Suzaku and XMM-Newton Observations of the Fornax cluster: Temperature and Metallicity Distribution
Hideyoshi Murakami, Madoka Komiyama, Kyoko Matsushita, Ryo Nagino, Takuya Sato, Kosuke Sato, Madoka Kawaharada, Kazuhiro Nakazawa, Takaya Ohashi, Yoh Takei
aa r X i v : . [ a s t r o - ph . C O ] O c t PASJ:
Publ. Astron. Soc. Japan , 1– ?? , c (cid:13) Suzaku and XMM-Newton Observations of the Fornax cluster:Temperature and Metallicity Distribution
Hideyoshi
Murakami , Madoka Komiyama , Kyoko Matsushita , Ryo Nagino ,Takuya Sato , Kosuke Sato , Madoka Kawaharada , Kazuhiro Nakazawa ,Takaya Ohashi , and Yoh Takei , Department of Physics, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku, Tokyo [email protected] Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Chuo-ku, Sagamihara,Kanagawa 252-5210 Department of Physics, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033 Department of Physics, Tokyo Metropolitan University, 1-1 Minami-Osawa, Hachioji, Tokyo 192-0397 (Received 0 0; accepted )
Abstract
Suzaku observed a central region and five offset regions within 0.2 r in the Fornax cluster, a nearbypoor cluster, and XMM-Newton mapped the cluster with 15 pointings out to 0.3 r . The distributionsof O, Mg, Si, S, and Fe in the intracluster medium (ICM) were studied with Suzaku, and those of Fe andtemperature were studied with XMM. The temperature of the ICM gradually decreases with radius from1.3 keV at 0.04 r to 1 keV at 0.2–0.3 r . If the new solar abundances of Lodders et al. (2003) anda single-temperature plasma model are adopted, O, Mg, Si, S, and Fe show similar abundances: 0.4–0.6solar within 0.02–0.2 r . This Fe abundance is similar to those at 0.1–0.2 r in rich clusters and othergroups of galaxies. At 0.2–0.3 r , the Fe abundance becomes 0.2–0.3 solar. A two-temperature plasmamodel yields ICM abundances that are higher by a factor of 1.2–1.5, but gives similar abundance ratiosamong O, Mg, Si, S, and Fe. The northern region has a lower ICM temperature and higher brightnessand Fe abundance, whereas the southern region has a higher ICM temperature and lower brightness andFe abundance. These results indicate that the cD galaxy may have traveled from the north because ofrecent dynamical evolution. The cumulative oxygen- and iron-mass-to-light ratios (OMLR and IMLR)within 0.3 r are more than an order of magnitude lower than those of rich clusters and some relaxedgroups of galaxies. Past dynamical evolution might have hindered the strong concentration of hot gas inthe Fornax cluster’s central region. Scatter in the IMLR and similarity in the element abundances in theICM of groups and clusters of galaxies indicate early metal synthesis. Key words: galaxies:abundances — clusters of galaxies:intracluster medium — clusters:individual (theFornax cluster)
1. Introduction
Groups and poor clusters of galaxies represent thebuilding blocks of rich clusters and are the best labora-tories for the study of their thermal and chemical his-tory, which is governed by baryons. An important clue tothe evolution of galaxies is the elemental abundances inthe hot X-ray-emitting gas, e.g., the intracluster medium(ICM), in groups and clusters of galaxies. Metals in theICM have been synthesized by supernovae (SNe) in galax-ies. As a result, the ratios of metal mass in the ICM tothe total light from galaxies in clusters or groups, i.e.,the metal-mass-to-light ratios, are the key parameters ininvestigating the chemical evolution of the ICM.Studies of the scaling relations in the clusters of galax-ies have revealed strong deviations in the observed rela-tions from the predictions based on self-similar collapse(Ponman et al. 1999; Ponman et al. 2003; Finoguenovet al. 2007; Rasmussen & Ponman 2009; Johnson et al.2009; Pratt et al. 2010). The gas density profiles in the central regions of groups and poor clusters are observed tobe shallower than those in the self-similar model, and therelative entropy level is correspondingly higher than thatin rich clusters. These deviations are considered to be bestcharacterized by the injection of energy (pre-heating) intothe gas before the clusters collapse (Kaiser 1991). Basedon ROSAT and ASCA data, Ponman et al. (2003) showedthat groups and clusters have significant excess entropyat r . Voit et al. (2003) have predicted that a smooth-ing of the gas density due to pre-heating in infalling sub-haloes would boost the entropy production at the accre-tion shock of clusters, and an excess of entropy is gener-ated in the cluster outskirts. This effect due to smoothaccretion should be more important for poorer systems.However, Borgani et al. (2005) discussed that this entropyamplification effect can be reduced by cooling. Recently,using XMM data, Pratt et al. (2010) showed that at r ,the mass dependence of the entropy excess disappeared.Sun et al. (2009) studied the entropy profiles of groups ofgalaxies observed with Chandra and found that the dif- Murakami et al. [Vol. ,ference in the entropy excess at r between groups andclusters is not as large as that by Ponman et al. (2003).The stellar and gas mass fractions within r depend onthe total system mass (Vikhlinin et al. 2006; Arnaud etal. 2007; Sun et al. 2009; Giodini et al. 2009). These stud-ies found that the stellar-to-total-mass ratios within r of the groups are much larger than those in the clusters,whereas the gas mass fraction increases with the systemmass.These poorer systems also differ from richer systems inthat their iron-mass-to-light ratios (IMLR) are systemati-cally smaller than those in rich clusters (Makishima et al.2001). The metal distribution in the ICM is a tracer ofthe history of gas heating, because both metal enrichmentand heating timescales determine the metal distribution inthe ICM. The Chandra and XMM observations of nearbygroups of galaxies with cool cores found that the Fe abun-dances of the groups declines with the radius and metal-mass-to-light ratios of groups are much smaller than thoseof the clusters (Rasmussen & Ponman 2007; Finoguenovet al. 2007; Rasmussen & Ponman 2009; Johnson et al.2011). On the basis of the Chandra data, Rasmussen &Ponman (2009) discussed the effect of feedback and thehistory of the more extended star formations in less mas-sive systems. With XMM observations, Johnson et al.