Stellar Archaeology is the endeavor to constrain the properties of the first stars by scrutinizing the chemical abundance patterns in the most metal-poor, and therefore presumably oldest, stars in the Milky Way and nearby galaxies within the Local Group (Beers & Christlieb 2005; Frebel 2010). Such a near-field cosmological approach nicely complements the traditional far-field cosmology based on high-redshift observations (Freeman & Bland-Hawthorn 2002). The first galaxies may have left behind a number of local fossils as well. (i) Some of the numerous dwarf galaxies in the Local Group may constitute the survivors of the first galaxies. In this regard, the ultra-faint dwarf (UFD) galaxies, recently discovered in the SDSS, are of particular promise. (ii) The first galaxies likely were the formation sites for the first low-mass Pop II stars (e.g., Tumlinson 2010). These eventually found their way into the halo, and possibly bulge, of our Galaxy through its complex hierarchical assembly process. (iii) Finally, a subset of the first galaxies may have provided the birth places for old, metal-poor globular clusters (GCs), which again might have been incorporated into our MW (Bromm & Clarke 2002; Kravtsov & Gnedin 2005; Brodie & Strader 2006; Boley et al. 2009). We focus on the first issue, as it is of most direct relevance for this review.
7.1. Ultrafaint Dwarf Galaxies
The newly discovered UFD galaxies are the intrinsically least luminous members (Ltot 105 L) of the Local Group (Kirby et al. 2008; Martin, de Jong & Rix 2008). Due to their simple assembly history, they can be regarded as the closest local relatives to the first galaxies. They are believed to have had only one or few early star formation events, but have been quiescent ever since (Tolstoy, Hill & Tosi 2009). Hence, they should reflect the signatures of the earliest stages of chemical enrichment in their population of low-mass stars. As opposed to the MW halo, which was assembled through numerous merger and accretion events, the lowest luminosity dwarfs provide us with a much cleaner fossil record of early star and galaxy formation. With their small number of stars (of order a few hundred), the UFDs may allow us to carry out a virtually complete census of their stellar content (Simon et al. 2011). Medium-resolution spectroscopic studies have shown that all of the UFDs have large [Fe/H] spreads of ~ 1 dex or more (Kirby et al. 2008; Norris et al. 2010), reaching below [Fe/H] = -3.0. Moreover, some of them have average metallicities as low as < [Fe/H] > ~ -2.6, which is lower than the values found in the most metal-poor GCs. The abundances of dwarf galaxy stars closely resemble those found in similarly metal-poor Galactic halo stars. Overall, this suggests that chemical evolution proceeded very similarly at the early times which are probed with the most metal-poor, and thus presumably the oldest, stars in a given system (Frebel & Bromm 2011). The same chemical behavior has also been found in Sculptor, a more luminous, classical dwarf spheroidal (dSph) galaxy, at [Fe/H] ~ -3.8 (Frebel et al. 2010). However, at higher metallicity ([Fe/H] > ~ -2.5), the Sculptor stellar ([ / Fe]-) abundances deviate with respect to the behavior of Galactic halo stars (Geisler et al. 2005), indicating a different evolutionary timescale and multiple star-formation events (Tolstoy et al. 2004).
7.2. Theoretical Models
There is widespread consensus that the UFDs may provide us with the Rosetta Stone for galaxy formation, given their relative simplicity. It is therefore very tempting to theoretically model their formation process. When did they form, and how do they fit into the hierarchical CDM cosmology? What kind of star formation history did they experience, and, related to this, how many SNe did contribute to their complement of metals? This field is still very young, and it is likely that progress over the next few years will be rapid. Here, we only provide a few comments to illustrate the flavor of the developing argument.
7.2.1. FORMATION SITE Currently, two main ideas for the origin of the UFDs are discussed in the literature. One class of models invokes H2-cooling minihalos (Bovill & Ricotti 2009, 2011; Salvadori & Ferrara 2009). The models couple a representation of the evolving dark matter distribution, either from cosmological simulations or from Press-Schechter type techniques, with a recipe for star formation and feedback, and can successfully explain the broad observational properties of the UFD population (see Figure 14). The suggested antecedents of the UFDs would then have been minihalos with masses M 107 - 108 M, close to the threshold where atomic cooling sets in. A challenge for these models comes from the highly-resolved, ab initio simulations discussed in Section 4. The underlying question again is where second-generation star formation can occur, already in minihalos or only in the next stage of hierarchical assembly, the atomic cooling halos (see the discussion in Section 2). Within the minihalo scenario, the same system would have to first lead to the explosion of Pop III SNe, subsequently reassemble the enriched gas inside their shallow potential well despite strong negative feedback effects, and finally trigger a second generation of star formation. The strength of the negative feedback crucially depends on the Pop III IMF; the more top-heavy it is, the longet the delay time between first and second generation star formation. For the minihalo model as UFD progenitors to work, one has to assume that the first stars typically were not too massive.
