Carbon-enhanced metal-poor stars in different environments
aa r X i v : . [ a s t r o - ph . GA ] D ec Astronomische Nachrichten, 2 January 2018
Carbon-enhanced metal-poor stars in di ff erent environments Stefania Salvadori , , ,⋆ ´Asa Sk ´ulad´ottir and Matteo de Bennassuti GEPI, Observatoire de Paris, PSL Research University, CNRS, Univ Paris Diderot, Sorbonne Paris Cit, Place JulesJanssen, 92195 Meudon, France Kapteyn Astronomical Institute, University of Groningen, Landleven 12 9747 AD, Groningen, The Netherlands INAF-Osservatorio Astronomico di Roma, Via di Frascati 33, I-00040 Monte Porzio Catone, ItalyReceived XXXX, accepted XXXXPublished online XXXX
Key words cosmology – galaxy formation – dwarf galaxies – metal enrichmentThe origin of carbon-enhanced metal-poor (CEMP) stars and their possible connections with the chemical elements pro-duced by the first stellar generations is still highly debated. We briefly review observations of CEMP stars in di ff erentenvironments (Galactic stellar halo, ultra-faint and classical dwarf galaxies) and interpret their properties using cosmolog-ical chemical-evolution models for the formation of the Local Group. We discuss the implications of current observationsfor the properties of the first stars, clarify why the fraction of carbon-enhanced to carbon-normal stars varies in dwarfgalaxies with di ff erent luminosity, and discuss the origin of the first CEMP(-no) star found in the Sculptor dwarf galaxy. Copyright line will be provided by the publisher
In the Local Group, spectroscopic studies of ancient indi-vidual stars, provide us with the unique opportunity to un-cover the chemical enrichment of the interstellar mediumwhen the Universe was less than 1 Gyr old. Thus, the fossilimprint of extinguished first stars can be found in these oldstellar populations.For more than a decade, the chemical signature of primor-dial pair instability supernovae (SN) with masses M ∗ = (140 − M ⊙ (Heger & Woosley 2002) have been lookedfor among the most metal-poor stars at [Fe / H] < − invain (e.g. Cayrel et al. 2004). Still, state-of-the-art numer-ical simulations continue to predict that the first stars werelikely very massive, M ∗ = (10 − M ⊙ (e.g. Hiranoet al. 2014). Cosmological chemical evolution models forthe Milky Way formation provide an explanation for sucha tension between numerical findings and observations.“Second-generation” stars formed in environments pollutedby pair instability SN only , are predicted to be extremelyrare with respect to the overall Galactic halo population, andto have [Fe / H] > − rare halo star at [Fe / H] ≈ − .
5, likely showingthe chemical signature of pair instability supernovae (Aokiet al. 2014), might be the first indication that this is the case.Thus, to catch these elusive relics we need to increase cur-rent stellar samples. Where we can find, instead, the chemi-cal signature of less massive and less energetic first stars?During the last few years an increasing number of carbon- ⋆ Corresponding author: [email protected] enhanced metal-poor (CEMP) stars, with [C / Fe] > . / H] < −
2, have been found in the Galactic stellar haloand in nearby dwarf galaxies. The most metal-poor amongthem, at [Fe / H] < −
3, do not typically show enhance-ment in slow (or rapid) neutron capture elements producedby Asymptotic Giant Branch stars. Furthermore, they arenot associated to binary systems, suggesting that their C-excess is likely representative of their environment of for-mation (Norris et al. 2010). These “CEMP-no” stars be-come more frequent as we move towards lower [Fe / H],and their C-excess gradually increases (Fig. 1). CEMP-nostars at [Fe / H] < − M ∗ = (10 − M ⊙ primordial faint SN that de-veloped mixing and fallback (e.g. Iwamoto et al. 2005).A relatively good agreement with data is also obtained bymodels of primordial (or low-metallicity) ”spinstars”, M ∗ = (40 − M ⊙ , that experience mixing and mass loss becauseof their high rotational velocity (e.g. Meynet et al. 2006). Inconclusion, available observations support the idea of a linkbetween CEMP-no stars and moderately massive first stars. In Fig. 1 we show a collection of Carbon measurementsin metal-poor stars dwelling in the oldest component ofthe Local Group: the Galactic stellar halo, ultra-faint dwarfgalaxies, and the dwarf spheroidal galaxy Sculptor. Ac-cording to hierarchical structure formation models, dwarfgalaxies are expected to be the building blocks of stellarhaloes (e.g. Helmi et al. 2008). Thus we are comparing
Copyright line will be provided by the publisher
Salvadori, Skuladottir & de Bennassuti: Carbon-enhanced metal-poor stars in di ff erent environments Fig. 1: Compilation of stars with measured [C / Fe] and [Fe / H] - see Fig. 1 of Salvadori et al. (2015) for references.
Left :stars in the Galactic halo ( squares ) and ultra-faint dwarfs ( circles ). CEMP-no stars are shown with dark filled symbols( black, orange ), light filled symbols show CEMP stars with no available measurements of slow (rapid) neutron captureelements, and open symbols all the remaining stars.