(2011) found a difference in the abundance profiles of thecool core and non-cool core groups and discussed the effectof mixing driven by active galactic nuclei (AGN) withinthe central regions.O and Mg are predominantly synthesized in SN II,whereas Fe and Si are synthesized in both SN Ia andSN II. Therefore, abundance measurements spanningthe range of species from O to Fe are required for theunambiguous determination of the formation history ofmassive stars. XMM-Newton provided a means of con-straining the O and Mg abundances of some systems(Matsushita et al. 2003; Tamura et al. 2003; Matsushitaet al. 2007b; Simionescu et al. 2009). However, reliableresults have been obtained only for the central regions ofvery bright clusters or groups of galaxies dominated by cDgalaxies. The X-ray imaging spectrometer (XIS; Koyamaet al. 2007) onboard Suzaku (Mitsuda et al. 2007) offersan improved line spread function because of its very smalllow-pulse-height tail in the energy range below 1 keV cou-pled with a very low background. Therefore, especially forregions of low surface brightness or equivalent width, XISprovides better sensitivity to O lines. The instrumental Alline of the MOS detectors on XMM-Newton causes prob-lems in measuring the Mg abundance in somewhat faintersystems.The oxygen-mass-to-light ratios (OMLRs) as well as theIMLRs of several clusters of galaxies and several groups ofgalaxies out to 0.2–0.3 r were measured with Suzakusatellite (Matsushita et al. 2007a; Tokoi et al. 2008; Satoet al. 2007; Sato et al. 2008; Sato et al. 2009a; Sato etal. 2009b; Komiyama et al. 2009; Sato et al. 2010). Thedifference in the OMLRs between groups and clusters isa factor of about 3–6, and tends to be larger than theIMLR difference, which is a factor of 2–3 (Komiyama et al. 2009).The Fornax cluster is a nearby poor cluster with anICM temperature of 1.3–1.5keV (Scharf et al. 2005). TheX-ray emission shows an asymmetric spatial distribution,and the cD galaxy, NGC 1399, is offset from the center(Paolillo et al. 2002; Scharf et al. 2005), which may be re-lated to large-scale dynamical evolution such as infall mo-tions of galaxies into the cluster (Dunn & Jerjen 2006).The Chandra observations suggest that relative motionmay occur between NGC 1399 and the ICM, and that thesecond brightest elliptical galaxy, NGC 1404, is movingsupersonically in the ICM (Scharf et al. 2005; Machaceket al. 2005). The Fe and Si abundances of the ICM within ∼ r de-rived from early Suzaku observations of two fields of theFornax cluster (Matsushita et al. 2007a) are the small-est among those in the groups of galaxies observed withSuzaku.In this paper, we describe our study of the ICM ofthe Fornax cluster for regions within 0.035–0.2 r ob-served with Suzaku and the temperature and Fe abun-dance out to 0.3 r observed with XMM-Newton. InSection 2, we summarize the observations and data prepa-ration. Section 3 describes our analysis of the data, andin Section 4, the temperature and the O, Mg, Si, S, andFe abundances are determined. We discuss our results inSection 5.We use the Hubble constant H = 70 km s − Mpc − .The distance to the Fornax cluster is D L = 19 . ′ corresponds to 5.70 kpc. The virial radius, r =1 . h − p k h T i /
10 keV Mpc (Markevitch et al. 1998;Evrard et al. 1996), is about 1 Mpc for the average tem-perature k h T i = 1 .
2. Observations
Suzaku performed six pointing observations of theFornax cluster, as summarized in Table 1. The first ob-servation (hereafter, Center field) was carried out on 2005September with the pointing direction 2 ′ south and 1 ′ eastof NGC 1399. The second one (hereafter, North field) wascentered 13 ′ north and 4 ′ east of NGC 1399, and was car-ried out on 2006 January. Four additional observationswere centered ∼ ′ north (hereafter, Far North field),and ∼ ′ –27 ′ south, northwest, and northeast (hereafter,South, North West, and North East fields, respectively)of NGC 1399. The left panel of Figure 1 shows a 0.5–4 . ′ , or ∼
240 kpc toward the north from NGC 1399: this distancecorresponds to 0.24 r . To constrain emissions from ourGalaxy, we also observed two fields offset of ∼ × × XSELECT(Ver. 2.4a) . The analysis was performed us-ing
HEAsoft(Ver. 6.6.3) and
XSPEC(Ver. 11.3.2ag) .After the standard data selection criteria are applied, theexposure times of the four offset fields are 35–56 ks. Thoseof the Galactic1 and Galactic2 fields are ∼
20 ks.The spectra of the Center and North fields were accu-mulated within concentric rings, 6 ′ –13 ′ and 13 ′ –26 ′ , cen-tered on NGC 1399. The spectra of the Far North, South,North West and North East fields were accumulated overthe field of view of XIS. Each spectrum was binned toobserve details in metal lines, and each spectral bin con-tained 50 or more counts.The response of the X-ray telescope (XRT) and XISfor each spectrum was calculated using the xisrmfgen re-sponse matrix file (RMF) generator, version .The ancillary response files (ARF) were calculated using xissimarfgen (Ishisaki et al. 2007), version ,assuming flat emission, because the Fornax cluster is muchmore extended than the field of view of the XIS, andARFs assuming flat-sky emission, a β − model profile, anda point-source are almost the same within the energyrange of 0.4 to 5 keV, except for normalization. Slightdegradation of the energy resolution was considered inthe RMF, and decrease in the low-energy transmission ofthe XIS optical blocking filter (OBF) was included in theARF.The non-X-ray background (NXB) was subtracted fromthe spectra using a database of night Earth observations(Tawa et al. 2008). We used the spectra within an energyrange of 0.4–5.0 keV, because above 5 . We analyzed 15 pointing XMM-Newton observations ofthe Fornax cluster (Table 1).The middle and right panels of Figure 1 show a 0.8–1.2keV MOS image. The NXB and Cosmic X-ray background(CXB) were not subtracted because this energy band isdominated by the ICM emission. The observed regioncovers out to 0.3 r in the east, north and southwestdirections. We used the PN, MOS1, and MOS2 detectorsand selected events with patterns smaller than 5 and 13for the PN and MOS, respectively. To screen backgroundflares, we constructed count rate histograms of PN andMOS. Then, we fitted each histogram with a Gaussianand selected the time within 2.5 σ of the mean for eachhistogram. The total exposures after background flaresare summarized in Table 1. Because the observations offields G and L were completely dominated by very highbackground flares, no exposure time remained.Spectra were accumulated in pie regions of north, east,south, and west, centered on NGC 1399 as summarizedin Figure 1. We also accumulated spectra in the square Table 2.
The results of the spectral fitting of theGalactic emission with Suzaku.
Region kT kT Ratio of χ /d.o.f.(keV) (keV) Norm ∗ -Galactic1 0 . +0 . − . . +0 . − . . +0 . − . / . +0 . − . . +0 . − . . +0 . − . / ∗ The ratio of normalizations of higher and lower temperaturecomponents. regions of 0.1 r × r (17.5 ′ × ′ ). Each squareregion surrounding NGC 1399 is divided into two: one is asmall square region 0.05 r × r and the other, asshown in Figure 1. When accumulating spectra, luminouspoint sources and NGC 1404 were excluded. Although wehave not screened hot chips of MOS (Kuntz & Snowden2008), we have verified that exclusions of the hot chipsdo not affect any results. RMF and ARF correspondingto each spectrum were calculated using SAS v8.0.0. Thespectral analysis also used the XSPEC v11.3.2ag package.
3. Spectral Analysis
To estimate the Galactic emission, we first analyzed thetwo Galactic fields observed with Suzaku. The Galacticemission, which includes the local hot bubble, the MilkyWay halo, and solar wind charge exchange, is empiricallyfitted with a two-temperature plasma model with redshift= 0 (Lumb et al. 2002; Yoshino et al. 2009). Therefore,we used the two-temperature APEC thermal model forthe Galactic emission, with a power-law model for theCXB, and fitted the observed spectra from the Galactic1and Galactic2 fields. The temperature and normalizationof the two components in the Galactic emission were leftfree, with the metal abundance fixed to the solar level.CXB was modeled by a power-law spectrum with a photonindex Γ = 1 .
