metal enrichmentoftheicm: a3-dpictureofchemical and ... · 2consejo nacional de investigaciones...

24
arXiv:astro-ph/0603270v1 10 Mar 2006 Mon. Not. R. Astron. Soc. 000, 1–24 (2005) Printed 10 June 2018 (MN L A T E X style file v2.2) Metal Enrichment of the ICM: a 3-D Picture of Chemical and Dynamical Properties Sof´ ıa A. Cora 1,21 Facultad de Ciencias Astron´ omicas y Geof´ ısicas de la Universidad Nacional de La Plata and Instituto de Astrof´ ısica de La Plata, Observatorio Astron´ omico, Paseo del Bosque S/N, 1900 La Plata, Argentina 2 Consejo Nacional de Investigaciones Cient´ ıficas y T´ ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop a model for the metal enrichment of the intracluster medium (ICM) that combines a cosmological non-radiative hydrodynamical N-Body/SPH simulation of a cluster of galaxies, and a semi-analytic model of galaxy formation. The novel feature of our hybrid model is that the chemical properties of the diffuse gas in the underlying simulation are dynamically and consistently generated from stars in the galaxies. We follow the production of several chemical elements, provided by low- and intermediate- mass stars, core collapse and type Ia supernovae. We analyse the spatial distribution of metals in the ICM, investigate the way in which the chemical enrichment proceeds, and use iron emissivity as a tracer of gas motions. The similar abundance patterns developed by O and Fe indicate that both types of SNe pollute the ICM in a simi- lar fashion. Their radial abundance profiles are enhanced in the inner 100 h -1 kpc in the last Gyr because of the convergence of enriched gas clumps to the cluster centre; this increment cannot be explained by the metal ejection of cluster galaxies which is quite low at the present epoch. Our results support a scenario in which part of the central intracluster gas comes from gas clumps that, in the redshift range of z 0.2 to 0.5, have been enriched to solar values and are at large distances from the cluster centre (from 1 to 6 h -1 Mpc) moving at very high velocities (from 1300 to 2500 km s -1 ). The turbulent gas motions within the cluster, originated in the inho- mogeneous gas infall during the cluster assembly, are manifested in emission-weighted velocity maps as gradients that can be as large as 1000 km s -1 over distances of a few hundred kpc. Gradients of this magnitude are also seen in velocity distributions along sightlines through the cluster centre. Doppler shifting and broadening suffered by the Fe K α 6.7 keV emission line along such sightlines could be used to probe these gas large-scale motions when they are produced within an area characterised by high iron line emissivity. Key words: galaxies: formation - galaxies: evolution - galaxies: cluster: general 1 INTRODUCTION Clusters of galaxies are the largest virialised objects in the Universe. Within the hierarchical structure formation sce- nario of the cold dark matter (CDM) model, they arise from the rare highest peaks of the initial density perturbation field. During cluster assembly, the dissipative gas component is shock heated by the potential well of the cluster, reaching very high temperatures (T 10 78 K). This hot and diffuse gas radiates energy through thermal bremsstrahlung in the X-ray band of the electromagnetic spectrum. The gravita- tionally dominating collisionless dark matter component on the other hand shows considerable substructure, as revealed E-mail: [email protected] by high resolution simulations that follow cluster assembly (e.g. Springel et al. 2001 and references therein). These sub- structures are the surviving remnants of dark matter haloes that have fallen in at earlier times, and plausibly mark the location of the cluster galaxies. X-ray observations have first revealed the presence of hot gas trapped in the potential well of galaxy clusters, and provide a wealth of information about the dynamical, thermal and chemical properties of clusters. A strong fea- ture in their spectra is the presence of line emission pro- duced by highly ionised iron, mainly, the H- and He-like iron lines at 6.9 and 6.7 keV. The content of iron in the in- tracluster medium (ICM) is approximately one third of the solar value (Fukazawa et al. 1998; Ettori, Allen & Fabian 2001), suggesting that some of its hot gas must have orig-

Upload: others

Post on 14-Jul-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

arX

iva

stro

-ph

0603

270v

1 1

0 M

ar 2

006

Mon Not R Astron Soc 000 1ndash24 (2005) Printed 10 June 2018 (MN LATEX style file v22)

Metal Enrichment of the ICM a 3-D Picture of Chemical

and Dynamical Properties

Sofıa A Cora12⋆

1Facultad de Ciencias Astronomicas y Geofısicas de la Universidad Nacional de La Plata and Instituto de Astrofısica de La PlataObservatorio Astronomico Paseo del Bosque SN 1900 La Plata Argentina2Consejo Nacional de Investigaciones Cientıficas y Tecnicas Rivadavia 1917 Buenos Aires Argentina

ABSTRACTWe develop a model for the metal enrichment of the intracluster medium (ICM) thatcombines a cosmological non-radiative hydrodynamical N-BodySPH simulation of acluster of galaxies and a semi-analytic model of galaxy formation The novel featureof our hybrid model is that the chemical properties of the diffuse gas in the underlyingsimulation are dynamically and consistently generated from stars in the galaxies Wefollow the production of several chemical elements provided by low- and intermediate-mass stars core collapse and type Ia supernovae We analyse the spatial distributionof metals in the ICM investigate the way in which the chemical enrichment proceedsand use iron emissivity as a tracer of gas motions The similar abundance patternsdeveloped by O and Fe indicate that both types of SNe pollute the ICM in a simi-lar fashion Their radial abundance profiles are enhanced in the inner 100 hminus1 kpc inthe last Gyr because of the convergence of enriched gas clumps to the cluster centrethis increment cannot be explained by the metal ejection of cluster galaxies which isquite low at the present epoch Our results support a scenario in which part of thecentral intracluster gas comes from gas clumps that in the redshift range of z sim 02to sim 05 have been enriched to solar values and are at large distances from the clustercentre (from sim 1 to sim 6 hminus1 Mpc) moving at very high velocities (from sim 1300 tosim 2500 kmsminus1) The turbulent gas motions within the cluster originated in the inho-mogeneous gas infall during the cluster assembly are manifested in emission-weightedvelocity maps as gradients that can be as large as sim 1000 kmsminus1 over distances of afew hundred kpc Gradients of this magnitude are also seen in velocity distributionsalong sightlines through the cluster centre Doppler shifting and broadening sufferedby the Fe Kα 67 keV emission line along such sightlines could be used to probe thesegas large-scale motions when they are produced within an area characterised by highiron line emissivity

Key words galaxies formation - galaxies evolution - galaxies cluster general

1 INTRODUCTION

Clusters of galaxies are the largest virialised objects in theUniverse Within the hierarchical structure formation sce-nario of the cold dark matter (CDM) model they arise fromthe rare highest peaks of the initial density perturbationfield During cluster assembly the dissipative gas componentis shock heated by the potential well of the cluster reachingvery high temperatures (T sim 107minus8 K) This hot and diffusegas radiates energy through thermal bremsstrahlung in theX-ray band of the electromagnetic spectrum The gravita-tionally dominating collisionless dark matter component onthe other hand shows considerable substructure as revealed

⋆ E-mail sacorafcaglpunlpeduar

by high resolution simulations that follow cluster assembly(eg Springel et al 2001 and references therein) These sub-structures are the surviving remnants of dark matter haloesthat have fallen in at earlier times and plausibly mark thelocation of the cluster galaxies

X-ray observations have first revealed the presence ofhot gas trapped in the potential well of galaxy clustersand provide a wealth of information about the dynamicalthermal and chemical properties of clusters A strong fea-ture in their spectra is the presence of line emission pro-duced by highly ionised iron mainly the H- and He-likeiron lines at 69 and 67 keV The content of iron in the in-tracluster medium (ICM) is approximately one third of thesolar value (Fukazawa et al 1998 Ettori Allen amp Fabian2001) suggesting that some of its hot gas must have orig-

ccopy 2005 RAS

2 S A Cora

inated in the galaxies that reside in the cluster The pro-cesses considered for the supply of this enriched gas includegalactic winds driven by supernovae explosions (White 1991Renzini 1997) ram pressure stripping (Mori amp Burkert2000) early enrichment by hypernovae associated with pop-ulation type III stars (Loewenstein 2001) and intraclusterstars (Zaritsky Gonzalez amp Zabludoff 2004)

Spatially resolved maps of the temperature andmetal abundance distributions in the intracluster mediumhave been obtained by X-ray surveys carried out withROSAT ASCA BeppoSAX (Finoguenov David ampPonman 2000 Finoguenov Arnaud amp David 2001De Grandi amp Molendi 2001) and more recently withXMM-Newton and Chandra (Gastaldello amp Molendi2002 Finoguenov et al 2002 Sanders amp Fabian 2002Matsushita Finoguenov amp Bohringer 2003 Tamura et al2004) As a result abundance profiles of several chemicalspecies with different origins have been obtained namelyfor O and Si generated by core collapse supernovae (SNeCC) and Fe mainly produced by type Ia supernovae (SNeIa) These profiles reflect physical mixing processes in thebaryonic component due to gas cooling star formationmetal production and energetic and chemical feedback allcoupled to the hierarchical build up of the galaxies and theclusters These observations can therefore be very valuablefor constraining models of galaxy formation and chemicalenrichment of the ICM allowing to evaluate the relativeimportance of different types of supernovae for metalpollution The spatial distribution of metals obtained frommodels of the ICM chemical enrichment can be analysed byconstructing X-ray weighted metal maps which trace theinteractions between the ICM and cluster galaxies thuswhen they are compared with observed metal maps theyprovide information about the mechanisms involved in theenrichment process and may reveal the dynamical state of agalaxy cluster (Kapferer et al 2005 Domainko et al 2005)

X-ray observations also provide important informationabout the dynamics of the intracluster gas through radial ve-locity measurements in clusters (Dupke amp Bregman 2001)allowing in principle further tests of structure formationmodels Gaseous bulk flows affect the characteristics of X-ray spectra of the ICM by Doppler broadening and shiftingof emission lines The prominent Fe Kα 67 keV iron emis-sion line present in X-ray spectra of galaxy clusters couldbe used as a tracer of these motions This possibility wasevaluated by Sunyaev et al (2003) who showed that futureX-ray missions like CONSTELLATION-X and XEUS withan energy resolution of a few eV should be able to detectseveral Doppler-shifted components of this emission line inthe core of the cluster

All this wealth of data calls for a theoretical interpre-tation of the ICM chemical enrichment in the framework ofa detailed cluster formation model The approaches used sofar to this end include semi-analytic models of galaxy forma-tion coupled to N-Body simulations of galaxy clusters (DeLucia Kauffmann amp White 2004 Nagashima et al 2005)hydrodynamical simulations of cluster formation which in-clude self-consistently star formation SNe feedback andmetal enrichment from SNe CC and Ia (eg Valdarnini 2003Tornatore et al 2004) and an intermediate approach thatcombines N-Body and hydrodynamic simulations togetherwith a phenomenological galaxy formation model and a pre-

scription of the effect under study such as galactic winds andmerger-driven starbursts (Kapferer et al 2005) and ram-pressure stripping (Schindler et al 2005 Domainko et al2005) In this work we present a different intermediate ap-proach which combines a cosmological non-radiative hydro-dynamical N-BodySPH simulation of a cluster of galax-ies with a semi-analytic model of galaxy formation Thesimplicity of the semi-analytic model has the advantage ofreaching a larger dynamic range than fully self-consistenthydro-simulations at a far smaller computational cost Inparticular it allows us to explore more easily the range ofparameters that characterise appropriate chemical enrich-ment models

The long cooling time of the bulk of the gas in rich clus-ters justifies the assumption of a non-radiative gas (Evrard1990 Frenk et al 1999) However this time-scale becomessmaller than the Hubble time in the core of the clusterwhere the densities are higher Here is where the semi-analytic model plays an important part taking into accountradiative cooling and models for star formation chemi-cal enrichment and energetic feedback from galaxies Thesemi-analytic model used here is based on previous works(Springel et al 2001 De Lucia et al 2004 DL04 hereafter)but was extended with a new chemical implementation thattracks the abundance of different species resulting from massloss through stellar winds of low- intermediate- and high-mass stars and from the explosions of SNe CC and Ia su-pernovae The principal new feature of our model is how-ever the link between semi-analytic model results and thechemical enrichment of gas particles in the underlying N-BodySPH simulation We pollute gas particles with met-als ejected from the galaxies consistent with the modellingof the semi-analytic model These metals are then carriedaround and mixed by the hydrodynamic processes duringcluster formation Thus we are not limited to an analysisof mean chemical properties of the intracluster gas like inthe original model of DL04 Instead the enrichment of gasparticles allows a study of the spatial distribution of metalsin the ICM and to use iron emissivity as a tracer of gasmotions

The aim of our model is on one hand to understandthe way in which the chemical enrichment patterns char-acterising the intracluster gas develop and on the otherto detect the imprints of gas bulk motions in the shape ofX-ray emission lines thus providing guidance for the inter-pretation of future spectroscopic X-ray data The first partof the present study is therefore devoted to an analysis ofchemical enrichment of the ICM focusing on the spatial dis-tribution of different chemical elements In a second part weanalyse the link between the occurrence of multiple com-ponents in the Fe Kα 67 keV emission line and the rangeof velocities of the gas that generates this emission Thisis directly relevant to the question whether metal emission-line-weighted velocity maps of the cluster could be used toinfer the spatial coherence of these bulk motions Gatheringall the information available in the form of projected mapsand detailed spectra along lines of sight through the ICMwe are able to extend the information in 2-D provided byobserved projected distributions of gas temperature abun-dances and surface brightness into a 3-D picture of the ICMproperties

This paper is organised as follows Section 2 describes

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 3

our hybrid model used to study the ICM chemical enrich-ment summarises the properties of the hydrodynamical sim-ulation used and presents the main features of the semi-analytic model Section 3 contains the details of the chem-ical model implemented in the semi-analytic model It alsodescribes the way in which metals released by stellar masslosses and supernovae explosions are spread among gas par-ticles of the N-BodySPH simulation In Section 4 we fix theparameters that characterise our version of the semi-analyticmodel and compare its results with local observations Thefollowing three sections present the main results of this workSection 5 gives the radial abundance profiles of differentchemical elements comparing them with observations InSection 6 we visualise the thermodynamical and chemicalproperties of the ICM by projecting the corresponding mass-weighted or emission-weighted related quantities These twotechniques (radial profiles and projected maps) are used toanalyse the history of the ICM chemical enrichment as wellas the dynamical evolution of the intracluster gas as dis-cussed in Section 7 Section 8 describes the use of Fe Kα

67 keV emission line as a probe of the ICM dynamics Fi-nally in Section 9 we give a summary and discussion of ourresults

Throughout this paper we express our model chemicalabundances in terms of the solar photospheric ones recentlyrecalibrated by Asplund Grevesse amp Sauval (2005) Whennecessary we refer the chemical abundances obtained in dif-ferent observational works to these standard solar values

2 HYBRID MODEL OF THE ICM CHEMICALENRICHMENT

Our hybrid model for studying the chemical enrichmentof the intracluster medium consists of a combination of anon-radiative N-BodySPH simulation of a galaxy clusterand a semi-analytic model of galaxy formation Dark mat-ter haloes and substructures that emerge in the simulationare tracked by the semi-analytic code and used to gener-ate the galaxy population (Springel et al 2001) The keyadditional component of our model is that we also use thediffuse gas component of the hydrodynamical simulation tofollow the enrichment process By enriching the gas parti-cles locally around galaxies we can account for the spreadingand mixing of metals by hydrodynamical processes therebyobtaining a model for the evolution of the spatial distri-bution of metals in the ICM This in turn allows a moredetailed analysis of the ICM metal enrichment than possi-ble with previous models based on semi-analytic techniques(Kauffmann amp Charlot 1998 DL04) In this section we pro-vide a detailed description of both parts of the modelling

21 Non-radiative N-BodySPH clustersimulation

In the present study we analyse a non-radiative hydro-dynamical simulation of a cluster of galaxies The clusterwas selected from the GIF-ΛCDM model (Kauffmann et al1999) characterised by cosmological parameters Ωo=03ΩΛ=07 and Ho = 100 h kmsminus1 Mpcminus1 with h = 07 ithas spectral shape Γ = 021 and was cluster-normalised to

σ8 = 09 The second most massive cluster in this parent cos-mological simulation (of mass Mvir = 84times1014 hminus1 M⊙) wasresimulated with higher resolution in the Lagrangian regionof the object and its immediate surroundings (Springel et al2001) with a number Nhr = 1999978 of high resolutionparticles Within this region each particle was split intodark matter (mdm = 118 times 109 hminus1M⊙) and gas (mgas =182times108 hminus1M⊙) according to Ωb = 004 and ΩDM = 026consistent with the baryon density Ωbh

2 = 002 required byBig-Bang nucleosynthesis constraints However the identifi-cation of dark matter haloes for the semi-analytic model wasbased only on the dark matter particles with their mass in-creased to its original value The boundary region wheremass resolution degrades with increasing distance fromthe cluster extends to a total diameter of 1413 hminus1 MpcThe simulation was carried out with the parallel tree N-BodySPH code GADGET (Springel et al 2001 Springel2005) The starting redshift and the gravitational soften-ing length are zstart = 50 and ε = 30 hminus1 kpc respectivelyWe note that this simulation did not include gas coolingThis is a reasonable approximation for cluster simulationssince the bulk of the halo gas has very long cooling timesHowever the cooling time becomes shorter than the age ofthe Universe in a central region with radius sim 100minus200 kpcfrom where the bulk of X-ray radiation is emitted (Sarazin1986)

22 Semi-analytic model of galaxy formation

The properties of galaxies in the semi-analytic model aredetermined by the included physical processes We considerthe effects of cooling of hot gas due to radiative losses starformation feedback from supernovae explosions metal pro-duction and merging of galaxies Except for the metal pro-duction the parametrization of these physical processes cor-responds to the model described by Springel et al (2001)However we modified the model to track the enrichmentcycle of metals between the different baryonic componentsof the haloes ie cold gas hot diffuse gas and stars in themost consistent way possible within our framework Our spe-cific chemical implementation is similar to the semi-analyticmodel discussed by DL04 In the present work we refine thechemical enrichment prescription of DL04 by tracking themass evolution of different chemical elements as providedby three kinds of sources low- and intermediate-mass starscore collapse supernovae and type Ia supernovae The firstgroup of sources yields metals through mass losses and stel-lar winds In the following we briefly summarise the detailsof the semi-analytic code In Section 3 we then describe thenew chemical implementation we introduce in this work

Based on the merging trees the mass of hot gas is calcu-lated at the beginning of the evolution between consecutiveoutputs of the simulation We assume that the hot gas al-ways has a distribution that parallels that of the dark matterhalo whose virial mass changes from one output to anotherdue to the hierarchical growth of structure Once a fractionof hot gas has cooled and star formation and feedback pro-cesses are triggered the mass of hot gas is given by

Mhot = fb Mvir minus

sumi

[M(i)stellar +M

(i)cold] (1)

where Mvir is the virial mass of the dark matter halo Mhot is

ccopy 2005 RAS MNRAS 000 1ndash24

4 S A Cora

the mass of the hot gas associated with it and Mstellar andMcold are the masses of stars and cold gas of each galaxycontained in the halo The virial mass is given by Mvir =100H2R3

virG This is the mass enclosed by the virial radiusRvir which is defined as the radius of a sphere with meandensity 200ρcrit centred on the most bound particle of thegroup Here ρcrit is the critical density The virial velocity isgiven by V 2

vir = GMvirRvirThe mass of hot gas that cools at each snapshot is given

by the cooling rate

dMcool

dt= 4πρgr

2cool

drcooldt

(2)

where ρg is the density profile of an isothermal sphere thathas been assumed for the distribution of hot gas within adark matter halo and rcool is the cooling radius The localcooling time is defined as the ratio of the specific thermal en-ergy content of the gas and the cooling rate per unit volumeΛ(TZ) The latter depends quite strongly on the metallicityZ of the hot gas and the temperature T = 359(Vvirkm sminus1)of the halo and is represented by the cooling functions com-puted by Sutherland amp Dopita (1993)