Figure 14. Stellar archaeology with dwarf galaxies. Shown are average Fe abundances vs. total luminosities for dwarf galaxies as predicted by a semi-analytical merger tree model. Different colors indicate the baryon fraction at the time of formation, expressed relative to the cosmic mean: fb / fb > 0.5 (blue dots), 0.1 < fb / fb < 0.5 (green), and fb / fb < 0.1 (yellow). The symbols with erroe bars denote observational data from Kirby et al. (2008). Within this model, the ultra-faint (UF) dwarf galaxies are the fossils of minihalos with (virial) masses close to the limit where atomic cooling would set in (M 107 - 108 M). The classical dwarf spheroidals, such as the prototypical Sculptor system, would then be descendants of more massive dark matter halos. Adopted from Salvadori & Ferrara (2009).
The above challenge provides the motivation for the competing model to explain the origin of UFDs (Maccio et al. 2010; Frebel & Bromm 2011). In the atomic cooling halo pathway, the sites for first and second-generation star formation are decoupled (see Figure 1), thus alleviating the problem of admitting local Pop III pre-enrichment.
7.2.2. ENRICHMENT MODE An important clue to the true nature of the UFD formation site could come from a knowledge of the chemical enrichment mode. Did enrichment in the UFD progenitors occur in one intial burst, to be completely shut-off subsequently, or continuously, spread out over an extended star formation and SN history? The first possibility has been termed "one-shot" chemical enrichment by Frebel & Bromm (2011). The answer to this question would provide us with important clues about the strength of the feedback in the first galaxies. If this feedback was sufficiently violent to disrupt the first galaxy already after its initial starburst, blowing all remaining gas into the general IGM, "one-shot" conditions would be realized. The simulations have not yet answered this question with any degree of certainty, but one can look for the chemical signature of such burst-like enrichment in the stellar content of the UFDs (Frebel & Bromm 2011). Their surviving Pop II stars would then preserve the yields from the initial Pop III SNe that had occurred in the progenitor minihaloes without any subsequent enrichment from events that operated on timescales longer than the short dynamical time that governs the formation of the starburst, such as type Ia SNe or AGB winds. Specifically, one would expect high [ / Fe] values for all stars in the UFD, and low n-capture abundances due to the absence of any s-process contribution from AGB stars.
An important caveat is that a subset of those Pop II stars might have experienced post-processing of their surface abundance, e.g., via mass transfer from a binary companion or dredge-up events during later stages of stellar evolution. A possible strategy to circumvent this problem is to realize that almost all stars form in clusters. A properly defined multi-dimensional abundance space could thus uniquely identify the primordial signature through this clustering effect (Bland-Hawthorn et al. 2010).
7.2.3. LESSONS LEARNED Currently, the lowest luminosity dwarfs are consistent with the one-shot criteria, but the data is still very sparse, and the case therefore remains inconclusive. The hope is that high-resolution spectroscopy of more UFD stars will soon become available. The abundance ratios in most individual stars reflect an enrichment history that is dominated by core-collapse SNe, even in the higher metallicity regime ([Fe/H] ~ -2.0). The latter is dominated by SN Ia enrichment in the more luminous classical dSphs. The observed spread in Fe and other elements may suggest that mixing in the UFD progenitors was not very efficient, at least on scales of 10 pc, whereas mixing on smaller scales may have been almost complete, if the simulations discussed in Section 4 are correct. The suggested signature from clustered star formation in the first galaxies may again help to constrain the mixing efficieny on different length scales (Bland-Hawthorn et al. 2010). Without inhomogeneous mixing, all stars should have nearly identical abundances, similar to what is found in globular cluster. We can thus tentatively infer that GCs must have formed in more massive haloes where turbulent mixing would have been much more efficient.
As additional abundances of individual dwarf galaxy stars become available, abundance gradient studies of the UFD galaxies should shed further light on the mixing efficiency. Stronger gravitational fields in the center of a system would drive more turbulence that in turn would induce mixing. Because the UFDs are ideal testbeds for various feedback processes, it will also be interesting to study the carbon abundances in these systems. Carbon, as well as oxygen, may have been a key cooling agent inside the first galaxies (Bromm & Loeb 2003a). Although one extremely carbon-rich star (with [Fe/H] ~ -3.5) has recently been found in Segue 1 (Norris et al. 2010a), low stellar C abundances, if ever found, would greatly weaken the theory of fine-structure line cooling for driving the transition to low-mass star formation.