4. The normalization of the power-law wasallowed to be free. The model spectra, except for thelower-temperature APEC component, were subjected to acommon interstellar absorption N H , fixed at the Galacticvalue in the direction of each field by Dickey & Lockman(1990).The results of the spectral fits are shown in Table 2.Except for the instrumental Al line at 1.5 keV, this modelreproduced the spectra well. The derived temperaturesof the two APEC components and the ratio of the nor-malizations of the two Galactic fields are consistent witheach other. The derived temperatures, ∼ ∼ We first assumed an ICM in each region consisting ofa single-temperature vAPEC (Smith et al. 2001) model(hereafter, the 1T model). We fitted all the XIS spectraof each region with the 1T model for the ICM, a power-law model for the CXB, and the two-temperature APECmodel for the Galactic components. The temperature, Murakami et al. [Vol. ,
Fig. 1. (left) Suzaku-XIS (0.5–4.0 keV) image of the Fornax cluster. The NXB was subtracted and the difference in exposuretimes was corrected. The Cosmic X-ray background (CXB) was not subtracted. (middle) Raw XMM-MOS image (0.8–1.2 keV)of the Fornax cluster. Magenta circles correspond to field of view of the XMM observations. Red squares indicate the four offsetobservations with Suzaku. Blue pie and black square regions summarize the accumulation area of spectral analysis. (right) Exposure-and vignetting-corrected and adaptively smoothed XMM-MOS image (0.8–1.2 keV). The NXB and CXB were not subtracted.
Table 1.
Suzaku and XMM observations of the Fornax cluster and background fields
Fields Seq. No. (R.A., Dec.) in J2000.0 Date of obs. Exp. time (after screenings)Suzaku observations of the Fornax clusterCenter 100020010 (3 h m . s − ◦ ′ . ′′
5) 2005/09/13 76 ksNorth 800002010 (3 h m . s − ◦ ′ . ′′
6) 2006/01/05 78 ksFar North 802021010 (3 h m . s − ◦ ′ . ′′
4) 2008/01/14 56 ksSouth 803006010 (3 h m . s − ◦ ′ . ′′
1) 2008/07/15 35 ksNorth West 803007010 (3 h m . s − ◦ ′ . ′′
7) 2008/07/16 41 ksNorth East 803008010 (3 h m . s − ◦ ′ . ′′
1) 2008/07/17 41 ksSuzaku observations of the Fornax Galactic fieldGalactic1 802037010 (3 h m . s − ◦ ′ . ′′
0) 2007/06/28 20 ksGalactic2 802040010 (3 h m . s − ◦ ′ . ′′
8) 2007/06/29 21 ksXMM observations of the Fornax cluster MOS1, MOS2, PNA 0550930101 (3 h m s .4, -35 ◦ ′ ′′ .2) 2008/06/28 10.6, 11.2, 7.0 ksB 0550930201 (3 h m s .2, -34 ◦ ′ ′′ .2) 2008/06/27 7.6, 6.7, 3.5 ksC 0550930301 (3 h m s .1, -34 ◦ ′ ′′ .8) 2008/07/17 11.4, 11.6, 8.0 ksD 0550930401 (3 h m s .0, -35 ◦ ′ ′′ .8) 2009/02/09 14.5, 15.1, 11.7 ksE 0550930501 (3 h m s .8, -35 ◦ ′ ′′ .6) 2009/02/23 18.1, 17.8, 14.6 ksF 0550930601 (3 h m s .0, -35 ◦ ′ ′′ .0) 2009/02/24 17.7, 17.9, 13.7 ksG 0550930701 (3 h m s .1, -35 ◦ ′ ′′ .4) 2009/02/24 0, 0, 0 ksJ 0550931001 (3 h m s .9, -35 ◦ ′ ′′ .0) 2008/06/25 19.1, 19.4, 11.7ksL 0550931201 (3 h m s .6, -35 ◦ ′ ′′ .4) 2008/06/25 0, 0, 0 ksN 0550931401 (3 h m s .5, -35 ◦ ′ ′′ .8) 2008/06/26 11.0, 11.4, 8.0 ksNGC 1399 0400620101 (3 h m s .1, -35 ◦ ′ ′′ .0) 2006/08/23 99.6, 102.7, 53.2 ksNGC 1404 0304940101 (3 h m s .9, -35 ◦ ′ ′′ .8) 2005/07/30 24.6, 15.0, 17.0 ksLP 944-20 0055140101 (3 h m s .60, -35 ◦ ′ ′′ .0) 2001/01/07 43.0, 43.2, 36.5 ksRXJ 0337-3457 0210480101 (3 h m s .70, -34 ◦ ′ ′′ .0) 2005/01/04 44.3, 44.4, 37.9 ksNGC 1386 0140950201 (3 h m s .4, -35 ◦ ′ ′′ .0) 2002/12/29 15.9, 15.9, 12.8 kso. ] Suzaku and XMM-Newton observations of the Fornax cluster 5 c oun t s k e V − Far North 10.5 2 50.60.811.21.4 χ Energy (keV) 100010 c oun t s k e V − South 10.5 2 50.60.811.21.4 χ Energy (keV)100010 c oun t s k e V − North West 10.5 2 50.60.811.21.4 χ Energy (keV) 100010 c oun t s k e V − North East 10.5 2 50.60.811.21.4 χ Energy (keV)
Fig. 2.
Spectra of the four offset fields from the XIS-1 instruments fitted with the 1T (blue) or 2T (red) model for the ICM (solidlines) and background components (dashed lines) including the power-law and the Galactic components. Lower panels show thedata-to-model ratios for the 1T (blue crosses) and the 2T (red diamonds) model fits.