The star formation rate is given by

dM⋆

dt=

αMcold

tgxdyn (3)

where tgxdyn = 01RvirVvir is the dynamical time of thegalaxy and α is a dimensionless parameter that regulatesthe efficiency of star formation We adopt the variable starformation efficiency introduced by DL04 that depends on theproperties of the dark halo α(Vvir) = α0 (Vvir220 kmsminus1)nwith α0 and n as free parameters

Each star formation event generates a stellar mass∆M⋆ = M⋆∆TNsteps where Nstep = 50 are timesteps ofequal size used to subdivide the intervals between simulationoutputs ∆T and to integrate the differential equations thatdescribe the changes in mass and metals of each baryoniccomponent over these timespans Each solar mass of starsformed leads to a number ηCC of core collapse supernovaeThis class of supernovae includes those of type Ibc and IIthe former being generated when the progenitor looses itshydrogen-rich envelope before the explosion Ordinary typeII explosions are the most abundant ones however The en-ergy ESNCC released by each core collapse supernova is as-sumed to reheat some of the cold gas of a galaxy with amass of

∆Mreheat =4

3ǫηCC ESNCC

V 2vir

∆M⋆ (4)

where ǫ is a dimensionless parameter that regulates the ef-ficiency of the feedback process

The fate of the reheated gas is quite uncertain Weadopt a model in which the transfer from the cold to the hotphase does induce galactic outflows ie the reheated mass iskept within its host halo This model is referred to as reten-tion model The resulting cycle of metal enrichment is con-sistent with the way in which metals provided by the semi-analytic model are deposited in the gas particles around eachgalaxy in the underlying N-BodySPH simulation describedin Section 32 The effect of other feedback schemes (ejectionand wind models) on the chemical enrichment of the ICMis discussed by DL04 They found that these three prescrip-tions (including the retention model) predict a similar time

evolution of the ICM metal pollution which mainly occursat redshifts larger than 1 As we show in the present studygas dynamical processes driven during the cluster assemblyoccurring at present epochs contribute significantly to de-termine the spatial distribution of metals in the ICM verylikely erasing any signature of the feedback scheme involved

In a hierarchical scenario of structure formation merg-ers of galaxies are a natural consequence of the accretionand merger processes of dark matter haloes in which they re-side We directly use the simulation outputs to construct themerger histories of dark haloes To this end we first identifydark haloes as virialised particle groups by a friend-of-friend(FOF) algorithm The SUBFIND algorithm (Springel et al2001) is then applied to these groups in order to find self-bound dark matter substructures and the resulting set ofgravitationally bound structures is tracked over time to yieldmerging history trees

In this subhalo scheme we distinguish three types ofgalaxies when tracking galaxy formation along the mergingtrees The largest subhalo in a FOF group hosts the lsquocentralgalaxyrsquo of the group its position is given by the most boundparticle in that subhalo Central galaxies of other smallersubhaloes that are contained in a FOF group are referred toas lsquohalo galaxiesrsquo The (sub)haloes of these galaxies are stillintact after falling into larger structures The third group ofgalaxies comprises lsquosatellitersquo galaxies This type of galaxiesis generated when two subhaloes merge and the galaxy ofthe smaller one becomes a satellite of the remnant subhaloThese galaxies are assumed to merge on a dynamical time-scale with the halo-galaxy of the new subhalo they residein (Springel et al 2001 DL04) In these previous versionsof the semi-analytic model satellite galaxy positions weregiven by the most-bound particle identified at the last timethey were still a halo galaxy In the present work we insteadassume a circular orbit for these satellite galaxies with a ve-locity given by the virial velocity of the parent halo and de-caying radial distance to the corresponding central galaxyThis provides a more robust estimate of the position of satel-lite galaxies consistent with the dynamical friction formulaused (Binney amp Tremaine 1987)

With respect to the spectro-photometric properties ofgalaxies we apply the calculations made by DL04 con-sidering evolutionary synthesis models that depend on themetallicity of the cold gas from which the stars formed(Bruzual amp Charlot 1993) Our morphological classifica-tion of galaxies is based on the criterion adopted bySpringel et al (2001) who used lsquoshiftedrsquo values of theHubble-type T of galaxies to obtain a good morphology den-sity relation Thus we classify as S0 galaxies those with0 lt T lt 5 and as elliptical and spirals galaxies those withlower and higher values respectively

3 CHEMICAL IMPLEMENTATION

One of the main aims of this work is to reach a better un-derstanding of how the chemical enrichment of the ICMproceeds by establishing a connection between the spatialdistribution of several properties of the ICM such as sur-face mass density mass-weighted temperature distributionX-Ray surface brightness and chemical abundances

Recent observations of radial abundance pro-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 5

files of different elements (Finoguenov et al 2000De Grandi amp Molendi 2001 Gastaldello amp Molendi 2002De Grandi et al 2003 Tamura et al 2004) provide valuableconstraints on the physical processes involved especiallythose related to the feedback mechanisms that inject metalsinto the diffuse phase The available observational datacalls for a model that can explain the spatial distributionof different chemical elements Thus it is essential toimplement a chemical model that takes into account severaldifferent elements and discriminates between differentsources such as SNe CC and SNe Ia

In our model information about the spatial distributionof the ICM properties can be obtained from the gas par-ticles in the hydro-simulation However the non-radiativesimulation used only give the thermodynamic properties ofthe diffuse hot gas In order to obtain the chemical prop-erties of gas particles we have to establish a link betweenthe chemical implementation in the semi-analytic model andthe metal enrichment of the gas particles In the followingwe will first describe in detail how we model the enrichmentcycle of metals among the different baryonic components inthe semi-analytic model and then explain how these quan-tities are tied to the gas dynamics in the simulation

31 Enrichment cycle of chemical elements in thesemi-analytic model

In our model stars can contaminate the cold and hot gasbecause of mass loss during their stellar evolution and metalejection at the end of their lives The hot gas has primordialabundances initially (76 per cent of hydrogen and 24 percent of helium) but becomes chemically enriched as a resultof the transfer of contaminated cold gas to the hot phase dueto reheating by supernovae explosions This chemical enrich-ment has a strong influence on the amount of hot gas thatcan cool since we are using metal dependent cooling ratesThis process in turn influences the star formation activitywhich is ultimately responsible for the chemical pollution

Our semi-analytic model considers mass losses by starsin different mass ranges from low and intermediate stars toquasi-massive and massive stars taking into account stellarlifetimes Massive stars give raise to SNe CC Ejecta fromSNe Ia are also included The cold gas component of eachgalaxy becomes gradually more chemically contaminated asstar formation proceeds and consequently new stars thatare formed are progressively more metal rich This fact callsfor the use of metallicity dependent stellar yields Howevervariations of yields with stellar metallicities are very smallso we only consider the mass dependence of stellar yieldsWe adopt yields of different chemical elements estimated forstars with solar heavy element abundance

Thus for each stellar mass ∆M⋆ formed we determinethe fraction of mass of stars contained in a given mass rangeby assuming an Initial Mass Function (IMF) Φ(M) Themass of chemical element j ejected by stars with masses inthe range centred in mk with lower and upper limits ml

k

and muk is given by

∆M j

ejk= [Rk X

j + Y j

k ] ∆M⋆ (5)

where Xj is the solar abundance of element j and Rk andY j

k are the recycled fraction and the yield of element j for

the k-th mass range respectively The recycled fraction isgiven by

Rk =

mu

kint

ml

k

Φ(M) rk dM

with rk defined as the difference between the mass mk ofthe star at birth and its remnant mass after mass loss dueto stellar winds andor supernova explosions The yield ofchemical element j newly formed is

Y j

k =

mu

kint

ml

k

Φ(M) pjk dM

where pjk is the stellar yield of a chemical element j producedby a star with mass in the range centred around mk Thevariable j may refer to H He or chemical species with atomicnumber larger than 2

The mass of species j ejected by each SN Ia is labelledmj

Ia In this case we are dealing with the total mass ejectedbecause this type of supernovae leaves no remnants thus

∆M j

ej(Ia)k

= ηIak mj

Ia ∆M⋆ (6)

where ηIak is the number of SNe Ia per stellar mass formedin the mass interval k which depend on the model adoptedfor this type of SNe Note that SNe Ia do not contribute toH or He hence M j

ej(Ia)kis zero when referring to these two

chemical elementsThe transport of different chemical species between hot

gas cold gas and stars can be expressed as variation ofthe mass of the chemical element j present in each baryoniccomponent namely

∆M j

hot = minus∆Mcool Aj

hot +∆Mreheat Aj

cold (7)

∆M j

cold = +∆Mcool Aj

hot minus∆M⋆ Aj

cold(SF)

+∆M j

ej +∆M j

ej(Ia) minus∆Mreheat Aj

cold (8)

∆M j

stellar = +∆M⋆ Aj

cold(SF) minusRAjstar ∆M⋆ (9)

where Aj

B = M j

BMB is the abundance of different chemicalelements in each baryonic component which involves thetotal mass of the baryonic component MB and the mass ofspecies j contained in it M j

B The suffix B may refer to hotgas cold gas or stars Aj

cold(SF)denotes the abundance of the

element j in the cold gas at the time of birth of ∆M⋆ Theabove set of equations takes into account the accumulatedcontribution of different mass ranges that affect the baryoniccomponents at a given time as a result of the combinationof the star formation rate of each galaxy and the returntime-scale of the ejecta from all sources considered

32 Injection of metals in the diffuse gas

We describe now the procedure used to distribute the chem-ical elements generated by the galaxies in the semi-analyticmodel among the gas particles of the N-BodySPH simula-tion For each snapshot we identify the gas particles con-tained within spheres of radius 100 hminus1 kpc centred on each

ccopy 2005 RAS MNRAS 000 1ndash24

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 2: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

2 S A Cora

inated in the galaxies that reside in the cluster The pro-cesses considered for the supply of this enriched gas includegalactic winds driven by supernovae explosions (White 1991Renzini 1997) ram pressure stripping (Mori amp Burkert2000) early enrichment by hypernovae associated with pop-ulation type III stars (Loewenstein 2001) and intraclusterstars (Zaritsky Gonzalez amp Zabludoff 2004)

Spatially resolved maps of the temperature andmetal abundance distributions in the intracluster mediumhave been obtained by X-ray surveys carried out withROSAT ASCA BeppoSAX (Finoguenov David ampPonman 2000 Finoguenov Arnaud amp David 2001De Grandi amp Molendi 2001) and more recently withXMM-Newton and Chandra (Gastaldello amp Molendi2002 Finoguenov et al 2002 Sanders amp Fabian 2002Matsushita Finoguenov amp Bohringer 2003 Tamura et al2004) As a result abundance profiles of several chemicalspecies with different origins have been obtained namelyfor O and Si generated by core collapse supernovae (SNeCC) and Fe mainly produced by type Ia supernovae (SNeIa) These profiles reflect physical mixing processes in thebaryonic component due to gas cooling star formationmetal production and energetic and chemical feedback allcoupled to the hierarchical build up of the galaxies and theclusters These observations can therefore be very valuablefor constraining models of galaxy formation and chemicalenrichment of the ICM allowing to evaluate the relativeimportance of different types of supernovae for metalpollution The spatial distribution of metals obtained frommodels of the ICM chemical enrichment can be analysed byconstructing X-ray weighted metal maps which trace theinteractions between the ICM and cluster galaxies thuswhen they are compared with observed metal maps theyprovide information about the mechanisms involved in theenrichment process and may reveal the dynamical state of agalaxy cluster (Kapferer et al 2005 Domainko et al 2005)

X-ray observations also provide important informationabout the dynamics of the intracluster gas through radial ve-locity measurements in clusters (Dupke amp Bregman 2001)allowing in principle further tests of structure formationmodels Gaseous bulk flows affect the characteristics of X-ray spectra of the ICM by Doppler broadening and shiftingof emission lines The prominent Fe Kα 67 keV iron emis-sion line present in X-ray spectra of galaxy clusters couldbe used as a tracer of these motions This possibility wasevaluated by Sunyaev et al (2003) who showed that futureX-ray missions like CONSTELLATION-X and XEUS withan energy resolution of a few eV should be able to detectseveral Doppler-shifted components of this emission line inthe core of the cluster

All this wealth of data calls for a theoretical interpre-tation of the ICM chemical enrichment in the framework ofa detailed cluster formation model The approaches used sofar to this end include semi-analytic models of galaxy forma-tion coupled to N-Body simulations of galaxy clusters (DeLucia Kauffmann amp White 2004 Nagashima et al 2005)hydrodynamical simulations of cluster formation which in-clude self-consistently star formation SNe feedback andmetal enrichment from SNe CC and Ia (eg Valdarnini 2003Tornatore et al 2004) and an intermediate approach thatcombines N-Body and hydrodynamic simulations togetherwith a phenomenological galaxy formation model and a pre-

scription of the effect under study such as galactic winds andmerger-driven starbursts (Kapferer et al 2005) and ram-pressure stripping (Schindler et al 2005 Domainko et al2005) In this work we present a different intermediate ap-proach which combines a cosmological non-radiative hydro-dynamical N-BodySPH simulation of a cluster of galax-ies with a semi-analytic model of galaxy formation Thesimplicity of the semi-analytic model has the advantage ofreaching a larger dynamic range than fully self-consistenthydro-simulations at a far smaller computational cost Inparticular it allows us to explore more easily the range ofparameters that characterise appropriate chemical enrich-ment models

The long cooling time of the bulk of the gas in rich clus-ters justifies the assumption of a non-radiative gas (Evrard1990 Frenk et al 1999) However this time-scale becomessmaller than the Hubble time in the core of the clusterwhere the densities are higher Here is where the semi-analytic model plays an important part taking into accountradiative cooling and models for star formation chemi-cal enrichment and energetic feedback from galaxies Thesemi-analytic model used here is based on previous works(Springel et al 2001 De Lucia et al 2004 DL04 hereafter)but was extended with a new chemical implementation thattracks the abundance of different species resulting from massloss through stellar winds of low- intermediate- and high-mass stars and from the explosions of SNe CC and Ia su-pernovae The principal new feature of our model is how-ever the link between semi-analytic model results and thechemical enrichment of gas particles in the underlying N-BodySPH simulation We pollute gas particles with met-als ejected from the galaxies consistent with the modellingof the semi-analytic model These metals are then carriedaround and mixed by the hydrodynamic processes duringcluster formation Thus we are not limited to an analysisof mean chemical properties of the intracluster gas like inthe original model of DL04 Instead the enrichment of gasparticles allows a study of the spatial distribution of metalsin the ICM and to use iron emissivity as a tracer of gasmotions

The aim of our model is on one hand to understandthe way in which the chemical enrichment patterns char-acterising the intracluster gas develop and on the otherto detect the imprints of gas bulk motions in the shape ofX-ray emission lines thus providing guidance for the inter-pretation of future spectroscopic X-ray data The first partof the present study is therefore devoted to an analysis ofchemical enrichment of the ICM focusing on the spatial dis-tribution of different chemical elements In a second part weanalyse the link between the occurrence of multiple com-ponents in the Fe Kα 67 keV emission line and the rangeof velocities of the gas that generates this emission Thisis directly relevant to the question whether metal emission-line-weighted velocity maps of the cluster could be used toinfer the spatial coherence of these bulk motions Gatheringall the information available in the form of projected mapsand detailed spectra along lines of sight through the ICMwe are able to extend the information in 2-D provided byobserved projected distributions of gas temperature abun-dances and surface brightness into a 3-D picture of the ICMproperties

This paper is organised as follows Section 2 describes

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 3

our hybrid model used to study the ICM chemical enrich-ment summarises the properties of the hydrodynamical sim-ulation used and presents the main features of the semi-analytic model Section 3 contains the details of the chem-ical model implemented in the semi-analytic model It alsodescribes the way in which metals released by stellar masslosses and supernovae explosions are spread among gas par-ticles of the N-BodySPH simulation In Section 4 we fix theparameters that characterise our version of the semi-analyticmodel and compare its results with local observations Thefollowing three sections present the main results of this workSection 5 gives the radial abundance profiles of differentchemical elements comparing them with observations InSection 6 we visualise the thermodynamical and chemicalproperties of the ICM by projecting the corresponding mass-weighted or emission-weighted related quantities These twotechniques (radial profiles and projected maps) are used toanalyse the history of the ICM chemical enrichment as wellas the dynamical evolution of the intracluster gas as dis-cussed in Section 7 Section 8 describes the use of Fe Kα

67 keV emission line as a probe of the ICM dynamics Fi-nally in Section 9 we give a summary and discussion of ourresults

Throughout this paper we express our model chemicalabundances in terms of the solar photospheric ones recentlyrecalibrated by Asplund Grevesse amp Sauval (2005) Whennecessary we refer the chemical abundances obtained in dif-ferent observational works to these standard solar values

2 HYBRID MODEL OF THE ICM CHEMICALENRICHMENT

Our hybrid model for studying the chemical enrichmentof the intracluster medium consists of a combination of anon-radiative N-BodySPH simulation of a galaxy clusterand a semi-analytic model of galaxy formation Dark mat-ter haloes and substructures that emerge in the simulationare tracked by the semi-analytic code and used to gener-ate the galaxy population (Springel et al 2001) The keyadditional component of our model is that we also use thediffuse gas component of the hydrodynamical simulation tofollow the enrichment process By enriching the gas parti-cles locally around galaxies we can account for the spreadingand mixing of metals by hydrodynamical processes therebyobtaining a model for the evolution of the spatial distri-bution of metals in the ICM This in turn allows a moredetailed analysis of the ICM metal enrichment than possi-ble with previous models based on semi-analytic techniques(Kauffmann amp Charlot 1998 DL04) In this section we pro-vide a detailed description of both parts of the modelling

21 Non-radiative N-BodySPH clustersimulation

In the present study we analyse a non-radiative hydro-dynamical simulation of a cluster of galaxies The clusterwas selected from the GIF-ΛCDM model (Kauffmann et al1999) characterised by cosmological parameters Ωo=03ΩΛ=07 and Ho = 100 h kmsminus1 Mpcminus1 with h = 07 ithas spectral shape Γ = 021 and was cluster-normalised to

σ8 = 09 The second most massive cluster in this parent cos-mological simulation (of mass Mvir = 84times1014 hminus1 M⊙) wasresimulated with higher resolution in the Lagrangian regionof the object and its immediate surroundings (Springel et al2001) with a number Nhr = 1999978 of high resolutionparticles Within this region each particle was split intodark matter (mdm = 118 times 109 hminus1M⊙) and gas (mgas =182times108 hminus1M⊙) according to Ωb = 004 and ΩDM = 026consistent with the baryon density Ωbh

2 = 002 required byBig-Bang nucleosynthesis constraints However the identifi-cation of dark matter haloes for the semi-analytic model wasbased only on the dark matter particles with their mass in-creased to its original value The boundary region wheremass resolution degrades with increasing distance fromthe cluster extends to a total diameter of 1413 hminus1 MpcThe simulation was carried out with the parallel tree N-BodySPH code GADGET (Springel et al 2001 Springel2005) The starting redshift and the gravitational soften-ing length are zstart = 50 and ε = 30 hminus1 kpc respectivelyWe note that this simulation did not include gas coolingThis is a reasonable approximation for cluster simulationssince the bulk of the halo gas has very long cooling timesHowever the cooling time becomes shorter than the age ofthe Universe in a central region with radius sim 100minus200 kpcfrom where the bulk of X-ray radiation is emitted (Sarazin1986)

