Pre-galactic metal enrichment - The chemical signatures of the first stars

Abstract

The emergence of the first sources of light at redshifts of z ~ 10-30
signaled the transition from the simple initial state of the Universe to one of
increasing complexity. We review recent progress in our understanding of the
formation of the first stars and galaxies, starting with cosmological initial
conditions, primordial gas cooling, and subsequent collapse and fragmentation.
We emphasize the important open question of how the pristine gas was enriched
with heavy chemical elements in the wake of the first supernovae. We conclude
by discussing how the chemical abundance patterns conceivably allow us to probe
the properties of the first stars and subsequent stellar generations, and allow
us to test models of early metal enrichment.

Full text

Page 1

arXiv:1101.4024v1 [astro-ph.CO] 20 Jan 2011
Pre-galactic metal enrichment –
The chemical signatures of the first stars
Torgny Karlsson∗
Sydney Institute for Astronomy,
School of Physics,
The University of Sydney,
NSW 2006,
Australia
Visiting Research Fellow,
University of Oxford,
Oxford, OX1 3RH,
UK
Volker Bromm
Department of Astronomy and Texas Cosmology Center,
University of Texas at Austin,
Austin, TX 78712,
USA
Joss Bland-Hawthorn
Sydney Institute for Astronomy,
School of Physics,
The University of Sydney,
NSW 2006,
Australia
Leverhulme Visiting Professor,
University of Oxford,
Oxford, OX1 3RH,
UK
(Dated: December 2010)
The emergence of the first sources of light at redshifts of z ∼ 10 − 30 signaled the tran-
sition from the simple initial state of the Universe to one of increasing complexity. We
review recent progress in our understanding of the formation of the first stars and galax-
ies, starting with cosmological initial conditions, primordial gas cooling, and subsequent
collapse and fragmentation. We emphasize the important open question of how the pris-
tine gas was enriched with heavy chemical elements in the wake of the first supernovae.
We conclude by discussing how the chemical abundance patterns conceivably allow us
to probe the properties of the first stars and subsequent stellar generations, and allow
us to test models of early metal enrichment.
∗torgny@physics.usyd.edu.au

Page 2

2
CONTENTS
I. Introduction2
II. Formation of the first stars
A. Population III: The first mode
B. Population III: The second mode
5
5
7
III. Chemical feedback
A. Final fate of the first stars
B. Nucleosynthesis in the first stars
1. Asymptotic giant branch stars: m/M⊙<
2. Electron capture supernovae: 9 ≤ m/M⊙< 10
3. Core collapse supernovae: 10 ≤ m/M⊙< 25
4. Faint supernovae and hypernovae: 25 ≤ m/M⊙< 40
5. Pair-instability supernovae: 140 ≤ m/M⊙< 260
6. Supermassive core collapse supernovae: 260 ≤ m/M⊙<
C. Metal transport and mixing
1. Early transport: The supernova explosion
2. Late transport: Turbulence
D. Transition to Population II star formation
9
9
11
11
13
13
13
15
15
16
16
19
21
∼9
∼105
IV. Tracers of pre-galactic metal enrichment – Concepts
A. Chemical fingerprints (microscopic processes)
B. Chemical signatures (macroscopic processes)
C. First versus second generation stars
23
23
23
24
V. In search of the earliest chemical signatures
A. Sites of early star formation
B. Secular evolution, blurring and churning
24
24
26
VI. Chemical signatures from the Galactic halo
A. General behaviour
B. Scatters, trends and mixing
C. Outlier objects and carbon-enhanced metal-poor stars
D. Constraining the primordial IMF
1. The low-mass end of the IMF
2. The high-mass end of the IMF
26
26
28
29
31
31
32
VII. Chemical signatures from low mass galaxies
A. Dwarf galaxies
B. Damped Lyman-α systems
33
33
35
VIII. The impact of star clusters
A. Chemical homogeneity and the distribution of cluster masses
B. The Galactic halo
C. Dwarf galaxies
1. Signatures of an ancient star cluster in the Sextans dwarf spheroidal galaxy?
35
35
36
38
40
IX. Peering into the future 40
Acknowledgments42
References42
I. INTRODUCTION
One of the key goals in modern cosmology is to understand the formation of the first generations of stars and the
assembly process of the first galaxies. With the advent of the first stars – referred to historically as Population III
(Pop III) – the Universe was rapidly transformed into an increasingly complex, hierarchical system, due to the energy
and heavy element input from stellar sources and accreting black holes (Barkana and Loeb, 2001; Bromm and Larson,
2004; Ciardi and Ferrara, 2005; Miralda-Escud´ e, 2003). Anisotropies in the cosmic microwave background (CMB)
allow us to probe the state of the Universe 370,000 years after the Big Bang and provide us with some details of
early structure formation. With the best available ground- and space-based telescopes, we can probe cosmic history
all the way from the present-day Universe to roughly a billion years after the Big Bang. In between lies the remaining
frontier, and the first stars and galaxies are the signposts of this early, formative epoch.

