Suzaku observation of a high entropy cluster Abell 548W
Kazuhiro Nakazawa, Yuichi Kato, Liyi Gu, Madoka Kawaharada, Motokazu Takizawa, Yutaka Fujita, Kazuo Makishima
aa r X i v : . [ a s t r o - ph . H E ] F e b Publ. Astron. Soc. Japan (2014) 00(0), 1–10doi: 10.1093/pasj/xxx000 Suzaku observation of a high entropy clusterAbell 548W
Kazuhiro Nakazawa , Yuichi Kato , Liyi Gu , Madoka Kawaharada ,Motokazu Takizawa , Yutaka Fujita and Kazuo Makishima Department of Physics, the University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113–0033 SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, TheNetherlands Tsukuba Space Center, Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba,Ibaraki 305-8505, Japan Department of Physics, Yamagata University, 1-4-12 Kojirakawa-machi, Yamagata 990-8560 Department of Earth and Space Science, Graduate School of Science, Osaka University, 1-1Machikaneyama-cho, Toyonaka, Osaka 560-0043, Japan MAXI Team, Institute of Physical and Chemical Research, 2-1 Hirosawa, Wako, Saitama351-0198 ∗ E-mail: [email protected]
Received h reception date i ; Accepted h acception date i Abstract
Abell 548W, one of the galaxy clusters located in the Abell 548 region, has about an orderof magnitude lower X-ray luminosity compared to ordinal clusters in view of the well knownintracluster medium (ICM) temperature vs X-ray luminosity ( kT - L X ) relation. The cluster hostsa pair of diffuse radio sources to the north west and north, both about ′ apart from thecluster center. They are candidate radio relics, frequently associated with merging clusters.A Suzaku deep observation with exposure of 84.4 ks was performed to search signatures formerging in this cluster. The XIS detectors successfully detected the ICM emission out to ′ from the cluster center. The temperature is ∼ . keV around its center, and ∼ keV at theoutermost regions. The hot region ( ∼ keV) aside the relic candidates shifted to the clustercenter reported by XMM-Newton was not seen in the Suzaku data, although its temperatureof 3.6 keV itself is higher than the average temperature of 2.5 keV around the radio sources.In addition, a signature of a cool ( kT ∼ . keV) component was found around the north westsource. A marginal temperature jump at its outer-edge was also found, consistent with thecanonical idea of shock acceleration origin of the radio relics. The cluster has among thehighest central entropy of ∼ keV cm and is one of the so-called low surface brightnessclusters. Taking into account the fact that its shape itself is relatively circular and smooth andalso its temperature structure is nearly flat, possible scenarios for merging is discussed. Key words: galaxies: clusters: individual: Abell 548W — galaxies: clusters: intracluster medium —X-rays: galaxies: clusters
Merging clusters of galaxies is an aspect of gravitational grow-ing of the large scale structure. While there are many “ap- parently circular and relaxed” clusters, 10–20% of the clus-ters show evidences for on-going merger, consistent with thedynamical timescale of a cluster ( ∼ years) compared to c (cid:13) Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
Hubble time ( ∼ years). The intracluster medium (ICM),the vast hot plasma emitting X-rays, permeating the gravita-tional potential of the cluster is strongly affected by the mergerevents. In X-rays, many mergers have complicated shapes, of-ten strongly elongated, with signatures of complex temperaturestructure. Entropy of the ICM of mergers are often high, con-sidered to be due to heating by shock wave as well as mixing ofthe outer high entropy gas with inner low entropy ones.Among the 33 flux-limited non-biased samples ofREXCESS, three (Abell 2399, Abell 3771 and Abell 2328 )show very low surface brightness (LSB), indicating high en-tropy of the ICM. They all lack bright central core, apparentlyelongated and/or have sub-structures, showing that they aredynamically young (B¨oringer et al. 2007). Another template ofthis type of cluster is Abell 76, which has a very low surfacebrightness and a complicated structure (Ota et al. 2014). By theSuzaku observations, its ICM with a temperature of ∼ keV isshown to have very high entropy of ∼ keV cm in its center.These clusters are outliers in cluster scaling relations, such asICM temperature vs X-ray luminosity ( kT - L X ) relation andtheir origin is not clear yet. Thus, it is important to understandthe nature of these LSB clusters as extreme cases.Abell 548W (or Abell 