22 Semi-analytic model of galaxy formation

The properties of galaxies in the semi-analytic model aredetermined by the included physical processes We considerthe effects of cooling of hot gas due to radiative losses starformation feedback from supernovae explosions metal pro-duction and merging of galaxies Except for the metal pro-duction the parametrization of these physical processes cor-responds to the model described by Springel et al (2001)However we modified the model to track the enrichmentcycle of metals between the different baryonic componentsof the haloes ie cold gas hot diffuse gas and stars in themost consistent way possible within our framework Our spe-cific chemical implementation is similar to the semi-analyticmodel discussed by DL04 In the present work we refine thechemical enrichment prescription of DL04 by tracking themass evolution of different chemical elements as providedby three kinds of sources low- and intermediate-mass starscore collapse supernovae and type Ia supernovae The firstgroup of sources yields metals through mass losses and stel-lar winds In the following we briefly summarise the detailsof the semi-analytic code In Section 3 we then describe thenew chemical implementation we introduce in this work

Based on the merging trees the mass of hot gas is calcu-lated at the beginning of the evolution between consecutiveoutputs of the simulation We assume that the hot gas al-ways has a distribution that parallels that of the dark matterhalo whose virial mass changes from one output to anotherdue to the hierarchical growth of structure Once a fractionof hot gas has cooled and star formation and feedback pro-cesses are triggered the mass of hot gas is given by

Mhot = fb Mvir minus

sumi

[M(i)stellar +M

(i)cold] (1)

where Mvir is the virial mass of the dark matter halo Mhot is

ccopy 2005 RAS MNRAS 000 1ndash24

4 S A Cora

the mass of the hot gas associated with it and Mstellar andMcold are the masses of stars and cold gas of each galaxycontained in the halo The virial mass is given by Mvir =100H2R3

virG This is the mass enclosed by the virial radiusRvir which is defined as the radius of a sphere with meandensity 200ρcrit centred on the most bound particle of thegroup Here ρcrit is the critical density The virial velocity isgiven by V 2

vir = GMvirRvirThe mass of hot gas that cools at each snapshot is given

by the cooling rate

dMcool

dt= 4πρgr

2cool

drcooldt

(2)

where ρg is the density profile of an isothermal sphere thathas been assumed for the distribution of hot gas within adark matter halo and rcool is the cooling radius The localcooling time is defined as the ratio of the specific thermal en-ergy content of the gas and the cooling rate per unit volumeΛ(TZ) The latter depends quite strongly on the metallicityZ of the hot gas and the temperature T = 359(Vvirkm sminus1)of the halo and is represented by the cooling functions com-puted by Sutherland amp Dopita (1993)

The star formation rate is given by

dM⋆

dt=

αMcold

tgxdyn (3)

where tgxdyn = 01RvirVvir is the dynamical time of thegalaxy and α is a dimensionless parameter that regulatesthe efficiency of star formation We adopt the variable starformation efficiency introduced by DL04 that depends on theproperties of the dark halo α(Vvir) = α0 (Vvir220 kmsminus1)nwith α0 and n as free parameters

Each star formation event generates a stellar mass∆M⋆ = M⋆∆TNsteps where Nstep = 50 are timesteps ofequal size used to subdivide the intervals between simulationoutputs ∆T and to integrate the differential equations thatdescribe the changes in mass and metals of each baryoniccomponent over these timespans Each solar mass of starsformed leads to a number ηCC of core collapse supernovaeThis class of supernovae includes those of type Ibc and IIthe former being generated when the progenitor looses itshydrogen-rich envelope before the explosion Ordinary typeII explosions are the most abundant ones however The en-ergy ESNCC released by each core collapse supernova is as-sumed to reheat some of the cold gas of a galaxy with amass of

∆Mreheat =4

3ǫηCC ESNCC

V 2vir

∆M⋆ (4)

where ǫ is a dimensionless parameter that regulates the ef-ficiency of the feedback process

The fate of the reheated gas is quite uncertain Weadopt a model in which the transfer from the cold to the hotphase does induce galactic outflows ie the reheated mass iskept within its host halo This model is referred to as reten-tion model The resulting cycle of metal enrichment is con-sistent with the way in which metals provided by the semi-analytic model are deposited in the gas particles around eachgalaxy in the underlying N-BodySPH simulation describedin Section 32 The effect of other feedback schemes (ejectionand wind models) on the chemical enrichment of the ICMis discussed by DL04 They found that these three prescrip-tions (including the retention model) predict a similar time

evolution of the ICM metal pollution which mainly occursat redshifts larger than 1 As we show in the present studygas dynamical processes driven during the cluster assemblyoccurring at present epochs contribute significantly to de-termine the spatial distribution of metals in the ICM verylikely erasing any signature of the feedback scheme involved

In a hierarchical scenario of structure formation merg-ers of galaxies are a natural consequence of the accretionand merger processes of dark matter haloes in which they re-side We directly use the simulation outputs to construct themerger histories of dark haloes To this end we first identifydark haloes as virialised particle groups by a friend-of-friend(FOF) algorithm The SUBFIND algorithm (Springel et al2001) is then applied to these groups in order to find self-bound dark matter substructures and the resulting set ofgravitationally bound structures is tracked over time to yieldmerging history trees

In this subhalo scheme we distinguish three types ofgalaxies when tracking galaxy formation along the mergingtrees The largest subhalo in a FOF group hosts the lsquocentralgalaxyrsquo of the group its position is given by the most boundparticle in that subhalo Central galaxies of other smallersubhaloes that are contained in a FOF group are referred toas lsquohalo galaxiesrsquo The (sub)haloes of these galaxies are stillintact after falling into larger structures The third group ofgalaxies comprises lsquosatellitersquo galaxies This type of galaxiesis generated when two subhaloes merge and the galaxy ofthe smaller one becomes a satellite of the remnant subhaloThese galaxies are assumed to merge on a dynamical time-scale with the halo-galaxy of the new subhalo they residein (Springel et al 2001 DL04) In these previous versionsof the semi-analytic model satellite galaxy positions weregiven by the most-bound particle identified at the last timethey were still a halo galaxy In the present work we insteadassume a circular orbit for these satellite galaxies with a ve-locity given by the virial velocity of the parent halo and de-caying radial distance to the corresponding central galaxyThis provides a more robust estimate of the position of satel-lite galaxies consistent with the dynamical friction formulaused (Binney amp Tremaine 1987)

With respect to the spectro-photometric properties ofgalaxies we apply the calculations made by DL04 con-sidering evolutionary synthesis models that depend on themetallicity of the cold gas from which the stars formed(Bruzual amp Charlot 1993) Our morphological classifica-tion of galaxies is based on the criterion adopted bySpringel et al (2001) who used lsquoshiftedrsquo values of theHubble-type T of galaxies to obtain a good morphology den-sity relation Thus we classify as S0 galaxies those with0 lt T lt 5 and as elliptical and spirals galaxies those withlower and higher values respectively

3 CHEMICAL IMPLEMENTATION

One of the main aims of this work is to reach a better un-derstanding of how the chemical enrichment of the ICMproceeds by establishing a connection between the spatialdistribution of several properties of the ICM such as sur-face mass density mass-weighted temperature distributionX-Ray surface brightness and chemical abundances

Recent observations of radial abundance pro-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 5

files of different elements (Finoguenov et al 2000De Grandi amp Molendi 2001 Gastaldello amp Molendi 2002De Grandi et al 2003 Tamura et al 2004) provide valuableconstraints on the physical processes involved especiallythose related to the feedback mechanisms that inject metalsinto the diffuse phase The available observational datacalls for a model that can explain the spatial distributionof different chemical elements Thus it is essential toimplement a chemical model that takes into account severaldifferent elements and discriminates between differentsources such as SNe CC and SNe Ia

In our model information about the spatial distributionof the ICM properties can be obtained from the gas par-ticles in the hydro-simulation However the non-radiativesimulation used only give the thermodynamic properties ofthe diffuse hot gas In order to obtain the chemical prop-erties of gas particles we have to establish a link betweenthe chemical implementation in the semi-analytic model andthe metal enrichment of the gas particles In the followingwe will first describe in detail how we model the enrichmentcycle of metals among the different baryonic components inthe semi-analytic model and then explain how these quan-tities are tied to the gas dynamics in the simulation

31 Enrichment cycle of chemical elements in thesemi-analytic model

In our model stars can contaminate the cold and hot gasbecause of mass loss during their stellar evolution and metalejection at the end of their lives The hot gas has primordialabundances initially (76 per cent of hydrogen and 24 percent of helium) but becomes chemically enriched as a resultof the transfer of contaminated cold gas to the hot phase dueto reheating by supernovae explosions This chemical enrich-ment has a strong influence on the amount of hot gas thatcan cool since we are using metal dependent cooling ratesThis process in turn influences the star formation activitywhich is ultimately responsible for the chemical pollution

Our semi-analytic model considers mass losses by starsin different mass ranges from low and intermediate stars toquasi-massive and massive stars taking into account stellarlifetimes Massive stars give raise to SNe CC Ejecta fromSNe Ia are also included The cold gas component of eachgalaxy becomes gradually more chemically contaminated asstar formation proceeds and consequently new stars thatare formed are progressively more metal rich This fact callsfor the use of metallicity dependent stellar yields Howevervariations of yields with stellar metallicities are very smallso we only consider the mass dependence of stellar yieldsWe adopt yields of different chemical elements estimated forstars with solar heavy element abundance

Thus for each stellar mass ∆M⋆ formed we determinethe fraction of mass of stars contained in a given mass rangeby assuming an Initial Mass Function (IMF) Φ(M) Themass of chemical element j ejected by stars with masses inthe range centred in mk with lower and upper limits ml

k

and muk is given by

∆M j

ejk= [Rk X

j + Y j

k ] ∆M⋆ (5)

where Xj is the solar abundance of element j and Rk andY j

k are the recycled fraction and the yield of element j for

the k-th mass range respectively The recycled fraction isgiven by

Rk =

mu

kint

ml

k

Φ(M) rk dM

with rk defined as the difference between the mass mk ofthe star at birth and its remnant mass after mass loss dueto stellar winds andor supernova explosions The yield ofchemical element j newly formed is

Y j

k =

mu

kint

ml

k

Φ(M) pjk dM

where pjk is the stellar yield of a chemical element j producedby a star with mass in the range centred around mk Thevariable j may refer to H He or chemical species with atomicnumber larger than 2

The mass of species j ejected by each SN Ia is labelledmj

Ia In this case we are dealing with the total mass ejectedbecause this type of supernovae leaves no remnants thus

∆M j

ej(Ia)k

= ηIak mj

Ia ∆M⋆ (6)

where ηIak is the number of SNe Ia per stellar mass formedin the mass interval k which depend on the model adoptedfor this type of SNe Note that SNe Ia do not contribute toH or He hence M j

ej(Ia)kis zero when referring to these two

chemical elementsThe transport of different chemical species between hot

gas cold gas and stars can be expressed as variation ofthe mass of the chemical element j present in each baryoniccomponent namely

∆M j

hot = minus∆Mcool Aj

hot +∆Mreheat Aj

cold (7)

∆M j

cold = +∆Mcool Aj

hot minus∆M⋆ Aj

cold(SF)

+∆M j

ej +∆M j

ej(Ia) minus∆Mreheat Aj

cold (8)

∆M j

stellar = +∆M⋆ Aj

cold(SF) minusRAjstar ∆M⋆ (9)

where Aj

B = M j

BMB is the abundance of different chemicalelements in each baryonic component which involves thetotal mass of the baryonic component MB and the mass ofspecies j contained in it M j

B The suffix B may refer to hotgas cold gas or stars Aj

cold(SF)denotes the abundance of the

element j in the cold gas at the time of birth of ∆M⋆ Theabove set of equations takes into account the accumulatedcontribution of different mass ranges that affect the baryoniccomponents at a given time as a result of the combinationof the star formation rate of each galaxy and the returntime-scale of the ejecta from all sources considered

32 Injection of metals in the diffuse gas

We describe now the procedure used to distribute the chem-ical elements generated by the galaxies in the semi-analyticmodel among the gas particles of the N-BodySPH simula-tion For each snapshot we identify the gas particles con-tained within spheres of radius 100 hminus1 kpc centred on each

ccopy 2005 RAS MNRAS 000 1ndash24

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 3: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 3

our hybrid model used to study the ICM chemical enrich-ment summarises the properties of the hydrodynamical sim-ulation used and presents the main features of the semi-analytic model Section 3 contains the details of the chem-ical model implemented in the semi-analytic model It alsodescribes the way in which metals released by stellar masslosses and supernovae explosions are spread among gas par-ticles of the N-BodySPH simulation In Section 4 we fix theparameters that characterise our version of the semi-analyticmodel and compare its results with local observations Thefollowing three sections present the main results of this workSection 5 gives the radial abundance profiles of differentchemical elements comparing them with observations InSection 6 we visualise the thermodynamical and chemicalproperties of the ICM by projecting the corresponding mass-weighted or emission-weighted related quantities These twotechniques (radial profiles and projected maps) are used toanalyse the history of the ICM chemical enrichment as wellas the dynamical evolution of the intracluster gas as dis-cussed in Section 7 Section 8 describes the use of Fe Kα

67 keV emission line as a probe of the ICM dynamics Fi-nally in Section 9 we give a summary and discussion of ourresults

Throughout this paper we express our model chemicalabundances in terms of the solar photospheric ones recentlyrecalibrated by Asplund Grevesse amp Sauval (2005) Whennecessary we refer the chemical abundances obtained in dif-ferent observational works to these standard solar values

2 HYBRID MODEL OF THE ICM CHEMICALENRICHMENT

Our hybrid model for studying the chemical enrichmentof the intracluster medium consists of a combination of anon-radiative N-BodySPH simulation of a galaxy clusterand a semi-analytic model of galaxy formation Dark mat-ter haloes and substructures that emerge in the simulationare tracked by the semi-analytic code and used to gener-ate the galaxy population (Springel et al 2001) The keyadditional component of our model is that we also use thediffuse gas component of the hydrodynamical simulation tofollow the enrichment process By enriching the gas parti-cles locally around galaxies we can account for the spreadingand mixing of metals by hydrodynamical processes therebyobtaining a model for the evolution of the spatial distri-bution of metals in the ICM This in turn allows a moredetailed analysis of the ICM metal enrichment than possi-ble with previous models based on semi-analytic techniques(Kauffmann amp Charlot 1998 DL04) In this section we pro-vide a detailed description of both parts of the modelling

21 Non-radiative N-BodySPH clustersimulation

In the present study we analyse a non-radiative hydro-dynamical simulation of a cluster of galaxies The clusterwas selected from the GIF-ΛCDM model (Kauffmann et al1999) characterised by cosmological parameters Ωo=03ΩΛ=07 and Ho = 100 h kmsminus1 Mpcminus1 with h = 07 ithas spectral shape Γ = 021 and was cluster-normalised to

σ8 = 09 The second most massive cluster in this parent cos-mological simulation (of mass Mvir = 84times1014 hminus1 M⊙) wasresimulated with higher resolution in the Lagrangian regionof the object and its immediate surroundings (Springel et al2001) with a number Nhr = 1999978 of high resolutionparticles Within this region each particle was split intodark matter (mdm = 118 times 109 hminus1M⊙) and gas (mgas =182times108 hminus1M⊙) according to Ωb = 004 and ΩDM = 026consistent with the baryon density Ωbh

2 = 002 required byBig-Bang nucleosynthesis constraints However the identifi-cation of dark matter haloes for the semi-analytic model wasbased only on the dark matter particles with their mass in-creased to its original value The boundary region wheremass resolution degrades with increasing distance fromthe cluster extends to a total diameter of 1413 hminus1 MpcThe simulation was carried out with the parallel tree N-BodySPH code GADGET (Springel et al 2001 Springel2005) The starting redshift and the gravitational soften-ing length are zstart = 50 and ε = 30 hminus1 kpc respectivelyWe note that this simulation did not include gas coolingThis is a reasonable approximation for cluster simulationssince the bulk of the halo gas has very long cooling timesHowever the cooling time becomes shorter than the age ofthe Universe in a central region with radius sim 100minus200 kpcfrom where the bulk of X-ray radiation is emitted (Sarazin1986)

22 Semi-analytic model of galaxy formation

The properties of galaxies in the semi-analytic model aredetermined by the included physical processes We considerthe effects of cooling of hot gas due to radiative losses starformation feedback from supernovae explosions metal pro-duction and merging of galaxies Except for the metal pro-duction the parametrization of these physical processes cor-responds to the model described by Springel et al (2001)However we modified the model to track the enrichmentcycle of metals between the different baryonic componentsof the haloes ie cold gas hot diffuse gas and stars in themost consistent way possible within our framework Our spe-cific chemical implementation is similar to the semi-analyticmodel discussed by DL04 In the present work we refine thechemical enrichment prescription of DL04 by tracking themass evolution of different chemical elements as providedby three kinds of sources low- and intermediate-mass starscore collapse supernovae and type Ia supernovae The firstgroup of sources yields metals through mass losses and stel-lar winds In the following we briefly summarise the detailsof the semi-analytic code In Section 3 we then describe thenew chemical implementation we introduce in this work

Based on the merging trees the mass of hot gas is calcu-lated at the beginning of the evolution between consecutiveoutputs of the simulation We assume that the hot gas al-ways has a distribution that parallels that of the dark matterhalo whose virial mass changes from one output to anotherdue to the hierarchical growth of structure Once a fractionof hot gas has cooled and star formation and feedback pro-cesses are triggered the mass of hot gas is given by

Mhot = fb Mvir minus

sumi

[M(i)stellar +M

(i)cold] (1)

where Mvir is the virial mass of the dark matter halo Mhot is

ccopy 2005 RAS MNRAS 000 1ndash24

4 S A Cora

the mass of the hot gas associated with it and Mstellar andMcold are the masses of stars and cold gas of each galaxycontained in the halo The virial mass is given by Mvir =100H2R3

virG This is the mass enclosed by the virial radiusRvir which is defined as the radius of a sphere with meandensity 200ρcrit centred on the most bound particle of thegroup Here ρcrit is the critical density The virial velocity isgiven by V 2

vir = GMvirRvirThe mass of hot gas that cools at each snapshot is given

by the cooling rate

dMcool

dt= 4πρgr

2cool

drcooldt

(2)

where ρg is the density profile of an isothermal sphere thathas been assumed for the distribution of hot gas within adark matter halo and rcool is the cooling radius The localcooling time is defined as the ratio of the specific thermal en-ergy content of the gas and the cooling rate per unit volumeΛ(TZ) The latter depends quite strongly on the metallicityZ of the hot gas and the temperature T = 359(Vvirkm sminus1)of the halo and is represented by the cooling functions com-puted by Sutherland amp Dopita (1993)

The star formation rate is given by

dM⋆

dt=

αMcold

tgxdyn (3)

where tgxdyn = 01RvirVvir is the dynamical time of thegalaxy and α is a dimensionless parameter that regulatesthe efficiency of star formation We adopt the variable starformation efficiency introduced by DL04 that depends on theproperties of the dark halo α(Vvir) = α0 (Vvir220 kmsminus1)nwith α0 and n as free parameters

Each star formation event generates a stellar mass∆M⋆ = M⋆∆TNsteps where Nstep = 50 are timesteps ofequal size used to subdivide the intervals between simulationoutputs ∆T and to integrate the differential equations thatdescribe the changes in mass and metals of each baryoniccomponent over these timespans Each solar mass of starsformed leads to a number ηCC of core collapse supernovaeThis class of supernovae includes those of type Ibc and IIthe former being generated when the progenitor looses itshydrogen-rich envelope before the explosion Ordinary typeII explosions are the most abundant ones however The en-ergy ESNCC released by each core collapse supernova is as-sumed to reheat some of the cold gas of a galaxy with amass of

∆Mreheat =4

3ǫηCC ESNCC

V 2vir

∆M⋆ (4)

where ǫ is a dimensionless parameter that regulates the ef-ficiency of the feedback process