Page 3

3
To simulate the build-up of the first stellar systems, we have to address the feedback from the very first stars on
the surrounding intergalactic medium (IGM), and the formation of the second generation of stars out of material that
was enriched by the the first stellar generation. There are a number of reasons why addressing the feedback from the
first stars and understanding second-generation star formation is crucial:
• Over the past fifty years, there has been extraordinary progress in our understanding of the origin and evolution
of the chemical elements over cosmic time. But there remains considerable uncertainty on the physical processes
that seeded many elements in the first stars.
• The initial burst of Pop III star formation may have been rather brief due to the strong negative feedback
effects that likely acted to self-limit this formation mode (Greif and Bromm, 2006; Yoshida et al., 2004).
Second-generation star formation, therefore, may well have been cosmologically dominant compared to Pop III
stars. A subset of the second-generation stars with masses<
These stars provide an indirect window into the Pop III era, as they reflect the enrichment from a single, or
at most a small multiple of supernova (SN) events (Beers and Christlieb, 2005; Frebel et al., 2007; Karlsson,
2005, 2006; Karlsson and Gustafsson, 2005; Karlsson et al., 2008; Salvadori et al., 2007; Tumlinson, 2006). By
scrutinizing their chemical abundance patterns, we can extract empirical constraints which we apply to super-
computer simulations, which in turn allow us to derive theoretical abundance yields to be compared with the
data.
∼0.8 M⊙must have survived to the present day.
• The first steps in the hierarchical build-up of structure provide us with a simplified laboratory for studying
galaxy formation. Did the reionization epoch influence the subsequent evolution of galaxies? What was the
role of galactic winds in the first galaxies? Did star formation occur in bursts, or in a steady, self-regulated
mode? How were the first nuclear black holes seeded? We can probe these formative times by reconstructing
the conditions in the first galaxies from the chemical signatures1in the most ancient stars.
Existing and planned observatories, such as the Hubble Space Telescope (HST), the 8-10m class telescopes and
the James Webb Space Telescope (JWST), planned for launch in 2015, yield data on stars and quasars less than a
billion years after the Big Bang. The ongoing Swift gamma-ray burst (GRB) mission provides us with a window into
massive star formation at the highest redshifts (Bromm and Loeb, 2002, 2006; Lamb and Reichart, 2000). Measure-
ments of the near-IR cosmic background radiation, both in terms of the spectral energy distribution and the angular
fluctuations provide additional constraints on the overall energy production due to the first stars (Dwek et al., 2005;
Fernandez and Komatsu, 2006; Kashlinsky et al., 2005; Magliocchetti et al., 2003; Santos et al., 2002). Focusing on
the nearby Universe, the Gaia mission (Perryman et al., 2001) will provide six-dimensional phase-space information
on no less than one billion stars in the Milky Way and projects like HERMES (Barden et al., 2010) and APOGEE
(Allende Prieto et al., 2008), will measure detailed chemical abundances for millions of Galactic stars, out of which
some fraction likely will probe the very first metal enrichment phase. Understanding the formation of the first stars
and galaxies is thus of great interest to observational studies conducted both at high redshifts and in our local Galactic
neighborhood.
In this review, we will focus on the interplay between primordial star formation and pre-galactic metal enrichment,
since it provides one of the main feedback mechanisms that shaped the early IGM (Madau et al., 2001). The chemical
feedback from the first SNe had a far-reaching impact on early cosmic history (Ciardi and Ferrara, 2005). Our present
understanding is that the character of star formation changed from the early, high-mass dominated Pop III mode to
the more normal, lower-mass Pop II mode once a critical level of enrichment had been reached, the so-called critical
metallicity, Zcrit∼ 10−4Z⊙(Omukai, 2000; Bromm et al., 2001; Bromm and Loeb, 2003; Schneider et al., 2003). It
is then crucially important to understand the topology of early metal enrichment, and when a certain region in the
Universe becomes supercritical. In general, metals produced by Pop III SNe will initially be dispersed into the IGM,
and a fraction of them will later be re-incorporated into more massive systems during bottom-up structure formation.
The metal-enriched protogalaxies then drive further heavy element synthesis and ejection into the intergalactic medium
on a more massive scale. They are also candidate drivers of the early hydrogen reionization that has been inferred
1In this review, we make a clear distinction between a chemical fingerprint and a chemical signature (see Sec. IV). The term chemical
fingerprint is used when the elemental abundances provide direct evidence of a specific reaction mechanism: the r-process or triple-α
process, for example. This review is concerned with identifying the chemical signatures of the first stars in the surface abundances of
the oldest stellar populations. A chemical signature is inherently more complex because the elemental yields expelled from a dying star
are likely to depend on more than one physical process. The signature then reflects multiple parameters, such as stellar mass, rotation,
explosion energy, and the amount of fallback onto the remnant).