548b) is one of the 3 major cluster-sized diffuse X-ray sources detected in the Abell 548 region(e.g. Davis et al. 1995). It has a redshift of z = 0 . (Solovyeva et al. 2008), with the ICM temperature of kT ∼ . keV and L X = 12 . ± . × h − erg s − cm − (or . ± . × h − erg s − cm − ) at 0.1–2.4 keV (Davis etal. 1995). Here, Hubble constant is H = 50 × h = 70 × h km s − Mpc − . Notably, the luminosity is an order of mag-nitude smaller compared to other clusters with similar temper-ature (see, e.g. kT - L X relation figure 12 of Fukazawa et al.2004). This property make Abell 548W a typical LSB cluster.There are two bright elliptical galaxies in its center, ′′ apart, and the X-ray peaks possibly associated with them. Onthe other hand, its X-ray morphology is in general circular,distinct from other LSB clusters. The cluster is known tohave high velocity dispersion of σ v = 1300 km s − (Solovyevaet al. 2008). From the well known kT - σ relation kT =( σ v . − ) . (Xue et al. 2000), this indicates kT ∼ keVwhich is apparently too high. Solovyeva et al. (2008) attributedthis inconsistency as a result of line-of-sight merger with a ve-locity shift of ∼ km s − .In the north west (NW) and north directions, ′ – ′ apartfrom the cluster center, there are two diffuse radio emission ob-served at 1.4 GHz (Feretti et al. 2006). A source to the NW hasa flux density of ± mJy at 1.4 GHz and the source to thenorth ± Jy. Both sources are polarized by ∼ % and havea steep spectra of α = − ± . Their origin is not clear, but these Also named as RXC J2157.4-0747, RXC J2129.8-5048 and RXC J2-48.1-1750, respectively. properties as well as apparent non-association with any galaxymake them good candidates as cluster radio relics (Feretti et al.2006).By analyzing the XMM-Newton (hereafter XMM) data,Solovyeva et al. (2008) reported a hot region located at r = 4 ′ – ′ from the cluster center, aside (and not within) the relic can-didate regions. In merger scenario, hot region is in many casescoincident in position with relics (e.g. Akamatsu and Kawahara2013), with only a few exceptions (e.g. Ogrean et al. 2013),which make this result rather confusing. To explain the resultsby XMM, the authors made a merger model, in which a part ofthe shocked region shows radio emission while other parts donot, and these two regions are slightly overlapping, i.e. a rathercomplicated geometry.To resolve the dynamical status of this cluster, a high-sensitivity observation out to the relic region and farther isneeded. In this paper, we revisited this issue with Suzaku(Mitsuda et al. 2007) utilizing its high sensitivity for low sur-face brightness diffuse emission. We use H = 70 km s − Mpc − in the following sections. Distance to the cluster isthus 182 Mpc, and ′ angular distance corresponds to 55.2 kpc.The solar abundances are normalized to those of Anders andGrevesse (1989). Observation parameters and obtained imageis discussed in section 2, spectral analysis in section 3, followedby discussion (section 4) and summary (section 5). Otherwisenoted, all the error-bars are shown in 90% confidence level. Using Suzaku, we observed this source from 14 to 16February 2013 in a single pointing aimed at (Ra, Dec) =(86.233 ◦ , − . ◦ ), a little offset to NW from the cluster cen-ter to locate the two relics near the center of the field of view(FOV). The XIS instrument (Koyama et al. 2007) was operatedin the full window mode with the spaced-raw charge injection(Uchiyama et al. 2009). The data was processed via revision 2.8pipe-line, and screened with nominal parameters as follows: notwithin nor right after the South Atlantic Anomaly (SAA HXD = 0 and T SAA HXD > s), apart from dark and sun-litearth (ELV > ◦ and DYE ELV > ◦ ), and not within lowgeomagnetic cutoff rigidity region (COR > GV c − ). Totaleffective exposure thus obtained was 84.4 ks.The X-ray events of the two front-illuminated (FI) CCDs(XIS0 and XIS3) are combined to obtain an image in the 0.7–7.0 keV band, as shown in figure 1. Here, the non-X-ray back-ground (NXB) image was produced using the software xisnxb-gen (Tawa et al. 2008). Vignetting effect is corrected using theflat image made by xissim (Ishisaki et al. 2007). The ICM emis-sion was apparently detected out to the relic position. Actually,from the following spectral analysis, we detect ICM X-rays outto ∼ ′ from the cluster center (see the next section). ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 NorthRadio source North West (NW)Radio source12 3 4 5
Fig. 1.
