Hydrodynamical Simulations of Galaxy Clusters with Galcons

Abstract

We present our recently developed {\em galcon} approach to hydrodynamical cosmological simulations of galaxy clusters - a subgrid model added to the {\em Enzo} adaptive mesh refinement code - which is capable of tracking galaxies within the cluster potential and following the feedback of their main baryonic processes. Galcons are physically extended galactic constructs within which baryonic processes are modeled analytically. By identifying galaxy halos and initializing galcons at high redshift ($z \sim 3$, well before most clusters virialize), we are able to follow the evolution of star formation, galactic winds, and ram-pressure stripping of interstellar media, along with their associated mass, metals and energy feedback into intracluster (IC) gas, which are deposited through a well-resolved spherical interface layer. Our approach is fully described and all results from initial simulations with the enhanced {\em Enzo-Galcon} code are presented. With a galactic star formation rate derived from the observed cosmic star formation density, our galcon simulation better reproduces the observed properties of IC gas, including the density, temperature, metallicity, and entropy profiles. By following the impact of a large number of galaxies on IC gas we explicitly demonstrate the advantages of this approach in producing a lower stellar fraction, a larger gas core radius, an isothermal temperature profile in the central cluster region, and a flatter metallicity gradient than in a standard simulation.

Full text

Page 1

arXiv:1004.3839v2 [astro-ph.CO] 11 May 2010
Hydrodynamical Simulations of Galaxy Clusters with
Galcons
Yinon Arieli, Yoel Rephaeli & Michael L. Norman
May 12, 2010
* accepted for publication in The Astrophysical Journal
Abstract
We present our recently developed galcon approach to hydrodynamical cosmological
simulations of galaxy clusters - a subgrid model added to the Enzo adaptive mesh re-
finement code - which is capable of tracking galaxies within the cluster potential and
following the feedback of their main baryonic processes. Galcons are physically extended
galactic constructs within which baryonic processes are modeled analytically. By iden-
tifying galaxy halos and initializing galcons at high redshift (z ∼ 3, well before most
clusters virialize), we are able to follow the evolution of star formation, galactic winds,
and ram-pressure stripping of interstellar media, along with their associated mass, metals
and energy feedback into intracluster (IC) gas, which are deposited through a well-resolved
spherical interface layer. Our approach is fully described and all results from initial simu-
lations with the enhanced Enzo-Galcon code are presented. With a galactic star formation
rate derived from the observed cosmic star formation density, our galcon simulation better
reproduces the observed properties of IC gas, including the density, temperature, metal-
licity, and entropy profiles. By following the impact of a large number of galaxies on IC
gas we explicitly demonstrate the advantages of this approach in producing a lower stellar
fraction, a larger gas core radius, an isothermal temperature profile in the central cluster
region, and a flatter metallicity gradient than in a standard simulation.
1

Page 2

1Introduction
Hydrodynamical simulations of galaxy clusters, incorporating semi-analytic models for star
formation and galactic feedback processes, show an appreciable level of inconsistency with
observational results. This is particularly apparent in the simulated properties of intracluster
(IC) gas - temperature, metallicity and entropy profiles - and the stellar mass density at high
redshift (see the review by Borgani et al. 2008, and references therein). Statistical properties
of clusters, such as X-ray luminosity-temperature, entropy-temperature, and mass-temperature
relations, seem also to be discrepant when compared with high-precision optical and X-ray
observations (e.g., Kay et al. 2007, Nagai et al. 2007, Tornatore et al. 2007, Kapferer et
al. 2007; for a recent review, see Borgani et al. 2008). The mismatch between simulated and
observed cluster properties (e.g., Evrard & Henry 1991, Cavaliere, Menci & Tozzi 1998, Tozzi &
Norman 2001) is largely due to insufficient accounting for essential physical processes, unrealistic
simplifications of the evolution of star formation and feedback processes, and inadequate level
of spatial resolution.
Some of the relevant physical processes that affect cluster properties and the dynamical and
thermal state of IC gas are mergers of subclusters, galactic winds, ram-pressure stripping and
gravitational drag. These have been partly implemented (e.g., Kapferer et al. 2005, Domainko
et al. 2005, Bruggen & Ruszkowski 2005, Sijacki & Springel 2006, Kapferer et al. 2007)
with some success in predicting IC gas properties. Different combinations of these processes
and the various ways they are included in simulation codes generally result in quite different
gas properties. An example is star formation (SF), whose self-consistent modeling (in cluster
simulations) requires a prohibitively high level of spatial resolution which cannot be achieved
with the current computing resources. Because of this limitation, most current simulations
use a SF prescription that follows the formation of collisionless star ‘particles’ in a running
simulation (e.g., Cen & Ostriker 1992; Nagai & Kravtsov 2005), an approach which leads
to an overestimation of the evolution of the SF rate (SFR) (Nagamine et al. 2004), and a
higher than expected stellar to gas mass ratio. In addition, as we will show, in this particular
implementation of SF the impact of the process remains spatially localized, resulting much
2