Right : C and Fe measurements for stars in the Sculptor dwarf galaxy.systems that likely experienced similar early star-formationhistories. In Fig. 1 (left) we can first note that in the stel-lar halo C-enhanced and C-normal stars already co-exist at[Fe / H] ≈ − .
75. A key question is then: what physical mech-anism determines the formation of these di ff erent classes ofstars? Are all CEMP-no stars second-generation objects?We can also note that many CEMP-no stars have been foundin ultra-faint dwarf galaxies, the faintest and most metal-poor satellites of the Milky Way, which have total luminosi-ties L < L ⊙ . In these systems, the [C / Fe] vs [Fe / H] mea-surements follow the same trend observed in the Galactichalo, but the fraction of CEMP-no stars at [Fe / H] < − higher (Salvadori et al. 2015).Deep color-magnitude diagrams of ultra-faint dwarfs showthat these galaxies are dominated by >
12 Gyr old stars(Brown et al. 2014), confirming that they might be the liv-ing fossils of the first galaxies, which formed prior the endof reionization and hosted the first stars (Bovill & Ricotti2009; Salvadori & Ferrara 2009).In the more luminous “classical” dwarf spheroidal galaxySculptor ( L ≈ . L ⊙ ) instead, CEMP-no stars are muchmore rare than in the Galactic halo and in ultra-faint dwarfs(Fig. 1, right). In spite of very accurate and intense searches,no CEMP stars have been (yet) found at the lowest [Fe / H](see Sk´ulad´ottir et al. 2015). This result is quite surprisingsince Sculptor is also dominated by an old stellar popula-tion, and thus first stars are expected to be formed in thisgalaxy. If CEMP-no stars trace the early chemical enrich-ment by the first stars, why they are not observed in Sculp-tor? The puzzle became even more intricate after the dis-covery of first CEMP-no star in Sculptor (Sk´ulad´ottir et al. 2015). In fact, this star has an unusually high [Fe / H] ≈ − / H] absent or hidden in Sculptor?
To interpret these observations in terms of primordial cos-mic star-formation and early galaxy evolution, we can usecosmological models for the build-up of the Local Group.Such a statistical tools, which catch the essential physicsof early galaxy formation, follow the star-formation andchemical evolution of the Milky Way (Salvadori et al. 2007;de Bennassuti et al. 2014) and nearby dwarf galaxies (Sal-vadori & Ferrara 2009; Salvadori et al. 2015) along theirpossible merger histories by resolving star-formation in10 . M ⊙ mini-haloes (Salvadori & Ferrara 2009). The min-imum halo mass to form stars is assumed to increase withcosmic time to account for the gradual e ff ect of reioniza-tion (Salvadori et al. 2014). First stars with a variable massdistribution are assumed to form when the amount of dustand metals in the star-forming gas is lower than the criticalvalue to allow gas fragmentation, Z cr < − Z ⊙ (de Bennas-suti et al. 2014). Otherwise, “normal” stars form accordingto a Larson initial mass function (IMF). Stars with di ff er-ent masses contribute to the chemical enrichment in theirproper time scales, and SN explosions are assumed to eject Copyright line will be provided by the publisher sna header will be provided by the publisher 3
Fig. 2: The observed ( points ) and simulated ( histograms )MDFs for Galactic halo stars obtained by assuming di ff er-ent mass ranges for the first stars (see labels). We show C-enhanced ( violet ) and C-normal stars ( blue ). Shaded regionsare ± σ dispersion among 50 Milky Way merger histories.metals and gas into the surrounding Milky Way environ-ment, where the heavy elements gets instantaneously mixed(see Salvadori et al. 2014 for the inhomogeneous metal en-richment treatment). Both the e ffi ciency of star-formationand the SN winds are fixed to reproduce the global proper-ties of the Milky Way and they are assumed to be the same for all star-forming haloes. The only exceptions are mini-haloes, in which the star-formation e ffi ciency is supposed tobe reduced to account for ine ff ective cooling by molecularhydrogen (Salvadori & Ferrara 2009; Salvadori et al. 2015).For the details we remind the reader to the original papers.Let’s first focus on Galactic halo stars. In Fig. 2 we com-pare the observed Metallicity Distribution Function (MDF)with model results obtained by assuming that first starsform according to a Larson IMF with di ff erent mass ranges: M ∗ = (10 − M ⊙ (top) and M ∗ = (10 − M ⊙ (bot-tom). We can immediately see that the low-Fe tail of theMDF, which is populated by CEMP-no stars, is extremelysensitive to the assumed mass range of the first stars (de Bennassuti et al. 2014). Our models show that a good matchto the observations requires M ∗ = (10 − M ⊙ (Fig. 2top). This means that the early metal-enrichment shouldbe dominated by primordial faint SN, which have masses M ∗ ≈ (10 − M ⊙ and produce large amounts of C andvery small of Fe (e.g. Iwamoto et al. 2005). When the con-tribution from energetic pair instability SN is also accountedfor (Fig 2, bottom), the chemical signature of faint SN iscompletely washed out and CEMP-no stars at [Fe / H] < − . / H] < − only . Aswe move towards higher [Fe / H], CEMP-no stars form in en-vironments polluted by both primordial faint SN and normaltype II SN, which start to dominate the chemical enrichmentat very high redshifts (Salvadori et al. 2014). Normal SNtypeII are the main pollutants of the inter-stellar mediumof formation of C-normal stars, which are predicted to ex-ist already at [Fe / H] ≈ − .