An X-ray image of Abell 548W in the 0.7–7.0 keV band obtained fromXIS0 and 3, after subtracting the NXB and corrected for vignetting and expo-sure. The 1.4 GHz radio image contours by NVSS survey in white is overlaidto show the location of diffuse radio sources. Green lines defines the spectralanalysis regions, with annular radius of ′ , ′ , ′ , ′ , ′ and further out,together with the extraction region of the 5 point sources. Green dot-dashedline represent the region ′ . around the NW relic analyzed in subsection4.4 In the image, we identified 5 contaminating point sourcesclearly visible. They are also visible in the XMM data, and wediscarded regions r < ′ from these sources. The source near-est to the center ∼ ′ offset to the west (source 1) is a hardsource presumably an AGN associated with weak radio signals.Another one ∼ ′ to the north (source 2) has similar properties.Unfortunately, there is a source located very near to the northrelic candidate (source 3) and another one near the edge of theNW one (source 4). If we include the signal from these sourcesin the following spectral analysis, both spectra get significantlyharder, so exclusion of the region around these sources are nec-essary. The 5th source located ∼ ′ to the west is also maskedout. From the XMM data, these sources are shown to have aflux from . × − (source 1) to . × − erg cm − s − (source 5) in the 2–10 keV band. After masking out these 5 sources, we defined regions with an-nular radius separated at r = 2 ′ , ′ , ′ , ′ , ′ and further outto ∼ ′ . Central region is circular, while the other regionsare made of two ◦ opening arcs, oriented to the north andNW relic candidates. The center cordinate follows that of theREFLEX catalog using ROSAT data (B¨ohringer et al. 2004). The NXB is generated and subtracted using xisnxbgen again,while the CXB is modeled using the flat arf made from xissi-marfgen (Ishisaki et al. 2007).In our spectral analysis, the new approach to handle theincreasing flickering pixel in the NXB template is applied .Because the NXB template has a longer exposure, the flicker-ing pixel detection is more sensitive and hence their numberis larger than those detected from a single observation. Themethod applies the flickering pixel lists generated from theNXB database to both the NXB template and the observationdata.At the time of this paper writing, effect of this “additionalflickering pixel” is not handled in the effective area estimationby xissimarfgen and we need to correct this effect manually.It can be approximately estimated by comparing the photoncounts before and after applying this new method with reason-able accuracy. It was as small as 0.5–3.5% for the front illumi-nated (FI) CCDs (XIS0 and 3), while was as large as 7–20% forthe back-illuminated (BI) CCD (XIS1). Thus, we simply scaledthe normalization of the FI CCD data to this difference ratio (i.e.0.5–3.5%), while letting the XIS1 normalization to be free.In Suzaku X-ray spectra, it is known that there are two fore-ground soft components in addition to the CXB: the local hotbubble (LHB) modeled by a thermal emission with a tempera-ture kT = 0 . keV, and another thermal kT = 0 . ∼ . keVcomponent called the milky way halo (MHW). To estimatethese celestial background component, we first fitted the spec-tra of the two outermost regions ( ′ < r < ′ , north andNW) simultaneously using a thermal component with kT =0 . keV (using apec code in xspec ), another thermal compo-nent with free kT and a Γ = 1 . fixed power-law. The thirdcomponent were modified with a fixed galactic absorption of N H = 0 . cm − derived using the w3nh service from NASA(Dickey & Lockman 1990). . Here we used the co-added spec-tra of XIS0 and 3 (hereafter FI spectra) in the 0.7–8.0 keV band,as well as those of the BI CCD (XIS1) in the 0.5–6.0 keV band.The fit became acceptable with χ / dof = 167 . / , althoughthe residual spectra showed clear trend for softer component.When the photon index was left free, it became Γ = . × − ergcm − s − str − , which is 28.4% higher than the value obtainedin the Suzaku Lockman-hole observation (ID 101002010).The CXB intensity fluctuates field to field and its levelcan be approximated by assuming a certain source distribu-tion. Following the method used in Nakazawa et al. (2009),which itself is based on Kushino et al. (2002), we estimated the90% confidence fluctuation level for this region. Assuming the NASA W3NH service: https://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl
Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
Table 1.