Murakami et al. [Vol. ,abundances, and normalization of the ICM component ofeach field were allowed to vary. Here, the metal abun-dances of He, C, and N were fixed to the solar values. Wedivided the other metals into six groups: O; Ne; Mg andAl; Si; S, Ar, and Ca; Fe and Ni. The spectral compo-nents except for the lower-temperature Galactic emissionwere subjected to a common interstellar absorption, N H ,which was allowed to vary for each instrument, consideringthe systematic uncertainties of contaminants on the XISdetectors. The temperatures of the two Galactic compo-nents were fixed at 0.1 keV and 0.3 keV, respectively, andthe ratio of the normalizations of the higher- and lower-temperature components was fixed at 0.1, which was thebest-fit value for the two Galactic fields.The results are shown in Table 3 and the fitted spectraof the four new offset fields are shown in Figure 2. Thederived reduced χ values were ∼
1, and the spectra of theFornax cluster were consistently reproduced by the sumof the ICM model and this Galactic component.We also applied a two-temperature model for the ICM(hereafter, the 2T model), where the abundances of eachmetal in the two components were assumed to have thesame value. The results are shown in Table 3 and Figure2. Compared to the 1T model, the reduced χ decreasedslightly by several percent, whereas, the metal abundancesincreased by several tens of percent. As shown in Table3, F-test probabilities support an addition of the secondtemperature component in most of the systems. However,considering the possible systematic uncertainties in theFe-L and background components, we cannot determinewhether the 2T model is better than the 1T model becausethe 1T model still represents the spectra fairly well.To study the systematic uncertainties in the modelingof the Galactic components, we fitted the spectra with the1T or 2T model for the ICM in a similar manner, exceptthat the normalizations of the two Galactic components ofeach field were allowed to be free. The temperature andabundances of Ne, Mg, Si, S, and Fe, and the weightedaverages of O/Fe, Ne/Fe, Mg/Fe, Si/Fe, and S/Fe werealmost the same within 10–20%, although the best-fit val-ues of the O abundances were changed by 0.1–0.2 solar. We used the spectra from MOS1, MOS2 and PN inthe energy ranges of 0.4–4.0 keV, but ignored the energyrange of 1.4–1.6 keV. The spectra in each region (pie andsquare) were fitted simultaneously with the 1T model forthe ICM, the Galactic emission, CXB, and NXB compo-nents. Here, the temperatures and normalizations of theGalactic components were fixed at the best-fit values de-rived from the spectral fitting of the Suzaku data. Theabundance ratios were also fixed at the weighted averagesof the four offset fields observed with Suzaku (Section 4.3).We modeled the NXB spectrum with a powerlaw/b modeland two Gaussians for the instrumental lines at 1.48 keV(Al) and 1.74 keV (Si). The powerlaw/b model is notfolded through the ARF, and differs from a power-law.We added two Gaussians at 0.56 keV and 0.65 keV forthe solar wind charge exchange. Figure 3 shows represen- . . c oun t s s − k e V − − − χ channel energy (keV) Fig. 3.
Representative spectra of a square region of MOS1(black) and PN (gray), fitted with the 1T ICM model. Dottedand other lines corresponds to the ICM components and back-ground, respectively. tative spectra fitted in this manner. We also fitted thespectra of deep fields in the same energy range and con-firmed that this background model reproduced the datawell. Most of the spectra were fitted with the 1T modelfor the ICM with reduced χ ∼
1. We also applied the2T model for the ICM. Within 0.05 r , the 2T modelshowed a significantly decreased χ . We adopted the Feabundance from the 2T model fits within 0 . r .
4. Results
Figure 4 summarizes the temperatures of the > ′ re-gions of Suzaku and the pie regions of XMM versus radiusof NGC 1399, in comparison with those from the central6 ′ of NGC 1399 by Matsushita et al. (2007a). The tem-peratures derived from XMM and Suzaku are mostly con-sistent with each other.The observed ICM temperatures are ∼ r which further decreases with radius to ∼ r . The azimuthal dependence of thetemperature is largest at 0.1-0.2 r : the ICM temper-atures in the western and southern regions are 1.2 ∼ ∼ Table 3.
The ICM temperature, χ , and elemental abundances derived from the spectral fits of Suzaku for the Fornax cluster. Field r model kT kT χ /d.o.f. F-test probability ∗ (arcmin) (keV) (keV)Center 6–13 1T 1 . +0 . − . . +0 . − . . +0 . − . × − North 6–13 1T 1 . +0 . − . . +0 . − . . +0 . − . × − North 13–26 1T 1 . +0 . − . . +0 . − . . +0 . − . × − North West 7–28 1T 1 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . × − North East 17–36 1T 1 . +0 . − . . +0 . − . . +1 . − . . +0 . − . . +0 . − . . +0 . − . r model O Ne Mg Si S Fe(arcmin) (solar) (solar) (solar) (solar) (solar) (solar)Center 6–13 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . Center 6–13 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North 6–13 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North 6–13 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North 13–26 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North 13–26 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North West 7–28 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North West 7–28 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . South 10–31 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . South 10–31 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North East 17–36 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . North East 17–36 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . Far North 20–42 1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . Far North 20–42 2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ∗ F-test probability for adding second temperature component the northeastern regions. We also derived a hardness ratiomap of 1.0–1.2 keV to 0.7–1.0 keV observed with XMM(Figure 5). The map shows similar ICM temperatures.The southwest and northwest regions tend to have higherICM temperatures than the northeast regions.
The radial profiles of the Fe abundances observed withSuzaku and the pie regions of XMM derived from the 1Tmodel fits are summarized in Figure 6. The values derivedfrom the XMM and Suzaku data are mostly consistentwith each other, although the error bars for Suzaku aresmaller than those for XMM.In the Suzaku data, at 0.03–0.2 r , the Fe abundancesof the North, North West, and North East fields derivedfrom the 1T model for the ICM are about 0.5–0.6 solar,which is higher than the 0.3–0.4 solar value of the Southfield. The 2T model for the ICM gives Fe abundances thatare systematically higher by ∼ r .As derived from the Suzaku observations, at 0.05–0.2 r ,the northern regions have higher Fe abundances of 0.5–0.6 solar, whereas the south and west regions have lowervalues of 0.3–0.4 solar. Beyond 0.2 r , the Fe abundancewith the 1T model decreases to ∼ ∼ . r hasa lower Fe abundance than the other regions at the sameradius from NGC 1399. Murakami et al. [Vol. , Fig. 5. (left) Temperature map of the ICM derived from spectral fitting of the XMM data of the square regions of 0.1 r × r (17.5 ′ × ′ ). Temperatures of the four small square regions of 0.05 r × r surrounding NGC 1399 are not plotted, becausethe 2T model is required to fit the spectra. (right) Hardness ratio map of 1.0–1.2 keV to 0.7–1.0 keV. White squares indicate theregions for the spectral analysis in the left panel. Green squares show the four offset regions (South, North West, North East, FarNorth) observed with Suzaku. (solar) F e a bund a n ce ( s o l a r) r / r NortheastSoutheastSouthwestNorthwest