The fate of the reheated gas is quite uncertain Weadopt a model in which the transfer from the cold to the hotphase does induce galactic outflows ie the reheated mass iskept within its host halo This model is referred to as reten-tion model The resulting cycle of metal enrichment is con-sistent with the way in which metals provided by the semi-analytic model are deposited in the gas particles around eachgalaxy in the underlying N-BodySPH simulation describedin Section 32 The effect of other feedback schemes (ejectionand wind models) on the chemical enrichment of the ICMis discussed by DL04 They found that these three prescrip-tions (including the retention model) predict a similar time

evolution of the ICM metal pollution which mainly occursat redshifts larger than 1 As we show in the present studygas dynamical processes driven during the cluster assemblyoccurring at present epochs contribute significantly to de-termine the spatial distribution of metals in the ICM verylikely erasing any signature of the feedback scheme involved

In a hierarchical scenario of structure formation merg-ers of galaxies are a natural consequence of the accretionand merger processes of dark matter haloes in which they re-side We directly use the simulation outputs to construct themerger histories of dark haloes To this end we first identifydark haloes as virialised particle groups by a friend-of-friend(FOF) algorithm The SUBFIND algorithm (Springel et al2001) is then applied to these groups in order to find self-bound dark matter substructures and the resulting set ofgravitationally bound structures is tracked over time to yieldmerging history trees

In this subhalo scheme we distinguish three types ofgalaxies when tracking galaxy formation along the mergingtrees The largest subhalo in a FOF group hosts the lsquocentralgalaxyrsquo of the group its position is given by the most boundparticle in that subhalo Central galaxies of other smallersubhaloes that are contained in a FOF group are referred toas lsquohalo galaxiesrsquo The (sub)haloes of these galaxies are stillintact after falling into larger structures The third group ofgalaxies comprises lsquosatellitersquo galaxies This type of galaxiesis generated when two subhaloes merge and the galaxy ofthe smaller one becomes a satellite of the remnant subhaloThese galaxies are assumed to merge on a dynamical time-scale with the halo-galaxy of the new subhalo they residein (Springel et al 2001 DL04) In these previous versionsof the semi-analytic model satellite galaxy positions weregiven by the most-bound particle identified at the last timethey were still a halo galaxy In the present work we insteadassume a circular orbit for these satellite galaxies with a ve-locity given by the virial velocity of the parent halo and de-caying radial distance to the corresponding central galaxyThis provides a more robust estimate of the position of satel-lite galaxies consistent with the dynamical friction formulaused (Binney amp Tremaine 1987)

With respect to the spectro-photometric properties ofgalaxies we apply the calculations made by DL04 con-sidering evolutionary synthesis models that depend on themetallicity of the cold gas from which the stars formed(Bruzual amp Charlot 1993) Our morphological classifica-tion of galaxies is based on the criterion adopted bySpringel et al (2001) who used lsquoshiftedrsquo values of theHubble-type T of galaxies to obtain a good morphology den-sity relation Thus we classify as S0 galaxies those with0 lt T lt 5 and as elliptical and spirals galaxies those withlower and higher values respectively

3 CHEMICAL IMPLEMENTATION

One of the main aims of this work is to reach a better un-derstanding of how the chemical enrichment of the ICMproceeds by establishing a connection between the spatialdistribution of several properties of the ICM such as sur-face mass density mass-weighted temperature distributionX-Ray surface brightness and chemical abundances

Recent observations of radial abundance pro-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 5

files of different elements (Finoguenov et al 2000De Grandi amp Molendi 2001 Gastaldello amp Molendi 2002De Grandi et al 2003 Tamura et al 2004) provide valuableconstraints on the physical processes involved especiallythose related to the feedback mechanisms that inject metalsinto the diffuse phase The available observational datacalls for a model that can explain the spatial distributionof different chemical elements Thus it is essential toimplement a chemical model that takes into account severaldifferent elements and discriminates between differentsources such as SNe CC and SNe Ia

In our model information about the spatial distributionof the ICM properties can be obtained from the gas par-ticles in the hydro-simulation However the non-radiativesimulation used only give the thermodynamic properties ofthe diffuse hot gas In order to obtain the chemical prop-erties of gas particles we have to establish a link betweenthe chemical implementation in the semi-analytic model andthe metal enrichment of the gas particles In the followingwe will first describe in detail how we model the enrichmentcycle of metals among the different baryonic components inthe semi-analytic model and then explain how these quan-tities are tied to the gas dynamics in the simulation

31 Enrichment cycle of chemical elements in thesemi-analytic model

In our model stars can contaminate the cold and hot gasbecause of mass loss during their stellar evolution and metalejection at the end of their lives The hot gas has primordialabundances initially (76 per cent of hydrogen and 24 percent of helium) but becomes chemically enriched as a resultof the transfer of contaminated cold gas to the hot phase dueto reheating by supernovae explosions This chemical enrich-ment has a strong influence on the amount of hot gas thatcan cool since we are using metal dependent cooling ratesThis process in turn influences the star formation activitywhich is ultimately responsible for the chemical pollution

Our semi-analytic model considers mass losses by starsin different mass ranges from low and intermediate stars toquasi-massive and massive stars taking into account stellarlifetimes Massive stars give raise to SNe CC Ejecta fromSNe Ia are also included The cold gas component of eachgalaxy becomes gradually more chemically contaminated asstar formation proceeds and consequently new stars thatare formed are progressively more metal rich This fact callsfor the use of metallicity dependent stellar yields Howevervariations of yields with stellar metallicities are very smallso we only consider the mass dependence of stellar yieldsWe adopt yields of different chemical elements estimated forstars with solar heavy element abundance

Thus for each stellar mass ∆M⋆ formed we determinethe fraction of mass of stars contained in a given mass rangeby assuming an Initial Mass Function (IMF) Φ(M) Themass of chemical element j ejected by stars with masses inthe range centred in mk with lower and upper limits ml

k

and muk is given by

∆M j

ejk= [Rk X

j + Y j

k ] ∆M⋆ (5)

where Xj is the solar abundance of element j and Rk andY j

k are the recycled fraction and the yield of element j for

the k-th mass range respectively The recycled fraction isgiven by

Rk =

mu

kint

ml

k

Φ(M) rk dM

with rk defined as the difference between the mass mk ofthe star at birth and its remnant mass after mass loss dueto stellar winds andor supernova explosions The yield ofchemical element j newly formed is

Y j

k =

mu

kint

ml

k

Φ(M) pjk dM

where pjk is the stellar yield of a chemical element j producedby a star with mass in the range centred around mk Thevariable j may refer to H He or chemical species with atomicnumber larger than 2

The mass of species j ejected by each SN Ia is labelledmj

Ia In this case we are dealing with the total mass ejectedbecause this type of supernovae leaves no remnants thus

∆M j

ej(Ia)k

= ηIak mj

Ia ∆M⋆ (6)

where ηIak is the number of SNe Ia per stellar mass formedin the mass interval k which depend on the model adoptedfor this type of SNe Note that SNe Ia do not contribute toH or He hence M j

ej(Ia)kis zero when referring to these two

chemical elementsThe transport of different chemical species between hot

gas cold gas and stars can be expressed as variation ofthe mass of the chemical element j present in each baryoniccomponent namely

∆M j

hot = minus∆Mcool Aj

hot +∆Mreheat Aj

cold (7)

∆M j

cold = +∆Mcool Aj

hot minus∆M⋆ Aj

cold(SF)

+∆M j

ej +∆M j

ej(Ia) minus∆Mreheat Aj

cold (8)

∆M j

stellar = +∆M⋆ Aj

cold(SF) minusRAjstar ∆M⋆ (9)

where Aj

B = M j

BMB is the abundance of different chemicalelements in each baryonic component which involves thetotal mass of the baryonic component MB and the mass ofspecies j contained in it M j

B The suffix B may refer to hotgas cold gas or stars Aj

cold(SF)denotes the abundance of the

element j in the cold gas at the time of birth of ∆M⋆ Theabove set of equations takes into account the accumulatedcontribution of different mass ranges that affect the baryoniccomponents at a given time as a result of the combinationof the star formation rate of each galaxy and the returntime-scale of the ejecta from all sources considered

32 Injection of metals in the diffuse gas

We describe now the procedure used to distribute the chem-ical elements generated by the galaxies in the semi-analyticmodel among the gas particles of the N-BodySPH simula-tion For each snapshot we identify the gas particles con-tained within spheres of radius 100 hminus1 kpc centred on each

ccopy 2005 RAS MNRAS 000 1ndash24

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 4: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

4 S A Cora

the mass of the hot gas associated with it and Mstellar andMcold are the masses of stars and cold gas of each galaxycontained in the halo The virial mass is given by Mvir =100H2R3

virG This is the mass enclosed by the virial radiusRvir which is defined as the radius of a sphere with meandensity 200ρcrit centred on the most bound particle of thegroup Here ρcrit is the critical density The virial velocity isgiven by V 2

vir = GMvirRvirThe mass of hot gas that cools at each snapshot is given

by the cooling rate

dMcool

dt= 4πρgr

2cool

drcooldt

(2)

where ρg is the density profile of an isothermal sphere thathas been assumed for the distribution of hot gas within adark matter halo and rcool is the cooling radius The localcooling time is defined as the ratio of the specific thermal en-ergy content of the gas and the cooling rate per unit volumeΛ(TZ) The latter depends quite strongly on the metallicityZ of the hot gas and the temperature T = 359(Vvirkm sminus1)of the halo and is represented by the cooling functions com-puted by Sutherland amp Dopita (1993)

The star formation rate is given by

dM⋆

dt=

αMcold

tgxdyn (3)

where tgxdyn = 01RvirVvir is the dynamical time of thegalaxy and α is a dimensionless parameter that regulatesthe efficiency of star formation We adopt the variable starformation efficiency introduced by DL04 that depends on theproperties of the dark halo α(Vvir) = α0 (Vvir220 kmsminus1)nwith α0 and n as free parameters

Each star formation event generates a stellar mass∆M⋆ = M⋆∆TNsteps where Nstep = 50 are timesteps ofequal size used to subdivide the intervals between simulationoutputs ∆T and to integrate the differential equations thatdescribe the changes in mass and metals of each baryoniccomponent over these timespans Each solar mass of starsformed leads to a number ηCC of core collapse supernovaeThis class of supernovae includes those of type Ibc and IIthe former being generated when the progenitor looses itshydrogen-rich envelope before the explosion Ordinary typeII explosions are the most abundant ones however The en-ergy ESNCC released by each core collapse supernova is as-sumed to reheat some of the cold gas of a galaxy with amass of

∆Mreheat =4

3ǫηCC ESNCC

V 2vir

∆M⋆ (4)

where ǫ is a dimensionless parameter that regulates the ef-ficiency of the feedback process

The fate of the reheated gas is quite uncertain Weadopt a model in which the transfer from the cold to the hotphase does induce galactic outflows ie the reheated mass iskept within its host halo This model is referred to as reten-tion model The resulting cycle of metal enrichment is con-sistent with the way in which metals provided by the semi-analytic model are deposited in the gas particles around eachgalaxy in the underlying N-BodySPH simulation describedin Section 32 The effect of other feedback schemes (ejectionand wind models) on the chemical enrichment of the ICMis discussed by DL04 They found that these three prescrip-tions (including the retention model) predict a similar time

evolution of the ICM metal pollution which mainly occursat redshifts larger than 1 As we show in the present studygas dynamical processes driven during the cluster assemblyoccurring at present epochs contribute significantly to de-termine the spatial distribution of metals in the ICM verylikely erasing any signature of the feedback scheme involved

In a hierarchical scenario of structure formation merg-ers of galaxies are a natural consequence of the accretionand merger processes of dark matter haloes in which they re-side We directly use the simulation outputs to construct themerger histories of dark haloes To this end we first identifydark haloes as virialised particle groups by a friend-of-friend(FOF) algorithm The SUBFIND algorithm (Springel et al2001) is then applied to these groups in order to find self-bound dark matter substructures and the resulting set ofgravitationally bound structures is tracked over time to yieldmerging history trees

In this subhalo scheme we distinguish three types ofgalaxies when tracking galaxy formation along the mergingtrees The largest subhalo in a FOF group hosts the lsquocentralgalaxyrsquo of the group its position is given by the most boundparticle in that subhalo Central galaxies of other smallersubhaloes that are contained in a FOF group are referred toas lsquohalo galaxiesrsquo The (sub)haloes of these galaxies are stillintact after falling into larger structures The third group ofgalaxies comprises lsquosatellitersquo galaxies This type of galaxiesis generated when two subhaloes merge and the galaxy ofthe smaller one becomes a satellite of the remnant subhaloThese galaxies are assumed to merge on a dynamical time-scale with the halo-galaxy of the new subhalo they residein (Springel et al 2001 DL04) In these previous versionsof the semi-analytic model satellite galaxy positions weregiven by the most-bound particle identified at the last timethey were still a halo galaxy In the present work we insteadassume a circular orbit for these satellite galaxies with a ve-locity given by the virial velocity of the parent halo and de-caying radial distance to the corresponding central galaxyThis provides a more robust estimate of the position of satel-lite galaxies consistent with the dynamical friction formulaused (Binney amp Tremaine 1987)

With respect to the spectro-photometric properties ofgalaxies we apply the calculations made by DL04 con-sidering evolutionary synthesis models that depend on themetallicity of the cold gas from which the stars formed(Bruzual amp Charlot 1993) Our morphological classifica-tion of galaxies is based on the criterion adopted bySpringel et al (2001) who used lsquoshiftedrsquo values of theHubble-type T of galaxies to obtain a good morphology den-sity relation Thus we classify as S0 galaxies those with0 lt T lt 5 and as elliptical and spirals galaxies those withlower and higher values respectively

3 CHEMICAL IMPLEMENTATION

One of the main aims of this work is to reach a better un-derstanding of how the chemical enrichment of the ICMproceeds by establishing a connection between the spatialdistribution of several properties of the ICM such as sur-face mass density mass-weighted temperature distributionX-Ray surface brightness and chemical abundances

Recent observations of radial abundance pro-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 5

files of different elements (Finoguenov et al 2000De Grandi amp Molendi 2001 Gastaldello amp Molendi 2002De Grandi et al 2003 Tamura et al 2004) provide valuableconstraints on the physical processes involved especiallythose related to the feedback mechanisms that inject metalsinto the diffuse phase The available observational datacalls for a model that can explain the spatial distributionof different chemical elements Thus it is essential toimplement a chemical model that takes into account severaldifferent elements and discriminates between differentsources such as SNe CC and SNe Ia

In our model information about the spatial distributionof the ICM properties can be obtained from the gas par-ticles in the hydro-simulation However the non-radiativesimulation used only give the thermodynamic properties ofthe diffuse hot gas In order to obtain the chemical prop-erties of gas particles we have to establish a link betweenthe chemical implementation in the semi-analytic model andthe metal enrichment of the gas particles In the followingwe will first describe in detail how we model the enrichmentcycle of metals among the different baryonic components inthe semi-analytic model and then explain how these quan-tities are tied to the gas dynamics in the simulation

31 Enrichment cycle of chemical elements in thesemi-analytic model

In our model stars can contaminate the cold and hot gasbecause of mass loss during their stellar evolution and metalejection at the end of their lives The hot gas has primordialabundances initially (76 per cent of hydrogen and 24 percent of helium) but becomes chemically enriched as a resultof the transfer of contaminated cold gas to the hot phase dueto reheating by supernovae explosions This chemical enrich-ment has a strong influence on the amount of hot gas thatcan cool since we are using metal dependent cooling ratesThis process in turn influences the star formation activitywhich is ultimately responsible for the chemical pollution

Our semi-analytic model considers mass losses by starsin different mass ranges from low and intermediate stars toquasi-massive and massive stars taking into account stellarlifetimes Massive stars give raise to SNe CC Ejecta fromSNe Ia are also included The cold gas component of eachgalaxy becomes gradually more chemically contaminated asstar formation proceeds and consequently new stars thatare formed are progressively more metal rich This fact callsfor the use of metallicity dependent stellar yields Howevervariations of yields with stellar metallicities are very smallso we only consider the mass dependence of stellar yieldsWe adopt yields of different chemical elements estimated forstars with solar heavy element abundance

Thus for each stellar mass ∆M⋆ formed we determinethe fraction of mass of stars contained in a given mass rangeby assuming an Initial Mass Function (IMF) Φ(M) Themass of chemical element j ejected by stars with masses inthe range centred in mk with lower and upper limits ml

k

and muk is given by

∆M j

ejk= [Rk X

j + Y j

k ] ∆M⋆ (5)

where Xj is the solar abundance of element j and Rk andY j

k are the recycled fraction and the yield of element j for

the k-th mass range respectively The recycled fraction isgiven by

Rk =

mu

kint

ml

k

Φ(M) rk dM

with rk defined as the difference between the mass mk ofthe star at birth and its remnant mass after mass loss dueto stellar winds andor supernova explosions The yield ofchemical element j newly formed is

Y j

k =

mu

kint

ml

k

Φ(M) pjk dM

where pjk is the stellar yield of a chemical element j producedby a star with mass in the range centred around mk Thevariable j may refer to H He or chemical species with atomicnumber larger than 2

The mass of species j ejected by each SN Ia is labelledmj

Ia In this case we are dealing with the total mass ejectedbecause this type of supernovae leaves no remnants thus

∆M j

ej(Ia)k

= ηIak mj

Ia ∆M⋆ (6)

where ηIak is the number of SNe Ia per stellar mass formedin the mass interval k which depend on the model adoptedfor this type of SNe Note that SNe Ia do not contribute toH or He hence M j

ej(Ia)kis zero when referring to these two

chemical elementsThe transport of different chemical species between hot

gas cold gas and stars can be expressed as variation ofthe mass of the chemical element j present in each baryoniccomponent namely

∆M j

hot = minus∆Mcool Aj

hot +∆Mreheat Aj

cold (7)

∆M j

cold = +∆Mcool Aj

hot minus∆M⋆ Aj

cold(SF)

+∆M j

ej +∆M j

ej(Ia) minus∆Mreheat Aj

cold (8)

∆M j

stellar = +∆M⋆ Aj

cold(SF) minusRAjstar ∆M⋆ (9)

where Aj

B = M j

BMB is the abundance of different chemicalelements in each baryonic component which involves thetotal mass of the baryonic component MB and the mass ofspecies j contained in it M j

B The suffix B may refer to hotgas cold gas or stars Aj

cold(SF)denotes the abundance of the

element j in the cold gas at the time of birth of ∆M⋆ Theabove set of equations takes into account the accumulatedcontribution of different mass ranges that affect the baryoniccomponents at a given time as a result of the combinationof the star formation rate of each galaxy and the returntime-scale of the ejecta from all sources considered

32 Injection of metals in the diffuse gas

We describe now the procedure used to distribute the chem-ical elements generated by the galaxies in the semi-analyticmodel among the gas particles of the N-BodySPH simula-tion For each snapshot we identify the gas particles con-tained within spheres of radius 100 hminus1 kpc centred on each

ccopy 2005 RAS MNRAS 000 1ndash24

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 5: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 5

files of different elements (Finoguenov et al 2000De Grandi amp Molendi 2001 Gastaldello amp Molendi 2002De Grandi et al 2003 Tamura et al 2004) provide valuableconstraints on the physical processes involved especiallythose related to the feedback mechanisms that inject metalsinto the diffuse phase The available observational datacalls for a model that can explain the spatial distributionof different chemical elements Thus it is essential toimplement a chemical model that takes into account severaldifferent elements and discriminates between differentsources such as SNe CC and SNe Ia

In our model information about the spatial distributionof the ICM properties can be obtained from the gas par-ticles in the hydro-simulation However the non-radiativesimulation used only give the thermodynamic properties ofthe diffuse hot gas In order to obtain the chemical prop-erties of gas particles we have to establish a link betweenthe chemical implementation in the semi-analytic model andthe metal enrichment of the gas particles In the followingwe will first describe in detail how we model the enrichmentcycle of metals among the different baryonic components inthe semi-analytic model and then explain how these quan-tities are tied to the gas dynamics in the simulation