Fitted results to the foreground components. components temperature (keV) normalizationLHB 0.08 (fixed) . +1 . − . MWH . +0 . − . . +0 . − . normalization in apec model, scaled to π (20) arcmin flat region. North 2’ < r < 4’
NW 2’ < r < 4’Center r < 2’ North 4’ < r < 7’NW 4’ < r < 7’ −5 −4 −3 c oun t s s − k e V − −4−2024 χ −5 −4 −3 c oun t s s − k e V − −4−2024 χ −5 −4 −3 c oun t s s − k e V − −4−2024 χ Center r < 2’2kT1kT
Fig. 2.
NXB subtracted XIS spectra fitted with the 1kT model of the inner5 regions. Black crosses are from the summed XIS0 and 3 (or FI) spectraand red ones are from XIS1. Spectral model consists of the two foregroundcomponents (LHB and MWH), the CXB, and the ICM emission. For clarity,the ICM component in the FI models are shown with solid lines, while all theother model components are in dotted lines. Top right panel stands for the2kT model results for the center spectra. See text for details. source cut flux of S c ∼ × − erg cm − s − at 2–10 keV,combined with the region area of 0.0158 degree , we get a num-ber of 15.5%, which cannot explain the observed difference.Combined with the soft spectra, we thus conclude that even inthese outermost region, the ICM component from Abell 548Wis detected.Because the ICM emission is contaminating out to the out-ermost region, we need to estimate appropriate CXB level byother information. Analysis of another Suzaku observation 6.6degree south to Abell 548W (ID 405059010), after excludingthe central soft source, showed a CXB normalization within0.6% to those of the Lockman-hole observation. We hence uti-lized this value as a base-line, i.e. . × − erg cm − s − str − , and handled the fluctuation separately. This gave the pa-rameters of the two foreground components (LHB and MWH)as listed in table 1. For all the inner regions, we used this valueas a fixed foreground component.With the CXB and the two foreground components derived, North 10’ < r < 13’NW 7’ < r < 10’ North 7’ < r < 10’NW 10’ < r < 13’
North 13’ < r < 16’NW 13’ < r < 16’ −5 −4 −3 c oun t s s − k e V − −4−2024 χ −5 −4 −3 c oun t s s − k e V − −4−2024 χ −5 −4 −3 c oun t s s − k e V − −4−2024 χ Fig. 3.