Page 4

4
from the relatively large optical depth in the CMB polarization measured with the Wilkinson Microwave Anisotropy
Probe (WMAP). This second wave of metal injection pollutes a larger fraction of the Universe, and sets the stage
for pervasive gravitational fragmentation in those dark halos that have avoided complete photoevaporation in the
emerging cosmic ionizing background.
A prime objective for current astrophysics is to predict the properties of the first galaxies with the goal of detecting
them with the JWST or an Extremely Large Telescope (ELT). What is the required minimum stellar and, by extension,
virial mass in order for them to be detectable? This will determine the number of sources accessible through deep
field campaigns. In turn, it will depend on the stellar population mix within the first galaxies. Is the fine-grain
mixing of the metals that fall into the center of the galactic potential well incomplete, so that Pop III stars would
form simultaneously with Pop I/II stars? Or is the turbulent mixing so efficient as to wipe out all remaining pockets
of primordial gas? Thus, arguably, the most important ingredient that defines the physical state of the first galaxies
is their metal content. This is clearly a very complex problem. Recently, it has become possible to attack it with
numerical simulations, utilizing the latest breakthroughs in algorithm development and supercomputing power.
Here, we raise a number of issues that are controversially discussed in the general literature, and appear in need of
clarification. Because of recent developments in galactic dynamics, we argue that it is not obvious where astronomers
should look for the most ancient stars. Published numerical simulations are already unclear on whether the oldest stars
are solely the preserve of the inner bulge (Bland-Hawthorn and Peebles, 2006; White and Springel, 2000) or spread
over the entire Galaxy (Brook et al., 2007; Scannapieco et al., 2006). Given that galaxies appear to grow inside-out
(Cooper et al., 2010; Zolotov et al., 2009), it is often assumed that the majority of these stars must be confined to
the inner bulge, consistent with its age (Zoccali et al., 2006). But regardless of the galaxy formation model, the
long-term dynamical evolution of the galaxy must also be considered, and this has important consequences for our
understanding of the early Milky Way (Sec. V.B).
The uncertainty in the birth site, and therefore the birth process, leads us to question another tenet of stellar
astrophysics. Does it necessarily follow that targeting the most metal-poor stars is an efficient route to learning about
the yields of the first stars? This needs some clarification. A low value of [Fe/H]2is no guarantee that a star is
ancient since it may reflect environmental conditions (e.g. low star formation efficiency, shallow potential well). But
radioactivity in some extremely metal poor stars provides clear independent evidence that many are indeed ancient
(Sneden et al., 2008).