31 Enrichment cycle of chemical elements in thesemi-analytic model

In our model stars can contaminate the cold and hot gasbecause of mass loss during their stellar evolution and metalejection at the end of their lives The hot gas has primordialabundances initially (76 per cent of hydrogen and 24 percent of helium) but becomes chemically enriched as a resultof the transfer of contaminated cold gas to the hot phase dueto reheating by supernovae explosions This chemical enrich-ment has a strong influence on the amount of hot gas thatcan cool since we are using metal dependent cooling ratesThis process in turn influences the star formation activitywhich is ultimately responsible for the chemical pollution

Our semi-analytic model considers mass losses by starsin different mass ranges from low and intermediate stars toquasi-massive and massive stars taking into account stellarlifetimes Massive stars give raise to SNe CC Ejecta fromSNe Ia are also included The cold gas component of eachgalaxy becomes gradually more chemically contaminated asstar formation proceeds and consequently new stars thatare formed are progressively more metal rich This fact callsfor the use of metallicity dependent stellar yields Howevervariations of yields with stellar metallicities are very smallso we only consider the mass dependence of stellar yieldsWe adopt yields of different chemical elements estimated forstars with solar heavy element abundance

Thus for each stellar mass ∆M⋆ formed we determinethe fraction of mass of stars contained in a given mass rangeby assuming an Initial Mass Function (IMF) Φ(M) Themass of chemical element j ejected by stars with masses inthe range centred in mk with lower and upper limits ml

k

and muk is given by

∆M j

ejk= [Rk X

j + Y j

k ] ∆M⋆ (5)

where Xj is the solar abundance of element j and Rk andY j

k are the recycled fraction and the yield of element j for

the k-th mass range respectively The recycled fraction isgiven by

Rk =

mu

kint

ml

k

Φ(M) rk dM

with rk defined as the difference between the mass mk ofthe star at birth and its remnant mass after mass loss dueto stellar winds andor supernova explosions The yield ofchemical element j newly formed is

Y j

k =

mu

kint

ml

k

Φ(M) pjk dM

where pjk is the stellar yield of a chemical element j producedby a star with mass in the range centred around mk Thevariable j may refer to H He or chemical species with atomicnumber larger than 2

The mass of species j ejected by each SN Ia is labelledmj

Ia In this case we are dealing with the total mass ejectedbecause this type of supernovae leaves no remnants thus

∆M j

ej(Ia)k

= ηIak mj

Ia ∆M⋆ (6)

where ηIak is the number of SNe Ia per stellar mass formedin the mass interval k which depend on the model adoptedfor this type of SNe Note that SNe Ia do not contribute toH or He hence M j

ej(Ia)kis zero when referring to these two

chemical elementsThe transport of different chemical species between hot

gas cold gas and stars can be expressed as variation ofthe mass of the chemical element j present in each baryoniccomponent namely

∆M j

hot = minus∆Mcool Aj

hot +∆Mreheat Aj

cold (7)

∆M j

cold = +∆Mcool Aj

hot minus∆M⋆ Aj

cold(SF)

+∆M j

ej +∆M j

ej(Ia) minus∆Mreheat Aj

cold (8)

∆M j

stellar = +∆M⋆ Aj

cold(SF) minusRAjstar ∆M⋆ (9)

where Aj

B = M j

BMB is the abundance of different chemicalelements in each baryonic component which involves thetotal mass of the baryonic component MB and the mass ofspecies j contained in it M j

B The suffix B may refer to hotgas cold gas or stars Aj

cold(SF)denotes the abundance of the

element j in the cold gas at the time of birth of ∆M⋆ Theabove set of equations takes into account the accumulatedcontribution of different mass ranges that affect the baryoniccomponents at a given time as a result of the combinationof the star formation rate of each galaxy and the returntime-scale of the ejecta from all sources considered

32 Injection of metals in the diffuse gas

We describe now the procedure used to distribute the chem-ical elements generated by the galaxies in the semi-analyticmodel among the gas particles of the N-BodySPH simula-tion For each snapshot we identify the gas particles con-tained within spheres of radius 100 hminus1 kpc centred on each

ccopy 2005 RAS MNRAS 000 1ndash24

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 6: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

6 S A Cora

galaxy Once the set of Ngas gas particles around a givengalaxy is found we apply our chemical enrichment modelto them based on the semi-analytic computation of galaxyformation The mass of chemical element j transfered to thehot phase is evenly distributed among the set of Ngas gasparticles as

∆mjgas = +

1

Ngas∆Mreheat A

j

cold (10)

This prescription arises from equation (7) which involvesthe hot gas component of the semi-analytic model Notethat the effect of gas cooling on gas particles is suppressed inour model because of the use of a non-radiative simulationThis process should however not change the mass fractionof different species since cooling would reduce the mass inthem proportionally to their abundances

The increment of the masses of chemical elements in agas particle in principle modifies its total mass However themass of gas particles does not change in the N-BodySPHsimulation Since we are interested in the resulting chemicalabundances of gas particles after this enrichment processwe redefine the masses of species j contained in gas parti-cles such that their sum gives the original mass of the gasparticle mgas as given in the simulation and the mass frac-tion of each chemical element keeps the value reached afterthe contribution given by Equation (10) has been added

The chemical elements injected into gas particles areredistributed among neighbouring particles simulating asmall-scale mixing process We apply a smoothing techniqueto the chemical properties of the gas particles within thevirial radius of the cluster considering 32 neighbours aroundeach of them The radius of the sphere containing theseneighbouring particles is less than sim 100 hminus1 kpc thereforethe large-scale distribution of metals in the ICM is not af-fected by this procedure After smoothing the mass of chem-ical species j of particle i is given by

mj

gasi =

32suml=1

mj

gaslρgasl W (rli hi) (11)

where mj

gasl and ρgasl are the mass of species j of the l-th neighbour and its density respectively W (rli hi) is thekernel used rli is the relative distance between particlesand hi is the smoothing length of particle i for which thesmoothed quantity is being estimated The smoothing kernelis the one used for SPH calculations in the N-BodySPHsimulation (Springel Yoshida amp White 2001) that has beenobtained from a B-spline This pocedure is applied at everyoutput of the simulation

4 CHARACTERISTICS OF THESEMI-ANALYTIC MODEL

Our semi-analytic model is characterised by several param-eters that regulate the way in which gas cooling star forma-tion supernovae feedback and galaxy mergers proceed Asmall number of them are considered free parameters andtheir appropriate values are determined by observationalconstraints These are the star formation efficiency α andthe feedback efficiency ǫ which regulate the star formationrate and the core collapse supernovae rate respectively and

a parameter characterising the contribution of supernovaetype Ia This section describes the fixed parameters adoptedand the conditions we impose to determine the free onesWe then compare some properties of our model with obser-vational results

41 Fixed parameters of the model

Our chemical implementation involves an IMF and stellaryields of different species We do not analyse the impactof changing these parameters on the chemical enrichmentof the ICM they are considered fixed as well as the baryonfraction fb For the later we adopt a value of 013 which waschosen to match that used in the N-BodySPH simulation

We consider a Salpeter IMF normalised by fixing thefraction ξ of stars with masses larger than 1M⊙ which arethe major contributors to the chemical enrichment (Porti-nari Chiosi amp Bressan 1998) Thus we obtain the condition

Muint

1M⊙

Φ(M)MdM = ξ (12)

with the upper limit Mu = 100M⊙ The lower limit Ml is de-termined by the value of ξ adopted such that the integrationbetween Ml and Mu equals unity We adopt ξ = 05 a highenough value to reach the right metallicity in all baryoniccomponents

For low- and intermediate-mass stars (08M⊙ M

5 minus 8M⊙) we adopt the stellar yields of Marigo (2001)and for quasi-massive (5M⊙ M 8M⊙) and massivestars (8M⊙ M 120M⊙) we use yields from modelsof Portinari et al (1998) The first two ranges of masseseject mainly 4He 12C 14N and possibly 16O through stel-lar winds The last two stellar mass ranges contribute to thechemical enrichment of the interstellar medium by stellarwinds and supernovae explosions triggered by core collapseSNe CC are the main contributors of αminuselements In our cal-culations we follow the production of 8 chemical elements(H 4He 12C 14N 16O 24Mg 28Si 56Fe) generated by starswith masses distributed in 27 mass ranges from 08M⊙ to100M⊙ The mass of the rest of the elements produced isstored in a separate variable

Since we consider yields from stars with solar metal-licity and taking into account that the solar heavy ele-ment abundance has been recalibrated to Z⊙ = 00122(Asplund et al 2005) we estimate average yields fromthose given for metallicities 002 and 0008 by Marigo(2001) and Portinari et al (1998) Taking into accountthat the combination of standard IMF and yields do notsatisfy observational constraints such as abundances insolar neighbourhood stars metallicity-mass relation andICM metallicity (Moretti Portinari amp Chiosi 2003) andconsidering the uncertainties that affect the stellar yields(Gibson Loewenstein amp Mushotzky 1997) Mg and O areincreased by a factor of 4 and 15 respectively When allmass ranges are considered we get a net yield of 0043(sim 35Z⊙) and a corresponding recycled fraction of 039

Integrating the normalised IMF between 8M⊙ and100M⊙ which is the mass range of stars leading to SNe CCgives ηCC = 0009 The energy ejected by each SNe CC isset to 12times1051 following the results by Woosley amp Weaver

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 7: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 7

(1995) on which the nucleosynthesis calculations of totalejecta of massive stars made by Portinari et al (1998) arebased

We include the contribution of SNe Ia which are char-acterised by high iron production (sim 06M⊙) and long delaytime from the formation of the progenitor to the supernovaexplosion The currently accepted scenarios for the occur-rence of SNe Ia are on one hand carbon deflagration in C-Owhite dwarfs in binary systems (single degenerate systems)and on the other mergers of two white dwarfs (double de-generate systems) The former would account for sim 80 percent of the events with mean time delays of 05minus3 Gyr whilethe later would constitute the remaining sim 20 per cent withlower time delays of sim 03 Gyr (Tutukov amp Yungelson 1994Yoshii Tsujimoto amp Nomoto 1996 Ruiz-Lapuente amp Canal1998 Hachisu Kato amp Nomoto 1999 Greggio 2005) Weadopt the single degenerate model to estimate the SNeIa rate following the scheme of Greggio amp Renzini (1983)where type Ia supernovae originate in binary systemswhose components have masses between 08 and 8M⊙Calculations are based on the formalism described inLia Portinari amp Carraro (2002) but assuming that SNe CCoriginate from single stars with masses larger than 8M⊙For the chemical contribution of SNe Ia we adopt the nu-cleosynthesis prescriptions from the updated model W7 byIwamoto et al (1999)

We use the stellar lifetime given byPadovani amp Matteucci (1993) to model the return time-scale of the ejecta from all sources considered beingspecially relevant for the single stars in the low- andintermediate-mass range and for the progenitors of SNeIa In the last case the mass range of secondary stars inbinary systems gives explosion times for SNe Ia comprisedbetween sim 29 times 107 and sim 14 times 1010 yrs with the SNeIa rate reaching a maximun within sim 01 minus 07 Gyr for asingle stellar population

This chemical model shares similar characteristics withthat implemented in the semi-analytic code developed byNagashima et al (2005) being an improvement with respectto other semi-analytic models (Kauffmann amp Charlot 1998Cole et al 2000 Somerville et al 2001 DL04) which typi-cally consider both the fraction of metals ejected by the starsformed and the recycled mass that has been locked in starsat their birth and re-ejected as free parameters

42 Parameters governing the SF and SNe CC

The parameters involved in the star formation and feedbackprocesses have a strong influence on the star formation rateThe SNe CC rate closely resembles the star formation ratedensity (SFR) because it is dominated by high mass starswith short lifetimes ( 3times107 Gyrs) The value of the indexn involved in the star formation law adopted (Equation 3)is set equal to the value suggested by DL04 n = 22 How-ever the values of α0 and ǫ used by DL04 for the retentionmodel do not constitute the best option for our semi-analyticmodel The different chemical model implemented in ourstudy demands a new determination of the appropriate val-ues We use as observational constraints the Milky Wayproperties the luminosity function the Tully-Fisher color-magnitude and mass-metallicity relations (Kauffmann et al1999 Springel et al 2001 DL04) obtaining the best fit for

(i) α0 = 01(ii) ǫ = 02

The agreement of our model results with the observationaldata used to constrain these free parameters is quite similarto that showed by the models of DL04 therefore we refer thereader to their paper for further discussion on this topic

The evolution of SFR in our model is broadly con-sistent with the observational results recently compiled bySomerville et al (2001) with the peak of the SFR howevershifted to higher redshifts This behaviour is similar to re-sults for the cosmic star formation history in hydrodynamicsimulations by Springel amp Hernquist (2003) (see their figure12) although the normalisation is less certain in our modelhere since our estimate of the comoving volume is made un-certain by the irregular shape of our high-resolution zonewithin the simulated volume which is used to estimate SFand SNe density rates

We note that our simulation method is really quite dif-ferent from that applied by Springel amp Hernquist (2003)The present work is based on a non-radiative sim-ulation combined with a semi-analytic model whileSpringel amp Hernquist (2003) use a fully dissipative simula-tion with a subresolution model for the ISM and feedback ef-fects However despite the huge differences in methodologythe principal qualitative results are pretty similar which isan encouraging consistency Even though our simulation issimilar in volume to simulation lsquoG5rsquo of Springel amp Hernquist(2003) the SFR density obtained from the semi-analyticmodel is higher than in lsquoG5rsquo specially at higher redshiftsbeing in better agreement with their simulations with highermass resolution (see their figure 10) One possibility to ex-plain this is that the semi-analytic model is less affected byresolution effects all the way down to the mass resolutionlimit of the haloes while this is not the case for the fullhydro-simulations While a halo with sim 30 particles will beable to yield a good representation of the SFR when thesemi-analytic model is applied one has to go substantiallyabove the halo detection threshold in the full hydrodynam-ical simulations in order to recover a numerically convergedresult for the hydrodynamics Effectively this means thatthe dynamic range of the semi-analytic method is higher

A convergence test of cluster simulations similar to theone used in the present study has been done by CiardiStoehr amp White (2003) They analysed the redshift evo-lution of the total star formation rate normalised to thetotal baryonic mass obtained by applying a semi-analyticmodel to simulations lsquoS1rsquo-lsquoS4rsquo performed by Springel et al(2001) They show that simulation lsquoS2rsquo with the samemass resolution than our hydrodynamical simulation con-verged at z sim 9 to the higher resolution simulation lsquoS3rsquo(mdm = 24 times 108 hminus1M⊙) which already accounts for allsignificant star formation Therefore the simulation used inour hybrid model is suitable for studying the chemical en-richment of the ICM

43 Parameters governing the SNe Ia rate

The contribution of SNe Ia to the metal production of themodel is characterised by the fraction A of binaries origi-nating this type of supernovae Apart from the iron contentthis parameter do not significantly affect the properties of

ccopy 2005 RAS MNRAS 000 1ndash24

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 8: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

8 S A Cora

the baryonic components in the model hence the constraintsimposed for determining the star formation and feedback ef-ficiencies are almost insensitive to the value of A This valuehas to satisfy other kind of requirements

During the last five years great efforts have been un-dertaken to detect SNe over cosmological distances with theaim of studying the evolution of SNe CC and SNe Ia rateswith redshift (Pain et al 1996 Cappellaro Evans amp Turatto1999 Hardin et al 2000 Pain et al 2002 Madgwick et al2003 Tonry et al 2003 Dahlen et al 2004 Strolger et al2004) We adopt A = 01 in order to reproduce this set ofobservations and to achieve the observed iron distributionin the ICM that will be discussed in detail in Section 5

The combination of the SF history of the model and thescenario chosen for the origin of SNe Ia naturally leads toSNe CC and SNe Ia rates whose ratio evolves with redshiftThe ratio between the number of SNe Ia and SNe CC hasa fairly constant value of sim 04 to z sim 07 similar to thebehaviour suggested by observations of Dahlen et al (2004)The time delay in SNe Ia explosions makes this ratio smallerat higher redshifts We find that it decreases tosim 025 at z sim

25 where SNe Ia rate reaches its maximum and continuesdecreasing afterwards manifesting the minor contributionof SNe Ia at large lookback times

44 Comparison of model results withobservations

Having specified the parameters that define our model wemake further comparisons with observations in order tocheck that it produces a proper circulation of metals amongthe different baryonic components

An important constraint on cluster chemical models isthe fraction of heavy elements ejected by both types of SNewhich end up in stars and in the intergalactic medium A re-cent study of metal enrichment of early type galaxies basedon the X-ray properties obtained with Chandra has beendone by Humphrey amp Buote (2005) The sample of galax-ies span sim 3 orders of magnitude in X-ray luminosity (LX)from group-dominant to low-LX galaxies We apply this se-lection criterium to the central galaxies in our simulationand estimate the iron abundance ZFe of the hot gas con-tained in the dark haloes in which they reside The threepanels in Figure 1 present the dependence of ZFe with lumi-nosities LX (which calculation is based on the formula pre-sented by Valdarnini 2002) and LB and the iron metallicity[FeH] of their stars they are compared with the values re-ported by Humphrey amp Buote (2005) The general trend ofthe properties of our model galaxies is consistent with thatshowed by their observational results On one hand thereis no evidence of a correlation between ZFe and LX or LBOn the other the iron abundance for the gas and stellarcomponents of each of our model galaxies closely follow theZFe(gas) = [FeH](stars) line around which many of theobserved galaxies are clustered This fact strongly supportsthe circulation of metals among the different baryonic com-ponents achieved by our model

The fraction of iron in hot gas and stars that is syn-thesized by different type of SNe is indicative of the dif-ferent time-scales of star formation and of pollution of thehot gas with SNe Ia products Adopting standard SNe Iaand CC stellar yields and performing a fitting analogous to

that of Gastaldello amp Molendi (2002) Humphrey amp Buote(2005) infer an SNIa iron enrichment fraction of sim 066 inthe hot gas of their observed galaxies and sim 035 in theirstars The model galaxies considered in Figure 1 are char-acterised by similar fractions from the total amount of ironoriginated in SNe Ia sim 65 per cent is contained in the hotgas of these galaxies while the remaining sim 35 per cent islocked in the stars The corresponding fractions for iron pro-duced by SNe CC are sim 62 and sim 38 per cent for the hotgas and stars respectively We see that the iron mass frac-tion in stars is a bit higher if we consider iron from SNeCC instead of that produced by SNe Ia This fact reflectsthe increment of the ratio between the number of SNe Iaand SNe CC to lower redshifts thus the hot gas continuesbeing polluted by SNe Ia ejecta after most of the stars havebeen formed However the bulk of the star formation haveoccurred when the hot gas have already been considerablypolluted by SNe Ia products as it is evident from the smalldifference between these fractions

A similar conclusion can be obtained from the fractionof iron originated in different sources that end up in thebaryonic components of our simulated cluster We have thatsim 78 per cent of the iron produced by SNe Ia is containedin the hot gas and sim 21 per cent is locked in the stars whilefor SNe CC we get sim 72 and sim 28 per cent for hot gas andstars respectively The interpretation about the influenceof the evolution of SNe rates on these percentages becomesclearer from the inspection of Figure 9 and the associateddiscussion which focalise on the ejected iron mass rate forboth types of supernovae produced by the cluster galaxies

Our simulated cluster has cold baryon fraction fccls =Mstellar(Mstellar + MICM) sim 0081 iron yield YclsFe =MFeMstellar = (MFe

ICM +MFestellar)Mstellar sim 32 times solar

and iron mass fraction fFe = MFeMvir sim 53times 10minus5 whereMstellar is the mass of stars of the cluster galaxies MICM isthe mass of the intracluster hot gas and MFe

stellar and MFeICM

are the masses of iron contained in these two baryonic com-ponents These values are compared with estimates froma combination of near-infrared properties of galaxies withinclusters for which X-ray imaging data is available (Lin Mohramp Standford 2003) The properties of galaxies are obtainedfrom the Two Micron All Sky Survey (2MASS) Lin et al(2003) present several relations for low redshift clusters(0016 z 009) spanning a range of X-ray emission-weighted temperatures of 21 keV kBTX 91 keV (withkB the Boltzmann constant) that corresponds to about anorder of magnitude in cluster virial mass (08minus9times1014M⊙)The simulated cluster has a virial temperature of kB T sim