The same as figure 2, but for the outer 6 regions. we fitted all spectra with a single temperature thermal emissionusing the apec code (hereafter 1kT model). Systematic uncer-tainties of the NXB are assumed to be 2.1% for the FI spectraand 4.9% for the XIS1, following table 7 of Tawa et al. (2008),which stands for “5–12 keV NXB reproducibility for 50 ks ex-posure bins”, modified to 90% confidence limit. Note that effectof NXB reproducibility is minor, as shown in the following re-sults. Fluctuation of the CXB is calculated in the same way asdescribed above. All spectra are shown in figure 2 and 3 andfitted results are listed in table 2. In the fitting, redshift is fixedat z = 0 . optically estimated by Solovyeva et al. (2008).Abundance ( A ) is also fixed at 0.3 of the solar value, a typicalvalue for a cluster. In our Suzaku data, spectra of the central r < ′ shows clear Fe-K line, while in the other regions it is notclear. If we set z and A free in the former fitting, they are de-rived as z = 0 . +0 . − . and A = 0 . +0 . − . , respectively, bothof which are consistent with the assumed value.Among the 11 fit results, the central region showed rel-atively large χ resulting in low null-hypothesis probability(NHP) of 0.4%, with clear concave residual suggesting itsmulti-temperature nature. When fitted with two temperaturethermal emission (hereafter 2kT model), the fit significantlyimproved with the f -statistics probability of 0.06%. Fit re-sults are shown in the top right panel of figure 2 and table 3.The hotter component dominates the spectra and its parametersare not much different from the 1kT fit results (e.g. only 0.42keV higher in temperature and 8% smaller in normalization),while the cooler component is minor (only 6% of the hot com-ponent normalization) and will be naturally explained as the ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 inter-stellar medium (ISM) of the central galaxies. Hereafter,we interpret the hotter component as the ICM.While the fits to other 8 spectra are acceptable in 90% confi-dence level, the remaining 2 spectra, NW ′ < r < ′ and north ′ < r < ′ , showed marginally poor fit with NHP of 6.3% and9.9%, respectively. Since the former one has very low statistics(due to masking out source 1), we could not go into its detail.The latter is almost acceptable in 90% confidence and thus westick to the 1kT fit for a moment. Later in subsection 4.2 wewill revisit the spectra. The temperature profile as a function of the distance from thecluster center is shown in figure 4. Average cluster temper-ature is ∼ . keV, in good agreement with the XMM result(Solovyeva et al. 2008). Even though the surface brightness ofthe source is low, thanks to the Suzaku low background, theICM temperature was determined out to r < ′ with muchbetter accuracy compared to XMM. For example, in the region ′ < r < ′ around the relic candidates, we obtained 90% con-fidence statistical error of ± . keV and ± . keV for theNW and north regions, respectively, while the XMM results in ′ < r < ′ . of north and NW co-added spectra has an errorof ± . keV (converted into the 90% confidence, from figure 8of Solovyeva et al. 2008). Note that all errors in their paper isshown in 68% confidence, while in this paper it is 90%.The ICM temperature is in first approximation flat, withsome symptom of fluctuation and marginal tendency for get-ting lower to the outer radius, which is seen in many clusters(e.g. Pratt et al. 2007). The ∼ keV hot region to the north at ′ < r < ′ reported by Solovyeva et al. (2008) was not detectedin Suzaku spectra, although we do see milder jump, as discussedin the next subsection. We quickly checked the XMM data andfound there is a local apparently hot region around source 2,which is almost excluded in our Suzaku analysis. Since theNXB of the XMM data is already a bit high in this region, wedid not go into farther detail on the XMM data and focus on ourSuzaku data in this paper. Here we focus on the temperature structure around the two reliccandidates. Since majority of the north relic candidate region ismasked out by source 3, here we focus on the NW arc.Although there is no evidence for ∼ keV hot region ataround ′ < r < ′ , there exists a temperature jump at r ∼ ′ ,from 3.60 keV at ′ < r < ′ to 2.52 keV at ′ < r < ′ an-nulus (see figure 4), which is qualitatively consistent with the (cid:38)(cid:68)(cid:81)(cid:71)(cid:76)(cid:71)(cid:68)(cid:87)(cid:72)(cid:53)(cid:72)(cid:79)(cid:76)(cid:70)(cid:3)(cid:85)(cid:72)(cid:74)(cid:76)(cid:82)(cid:81) k T ( k e V ) r (arcmin) Fig. 4.