Fig. 8. (left) Fe abundance map of the ICM derived from the spectral fitting of the XMM data of the square regions by the2T model fits for the four central small square regions of 0.05 r × r , and by the 1T model fits for the other regions of0.1 r × r . NGC 1399 is located at the center of the four small square regions. (right) Radial profile of the ICM of the squareregions in the left panel derived from the 2T model (filled circles with sold error bars) and 1T model (diamonds). Red, green, blueand magenta colors correspond to the square regions of northeast, southeast, southwest, and northwest, respectively, of NGC 1399. o. ] Suzaku and XMM-Newton observations of the Fornax cluster 9 O ( s o l a r) M g ( s o l a r) S i ( s o l a r) −3 S ( s o l a r) r/r O / F e M g / F e S i/ F e −3 S / F e r/r Fig. 9. (left) Abundance profiles of O, Mg, Si, and S in the Center (black), North (orange), North West (magenta), South (blue),North East (green), and Far North (red) fields derived from the 1T (open circles) or 2T (diamonds) model for the ICM. Centralthree radial bins are derived by Matsushita et al. (2007a). (right) Radial profiles of the O/Fe, Mg/Fe, Si/Fe, and S/Fe ratios in solarunits. The data within 0.02 r are weighted averages of the values of the two innermost radial bins. Using the Suzaku observations, we also derived theabundances of O, Ne, Mg, Si, and S (Table 3, Figure 9).From 0.05 r to 0.2 r , the Mg, Si, and S abundancesderived from the 1T model fits are 0.2–0.5 solar, and 0.3–0.6 solar from the 2T model fits. From the 1T model fits,the abundances of O are 0.2–0.4 solar, with fairly largeerror bars, and the 2T model yields O abundance valuesthat are larger by 0.1 solar. When the 1T model is used,the derived Ne abundance values tend to be lower thanthose of other elements, whereas with the 2T model, theNe abundances are also close to those of Fe.Right panel of Figure 9 summarizes the abundance ra-tios of α -elements divided by the Fe value in solar units.Because the abundances of α -elements and Fe are corre-lated, the errors of the abundance ratios were estimatedfrom confidence contours of each α -element against Fe.The 1T and 2T model fits give similar values for the O/Fe,Mg/Fe, Si/Fe, and S/Fe ratios. The derived abundanceratios are mostly consistent with the absence of radial gra-dients and azimuthal dependence, although the Si/Fe ra-tio shows a small negative radial gradient. The weightedaverages of the abundance ratios are calculated for thedata with similar radial ranges and are summarized inTable 4. The abundance ratios are mostly consistent withthe absence of radial dependence. When the values be-yond 6 ′ from the 1T model fits are averaged, the abun-dance ratios of O/Fe, Mg/Fe, Si/Fe, and S/Fe are about 0.7, 0.7, 0.8, and 0.9 in solar units, respectively, The 2Tmodel yields similar abundance ratios within 10–15%.The weighted averages of Ne/Fe ratios from the 1T and2T model fits differs by a factor of 3–4. This is becausethe K-shell lines of Ne are completely mixed with the Fe-Llines, and the Fe-L lines between the 1T and 2T modelsare different, which may cause a discrepancy. Therefore,the derived abundances of Ne might include fairly largesystematic uncertainties. We derived radial profiles of the X-ray surface bright-ness (left panel of Figure 10) by azimuthally averagingthe background-subtracted intensity in the 0.8–1.2 keVenergy band of MOS centered on NGC 1399, and on thecluster center derived by Paolillo et al. (2002), which is ∼ ′ northeast of NGC 1399. Here, luminous point sources anda region around NGC 1404 were excluded. For the profilecentered on the cluster center, a region around NGC 1399was also excluded. Beyond 10 ′ , the two radial profiles re-semble each other, whereas within 10 ′ , a sharp brightnesspeak centered on NGC 1399 appears. Similar to Paolilloet al. (2002), we fitted the radial profile centered on NGC1399 with a sum of three β -models that are all centeredon NGC 1399. As shown in Figure 10, the radial pro-file was roughly reproduced with these three β -models.Considering the similarity of the brightness profiles cen-tered on NGC 1399 and on the cluster center beyond 10 ′ ,the density profile of the ICM from the cluster center0 Murakami et al. [Vol. , Table 4.
Weighted averages of the abundance ratios in solar units derived from the Suzaku observations region model O/Fe Ne/Fe Mg/Fe Si/Fe S/Fe0 ′ –4 ′∗
2T 0 . +0 . − . – 0 . +0 . − . . +0 . − . . +0 . − . ′ –42 ′†
1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –42 ′†
2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –13 ′†
1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –13 ′†
2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –31 ′†
1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –31 ′†
2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –42 ′†
1T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ′ –42 ′†
2T 0 . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . ∗ weight average within 4 ′ from NGC 1399 from Matsushita et al. (2007a) † Weighted averages of abundance ratios at 6 ′ –42 ′ , 6 ′ –13 ′ (Center and North),7 ′ –31 ′ (South and North West), and 17 ′ –42 ′ (North East and Far North) −3 T e m p e r a t u r e ( k e V ) r / r NorthEastSouthWest Far NorthNorth EastSouthNorth WestXMM SuzakuCenterNorth
Fig. 4.
The ICM temperatures from Suzaku data versus ra-dius of NGC 1399 in units of r for the Center (black opencircles), North (orange open circles), Far north (red open cir-cle), North East (green open circle), South (blue open circle),and North West (magenta open circle) fields derived from the1T model fit for the ICM, and those from the 2T (black filledtriangles) model fits within the central two radial bins. Thedata of the innermost three radial bins are from Matsushitaet al. (2007a). The ICM temperatures of the pie regions ofXMM with the 1T model fits are plotted as diamonds. Red,green, blue and magenta correspond to the northern, eastern,southern and western pie regions, respectively. −3 . . . . F e a bund a n ce ( s o l a r) r / r NGC 507HCG 62NGC 5044
Fornax NorthFornax SouthFornax WestFornax East Far NorthNorth EastSouthNorth West Center
North
Fig. 6.
Radial profile of the Fe abundance of the ICM de-rived from the 1T model fits for the ICM using Suzaku data(open circles) and XMM data of the pie regions (diamonds).Colors have the same meanings as in Figure 4. Data of the in-nermost three radial bins are from Matsushita et al. (2007a).Here, those of the innermost two radial bins are derived fromthe 2T model fits. Region at 0.035–0.07 r of the Centerfield covers mostly the south of NGC 1399. should also be similar to that from NGC 1399.To study the azimuthal variation in the radial bright-ness profile, we divided the images centered on NGC 1399,into four sectors–north, west, south, and east– and derivedradial brightness profiles within each sector. The radialprofiles of the four sectors are plotted in the middle panelof Figure 10. At ∼ ′ ( ∼ . r ), the brightness levelsin the north and east sectors are higher by a factor of2–3 than those in the south and west sectors. We also de-rived radial brightness profiles in the four sectors centeredon the cluster center. The discrepancy in the brightnesslevels of the four sectors became smaller (right panel ofFigure 10).o. ] Suzaku and XMM-Newton observations of the Fornax cluster 11 − − . . c oun t r a t e [ c oun t s / s / a r c m i n ] r (arcmin) 0.1 1 10 − − . . c oun t r a t e [ c oun t s / s / a r c m i n ] r (arcmin) 0.1 1 10 − − . . c oun t r a t e [ c oun t s / s / a r c m i n ] r (arcmin) Fig. 10. (Left) Radial profiles of the surface brightness centered on NGC 1399 (black) and on the cluster center (red) by Paolilloet al. (2002), which is ∼ ′ northeast of NGC 1399. The solid line represents the best-fit triple β -model. (middle) Radial profiles ofnorth (red), west (magenta), east (green), and south (blue) sectors centered on NGC 1399. (right) The same as the middle panel,but centered on the cluster center. . F e a bund a n ce ( s o l a r) r/r . . F e T ( s o l a r) Fe (solar) Far NorthNorth EastSouthNorth WestCenter 6−13’North Fig. 7. (upper panel) Fe abundance of the ICM observedwith Suzaku derived from the 2T model fits plotted againstthose from the 1T model fits. Black cross, orange filledsquares, blue open triangle, magenta open square, greenstar, and red filled triangle correspond to the Center, North,South, Northwest, Northeast, and Far North fields, respec-tively. (lower panel) Radial profiles of the Fe abundance ofthe ICM derived from the 2T model fits of the Suzaku data.