72 keV The approximate values of the quantities mentionedpreviously obtained by Lin et al (2003) for a cluster withthis temperature are fccls sim 009 YclsFe sim 35 times solarand fFe sim 9times 10minus5 Model values are lower than these onesbut are within the error bars of the observations (see figure9 of Lin et al 2003)

The number abundance of iron relative to hydrogen forthe ICM respect to the solar value is [FeH]ICM sim 028 closeto the lower limit of observed ICM abundances which rangefrom sim 03 to sim 05 (eg Ettori et al 2001) The iron mass-to-light ratio is sim 0014 Γ⊙ well within the range obtainedby Renzini (1993) for h = 07 that is 00085 minus 0017 Γ⊙

The total B-band luminosity of the cluster is LB =35times 1012L⊙ The B-band and V-band mass-to-light ratios

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 9: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 9

40 41 42 43log LX ([erg s-1 ])

01

10

ZF

e (

Hot

Gas

)

0 2 4 6 8 10 12 14log LB (1010Lo)

01

10

ZF

e(H

ot G

as)

-04 -02 00 02 04[FeH] (Stars)

01

10

ZF

e (H

ot G

as)

Figure 1 Hot gas iron abundance ZFe as a function of the X-ray luminosity LX (left panel) B-band luminosity of the galaxy LB

(middle panel) and iron abundance of their stars [FeH] (right panel) Filled circles represent central galaxies in the simulation selectedaccording to the X-ray luminosities of their associated hot gas (398 lt log(LX) lt 435) as the range span by the sample of galaxiesanalysed by Humphrey amp Buote (2005) Their observational results are represented by empty squares and triangles where the laterindicate lower limit determinations of iron abundances The orbservational points are consistent with the ZFe(gas) = [FeH](stars) lineplotted in the right panel within their error bars which are not included in these plots for clarity (we refer the reader to the figures 5and 6 of Humphrey amp Buote 2005)

are ΓB sim 437 Γ⊙ and ΓV sim 345 Γ⊙ respectively Observa-tional determinations of these quantities give γV sim 175minus252(Carlberg et al 1996 Girardi 2000) and γB sim 200 minus 400(Kent amp Gunn 1983 Girardi 2000) The intra-cluster massto light ratio of our cluster is MICMLB sim 40Γ⊙ quite sim-ilar to the typical observed value of sim 36Γ⊙ indicated byMoretti et al (2003)

The analysis of the physical properties of gas particlesincluded in the hybrid model provides more detailed featuresof the ICM regarding the spatial distribution of differentchemical elements and the connection between the X-rayspectra generated by the diffuse gas with its large-scale mo-tions as we will show in the next sections

5 METAL DISTRIBUTION IN THE ICM

The increase of spectral and spatial resolution of thelast generation of X-ray satellites like ASCA and Bep-

poSAX and more recently Chandra and XMM-Newton hasprovided a great deal of information about cluster cen-tres In particular it has become possible to obtain de-tailed abundance measurements of several chemical elementswithin radial bins allowing a determination of their spatialdistribution (Finoguenov et al 2000 De Grandi amp Molendi2001 Gastaldello amp Molendi 2002 De Grandi et al 2003Tamura et al 2004) Radial distributions of these abun-dances support the presence of metallicity gradients in theICM a feature that has been shown to be common in groupsand clusters of galaxies These radial profiles are of great im-portance to infer the origin of the different species found inthe cluster

In our approach the spatial distribution of chemicaland thermodynamical properties of the ICM is provided bythe gas particles in the N-BodySPH simulation The hy-brid model combines the dynamics of gas particles with thechemical information supplied by the semi-analytic modelThus at each snapshot gas particles carry the information

about masses of H He and heavy elements with the possi-bility of isolating the contribution due to SNe CC and SNeIa These data are analysed by estimating radial profiles ofthe chemical elements abundances

Radial profiles are constructed by taking into accountthe gas particles lying inside the virial radius of the clus-ter We then divide the volume limited by this radius intoconcentric spherical shells centred on the dominant clustergalaxy each of them containing the same number N shell

gas ofparticles For each shell we estimate the mass-weighted andX-ray emission-weighted mean abundance relative to hydro-gen for several species The later is estimated using the FeKα 67 keV emission line which calculation for each gas par-ticle is described in section 6 These quantities are then ex-pressed in terms of the solar value These mean abundancesare assigned to the mean radius of each shell

Figure 2 shows the radial distributions of Fe Kα 67keV emission-line-weighted iron oxygen and silicon abun-dances by number for our model which are preferred tomass-weighted ones in order to compare with observationsThe radius is normalised with Rvir These results are char-acterised by the presence of negative gradients in the ra-dial distribution of all the chemical elements consideredThe emission-weighted values are sim 01 minus 015 dex higherthan mass-weighted abundances specially in the inner re-gion ( 01RRvir)

There are ambiguous results regarding the de-pendence of iron abundance on cluster temperatureBaumgartner et al (2005) find decreasing abundances fromsim 07 to sim 03 (FeH)⊙ in the temperature rangekBT sim 3 minus 10 keV while a roughly constant value ofsim 03 (FeH)⊙ is claimed by Renzini (2003) An interme-diate behaviour was detected previously by Fukazawa et al(1998) and Finoguenov et al (2001) with iron abundancesof sim 03 (FeH)⊙ for cool clusters and sim 02 (FeH)⊙ forhotter ones Thus it is better to compare observed and sim-

ccopy 2005 RAS MNRAS 000 1ndash24

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 10: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

10 S A Cora

-10

-05

00

05

[Fe

H]

-10

-05

00

05

[OH

]

001 010RRvir

-10

-05

00

05

[SiH

]

Figure 2 Fe Kα 67 keV emission-line-weighted abundance pro-files of Fe O and Si relative to H referred to the solar value(Asplund et al 2005) Total abundances with the contributionof both types of SNe are given by solid lines root mean squarestandard deviations are indicated by dotted lines Separate con-tributions of SNe CC and SNe Ia are shown by dashed anddashed-dotted lines respectively Symbols with error bars areobservational data from Tamura et al (2004) for hot clusters(kBT amp 6 keV filled circles) and medium-temperature clusters(3 kBT 6 keV open diamonds)

ulated results on cluster abundances within the same tem-perature range

Tamura et al (2004) analysed XMM-Newton observa-tions of 19 galaxy clusters They obtained accurate abun-dance distributions from deprojected spectra for clusters indifferent temperature ranges and for three spherical regionswith radii between 0-50 50-200 and 200-500 hminus1 kpc Oursimulated cluster has temperatures of the order of kBT sim 75keV (see mass-weighted and emission-weighted maps in Fig-ures 3 and 4) Then results from our model are comparedwith data for hot clusters (kBT amp 6 keV) and intermediate-temperature clusters (3 kBT 6 keV) represented inFigure 2 by filled circles and open diamonds respectivelyThe later set of clusters is used since they are better deter-mined providing better constraints to the behaviour of Oand Si Following De Grandi amp Molendi (2001) we estimateRvir for observed data from an ICM temperature-dependentformula

From Figure 2 we can see that the contribution of SNeIa (dashed-dotted lines) is necessary to recover the observedFe abundance distribution The contribution of Si from SNeIa is not as large as for the Fe production but helps to reachthe mean observed values However the error bars in the Sideterminations are much larger than those for Fe and do notestablish a reliable constraint for the model parameters Thisis also the case for O the large observational uncertaintiesin the determination of these chemical abundances for hotclusters allow any type of behaviour of the radial profiles

while data points for medium temperature clusters are con-sistent with a flat profile with a slight increasing trend withradius Note that the model abundances show radial pro-files with negative gradients for all the elements consideredHowever the uncertainties both in observed and simulatedabundances do not allow us to infer strong constraints fromthe comparison at this point Only the slopes of iron abun-dance profiles are better established by observational dataThe good agreement between the iron abundance distribu-tion in our model and that observed in hot clusters makesthis model suitable for further analysis We now carry outa deeper inspection of our results in order to gain insight inthe enrichment history and dynamical evolution of the ICM

6 MAPS OF ICM PROPERTIES

Our hybrid model for investigating the ICM metal enrich-ment provides information about the chemical and ther-modynamical properties of the diffuse intracluster gas interms of the gas particle properties (positions velocitiesdensity ρ and internal energy per unit mass u) of the un-derlying hydrodynamic simulation From these quantities wecan obtain their temperature as T = micrompkB (γ minus 1)uwhere micro = 06 is the mean molecular weight mp theproton mass and γ = 53 the adiabatic exponent Wecan learn about the density temperature and abundancedistributions and velocity structure of the ICM by con-structing maps of these properties using planar projectionsof the gas particles within the cluster virial radius Thiskind of analysis is matched to the recent progress in X-ray observations with Chandra and XMM-Newton whichprovides temperature and abundance maps of cluster cores(Schmidt et al 2002 Sanders amp Fabian 2002 Sanders et al2004 Fukazawa Kawano amp Kawashima 2004) In order tocompare with observations we will analyse not only mass-averaged properties but also X-ray emission-weighted onesAs a result of the high temperature and low density of theICM the most prominent feature in its X-ray spectrum isa blend of emission lines from iron (mainly FeXXV andFeXXVI) with photon energies between 67 and 7 keV

The first step in computing line-emission weighted prop-erties of the ICM from our simulations is to estimate theintensity of the Fe Kα 67 keV emission line for each gasparticle The atomic processes involved in the generationof the emission are quite sensitive to the hydrogen volumedensity nH temperature T and chemical abundances of thegas Following the procedure performed by Furlanetto et al(2004) we build a grid where these quantities are varied andwe compute the intensity of the emission line for each pointin the grid For this purpose we use the radiative-collisionalequilibrium code CLOUDY (Ferland 2000 Ferland et al2001) Given the thermodynamical conditions of the ICMwe can well approximate it as a low density hot plasma incollisional equilibrium without being affected by an ioniz-ing backgroud Therefore photoionization is discarded inthe models we generate with CLOUDY

The iron line emissivity of the gas particles that makeup the ICM is obtained by interpolating the emissivitiesassigned to our grid of gas states where the parametersdefining the three-dimensional grid in density tempera-ture and iron abundances cover the range of values exhib-

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 11: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 11

00001

00010

00100

n H [c

m-3

] (M

W)

00001

00010

00100

50 100 150 200

50 100 150 200

107

108

T [

K ]

(MW

)

10 7

10 8

50 100 150 200

50 100 150 200

01

10

Fe

H (

MW

)

01

10

50 100 150 200

50 100 150 200

10-8

10-6

10-4

10-2

100

IIo (

MW

)

10 -8

10 -6

10 -4

10 -2

10 0

50 100 150 200

50 100 150 200

Figure 3 Projection of mass-weighted (MW) properties of gas particles contained within a sphere of radius 1hminus1 Mpc centred on thedominant cluster galaxy hydrogen number density (top-left panel) temperature (top-right panel) iron abundance by number FeHreferred to the solar value (bottom-left panel) and emissivity of the Fe Kα 67 keV emission line normalised with the maximum mass-weighted emissivity Io (bottom-right panel)

ited by the gas inside the virial radius of the cluster atz = 0 An important detail of our chemical implementa-tion is that we follow the evolution of the abundances ofseveral species Hence instead of assuming solar metallicityfor gas particles or varying this quantity as a global param-eter like in other works (eg Sutherland amp Dopita 1993Furlanetto et al 2004) our grid of models depends on setsof abundances of the chemical elements considered in thesemi-analytic model As gas particles become more metalrich the abundances of all elements increase roughly propor-tionally to each other We take advantage of this feature toconstruct the sets of abundances characterised by the meanvalue of iron abundance within a certain range of valuesSince we only consider the most abundant elements in oursemi-analytic model the code CLOUDY automatically givessolar values (meteroritic abundances of Grevesse amp Anders(1989) with extensions by Grevesse amp Noels (1993)) for therest of the elements

The range of values covered by the grid parametersare 1 times 106 lt T lt 1 times 109 K for gas temperature and

316times 10minus6 lt nH lt 01 cmminus3 for the hydrogen volume den-sity This last range of values is estimated from the densityρ of gas particles in the N-BodySPH simulation and fromtheir hydrogen abundance AH

gas in our model In generalthis quantity ranges from the initial value of 076 to sim 070which is reached by the most metal rich particles Iron abun-dances by number range from FeHsim 5times10minus8 to sim 5times10minus5which are sim 18times 10minus3 and sim 18 times the solar value re-spectively

We choose density spacings ∆ log(nHcmminus3) = 025

Since emissivities are quite sensitive to temperature varia-tions we adopt temperature spacings of ∆ log(TK) = 01for 107K lt T lt 1085K and ∆ log(TK) = 025 otherwiseWe consider seven bins for the iron abundances with spac-ings ∆ log(FeH) = 05 from log(FeH) = minus775 to minus425In order to produce the whole grid of models in a systematicway we use a user-friendly package of IDL routines that fa-cilitates the use of the code CLOUDY (in version C96B4)called MICE (MPE IDL Cloudy Environment) developed byHenrik Spoon

ccopy 2005 RAS MNRAS 000 1ndash24

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 12: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

12 S A Cora

00001

00010

00100

nH [cm-3 ] (EW)

00001 00010 00100

50100

150200

50100

150200 10

7

108

T [ K ] (EW)

107 108

50100

150200

50100

150200 01

10

FeH (EW)

01 10

50100

150200

50100

150200

Figure 4 Projection of Fe Kα 67 keV emission line-weighted (EW) properties of gas particles contained within a sphere of radius1hminus1 Mpc centred on the dominant cluster galaxy hydrogen number density (left panel) temperature (middle panel) and iron abundanceby number FeH relative to the solar value (right panel)

Figure 3 shows the projected mass-weighted maps ofhydrogen density (top-left panel) temperature (top-rightpanel) iron abundance (bottom-left panel) and emissivity ofthe Fe Kα 67 keV line normalised with the maximum mass-weighted emissivity Io (bottom-right panel) for gas parti-cles within a radius of 1 hminus1 Mpc from the cluster centre atz = 0 From the first three maps we can connect the ironabundances with the distributions of gas density and tem-perature in the ICM We can see that those regions withhigh levels of chemical enrichment are associated with cooland high density areas this fact is particularly evident in thesmall substructures in the outskirts of the cluster that havebeen recently accreted These substructures do not producestrong X-ray emission due to their low temperature (fourthmap) and the line emission map results in rather smoothmaps following mainly the coarse features of the densitydistribution

Figure 4 is equivalent to Figure 3 but the Fe Kα 67 keVemission-line-weighted quantities are shown instead We cansee that the denser cooler and more metal rich substructuresfound in the outskirts of the mass-weighted maps of Figure 3do not appear in the corresponding plots of Figure 4 sincethe intensity of the lines emitted from these regions is verylow (see bottom-right panel of Figure 3)

7 ENRICHMENT HISTORY ANDDYNAMICAL EVOLUTION OF THE ICM

In order to understand the way in which the chemical enrich-ment patterns characterising the intracluster gas developwe analyse the temporal evolution of maps of the projectedchemical properties of the gas and the corresponding radialabundance profiles This allows us to evaluate the relativecontribution of SNe CC and Ia to the metal enrichment aswell as to investigate the dynamical evolution of the intr-acluster gas which plays a fundamental role in the spatialdistribution of metals

For this purpose we first consider gas particles lyingwithin 500 hminus1 kpc from the cluster centre at z = 0 and trace

their properties back in time Thus we follow the chemicalenrichment of small substructures forming at high redshiftand study how they progressively become part of the clusterand contribute to establishing its final distribution of met-als Figures 5 and 6 show mass-weighted iron abundances bynumber relative to the sun FeH (left column) the corre-sponding Fe Kα 67 keV emission-line-weighted abundances(middle column) and the mass-weighted temperature of thegas (right column) for five different redshifts from z sim 1 toz sim 01 At each redshift the cluster centre is defined asthe centre of the most massive progenitor of the cluster atz = 0 and the distances from it are estimated using comov-ing coordinates We can see that the ICM at z = 0 is formedby the accretion and merger of small substructures that atearly times (z sim 1) are spread within an extended regionof radius sim 9 hminus1 Mpc These gaseous building blocks arealready considerably contaminated by that time with ironabundances that reach values of FeH sim 03

From these figures we can also infer interesting aspectsabout the dynamics of the contaminated gaseous clumpsThose that are at sim 6hminus1 Mpc from the cluster centre atz sim 05 needs to move at more than sim 1000 km sminus1 to getinside 05 hminus1 Mpc by z = 0 A quantitative analysis of thedynamics of these gaseous clumps is presented in Figure 7which shows the dependence of the velocity of gas particleswith their clustercentric distances and its evolution with red-shift We identify the gas particles contained within sphericalshells of 05 hminus1 Mpc of thickness centred at the most mas-sive progenitor of the cluster at z sim 05 and estimate theirmean velocities and clustercentric distances at lower red-shifts In this way we follow the evolution of these proper-ties as the cluster is being assembled The different symbolsdenote the redshifts considered and the lines connect thesets of gas particles defined according to their clustercentricdistance at z sim 05 There are many aspects to emphasizefrom this plot At z sim 05 gas particles have very high ve-locities which range from sim 1300 to sim 2500km sminus1 Thehighest velocities correspond to gas particles lying betweensim 1 and sim 2hminus1 Mpc This is the same for redshifts z sim 03and sim 02 although the velocities acquired are smaller If

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 13: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 13

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 5 Evolution of iron abundance and temperature of gas particles that lie within 500 hminus1 kpc from the cluster centre at z = 0The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to the solar value (leftcolumn) the corresponding Fe Kα 67 keV emission-line-weighted abundance (middle column) and the mass-weighted temperature ofthe gas (right column) We give results for three redshifts z sim 1 (lookback time of 78 Gyr first row) z sim 05 (lookback time of 49 Gyrsecond row) and z sim 03 (lookback time of 36 Gyr third row) At each refshift the spatial coordinates are centred at the most massiveprogenitor of the cluster at z = 0 and are expressed in comoving scales

we follow the evolution with redshift of gas particles lyingmore than 4hminus1 Mpc at z sim 05 we see that they are accel-erated till they achieve the highest velocities at z sim 03 andsim 02 This occurs at a distance corresponding to the virialradius of the main progenitor which is of the order 1 hminus1

Mpc at z sim 05 and increases to sim 15 hminus1 Mpc at z = 0

Thus despite shell crossing it is evident that the dynamicsof gas particles is characterised by a general behaviour inwhich they are accelerated till they reach the main clustershock radius at a distance of the order of its virial radiusand once they have been incorporated to it they convergeto the centre with velocities of sim 300 kmsminus1 This turbulent

ccopy 2005 RAS MNRAS 000 1ndash24

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 14: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

14 S A Cora

001

010

100

FeH

001 010 100

50100

150200

50100

150200

104

105

106

107

108

T [ K ] (MW)

10 4 105 10 6 107 10 8

50100

150200

50100

150200

Figure 6 Same as Fig 5 but for redshifts z sim 02 (lookback time of 24 Gyr first row) and z sim 01 (lookback time of 12 Gyr secondrow)

velocity reached inside the virial radius and the infall thatdominates the gas outside it are consistent with the resultsobtained by Sunyaev et al (2003) based on the analysis ofcluster simulations at z = 0 (Norman amp Bryan 1999) Thecomplex dynamics of gas particles acquired during the hier-archical formation of the structure helps to explain the wayin which the metallicity profiles develop as we discuss laterin this section

Figure 8 shows the evolution of iron and oxygen abun-dance profiles which are determined for gas particles lyingwithin the innermost comoving 1hminus1 Mpc at z = 0 and ateach of the redshifts considered in Figures 5 and 6 The firstthree rows show the shape of the profiles for the combinedeffect of both types of SNe and for the separate contribu-tions of SNe Ia and CC respectively The plot in the last rowgives the evolution of the OFe ratio profile resulting fromthe influence of the combined effect of both types of SNeThe way in which the iron and oxygen abundance profilesdevelop are quite similar even though the former is mainlydetermined by the large amount of iron ejected by SNe Iawhile the latter is only determined by the SNe CC contri-bution This fact leads to an almost flat OFe ratio profileat z = 0 with a slight increasing trend at large radii Thecentral abundances of both Fe and O and the almost flat