Page 3

lower mass (including metals) and energy ejection out of cluster galaxies, and consequently
insufficient suppression of cooling and gas overdensity in cluster cores. This difficulty reflects
the complexity of structure and SF processes, underlying the fact that a full implementation
of these processes in hydrodynamical simulations is indeed a challenging task that nonetheless
motivates attempts to develop a new approach in cluster simulations.
Considerations of galaxy clustering and star formation episodes at high redshift and the
inclusion of heating to suppress gas cooling and condensation, lead us to identify Lyman Break
Galaxies (LBGs) at z ≥ 3 as early (‘pre-heating’) sources of IC gas. Implementation of longer
episodes of SF in these galaxies induces stronger winds. As a result, the amount of energy and
metal-rich gas ejected to IC space is higher, as required for consistency with observations. Gas
dispersal is further enhanced by ram-pressure stripping. Incorporating these baryonic processes
motivated us to develop a new approach in the description of the evolution of IC gas, one
that is based on the powerful adaptive mesh refinement (AMR) cosmological hydrodynamical
simulation code - Enzo (Bryan & Norman 1997, O’Shea et al. 2004). We have modified and
improved Enzo such that it is capable of following more realistically the hierarchical formation of
structure through the inclusion of the most essential physical phenomena. This is accomplished
by modeling the baryonic contents of galactic halos at high redshift by an extended ‘galaxy
construct’, which we refer to as galcon. The new Enzo-Galcon code does not require additional
computational resources compared to the original Enzo code. Because SF and feedback are
modeled analytically, the level of resolution required to achieve improved results (compared to
the standard simulation) is not extreme. This is the first time that most of the known processes
are included in a hydrodynamical non-adiabatic simulation which is also capable of achieving
high resolution (≤ 10 kpc). Initial results from the first implementation of our galcon approach
were briefly described by Arieli, Rephaeli & Norman (2008, hereafter ARN).
In this paper a more complete description is given of our galcon approach, and an expanded
analysis of the first simulations with the new code, including a wider range of IC gas properties
than presented in ARN. In Section 2 we quantitatively describe the main baryonic processes
included in our code. The galcon approach is introduced in Section 3, and results from the
3

Page 4

first Enzo-Galcon simulations are presented in Section 4, with a detailed comparison with
the corresponding results from a (‘standard’) Enzo simulation using popular SF and feedback
prescription. We end with a summary in Section 5.
2 Gas Dispersal Processes
The nearly cosmic metal abundances of IC gas attest to its largely galactic origin. Among the
processes that were suggested to explain the ejection of metals form galaxies are winds (De
Young 1978), ram-pressure stripping (Gunn & Gott 1972), galaxy-galaxy interactions (Gnedin
1998, Kapferer et al. 2005), AGN outflows (Moll et al. 2006) and supernovae in IC space
(Domainko et al. 2004). Observations seem to indicate that the two most important processes
that transfer interstellar (IS) media to IC space, thereby building up the IC gas mass and its
metal content, are galactic winds and ram-pressure stripping (e.g. Schindler 2008). In this
section we describe the procedures for implementing these processes in our numerical code.
2.1Galactic wind
SN explosions generate shocks that drive metal-enriched galactic winds into IC space. The
warm (shock-heated) IS media are further heated when winds from various galaxies merge
and the velocities of the individual shocks are randomized (i.e., effectively thermalized). An
observed morphological relation between the X-ray emitting gas and optical emission-line gas
(Strickland et al 2000, 2002, Martin et al. 2002 & Cecil et al. 2002) provides evidence for this
process. The large width of these lines shows that accelerated IS material can reach velocities
of several hundreds up to few thousands of km/s (Heckman et al. 2000). These and other
X-ray observations (e.g., Sanders et al. 2004, De Grandi et al. 2004 & Pratt et al. 2006) have
observationally established the expected galactic origin of metals in IC gas, and that the mean
metallicity is Z ∼ 0.3 − 0.4Z⊙, where Z⊙is solar metal abundance.
The SN rate reflects the galactic formation rate of high-mass stars. Thus, the SF history is
a basic feature of any model of energy and metallicity ejection out of cluster galaxies. Obser-
4

Page 5

vations of galactic winds (e.g., Heckman 2003) provide direct evidence for the relation between
mass and energy ejection rates and the SFR, ˙M∗,
˙Ew= ew˙M∗c2
˙Mw= βw˙M∗, (1)
where the parameters ewand βware energy and mass ejection efficiencies, respectively, and c is
the speed of light. These efficiency factors cannot be directly predicted from simple considera-
tions, but they can be roughly estimated from observations (e.g., Pettini et al 2001, Heckman
et al. 2001), from which typical values of ew= 5 × 10−6and βw= 0.25 are adopted (e.g., Cen
& Ostriker 1993, Leitherer et al. 1992).
In the galcon approach, the SF history of each galaxy is deduced from observations and input
to the simulation. This contrasts sharply with standard approaches for which the SF history is
an output of the model. By fitting multiband photometric measurements with simulated galaxy
spectra generated by a population synthesis model, the stellar mass density of galaxies can be
deduced up to high redshift (Brinchmann & Ellis 2000, Cole et al. 2001, Cohen 2002, Dickinson
et al. 2003, Fontana et al. 2003, Glazebrook et al. 2004). Several groups have estimated the
global stellar mass density in the redshift range z = 0 − 6 by using this technique (Hopkins
& Beacom 2006, Nagamine et al. 2004, and references therein). The results strongly indicate
that the SFR peaks at zp≃ 3, in accord with some semi-analytic models (e.g. Somerville et al.
2001, Menci et al. 2002), though in other works the SFR peaks at an earlier redshift, zp≥ 5
(Hernquist & Spirngel 2003, Nagamine et al. 2004). The global SFR density can be expressed
in the form (Nagamine et al. 2006)
˙ ρ∗= ˙ ρ0
?texp−t/τd/τd+ Atexp−t/τs/τs
?
, (2)
where t is age of the universe, τdand τsare the characteristic SF times of disk and spheroid
galaxies, respectively. The parameter A sets the scaling of the spheroid to disk stellar mass
ratio, and ˙ ρ0is determined by the overall normalization of the cosmic SFR. The mass density of
galactic halos can be calculated using the Press & Schechter (1974, hereafter PS) mass function
?M2
M1
ρg(z) =Mn(M,z)dM ,(3)
5