75 (Fig. 2; de Bennassuti et al.2014; Salvadori et al. 2015). In conclusion, the existence ofCEMP-no stars at [Fe / H] < − M ∗ = (10 − M ⊙ , must have dominatedthe early phases of chemical evolution.Thus, to investigate the incidence of CEMP-no stars indwarf galaxies with di ff erent luminosities, we can simplyassume that the first stars have all masses M ∗ = M ⊙ and evolve as faint SN (Salvadori et al. 2015). With thisworking-hypothesis we find that, at [Fe / H] < −
3, the aver-age fraction of CEMP-no stars with respect to the total fol-lows almost the same trend in all dwarf galaxies (see Fig. 3of Salvadori et al. 2015). This “universal” shape is a con-sequence of the underlying hierarchical Λ CDM model forstructure formation, according to which all galaxies built-up through merging of progenitor mini-haloes (e.g. Sal-vadori et al. 2010), where [Fe / H] < − / H] range strongly de-pends on the galaxy luminosity and it is one order of magni-tude higher in ultra-faint dwarfs than in the Sculptor dwarfgalaxies (Salvadori et al. 2015). This is due to the dramaticchange, with increasing luminosity, of the MDF of dwarfgalaxies as shown in Fig. 3. We can see that, on average, theMDFs of ultra-faint dwarfs cover a broader [Fe / H] rangethan Sculptor-like dwarfs. Furthermore, they are flatter, andthus contain more stars at [Fe / H] < −
3, where CEMP-nomostly reside. Such a shape is a consequence of the low star-formation rate of ultra-faint dwarfs, which are predicted tobe associated to low-mass mini-haloes (Salvadori & Ferrara2009; Salvadori et al. 2015). More luminous dwarf galax-ies, instead, result from the merging of these small systemsand more massive progenitors, which assembled at laterepochs from metal enriched regions of the Milky Way envi-ronment, and have higher star-formation e ffi ciencies. TheirMDFs are hence peaked and shifted towards higher [Fe / H],where CEMP-no stars can be more likely found.
Copyright line will be provided by the publisher
Salvadori, Skuladottir & de Bennassuti: Carbon-enhanced metal-poor stars in di ff erent environments Consequently, the fraction of stars at [Fe / H] < − <
3% of the total. Thus, we predict that CEMP-no stars at[Fe / H] < − hidden and thus more di ffi cult to catch. We can then ask: how muchshould we enlarge the current stellar sample to uncover thelow-Fe tail of the Sculptor MDF?Fig. 3: Observed ( points ) and predicted ( histograms ) MDFsof: (a) ultra-faint dwarf galaxies with L < L ⊙ ; (b)Bootes-, L ≈ L ⊙ , and (c) Sculptor-like dwarf galaxies, L ≈ L ⊙ . Labels indicate the number of measurements ineach [Fe / H] bin. We show the ± σ dispersion among 100Monte Carlo sampling of the average MDF to the numberof stars observed (see Salvadori et al. 2015). Fig. 4 shows how many stars at [Fe / H] < − / H] measurements. By using currentfacilities, we can follow-up stars down to V ≤
20. Thismight allow us to observe 12 ± / H] < − ≈
40% of which should be CEMP-no stars. With newgeneration instruments and telescopes, such as MOSAIC onthe ESO-Extremely Large Telescope, we will be able to dra-matically increase the statistics by observing all the stars inSculptor down to the main sequence turn-o ff (Evans et al.2015). In this case, we might be able to catch ≈
80 stars at[Fe / H] < − ≈
16 at [Fe / H] < − .
7, where the incidenceof CEMP-no stars should be 100% (Fig. 4c, see Salvadoriet al. 2015 for details). These experiments, therefore, willallow us to test the predominant role of faint primordial SNon early metal enrichment, along with the underlying hier-archical models for structure formation. Fig. 4: Number of stars at [Fe / H] < − / H] measure-ments. From top to bottom we show results for: i) the cur-rent statistics ii) stars with V ≤
20, and iii) V ≤
23. Shadedarea show the ± σ errors (see Fig. 3).In conclusion, Near-Field cosmology is a powerful strategyto (indirectly) study the properties of extinguished first starsand the physics of early galaxy formation. Larger stellarsamples are required to tightly constrain the mass spectrumof the very first stars. In the nearby future, we will haveat our disposal results from wide and deep spectroscopicsurveys, such as those discussed in this volume. By com-bining theoretical and observational e ff orts we can exploitthese data to unveil the first star properties. Thus, we areentering in the Golden-Era of Near-Field cosmology. Acknowledgements.
We are in debt with E. Tolstoy and R. Schnei-der. SS thanks the conference organizers for a very productivemeeting. She is grateful to NWO for her VENI grant: 639.041.233.
References