OFe ratio show that both types of SNe contribute in a sim-ilar fashion to every part of the cluster that is without apreference for SNe Ia to chemically enrich the inner regionsof the cluster and for SNe CC to pollute its outskirts asinferred from some observations (Finoguenov et al 2001)The main change in the profiles occurs in the central region( 100 hminus1 kpc) at late times by a progressive increase ofthe chemical abundances

The evolution of the radial abundance profiles of themain SNe Ia and CC products (iron and oxygen respec-tively) from z sim 1 to z = 0 can be explained by taking intoaccount the history of metal ejection of the cluster galaxiesas well as the dynamics of the gas associated to them whilegalaxies are being accreted onto the cluster

Figure 9 shows the iron mass ejection rate as a functionof redshift produced by SNe Ia (top panel) and SNe CC(bottom panel) that are contained in galaxies that lie withinshells of different distances from the cluster centre at z = 0Since we are focusing on a set of galaxies in the z = 0 clusterthe contribution of SNe CC to the ICM chemical enrichmentpeaks at higher redshifts (z sim 5minus7) than the star formationrate of the whole simulation (see figure 1 of Ciardi et al(2003) for a comparison of the redshift evolution of the starformation rate for field region simulations and for cluster

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 15: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 15

1000Mean Distance from cluster centre [Kpc h-1]

0

500

1000

1500

2000

2500

Mea

n V

eloc

ity r

espe

ct to

clu

ster

cen

tre

[km

s-1]

z=05

z=03

z=02

z=01

z=00

Figure 7 Mean velocities of gas particles contained withinspherical shells of 05hminus1 Mpc of thickness centred at the mostmassive progenitor of the cluster at z sim 05 as a function of theirmean clustercentric distances This dependence is estimated fordifferent redshifts as denoted by the symbols The lines connectthe sets of gas particles defined according to their clustercentricdistance at z sim 05

simulations) As we have already mentioned the SNe CCrate closely follows the star formation rate since the lifetimesof the stars involved are quite short Instead the ejectionrate of elements produced by SNe Ia reaches a maximumat z sim 3 minus 5 due to the return time distribution resultingfrom the adopted model for SNe Ia explosions The peaks inthe mass ejection rates are followed by a strong decline atlower redshifts for both types of SNe such that the ongoingchemical contamination is quite low at z = 0 This behaviouris more pronounced for those galaxies that are nearer to thecluster centre at the present time

Considering the accumulated mass of iron in the ICM atz = 0 we find that SNe Ia are responsible for sim 75 per centof the iron content of the ICM Interpretation of observedprofiles of abundance ratios extends this percentage up tosim 80 per cent (Gastaldello amp Molendi 2002) However bothresults are quite dependent of the assumed SNe Ia modelThe point that we have to emphasize is that the relativecontribution of SNe CC and SNe Ia to the iron content ofthe cluster is already one-third at z sim 2 when the accu-mulated iron masses provided by both sources peack Thelack of metal contribution by SNe explosions at late timesfrom galaxies that are near the cluster centre cannot explainthe increase of abundances in the inner 100 hminus1 kpc sincez sim 01 suggesting that dynamical processes are playing animportant role in the developement of abundance profiles

Figure 10 shows the evolution of mass-weighted ironabundances of gas particles contained within 1hminus1 Mpcfrom the cluster centre at redshifts z sim 1 sim 05 sim 03sim 02 sim 01 and z = 0 We clearly see from these plots thathighly enriched gas clumps are already present in the out-

000204060810

Fe

abun

danc

e

SNe CC + Ia

000204060810

Fe

abun

danc

e

SNe Ia

01 10000204060810

Fe

abun

danc

e

SNe CC

000204060810

O a

bund

ance SNe CC + Ia

000204060810

O a

bund

ance SNe Ia

01 10000204060810

O a

bund

ance SNe CC

01 10R [ h-1 Mpc ]

130

140

150

160

170

OF

e

SNe CC + Ia

Figure 8 Evolution of the radial profiles of iron and oxygenabundances by number relative to hydrogen and of the profileof the OFe ratio referred to the solar value The profiles aredetermined for the gas contained within the innermost comov-ing 1hminus1 Mpc at z sim 1 (dash dot dot line) z sim 05 (dash dotline) z sim 03 (dashed line) z sim 02 (dotted line) z sim 01 (thinsolid line) and z = 0 (thick solid line) The first row shows thecontributions of both types of SNe (CC and Ia) to the Fe andO abundances while the second and third rows show the sepa-rate contributions of SNe Ia and CC respectively The last rowpresents the OFe abundance ratios when both types of SNe areconsidered

skirts of the cluster at z sim 03 and that the iron abundanceprofile starts to settle down at z sim 02 The contaminatedgas clumps located near y sim 350 hminus1 kpc at sim 02 and closeto y sim 200 hminus1 kpc at sim 01 seem to converge to the clustercentre at z = 0 considerably increasing the iron abundancein the inner 100 hminus1 kpc Thus gas particles bound to thedark matter substructures of lsquohalo galaxiesrsquo (see definition insection 22) have been mainly enriched at early epochs whenthe metal ejection rates from these galaxies were higher andhave then fallen to the inner regions of the cluster driven bythe haloes of infalling galaxies This dynamical effect is quitelikely taking into account the analysis of gas dynamics basedon figures 5 6 and 7 A simple test of finding the positionof the progenitor of the central galaxy at different redshiftsreveals that these galaxies are at sim 100 hminus1 kpc at z sim 05reinforcing our conclusion based on gas dynamics

The scenario of chemical enrichment considered inour model indicates that gas dynamical effects play animportant role in the developement of abundance pat-terns However there are other sources of chemical ele-ments not included in our model that may help to in-terprete observational results such as intracluster stel-lar populations and ram-pressure stripping There is in-creasing evidence of the presence of intracluster starsbased on observations of diffuse light (Feldmeier 2004Zibetti et al 2005) and individual stars between clustergalaxies (Feldmeier 2003 Durrell et al 2002 Gal-Yam et al

ccopy 2005 RAS MNRAS 000 1ndash24

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 16: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

16 S A Cora

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe Ia products from

Central galaxy

Gx with r lt 05 h-1 Mpc

Gx with 05 lt r lt 1 h-1 Mpc

Gx with 1 lt r lt 15 h-1 Mpc

1 101+z

0001

0010

0100

1000

Ej

Fe

Mas

s R

ate

[Mo yr

-1 ]

SNe CC products

Figure 9 Iron mass ejection rate as a function of redshift pro-duced by SNe Ia (top panel) and SNe CC (bottom panel) con-tained in galaxies that lie within different distances from the clus-ter centre at z = 0 the dominant cluster galaxy (at the clus-ter centre thick solid line) galaxies within the spherical regionwith radius 05hminus1 Mpc without the central galaxy (dashed line)galaxies within the spherical shells between 05 and 1 hminus1 Mpc(dash dot line) and between 10 and 15hminus1 Mpc (dash dot dotdot line)

2003) High resolution N-BodySPH simulations are start-ing to yield predictions of the spatial and kinematic distri-butions for these stars (Willman et al 2004 Murante et al2004 Sommer-Larsen Romeo amp Portinari 2005) which holdimportant information about the assembly history of agalaxy cluster These numerical works show that a fractionof at least sim 10 per cent of unbound stars accumulate withinthe cluster as a result of stripping events and infall of galaxygroups that already contain unbound stars The impact ofintracluster stars in the chemical enrichment of the ICM hasbeen explored by Zaritsky Gonzalez amp Zabludoff (2004) byusing a simple model They demonstrate that this stellarcomponent makes a significant contribution to the iron con-tent of the ICM The metal pollution due to ram-pressurestripping has been recently investigated by Schindler et al(2005) and Domainko et al (2005) These authors arguethat the efficiency of this enrichment mechanism is higherthan that of supernovae driven galactic winds producinga centrally concentrated metal distribution in massive clus-ters

8 PROBING THE ICM DYNAMICS WITHMETALS

In the previous section we have showed how the complexdynamical evolution of intracluster gas influences the devel-opement of the abundance patterns of the ICM We nowdiscuss the potentiality of spectroscopic observations for de-

tecting such strong gas bulk motions sheding light on ourunderstanding of gas dynamics during the cluster formation

The spatial and spectral resolution of X-ray telescopedetectors onboard of present satellite missions allow themeasurement of metallicity maps and of high resolutionemission line spectra The combination of these data con-tributes enormously to improving our knowledge about theICM properties For example Fe Kα 67 keV emission-line-weighted maps contain valuable information about thermo-dynamical and chemical properties of the intracluster gasas discussed in Section 6 We can also learn about the kine-matics of the intracluster gas and its spatial correlation withother properties from maps of emission-line-weighted line-of-sight (LOS) velocities and from Fe Kα 67 keV emissionline spectra generated by the gas along LOSs through thecluster

Figure 11 shows projections of Fe Kα 67 keV emission-line-weighted LOS velocities of the gas particles onto thethree orthogonal planes x-y y-z and z-x Velocities are rel-ative to the cluster centre being colour-coded such that redand yellow areas correspond to gas moving towards the ob-server and green and blue represent gas receding from itOnly material contained within a sphere of radius 1hminus1 Mpccentred on the dominant galaxy of the cluster has been in-cluded These maps exhibit steep azimuthal and radial gra-dients revealing a much more complex structure than thosedepicted by the density temperature and iron abundancedistributions indicating the presence of large-scale turbulentmotions generated by inhomogeneous infall This is consis-tent with our results on the evolution of the dynamics of gasparticles based on the analysis of Figure 7 The sharp edgesdefined by red and green colours near the cluster centre arecontact discontinuities between gas that originated in differ-ent clumps Note that this is not a projection effect of gasparticles lying far from the cluster centre These contact dis-continuities are also seen in the projection of particles wellinside the cluster ( 50 hminus1 kpc)

The gradients visible in the velocity maps of Figure 11can be as large as sim 1000 kmsminus1 over distances of a few hun-dred kpc They are of the same order of magnitude as thoseobtained from accurate observations of the Centaurus clus-ter (Abell 3526) (Dupke amp Bregman 2001) We note thatthis cluster is a very good candidate for this kind of stud-ies being one of the closest X-ray bright clusters of galaxieswith an optical redshift of 00104 Its core has been spatiallyresolved by the Chandra Observatory using the ACIS-S de-tector (Sanders amp Fabian 2002) leading to temperature andabundance maps However there is not enough informationin the spectrum to isolate spectral lines and perform accu-rate gas velocity measurements This task is possible withthe detector GIS on board of ASCA The gas velocity distri-bution determined with this instrument (Dupke amp Bregman2001) shows a region associated with the subcluster Cen 45at sim 130 hminus1 kpc from the main group Cen 30 (centred onthe cD galaxy) with a radial velocity higher than the rest ofthe cluster by sim 1700 kmsminus1 These radial velocity measure-ments obtained from X-ray observations support the opti-cally determined velocity difference between Cen 45 and Cen30 of sim 1300 kmsminus1 (Stein et al 1997)

These gradients in radial velocities are observationallydetected by the Doppler shifting of the metal lines used toperform the spectral fits which are mainly driven by the Fe

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 17: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 17

01

10

FeH (MW)

01 10

50100

150200

50100

150200

Figure 10 Evolution of iron abundance of gas particles that lie within 1hminus1 Mpc from the cluster centre at redshifts z sim 1 (upperleft panel) z sim 05 (upper middle panel) z sim 03 (upper right panel) z sim 02 (lower left panel) z sim 01 (lower middle panel) and z = 0(lower right panel) The plots show the projection of mass-weighted iron abundance by number relative to hydrogen FeH respect to thesolar value At each refshift the spatial coordinates are centred at the most massive progenitor of the cluster at z = 0 and are expressedin comoving scales

400

200

0 -200

-400

LOS velocity [ km sec-1 ] (EW)

400 200 0 -200 -400

50100

150200

50100

150200

Figure 11 Projection of Fe Kα 67 keV emission-line-weighted line-of-sight (LOS) velocities of gas particles contained within a sphereof radius 1hminus1 Mpc centred on the dominant cluster galaxy x-y projection (left panel) y-z projection (middle panel) and z-x projection(right panel)

ccopy 2005 RAS MNRAS 000 1ndash24

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 18: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

18 S A Cora

Kα complex However a higher spectral resolution would al-low us to detect the different components in which a certainline is split due to the bulk motions of different magnitudealong a line-of-sight through the cluster Future X-ray mis-sions promise a substantial increase in energy resolutionpaving the way for the construction of a full 3-D picture ofthe ICM dynamics Sunyaev et al (2003) analyse this pos-sibility by calculating the synthetic spectra of the Fe Kα 67keV emission line along sightlines through a simulated clus-ter (Norman amp Bryan 1999) They show that this emissionline is split into multiple components over a range of energy∆E sim plusmn15 eV as a result of turbulent and bulk motionsof the gas They estimate that a spectral resolution of 4eV similar to the one planned for future missions would beenough for the detection of dynamical features in the ICMbased on emission line spectra

Following the study made by Sunyaev et al (2003) weanalyse the way in which the special features of the gas mo-tion that become evident in the velocity maps of Figure 11are reflected in the Fe Kα 67 keV emission line spectra gen-erated by the gas along lines-of-sight through the clusterThe spectra are computed by dividing the sightlines in binsof width 20 hminus1 kpc along directions perpendicular to theprojection planes The thermodynamical kinematical andemission properties assigned to each bin are obtained fromthe 64 closest gas particles to the bin centres using the SPHsmoothing technique The energy of the iron line emittedfrom each bin along the sighline is Doppler shifted propor-tionally to the smoothed LOS velocity inferred from the gasaround the bin Combining these shifted energy values withthe corresponding intensities of the line emission we obtainthe iron emission line spectra along the LOS On top ofthis we consider the broadening suffered by each line dueto thermal motions of the gas

Figures 12 and 13 show the resulting synthetic spec-tra for six LOSs along the z-axis through the x-y projectionof the simulated cluster Like in the construction of mapsthe computation of the spectra only comprises gas particleswithin the virial radius of the cluster Solid thick lines corre-spond to spectra that are affected by both Doppler shiftingand thermal broadening For comparison we also includein each panel the iron emission line affected only by ther-mal broadening (dash-dotted lines) The maximum valuesof these last ones are used to normalise both kinds of ironemission lines with and without Doppler shifting The ra-tios of these maxima with respect to the one correspondingto the cluster centre referred to as ImaxImax

o are also givenin each of the panels that show the spectra together withthe location where the LOSs intercept the ICM Below thesepanels both Figures 12 and 13 display the distribution of gasproperties along the corresponding sightlines emission-lineweighted LOS velocities used to generate the spectra andmass-weighted hydrogen density temperature iron abun-dance and normalised emissivity of the Kα 67 keV emissionline These distributions allow us to see how gas compo-nents characterised by different thermodynamical and chem-ical properties move at different velocities as well as whichparts of the line of sight contribute most to the spectrallines

An inspection of these distributions indicates that thegas reaches high iron abundances at high density and lowtemperature regions as it is also shown by the maps of

Figure 3 This situation is especially clear in the clustercore and in a subclump moving at high speed located atsim 500 hminus1 kpc from the cluster centre as it is evident fromthe first and second columns of Figure 12 The Dopplershifting and split of the emission line at the cluster coreare produced by the large-scale turbulent motions of thegas which produces a LOS velocity gradient from sim 200 tominus400 km sminus1 around the middle of the sightline This motionis reflected in the spectra because the intensity of the ironemission line is larger than 10 per cent the central valuewithin a radius of sim 100 hminus1 kpc from the cluster centreHowever this intensity decays quite abruptly at larger radiiand becomes undetectable (see also the mass-weighted emis-sion map in Figure 3) For this reason the subclump mov-ing at minus800 km sminus1 from the cluster centre does not pro-duce any signature in the first two Doppler shifted spectrashown If the emissivity were higher at the location of thissubstructure a peak would appear at sim 6 685eV The sec-ond and third spectra lines in figure 12 are produced bysightlines that intercept the cluster at a distance of 100 and360 hminus1 kpc from its centre respectively Note how the ratioImaxImax

o decreases as we get further from the cluster coreThe second spectrum shows a clear Doppler shifting andbroadening of the line while the third one is only slightly al-tered with respect to the reference broadened line in agree-ment with the LOS velocity distributions within the innersim 100 hminus1 kpc

Figure 13 presents three more synthetic spectra gener-ated from lines of sight that intercept the cluster throughareas with large offset velocities near its centre (see fig-ure 11) The first one shows a split of the emission line in twocomponents arising from the central velocity gradient fromsim minus200 to 100 kmsminus1 The line in the second spectrum iswidely Doppler shifted as a result of high bulk motions ofsim 500 km sminus1 in the central parts of the sightline In thethird emission line spectrum a second component becomesapparent as a result of gas moving at sim minus500 kmsminus1 As wenoted in the previous set of spectra in Figure 12 the stronggas motions of sim minus500 kmsminus1 and sim minus700 kmsminus1 occurringat sim 500 hminus1 Kpc from the centre of the lines of sight con-sidered in the second and third columns are not imprinted inthe corresponding spectra because of the low emissivity thatcharacterises this region The shifts and splits observed inall these spectra are of the order of sim 10minus15 eV This effectis too small to be resolved by present instruments since forexample the spectral resolution of the EPIC pn detector forthe Fe Kα 67 keV emission line is 150 eV

From this analysis we can conclude that only gas mo-tions that produce absolute values of LOS velocities higherthan sim 400 km sminus1 around the cluster core where the ironline emissivity is larger than 10 per cent the central valuecould become observable by future X-ray missions throughthe shifting and split of the line in two components dueto Doppler effects Lower energy lines like the Fe L-blendcentred around 1 keV could be also used for the presentpurpose In principle they should show much more struc-ture since they are strongly emitted in lower temperaturegas However gas with this characteristic lies far from thecentre of our hot cluster and the low line emissivity in theseareas prevent us from getting more information from theselines than that provided by the Fe Kα line at 67 keV These

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 19: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 19

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=00

y=00

ImaxIomax =10

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=-100

y=00

ImaxIomax =0099

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=300

y=200

ImaxIomax =0016

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 12 Fe Kα 67 keV emission line spectra along three lines of sight (LOSs) on the z-axis through the simulated cluster which isprojected onto the x-y plane and the distribution along the corresponding LOS of kinematical thermodynamical chemical and emissivityproperties of the gas that contributes to the spectra The upper panels show the effect of thermal broadening on the Fe emission line(thin dash-dotted curve) and the composition of thermal broadening and Doppler shift due to gas bulk motion with respect to the clustercentre (thick solid line) The maximum values of the broadened lines are used to normalise both kinds of iron emission lines with andwithout Doppler shift These panels indicate the ratio of this maximum with respect to the one corresponding to the cluster centreImaxImax

o and the x-y position where the LOS intercepts the ICM The five plots below each spectra show the distributions of gasproperties along the corresponding sightlines emission-line weighted LOS velocities used to generate the spectra and mass-weightedhydrogen density nH temperature T iron abundance FeH and normalised emissivity of the Kα 67 keV emission line IImax Velocitiesare expressed in km sminus1 nH in cmminus3 T in K and iron abundances are referred to the solar value The positions along the LOSs aregiven with respect to the cluster centre

ccopy 2005 RAS MNRAS 000 1ndash24

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 20: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

20 S A Cora

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

IIm

ax

-500 0 500LOS vel [km s-1 ]

x=-50

y=-300

ImaxIomax =0028

-500

0

500

1000

LOS

vel

00001

00010

00100

n H

107

108

T

01

10

Fe

H

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

00001

00010

00100

01000

10000

IIom

ax

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-100

ImaxIomax =0243

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

6680 6690 6700 6710 6720Delta Energy [eV]