5. Discussion
In Figure 11, the radial profiles of the Fe abundancein the Fornax cluster derived from the 1T and 2T modelfits are compared with those in other groups of galaxiesobserved with Suzaku: the NGC 507 group (Sato et al.2009a), HGC 62 group (Tokoi et al. 2008), the NGC 5044group (Komiyama et al. 2009), and the fossil group NGC1550 (Sato et al. 2010). Here, the Fe abundances of theFornax cluster at similar radial ranges from NGC 1399are averaged. Considering the possible uncertainties inthe Fe-L atomic data and background components, wecannot conclude that the 2T model fit is better than thatthe 1T model, because the two models yield similar χ values in the region with low surface brightness.The Fe abundance within 0.03 r from NGC 1399 isabout 1 solar. At 0.05 r of the Fornax cluster, the 1Tand 2T model fits give the Fe abundances of ∼ . ∼ . ∼ r which are derived from the 2Tmodel fits. At 0.1–0.2 r , the Fe abundances of theFornax cluster are 0.4 solar and 0.6 solar, from the 1T and2T model fits, respectively. The value from the 2T modelfits is close to those of the NGC 507 and NGC 5044 groups,although HCG 62, a compact group of galaxies, and NGC1550, a fossil group, tend to have smaller Fe abundanceswhen the 2T model fits are used. These values are alsosimilar to those of the clusters of galaxies (Leccardi &Molendi 2008; Matsushita 2011). There is no systematicdifference between the poor systems and the clusters ofgalaxies regarding the Fe abundance of the ICM at 0.1–0.2 r In Figure 11, the Fe abundances derived from theSuzaku data are compared with those of the best-fitregression relations from Chandra data (Rasmussen &Ponman 2007) and from XMM data (Johnson et al. 2011).2 Murakami et al. [Vol. ,Here, we rescaled for the differences in the definition ofthe solar abundance table and virial radius. Johnsonet al. (2011) showed that within ∼ r , the 2T fitgives abundances that are higher by a factor of ∼ r and somegroups have comparable abundance profiles with those ob-served with Suzaku. The radial profile of Fe abundance ofthe HCG 62 system with Suzaku is consistent with thoseof the Chandra groups. Therefore, a major part of thedifferences in the Fe abundance should be caused by thedifferences in the sample. Different assumptions regard-ing the abundance ratio, i.e. VAPEC, or APEC, may notbe responsible for the discrepancy, since the abundancepatterns of these groups of galaxies derived with Suzakudo not different greatly from the solar ratio. On the otherhand, as shown in Komiyama et al. (2009), in regionswith low surface brightness, uncertainties in the Galacticemission can also affect the derived Fe abundance. Figure 12 summarizes the radial profiles of the O/Fe,Mg/Fe, Si/Fe, and S/Fe ratios in five poor systems (theFornax cluster, the NGC 5044 group, the HCG 62 group,and the NGC 507 group, and a fossil group, NGC 1550),and clusters of galaxies, A262 ( kT ∼ kT ∼ r , the scatter inthe abundance ratios, except for the Mg/Fe ratios, amongthese systems is relatively small. The scatter in the Mg/Feratios may be due to the systematic uncertainties in theFe-L lines around the Mg-K lines.In these poor systems, the abundance ratios exhibit nosignificant radial dependence. The abundance pattern ofO/Mg/Si/S/Fe in the offset regions of the Fornax clusteris similar to those of the other groups of galaxies. Outto 0.3 r there is no hint of increase in the ratio of α -elements to Fe abundances in these poor systems. Theseresults indicate that both SN Ia and SN II products havebeen mixed into the ICM, and that the ratio of the twotypes of SN in the Fornax cluster is similar to those of theother groups of clusters of galaxies. Because most of metals in the ICM synthesized in galax-ies, the metal-mass-to-light ratio is a useful measure forstudying the chemical evolution of clusters of galaxies. . . . F e a bund a n ce ( s o l a r) r / r NGC 507HCG 62NGC 5044NGC 1550Fornax (Suzaku 1T)Fornax(XMM)
Fornax (Suzaku) cD clusters groups (Rasmussen et al. 2007)(Suzaku)NW(Suzaku)
Chandra groupsXMM groups (2T)XMM groups (1T)Fornax (Suzaku 2T)
Fig. 11.
Radial profiles of the Fe abundance in the Fornaxcluster observed with Suzaku derived from the 1T model (redfilled circles with solid lines) and the 2T model (red open cir-cles with dotted lines) and with XMM from the 1T model(red diamonds). Here, the Fe abundances with similar ra-dial ranges are averaged. Those of the NGC 507 group (or-ange filled squares; Sato et al. 2009a), HCG 62 group (cyanfilled triangles; Tokoi et al. 2008), NGC 5044 group (ma-genta crosses; Komiyama et al. 2009), and a fossil group,NGC 1550 (green stars; Sato et al. 2010). The weightedaverage of relaxed clusters with a cD galaxy at their centerobserved with XMM (blue filled circles;Matsushita 2011) andthe best-fit regression relations for groups of galaxies observedwith Chandra (black solid line: Rasmussen & Ponman 2007)and cool-core groups observed with XMM (2T:black dashedline,1T:black dotted line; Johnson et al. 2011) are also plot-ted.
To estimate the metal mass, we used the radial bright-ness profile derived in section 4.4, and derived the gas den-sity and gas mass profiles. Then, integrated mass profilesof O and Fe were derived from the gas mass and abun-dance profiles using the 1T model fits. The 2T model fitsyielded 10–20% smaller normalizations of the ICM andseveral tens of % higher Fe abundances in the Fornax clus-ter compared with the 1T model fits. Therefore, the totalmetal mass may have systematic uncertainties of severaltens of percent, due to uncertainties in the temperaturestructure.Because the K-band luminosity of a galaxy correlateswell with the stellar mass, we calculated the luminosityprofile of the K-band. We collected K-band magnitudesof galaxies in 6 × box centered on NGC 1399 fromthe Two Micron All Sky Survey (2MASS). NGC 1399 hasan apparent magnitude of m K = 6 . L K /L K , ⊙ =11 . A K = 0 . ′ < r < ′ (0.86 r < r < . r ) is subtracted asthe background. Within 0.3 r of the Fornax cluster, ∼
10 early-type galaxies dominate the K-band luminosity.o. ] Suzaku and XMM-Newton observations of the Fornax cluster 13 . O / F e . M g / F e . S i/ F e . S / F e r/r Fornax AWM7A262N1550 N5044HCG62 N507
Fig. 12.
Radial profiles of O/Fe, Mg/Fe, Si/Fe, and S/Feratios of the Fornax cluster (red) from the 1T (diamonds) and2T (filled circles) model fits. Here, these values are weightedaverages of similar radial ranges. Radial profiles of the fossilgroup NGC 1550 (green; Sato et al. 2010), the NGC 5044group (magenta;Komiyama et al. 2009), HCG 62 (light blue;Tokoi et al. 2008), and the NGC 507 group (orange; Sato et al.2009a), and those of the clusters of galaxies, AWM 7 (black;Sato et al. 2008), and A262 (blue;Sato et al. 2009b).