Conversely, a high value of [Fe/H] does not indicate that a star is of young or intermediate age. Solar abundances
are detected in sources out to the highest detectable redshifts because the dynamical times are very short in the cores
of galaxies (Freeman and Bland-Hawthorn, 2002; Hamann and Ferland, 1999). It is now well established that the
bulge, the halo and all dwarf galaxies show a spread in [Fe/H] and comprise stellar populations that are 10 Gyr or
older, equivalent to a formation redshift prior to z>
∼2. Given the large errors in the ages of old stars, the relative
fractions of these stars that formed before, during or after the reionization epoch is an open question. In other words,
there are no stellar age-metallicity relations that we can appeal to at the present time. We believe that this complex
situation can only be sorted out with a far greater understanding of the origin and evolution of the chemical elements.
Our working definition of what constitutes a “first star” needs clarification. A relatively small number of widely
dispersed stars (and conceivably star clusters) triggered the reionization epoch in the redshift interval z =∼ 15 − 20.
Evidently, these stars belong to the first stellar population, the Pop III. But we can extend the temporal definition
of a “first star” to include those that were the first to enrich their immediate environment regardless of epoch. Most
of the gas confined by collapsing dark matter will not have experienced any form of chemical enrichment. The stars
that formed in these regions during or after the reionization epoch are the first to enrich the local gas. This stellar
population is identified by location rather than by epoch but since it specifically is formed out of gas with Z < Zcrit, it
too is ascribed to the Pop III. As we shall see, however, due to the all-pervasive ionizing field, the two sub-populations
are thought to have been physically distinct. It is not known at what stage in cosmic time the last vestiges of
pristine gas finally succumbed to stellar enrichment, but it is not inconceivable that some Pop III stars formed billion
years after the reionization epoch. This raises the tantalizing prospect that we can identify such regions in direct
observations of the intermediate and high-redshift Universe (Johnson, 2010; Scannapieco et al., 2003). We anticipate
that Pop III stars will have distinct chemical signatures that can be identified in large stellar surveys.
We direct the reader to excellent reviews on early star formation from which we have drawn inspiration.
Bromm et al. (2009) discuss the state-of-the-art numerical simulations of primordial star formation and the as-
sembly of the first galaxies. Beers and Christlieb (2005) discuss the search for, and detailed observations of the
most-metal poor stars in the Galaxy while Tolstoy et al. (2009) focus on the kinematics and chemical abundances of
2[A/B] = log10(nA/nB)⋆− log10(nA/nB)⊙, where nXis the number density of element X.