00

05

10

-500 0 500LOS vel [km s-1 ]

x=00

y=-200

ImaxIomax =0039

-1000 -500 0 500 1000Position along LOS [h-1 kpc ]

Figure 13 Same as figure 12 for other set of LOSs that intercept the ICM through areas with large offset velocities present near thecluster centre as depicted by Figure 11

lines are probably much more usefull in a lower temperaturecluster

We would like to note that the hydrodynamical sim-ulation considered by Sunyaev et al (2003) ignores radia-tive cooling which affects the thermal structure in clustercores and they assume constant iron abundances to calcu-late the iron emission line spectra Our simulation suffersfrom the first drawback as well however the critical advan-tage of our hybrid model over approaches like the one used

by Sunyaev et al (2003) is that the chemical abundancesof gas particles are dynamically and consistently generatedfrom the stars in the galaxies

Taking into account the information supplied by LOSvelocity maps and by spectra along selected sightlines wecan connect the global features of bulk motions in the gaswith the multiple components in the spectra that originateas a result of these motions Note that the latter featurecan be more easily observationally detected and quantified

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 21: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 21

Thus our results can constitute valuable help for the inter-pretation of observational data

9 SUMMARY AND CONCLUSIONS

We have presented an hybrid model for the chemical enrich-ment of the intracluster gas that allows a detailed analy-sis of the spatial distribution of chemical thermodynamicaland kinematical properties of the ICM Our model combinesa cosmological non-radiative hydrodynamical N-BodySPHsimulation of a cluster of galaxies and a semi-analytic modelof galaxy formation The spatial distribution of metals inthe ICM can be accounted for by the chemical enrichmentof the gas particles in the underlying N-BodySPH simu-lation This enrichment results from the metals ejected bythe galaxy population which are generated by applying thesemi-analytic model to the the dark matter haloes detectedin the outputs of the simulation This link between semi-analytic model results and the chemical enrichment of gasparticles leads to a consistent chemical enrichment schemefor the intracluster gas being the key ingredient to carryout the present project

The parameters that characterise our model are chosenaccording to several observational constraints The propercirculation of metals among the different baryonic compo-nents leads to a spatial distribution of chemical elementsin the intracluster gas that closely resemble observed abun-dance patterns as can be seen from the comparison of the ra-dial abundance profiles with data from Tamura et al (2004)(Figure 2) The corresponding mean iron abundance of theintracluster medium at z = 0 is [FeH] sim 028 It has beenreached with the contribution of SNe Ia that provide sim 75per cent of the total ICM iron content at z = 0 While thispercentage is quite dependent on the parameters that char-acterise the chemical model implemented what reflects thecirculation of metals is the fraction of elements from differ-ent sources that end up in the ICM as opposed to in starsThe fraction of iron originated in SNe Ia that is containedin the ICM is sim 078 while that locked in the stars of clustergalaxies is sim 021 These fractions are slightly different forthe iron ejected by SNe II being sim 072 and sim 028 for hotgas and stars respectively The increment of the proportionof SNe II products in stars reflects the relative time-scalesfor star formation and metal pollution by different types ofsupernovae

After showing that our model reproduces reasonablywell the distribution of metals in the ICM at z = 0 we anal-yse the evolution of chemical and dynamical properties ofthe intracluster gas in order to understand the physical pro-cesses involved in the determination of its final abundancepatterns The aim of this study is also to provide hints thathelps in gaining information about the characteristics of gasbulk motions from spectroscopic observations We analysethe rich information provided by our hybrid model in threeways

(i) by constructing radial abundance profiles of the ICM(ii) by building projected mass-weighted and Fe Kα 67

keV emission-line-weighted maps of chemical thermody-namical and kinematical properties of the diffuse gas

(iii) and by calculating synthetic spectra of the Fe Kα

67 keV emission line along sightlines through the simulatedcluster

The spatial distribution of chemical elements leads toradial abundance profiles with negative gradients irrespec-tive of the type of source from where they originated Thecentral enhancement of the Fe and O abundance profiles andthe flat profile of the OFe ratio support the notion thatthe intracluster gas is polluted in the same way by bothtypes of SNe From the analysis of the evolution of theseradial abundance profiles and the projected mass-weightediron abundance of gas particles at different redshifts we findthat a high level of enrichment of the ICM is already reachedat z sim 1 Despite the delay time for the ejection of metalsinherent to SNe Ia the contribution of both types of SNe tothe ICM chemical enrichment peaks at z sim 2

The lack of evolution of mean iron abundances up toz sim 1 is in agreement with observations of distant clusters(Tozzi et al 2003) However the final abundance profile ofthe cluster gets steeper within the inner sim 100hminus1 Kpc dur-ing the last Gyr (see the increments between z sim 01 andz = 0 in Figure 8) This cannot be accounted for the metalproduction of the cluster galaxies The peaks in the ratesof the ejected iron mass from these galaxies are followedby a strong decline towards lower redshifts for both typesof SNe such that the ongoing chemical contamination atz = 0 is quite low this behaviour is more pronounced forgalaxies that are nearer to the cluster centre at the presenttime The combination of all these results indicates that theenhancement of central abundances produced at low red-shift cannot be explained by metals ejected during this timeinterval from galaxies that lie close to the cluster centreIndeed galaxies within a sphere of radius 500 hminus1 kpc cen-tred on the dominant cluster galaxy do not contribute to themetal enrichment at recent epochs Instead a scenario ap-pears favoured in which gas particles are primarily enrichedat high redshifts and are subsequently driven to the clustercentre by bulk motions in the intracluster gas Hence themain cause that contribute to this central abundances en-hancement is the dynamical evolution of gas particles duringthe cluster formation

Gas particles develop very high velocities ranging fromsim 1300 to sim 2500 kmsminus1 when they are farther thansim 1 hminus1 Mpc from the cluster centre at z gt 02 Oncethese high velocity gas clumps cross this typical clustershock radius which correspond approximately to the virialradius of the most massive progenitor of the cluster atthese redshifts they slow down ending up within the innersim 500hminus1 Kpc of the cluster at z = 0 The inhomogeneousgas infall produces contact discontinuities that are mani-fested as sharp edges in the emission-line-weighted line-of-sight velocity maps These maps provide useful informationabout the gas dynamics in the ICM and its dependence onspatial position within the cluster They reveal a quite com-plex structure of the gas dynamics including the presenceof strong radial and azimuthal gradients with values as largeas sim 1000 kmsminus1 Gradients of this magnitude are also seenin velocity distributions along sightlines through the clus-ter centre Doppler shifting and broadening suffered by theFe Kα 67 keV emission line along sightlines could be usedto probe these gas bulk motions when they are producedwithin an area characterised by high iron line emissivity

ccopy 2005 RAS MNRAS 000 1ndash24

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 22: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

22 S A Cora

Velocities of sim 400 km sminus1 typically found within a radiusof 100 hminus1 kpc from the cluster core produce Doppler shiftsof the line of the order of sim 10minus15 eV The line can be splitin multiple components depending on the structure of thevelocity field We find that the higher velocities reached bythe intracluster gas beyond 500 hminus1 kpc or those aquired byinfalling subclumps located at such large distances from thecluster centre do not produce a detectable signature in thespectral lines because of the rather low iron line emissivityin the outskirts of the cluster

The combination of all these results strongly supportsthe fact that abundance patterns of the ICM are the resultof gas mixing because of hydrodynamical processes duringcluster formation being a considerable amount of the gasmainly enriched at high redshifts before the cluster havetime to virialise The central enhancement of Fe and O abun-dance profiles might be considered as produced by a partic-ular situation taking place in our simulation in which twobig highly enriched gas subclumps converge at the clustercentre at the present epoch However this kind of behavi-uor is consistent with the hierarchical formation scenario inwhich accretion and merging of substructures constitute themain processes driving the formation and evolution of clus-ters of galaxies Thus the specific shape of metallicity pro-files might well be a consequence of the combination of twotime-scales one related to the metal production of galaxyclusters and the other to the metal mixing In this last caseboth the effect of gas dynamics galaxy mergers and the effi-ciency in the diffusion of metals due to energetic and chemi-cal feedback associated with the star formation process playan important role Other enrichment processes not takeninto account in our model such as metal pollution by in-tracluster stars and ram-pressure stripping may contributesignificantly to recover the observed abundance patterns re-inforcing the effect of gas dynamics discussed here

Our analysis techniques in this study yield a 3-D pic-ture of the chemical and dynamical properties of the ICMThe information about the latter can be extracted from themetal content of the intracluster gas Our results show thatgas bulk motions are imprinted in the shape of X-ray lines ofheavy ions Since thermal broadening is small these featurescould be detected with the high-resolution spectrographs offuture X-ray missions (CONSTELLATION-X and XEUS)Such observations should be very important in determiningthe physical processes involved in the formation and evolu-tion of galaxy clusters contributing with clues that allow usto verify the enrichment scenario inferred from our model inwhich gas dynamics has a crucial effect

Finally even though all the processes considered in ourhybrid model are tuned to satisfy numerous observationalconstraints the implied global level of iron enrichment wefind is on the low side of the generally accepted observedrange which is around one-third the solar value This mayin part be due to the uncertainties in the stellar yields and toour poor knowledge of the strength and nature of the feed-back processes involved but could also point to a systematicproblem with the observational estimates Our model so faronly accounts for the effect of SNe driven outflows but itshould be possible to include other sources of feedback likethose from AGN as well This is left for future work andmay perhaps provide a solution to some of the puzzles stillpresent in understanding the observed cluster metallicities

ACKNOWLEDGMENTS

Simon White and Volker Springel are warmly thanked formaking possible this project and for very useful commentsand discussions Volker is specially acknowledged for provid-ing the simulation and postprocessing codes We are gratefulto the anonymous referee for useful comments and sugges-tions that improved the presentation of this work We thankGabriella De Lucia Patricia Tissera Felix Stoehr AminaHelmi Mariano Mendez Marcus Bruggen Diego GarcıaLambas Hernan Muriel Stefano Borgani Fabio GastaldelloLuca Tornatore and Naoki Yoshida for helpful and stimu-lating discussions Gabriella is also acknowledged for mak-ing available the version of the semi-analytic code used inDL04 on which our model is based We thank Laura Porti-nari for providing the tables of stellar yields and for usefulcomments Part of the calculations included in this workwere done with the publicly available programs CLOUDY

and MICE Fundacion Antorchas is gratefully acknowledgedfor the external postdoctoral fellowship that allowed SACto start this project and for the Reentry Grant awardedwhen returning to her home institution Part of this projectwas financially supported by a grant from Consejo Nacionalde Investigaciones Cientıficas y Tecnicas SAC thanks thehospitality of MPA during her postdoctoral stay at the in-stitute

REFERENCES

Aguerri JAL Gerhard OE Arnaboldi M NapolitanoNR Castro-Rodriguez N Freeman K C 2005 AJ 1292585

Asplund M Grevesse N Sauval A J 2005 in CosmicAbundances as Records of Stellar Evolution and Nucle-osynthesis ed TG Barnes III FN Bash ASP Confer-ence Series 336 25

Baumgartner WH Loewenstein M Horner DJMushotzky RF 2005 ApJ 620 680

Binney J Tremaine S 1987 Galactic Dynamics Prince-ton University Press

Bruzual G Charlot S 1993 ApJ 367 126Cappellaro E Evans R Turatto M 1999 AampA 351 459Carlberg RG Yee HKC Ellingston E Abraham RGravel P 1996 ApJ 462 32

Ciardi B Stoehr F White S D M 2003 MNRAS 3431101

Cole S Lacey CG Baugh CM Frenk CS 2000 MN-RAS 319 168

Dalhen T Strolger L-G Riess AG Mobasher B CharyR-R Conselice CJ Ferguson HC Fruchter AS etal 2004 ApJ 613 189

De Grandi S Molendi S 2001 ApJ 551 153De Grandi S Ettori S Longhetti M Molendi S 2003AampA 419 7

De Lucia G Kauffmann G White SDM 2004 MNRAS349 1101 (DL04)

Domainko W Mair M Kapferer W van Kampen E Kro-nberger T Schindler S Kimeswenger S Ruffert M etal 2005 AampA preprint (astro-ph0507605)

Dupke RA Bregman J N 2001 ApJ 562 266Durrell PR Ciardullo R Feldmeier JJ Jacoby GHSigurdsson S 2002 ApJ 570 119

ccopy 2005 RAS MNRAS 000 1ndash24

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 23: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

Metal Enrichment of the ICM 23

Ettori S Allen SW Fabian AC 2001 MNRAS 322187

Evrard AE 1990 ApJ 363 349Feldmeier JJ Ciardullo R Jacoby GH Durrell PR2003 ApJS 145 65

Feldmeier JJ Mihos JC Morrison HL Harding PKaib N Dubinski J 2004 ApJ 609 617

Ferland GJ 2000 in revista Mexicana de Astronomia yAstrofısica conference Series 153

Ferland GJ 2001 Hazy A Brief Introduction to Cloudy96 (httpwwwpaukyedu garycloudy)

Finoguenov A David LP Ponman TJ 2000 ApJ 544188

Finoguenov A Arnaud M David LP 2001 ApJ 555191

Finoguenov A Matsushita K Bohringer H Ikebe Y Ar-naud M 2002 AampA 381 21

Frenk CS White SDM Bode P Bond JR BryanGL Cen R Couchman HMP Evrard AE GnedinN Jenkins A et al 1999 ApJ 525 554

Fukazawa Y Makishima K Tamura T Ezawa H Xu HIkebe Y Kikuchi K Ohashi T 1998 PASJ 50 187

Fukazawa Y Kawano N Kawashima K 2004 ApJ 606L109

Furlanetto SR Schaye J Springel V Hernquist L 2004ApJ 606 221

Gal-Yam A Maoz D Guhathakurta P Filippenko AV2003 AJ 125 1087

Gastaldello F Molendi S 2002 ApJ 572 160Gibson BK Loewenstein M Mushotzky RF 1997 MN-RAS 290 623

Girardi M Borgani S Giuricin G Mardirossian FMezzetti M 2000 ApJ 530 62

Greggio L Renzini A 1983 AampA 118 217Greggio L 2005 AampA 441 1055Grevesse N Anders E 1989 Cosmic Abundances of Mat-ter AIP Conference Proceedings 183 p 1 Ed CJWaddington (New York AIP)

Grevesse N Anders A 1993 in Origin amp Evolution of theElements ed N Prantzos E Vangioni-Flam M Cassep 15 (Cambridge Cambrigde Univ Press)

Hachisu I Kato M Nomoto K 1999 ApJ 522 487Hardin D Afonso C Alard C Albert JN Amadon AAndersen J Ansari R Aubourg E Bareyre P BauerF et al 2000 AampA 362 419

Iwamoto K Brachwitz F Nomoto K Kishimoto NUmeda H Hix WH Thielemann F-K 1999 ApJS 125439

Humphrey PJ Buote DA 2005 ApJ preprint(astro-ph0504008)

Kapferer W Ferrari C Domainko W Mair M Kron-berger T Schindler S Kimeswenger S van Kampen Eet al 2005 AampA preprint (astro-ph0508107)

Kauffmann G Charlot S 1998 MNRAS 294 705Kauffmann G Colberg JM Diaferio A White SDM1999 MNRAS 303 188

Kent SM Gunn JE 1982 AJ 88 697Lia C Portinari L Carraro G 2002 MNRAS 330 821Lin Y-T Mohr JJ Standford SA 2003 ApJ 591 749Loewenstein M 2001 ApJ 557 L573Madgwick DS Hewett PC Mortlock DJ Wang L2003 ApJL 599 L33

Marigo P 2001 AampA 194 217Matsushita K Finoguenov A Bohringer H 2003 AampA401 443

Moretti A Portinari L Chiosi C 2003 AampA 408 431Mori M Burkert A 2000 ApJ 538 559Murante G Arnaboldi M Gerhard O Borgani S ChengLM Diaferio A Dolag K Moscardini L Tormen GTornatore L Tozzi P 2004 ApJ 607 L83

Nagashima M Lacey C Baugh CM Frenk CC 2005MNRAS 358 1247

Norman ML Bryan GL 1999 in Numerical Astro-physics 1998 eds S Miyama K Tomisaka Astrophysicsand Space Science Library vol 240 (Kluwer Boston) 106

Padovani P Matteucci F 1993 ApJ 41626Pain R Hook IM Deustua S Gabi S Goldhaber GGroom D Kim AG Kim MY Lee JC PennypackerCR et al 1996 ApJ 473 356

Pain R Fabbro S Sullivan M Ellis RS Aldering GAstier P Deustua SE Fruchter AS Goldhaber GGoobar A et al 2002 ApJ 577 120

Portinari L Chiosi C Bressan A 1998 AampA 334 505Renzini A 1993 ApJ 419 52Renzini A 1997 ApJ 488 35Renzini A 2003 in Clusters of Galaxies Probes ofCosmologicas Structure and Galaxy Evolution ed JSMulchaey A Dressler A Oemler as part of the CarnegieObservatories Astrophysics Series (Cambridge Cam-bridge Univ Press) 261

Ruiz-Lapuente P Canal R 1998 ApJL 497 L57Sanders JS Fabian AC 2002 MNRAS 331 273Sommer-Larsen J Romeo AD Portinari L 2005 MN-RAS 357 478

Sanders JS Fabian AC Allen SW Schmidt RW2004 MNRAS 349 952

Sarazin CL 1986 Rev Mod Phys 58 1Schindler S Kapferer W Domainko W Mair M vanKampen E Kronberger T Kimeswerger S Ruffert MMangete O Breitschwerdt D 2005 AampA 435 L25

Schmidt RW Fabian AC Sanders JS 2002 MNRAS337 71

Springel V Yoshida N White SDM 2001 New Astron-omy 79 117

Springel V White SDM Tormen G Kauffmann G2001 MNRAS 238 726

Springel V Hernquist L 2003 MNRAS 339 312Springel V 2005 MNRAS MNRAS 364 1105Somerville RS Primack JR Faber SM 2001 MNRAS320 504

Stein P Jerjen H Federspiel M 1997 AampA 327 952Strolger L-G Riess AG Dahlen T Livio M Panagia NChallis P Tonry JL Filippenko A V et al 2004 ApJ613 200

Sunyaev RA Norman ML Bryan GL 2003 AstronomyLetters 29 783

Sutherland RS Dopita MA 1993 ApJS 88 253Tamura T Kaastra JS Herder JWAden BleekerJAM Peterson JR 2004 AampA 420 135

Tonry JL Schmidt BP Barris B Candia P Challis PClocchiatti A Coil AL Filippenko AV et al 2003ApJ 594 1

Tornatore L Borgani S Matteucci F Recchi S Tozzi P2004 MNRAS 349 L19

ccopy 2005 RAS MNRAS 000 1ndash24

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions
Page 24: Metal EnrichmentoftheICM: a3-DPictureofChemical and ... · 2Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas, Rivadavia 1917, Buenos Aires, Argentina ABSTRACT We develop

24 S A Cora

Tozzi P Rosati P Ettori S Borgani S Mainieri V Nor-man C 2003 ApJ 593 705

Tutukov A V Yungelson L R 1994 MNRAS 268 871Valdarnini R 2002 ApJ 567741Valdarnini R 2003 MNRAS 339 1117White R 1991 ApJ 367 69Willman B Governato F Wadsley J Quinn T 2004MNRAS 355 159

Woosley SE Weaver TA 1995 ApJS 101 181Yoshii Y Tsujimoto T Nomoto K 1996 ApJ 462 266Zaritsky D Gonzalez AH Zabludoff AI 2004 ApJ 613L93

Zibetti S White SDM Schneider DP Brinkmann J2005 MNRAS 358 949

This paper has been typeset from a TEX LATEX file preparedby the author

ccopy 2005 RAS MNRAS 000 1ndash24

  • Introduction
    • Probing the ICM dynamics with metals
    • Summary and conclusions