NGC 1399 dominates the luminosity in the region r < ∼ . r . At r ∼ . r , NGC 1404, a bright galaxy,causes a break in the luminosity profiles. At ∼ r ,several early-type galaxies including NGC 1380 contributeto the luminosity profiles.The integrated mass-to-light ratios for O and Fe(OMLRs and IMLRs) using the K-band luminosities aresummarized in Figure 13. The error bars of the mass-to-light ratios include only abundance errors. The OMLRand IMLR profiles are not smooth because of several lumi-nous galaxies in the Fornax cluster. The profiles increasewith radius out to 0 . r . However, the IMLR profilebecame flat from 0.2 r to 0 . r , due to increase of K-band luminosity from several luminous early-type galax-ies. We extrapolated the best-fit three β -model of thesurface brightness of the ICM, assuming that Fe abun-dance of the ICM beyond 0.3 r is the same as in thatof the best-fit value at 0.2–0.3 r . Then, at 0.5 r , the IMLR may not increase very much (Figure 13).Figure 13 compares the derived OMLRs and IMLRsof the Fornax cluster with those of the NGC 5044 group(Komiyama et al. 2009), the fossil group NGC 1550 (Satoet al. 2010), and the Abell 262 cluster ( kT ∼ kT ∼ r , the IMLR and OMLRof the Fornax cluster are much smaller than those of theother systems. The similar IMLR profiles of the othersystems within 0.1 r indicate that the recent metalsupplies from the central galaxies are typical. The smallIMLR in the Fornax cluster indicates that the accumula-tion time scale of metals from the cD galaxy should beshorter, reflecting the fact that the cD galaxy is not lo-cated at the cluster center and moving. At 0.1–0.3 r ,the IMLR of the Fornax cluster is still an order of mag-nitude smaller than those of the rich systems, AWM 7,Abell 262, and NGC 1550. The IMLR of these three sys-tems increase with radius in the same way from 0.1 r to 0.3–0.5 r . However, the IMLRs of the NGC 5044group is constant from 0.1 r to 0.3 r , and at 0.3 r ,the IMLR of the NGC 5044 group is nearly an order ofmagnitude larger than that of the Fornax cluster. Theextrapolated value of the IMLR at 0.5 r of the Fornaxcluster are still over an order of magnitude smaller thanthose of the NGC 1550 group and Abell 262 cluster.Rasmussen & Ponman (2009) found that the IMLR de-rived from Chandra and Suzaku are consistent with eachother, despite the systematic difference in the derived Feabundance. This is because a larger Fe abundance is as-sociated with a smaller gas mass in lower temperaturegroups. Chandra observations revealed that NGC 1399 is mov-ing within the Fornax cluster (Scharf et al. 2005). TheX-ray emission of the northeast region of the Fornax clus-ter is brighter than that of the south region. The ICMtemperature of the southwest region is higher than thatof the northeast region. The lower ICM temperatureand the higher brightness of the northeast region indi-cate lower entropy in the ICM. The Fe abundance of thenortheast region at 0.1–0.2 r is higher than that of thesouthwest region at the same radius. Lower entropy andthe higher Fe abundance are usually observed in the coolcores. Therefore, the cD galaxy may have traveled fromthe center of the cluster to the south due to recent dy-namical evolution, as suggested by the optical dynamicalobservations by Dunn & Jerjen (2006).Recent dynamical evolution might have hindered thestrong concentration of hot gas in the central region ofthe Fornax cluster. The X-ray luminosity within 4 r e ofNGC 1399 is smaller by a factor of 20 than that of NGC5044 (Matsushita 2001; Nagino & Matsushita 2009). The4 Murakami et al. [Vol. , − − − . . I n t e g r a t e d I M L R r / r Abell 262AWM 7
Fornax allNGC 5044
NGC 1550 − − − . . I n t e g r a t e d O M L R r / r Abell 262
AWM 7
Fornax allNGC 5044
NGC 1550
Fig. 13.
Radial profiles of integrated IMLR (left panel) and OMLR (right panel) in the K-band of the Fornax cluster (red filledcircles), NGC 5044 group (purple open circle;Komiyama et al. 2009), NGC 1550 group (green crosses;Sato et al. 2010), Abell 262cluster (blue filled triangles;Sato et al. 2009b), and AWM 7 cluster (black open squares;Sato et al. 2008). Dotted line for the Fornaxcluster represent extrapolated values using the surface brightness and the Fe abundances within 0.3 r X-ray luminosity of NGC 1550 is also higher by an orderof magnitude than that of NGC 1399 (Fukazawa et al.2006). For NGC 1399, the central Fe peak is narrowerand the IMLR is much smaller than those of the othergroups. Some merging clusters also have smaller scaleof Fe peaks (Matsushita 2011). During cluster merging,mixing of the ICM could destroy the central Fe peak. TheFornax cluster is also in a stage of dynamical evolution,and may be in a phase of central Fe peak destruction.Then, recent supply of metals from NGC 1399 via stellarmass loss and SN Ia produces a smaller Fe abundancepeak than in the other groups.
The metal distribution in the ICM can be a powerfultracer of the history of gas heating in the early epoch,because the relative timing of metal enrichment and heat-ing should affect the present amount and distribution ofthe metals in the ICM. The observed IMLRs in poor sys-tems are scattered by an order of magnitude, whereas,the Fe abundance profiles are similar among these sys-tems and also similar to those of clusters. The Fornaxcluster has the smallest IMLR out to 0.3 r . If all galax-ies synthesize a similar amount of metals per unit stellarmass, the observed low IMLR and similar Fe abundancecompared with other systems indicate that a significantfraction of the Fe was synthesized in an early phase ofcluster evolution. If metal enrichment occurred before en-ergy injection, the poor systems would carry relativelysmaller metal mass with a smaller gas mass than rich clus-ters, whereas, the metal abundance would be quite similarto those in rich clusters. Dynamical evolutions may also change the gas distribution, but not the abundance of thegas. In contrast, if metal enrichment occurred after energyinjection, the metal mass becomes comparable to those inrich clusters and indicates a higher abundance reflectinga lower gas mass.Similar to rich clusters, most of the stellar light inthe Fornax cluster originates from bright old early-typegalaxies (Kuntschner 2000). The stellar metallicity and[Mg/Fe] ratios of the central regions of these galaxies aresimilar to those of similar size in rich clusters (Kuntschner2000). The O/Mg/Fe abundance pattern of the hot in-sterstellar medium (ISM) of NGC 1399 and NGC 1404 inthe Fornax cluster derived from the Suzaku observations(Matsushita et al. 2007a) are similar to those of NGC 4636(Hayashi et al. 2009) in the Virgo cluster. Reflection grat-ing spectrometer (RGS) observations onboard the XMMshowed that the O/Fe ratio of the ISM in NGC 1404 inthe Fornax cluster is similar to those of elliptical galaxiesin the Virgo cluster (Werner et al. 2009). Because the hotISM in elliptical galaxies results from the accumulation ofstellar mass loss and the ejecta from recent SNe Ia, theobserved similarity of the abundance patterns of the hotISM suggests that the stellar metallicity and SN Ia rate donot different greatly between the present elliptical galaxiesin the Fornax and Virgo clusters. These results indicatethat stars in elliptical galaxies in the Fornax cluster wereenriched in the same way as those in rich clusters.The observed higher stellar mass fraction and the lowergas mass fraction within r in poor systems, are some-times interpreted as demonstrating that the star forma-tion efficiency depends on the system mass. However, thesimilarity in abundances in the ICM between groups andclusters and a scatter in the metal-mass-to-light ratios cano. ] Suzaku and XMM-Newton observations of the Fornax cluster 15be better explained by early metal enrichment, as dis-cussed in Matsushita (2011), on the basis of the relativelyflat abundance profiles of Fe in the ICM in clusters ofgalaxies.
