper I )Honda, S., Aoki, W., Ando, H., Izumiura, H., Kajino, T., Kambe, E.,Kawanomoto, S., Noguchi, K., Okita, K., Sadakane, K., Sato, B.,Takada-Hidai, M., Takeda, Y., Watanabe, E., Beers, T. C., Norris,J. E., Ryan, S. G., 2004, ApJS 152, 113Johnson, J. A., Bolte, M. 2002 ApJ 579, 616Karlsson, T. 2006 ApJ 641, L41Ishimaru, Y. & Wanajo, S. 1999 ApJ 511, L33Ishimaru, Y. & Wanajo, S. 2000
The First Stars
A. Weiss, T. Abeland V. Hill eds. Springer, p. 189Ishimaru, Y., Wanajo, S., Aoki, W., Ryan, S. G. 2004 ApJ 600, 47Lawler, J.E., Sneden, C., Cowan, J.J. 2004 ApJ 604, 850Lucatello, S., Tsangarides, S., Beers, T. C., Carretta, E., Gratton, R.G., Ryan, S. G. 2005 ApJ 625, 825Madau, P., Rees, M. J., Volonteri, M., Haardt, F., Oh, S. P. 2004ApJ 604, 484McWilliam, A., Preston, G. W., Sneden, C., Searle, L. 1995, AJ 109,2757McWilliam, A. 1998 AJ 115, 1640Molaro, P. & Bonifacio, P. 1990, A&A 236, L5Nissen, P.E. & Schuster, W.J. 1997, A&A 326, 751Norris, J.E., Peterson, R.C., Beers, T.C. 1993 ApJ 415, 797Norris, J.E., Ryan, S.G., Beers, T.C. 2001 ApJ 561, 1034Norris, J., et al., 2007, ApJ, submittedPlez, B., Brett, J.M., Nordlund, ˚A.1992 A&A 256,551Prantzos, N., Hashimoto, M., Nomoto, K., 1990, A&A 234, 211Primas, F., Molaro, P., Castelli, F. 1994 A&A 290, 885Qian Y.Z., Wasserburg G. J. 2000, Phys. Rep., 333, 77Qian Y.Z., Wasserburg G. J. 2007, Phys. Rep., 444, 237Ryan, S.G., Norris, J.E., & Bessell, M.S. 1991, AJ 102, 303Ryan, S. G., Norris, J. E., Beers, T. C 1996, ApJ 471, 254Schlegel, D.J., Finkbeiner, D.P., Davis, M. 1998, ApJ 500, 525Sivarani, T., Beers, T.C., Bonifacio, P., et al. 2006, A&A, 459, 125(