Page 5

5
stars in dwarf galaxies, and Helmi (2008) describes the formation of the Galactic stellar halo and its connection to
dwarf galaxies. In this review, we focus on the chemical signatures of the first generations of SNe. These signatures
provide us with an important empirical probe to test our theories of high-redshift star formation with observations of
metal-poor stars in our cosmic neighborhood (Frebel et al., 2009; Karlsson et al., 2008; Tumlinson, 2006), an approach
often called Stellar Archaeology or Near-Field Cosmology (Bland-Hawthorn and Freeman, 2000).
II. FORMATION OF THE FIRST STARS
The first stars in the Universe likely formed roughly 200 Myr after the Big Bang (Abel et al., 2002; Bromm et al.,
1999, 2002), when the primordial gas was first able to cool and collapse into dark matter minihalos with masses of
the order of 106M⊙(see Fig. 1). These stars are believed to have been predominantly very massive, with masses
of the order of 100 M⊙, owing to the limited cooling ability of primordial gas, which in minihalos could only cool
through the radiation from H2molecules. While the initial conditions for the formation of these stars are, in principle,
known from precision measurements of cosmological parameters (Komatsu et al., 2010), Pop III star formation may
have occurred in different environments which may have allowed for different modes of star formation. Indeed, it has
become evident that Pop III star formation may actually consist of two distinct modes: one where the primordial gas
collapses into a dark matter (DM) minihalo, and one where the metal-free gas becomes significantly ionized prior to the
onset of gravitational runaway collapse (Johnson and Bromm, 2006). This latter mode of primordial star formation
was originally termed ‘Pop II.5’ (Greif and Bromm, 2006; Johnson and Bromm, 2006; Mackey et al., 2003). To more
clearly indicate that both modes pertain to metal-free star formation, we here follow the new classification scheme
(Johnson et al., 2008; McKee and Tan, 2008), in which the minihalo Pop III mode is termed Pop III.1, whereas the
second mode (formerly ‘Pop II.5’) is now called Pop III.2.
A. Population III: The first mode
The formation of the very first stars, Pop III.1 in the new terminology, can be understood with two basic ingredients:
ΛCDM structure formation and the atomic and molecular physics of the primordial hydrogen and helium. Here,
ΛCDM refers to a Universe composed of cold dark matter (CDM), but dominated by dark energy, possibly Einstein’s
cosmological constant (Λ). It has been recognized for some time that within variants of CDM, minihalos provide the
first star forming sites, where cooling relies on molecular hydrogen (Couchman and Rees, 1986; Haiman et al., 1996;
Tegmark et al., 1997). The properties of the dark matter must be drastically modified before one sees significant
deviations from this robust basic prediction, such as assuming warm dark matter (WDM) models, or self-annihilating
dark matter (Gao and Theuns, 2007; Spolyar et al., 2008; Yoshida et al., 2003). The question then is: What kind
of stars emerge during the H2-facilitated collapse of the pure H-He gas into the minihalo DM potential wells? This
problem, although still beyond our observational horizon, is much simpler than the corresponding star formation
process in the local, well observed, Universe, where the environment of giant molecular clouds is extremely complex.
Indeed, prior to the formation of the first stars, the Universe had no heavy elements, and therefore no dust to
complicate the physics of cooling and opacity. It was also likely that magnetic fields did not yet play a dynamically
significant role. Finally, the early post-recombination Universe was devoid of external ionizing radiation fields, and
strong drivers of turbulence, such that the collapse into the minihalos proceeded in a rather quiescent fashion. The
formation of the first stars inside minihalos, where we know the initial conditions as given by ΛCDM cosmology,
and where we have a complete understanding of the relevant physical processes, thus provides us with a well-posed
problem, amenable to rigorous numerical studies. Since the late 1990s, a number of groups (Abel et al., 2000, 2002;
Bromm et al., 1999, 2002; Nakamura and Umemura, 2001; O’Shea and Norman, 2007, 2008; Yoshida et al., 2006)
have simulated the formation of the first stars with sophisticated numerical algorithms, utilizing either smoothed-
particle hydrodynamics (SPH), or adaptive-mesh refinement (AMR) techniques. These calculations have converged
on a number of main results, leading to the current ‘standard model’ of first star formation, although important open
questions remain (see below).
The most important result, where there is general agreement, is that Pop III.1 stars were predominantly very
massive, with characteristic stellar masses of M⋆>
∼10 M⊙. It has, however, not yet been possible to self-consistently
simulate the assembly of an entire Pop III star, starting from realistic cosmological initial conditions. Recently, such
ab initio calculations (Yoshida et al., 2008a) have traced the evolution up to the point where a small protostellar core
has formed at the center of a minihalo. This initial hydrostatic core has a mass, M⋆∼ 10−2M⊙, very similar to
present-day, Pop I, protostellar seeds. The subsequent growth of the protostar through accretion, however, is believed