6. Summary and Conclusion
Suzaku and the XMM observed the Fornax cluster outto 0.2–0.3 r , and derived the temperature and O, Mg,Si, S, and Fe abundances in the ICM quite accurately.The ICM temperature decreases from 1.5 keV near thecD galaxy NGC 1399 to 1 keV in the outer region. TheFe abundance around NGC 1399 is about one solar, andthe central Fe abundance peak is narrower than that inthe groups and clusters with cool cores. The Fe abundancedrops to 0.3–0.5 solar at 0.1–0.2 r , which is similar tothe values in other groups and clusters. The abundanceratios, O/Fe, Mg/Fe, Si/Fe, and S/Fe, are close to thesolar ratio, similar to other groups and clusters of galaxies.The abundance pattern indicates that both SN Ia and SNII products have been mixed into the ICM, and the ratioof the two types of SN in the Fornax cluster is similar tothose of other groups and clusters of galaxies.The northeast region of NGC 1399 has a lowerICM temperature, higher abundances and higher surfacebrightness than the southwest regions. Therefore, the cDgalaxy may have traveled from the center of the cluster tothe south owing to recent dynamical evolution.Out to 0.3 r , the IMLR of the Fornax cluster is anorder of magnitude smaller than that of other groups andclusters. The metal distribution in the ICM can be usedas a tracer of the past history of heating and enrichment inclusters of galaxies. Scatter in the IMLR and similarity inthe abundances in the ICM indicate early metal synthesisin groups and clusters. References
Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta,53, 197Arnaud, M., Pointecouteau, E., & Pratt, G. W. 2007, A&A,474, L37Borgani, S., Finoguenov, A., Kay, S. T., Ponman, T. J.,Springel, V., Tozzi, P., & Voit, G. M. 2005, MNRAS, 361,233Buote, D. A. 2002, ApJ, 574, L135Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215Dunn, L. P., & Jerjen, H. 2006, AJ, 132, 1384Evrard, A. E., Metzler, C. A., & Navarro, J. F. 1996, ApJ,469, 494Finoguenov, A., Ponman, T. J., Osmond, J. P. F., & Zimer, M.2007, MNRAS, 374, 737Fukazawa, Y., Botoya-Nonesa, J. G., Pu, J., Ohto, A., &Kawano, N. 2006, ApJ, 636, 698Giodini, S., et al. 2009, ApJ, 703, 982Hayashi, K., Fukazawa, Y., Tozuka, M., Nishino, S.,Matsushita, K., Takei, Y., & Arnaud, K. A. 2009, PASJ,61, 1185Ishisaki, Y., et al. 2007, PASJ, 59, 113 Johnson, R., Ponman, T. J., & Finoguenov, A. 2009, MNRAS,395, 1287Johnson, R., Finoguenov, A., Ponman, T. J., Rasmussen, J.,& Sanderson, A. J. R. 2011, MNRAS, 413, 2467Kaiser, N. 1991, ApJ, 383, 104Komiyama, M., Sato, K., Nagino, R., Ohashi, T., &Matsushita, K. 2009, PASJ, 61, S337Koyama, K. et al. 2007, PASJ, 59, 23Kuntschner, H. 2000, MNRAS, 315, 184Kuntz, K. D., & Snowden, S. L. 2008, A&A, 478, 575Leccardi, A., & Molendi, S. 2008, A&A, 487, 461Lodders, K. et al. 2003, ApJ, 591, 1220Lumb, D. H., Warwick, R. S., Page, M., & De Luca, A. 2002,A&A, 389, 93Machacek, M., Dosaj, A., Forman, W., Jones, C., Markevitch,M., Vikhlinin, A., Warmflash, A., & Kraft, R. 2005, ApJ,621, 663Makishima, K., et al. 2001, PASJ, 53, 401Markevitch, M., Forman, W. R., Sarazin, C. L., & Vikhlinin, A.1998, ApJ, 503, 77Matsushita, K. 2001, ApJ, 547, 693Matsushita, K., Finoguenov, A., & B¨ohringer, H. 2003, A&A,401, 443Matsushita, K., et al. 2007a, PASJ, 59, 327Matsushita, K., B¨ohringer, H., Takahashi, I., & Ikebe, Y.2007b, A&A, 462, 953Matsushita, K. 2011, A&A, 527, A134Mitsuda, K., et al. 2007, PASJ, 59, 1Nagino, R., & Matsushita, K. 2009, A&A, 501, 157Paolillo, M., Fabbiano, G., Peres, G., & Kim, D.-W. 2002, ApJ,565, 883Ponman, T. J., Cannon, D. B., & Navarro, J. F. 1999, Nature,397, 135Ponman, T. J., Sanderson, A. J. R., & Finoguenov, A. 2003,MNRAS, 343, 331Pratt, G. W., et al. 2010, A&A, 511, A85Rasmussen, J., & Ponman, T. J. 2007, MNRAS, 380, 1554Rasmussen, J., & Ponman, T. J. 2009, MNRAS, 399, 239Sato, K., et al. 2007, PASJ, 59, 299Sato, K., Matsushita, K., Ishisaki, Y., Sasaki, S., Ohashi, T.,Yamasaki, N. Y., & Ishida, M. 2008, PASJ, 60, 333Sato, K., Matsushita, K., Ishisaki, Y., Yamasaki, N. Y.,Ishida, M., & Ohashi, T. 2009a, PASJ, 61, S353Sato, K., Matsushita, K., & Gastaldello, F. 2009b, PASJ, 61,S365Sato, K., Kawaharada, M., Nakazawa, K., Matsushita, K.,Ishisaki, Y., Yamasaki, N. Y., & Ohashi, T. 2010, PASJ,62, 1445Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500,525Scharf, C. A., Zurek, D. R., & Bureau, M. 2005, ApJ, 633, 154Simionescu, A., Werner, N., B¨ohringer, H., Kaastra, J. S.,Finoguenov, A., Br¨uggen, M., & Nulsen, P. E. J. 2009,A&A, 493, 409Sun, M., Voit, G. M., Donahue, M., Jones, C., Forman, W., &Vikhlinin, A. 2009, ApJ, 693, 1142Smith, R. K., Brickhouse, N. S., Liedahl, D. A., &Raymond, J. C. 2001, ApJL, 556, L91Tamura, T., Kaastra, J. S., Makishima, K., & Takahashi, I.2003, A&A, 399, 497Tawa, N., et al. 2008, PASJ, 60, 11Tokoi, K., et al. 2008, PASJ, 60, 3176 Murakami et al. [Vol. ,