Temperature profile of the ICM towards the NW (red) and the north(green) relic candidates. Error bars shown in cross are statistical 90% con-fidence limit, while the thin dashed lines stand for the quadrature sum ofboth the CXB and NXB fluctuations. Note that plot of north regions (green)are artificially shifted by +0 ′ . for clarity. Gray dot-dashed line from ′ – ′ stands for the location of the candidate relics. Thick black dashed lines isthe typical temperature profile given in Pratt et al. (2007) based on the XMMdata, plotted by assuming an average temperature of kT = 3 . keV. XMM result. Temperature ratio of the outer region comparedto the inner one is calculated to be . ± . . Here the erroris at statistical 90% confidence level, i.e. 1.65 σ . If we includethe CXB fluctuation effect, the error rises to 0.35. When cor-rected for the “average ICM temperature gradient” (e.g. Pratt etal. 2007), it become . ± . . Thus, the temperature rise issignificant right at . σ (or at one-side 5% confidence level)judging from the 1kT fit.Looking at figure 4, however, the temperature profile (shownin red) can be interpreted as a “dip” at ′ < r < ′ annulus.Thus, it is natural to conclude there is some “cooler” gas atthis region, rather than assuming hotter gas in the inner. The1kT fit to the NW ′ < r < ′ spectra actually shows smallpositive residual at around 0.8–1.0 keV and negative around1.5 keV. With the 2kT model, as shown in table 3, we have kT cool = 0 . +0 . − . keV and kT hot = 2 . +0 . − . keV, respec-tively (all errors are shown in quadrature sum). The χ / dofimproved to 89.5/98 from 105.0/100 of the 1kT fit, and f -test shows NHP of 0.06%. Thus, the spectra can be well ex-plained by a combination of minor ( ∼ % in its normalization) kT cool ∼ . keV and major kT hot ∼ . keV components. Inother words, there is no temperature jump if we think the hotcomponent is the main ICM.Spectra from other regions including relic candidates (bothNW and north, at annuli of ′ < r < ′ and ′ < r < ′ ) showssimilar residual but with less significance. Improvement of fitwith the 2kT model in view of f -statistics is only about 1–7% inNHP, and what is more the 1kT fit itself is acceptable in 90% Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
Table 2.
Temperature and normalization of the spectra from the 11 regions with the 1kT model fit.
CenterRegion kT keV norm χ /dof r < ′ . ± . ± . ± .
02 6 . ± . ± . ± . × − kT keV norm χ /dof ′ < r < ′ . +1 . − .
86 +0 . − . ± .
02 3 . +0 . − . ± . ± . × − ′ < r < ′ . +0 . − .
41 +0 . − . ± .
05 1 . ± . ± . ± . × − ′ < r < ′ . +0 . − .
26 +0 . − . ± .
04 8 . ± . ± . ± . × − ′ < r < ′ . +1 . − . . − . . − . . ± . ± . ± . × − ′ < r < ′ . +0 . − . . − . . − . . ± . +0 . − . . − . × − NorthRegion kT keV norm χ /dof ′ < r < ′ . +0 . − . . − . ± .
02 4 . +0 . − . . − . ± . × − ′ < r < ′ . +0 . − . . − . ± .
05 1 . +0 . − . . − . ± . × − ′ < r < ′ . +0 . − . . − . ± .
07 0 . ± . +0 . − . ± . × − ′ < r < ′ . +0 . − . . − . . − . . ± . +0 . − . ± . × − ′ < r < ′ . +1 . − . . − . . − . . ± . ± . +0 . − . × − N/A : kT errors are shown in 90% confidence level, with an order of statistical, CXB fluctuation and NXB fluctuation origins. : normalization in apec model, scaled to π × arcmin flat region. : Fit to the outermost region is combined one to the NW and north. In addition, the foreground components were set free. See text for detail. Table 3.
Temperature and normalization of the spectra from the selected 3 regions fitted with the 2kT model.
Region kT coool and kT hot keV norm cool and norm hot χ /dofCenter r < ′ . +0 . − . ± . ± .
00 0 . +0 . − . ± . ± . × − . +0 . − . . − . ± .
02 5 . ± . ± . ± . × − NW ′ < r < ′ . +0 . − . . − . ± .
00 0 . ± . ± . ± . × − . +0 . − . . − . ± .
07 0 . ± . ± . ± . × − North ′ < r < ′ . +1 . − . . − . ± .
00 0 . +0 . − . . − . ± . × − . +2 . − . . − . ± .
10 1 . +0 . − . . − . ± . × − confidence level. So the cool component will be existing allaround the candidate NW relic region, but with only marginalevidence with the Suzaku data.The kT ∼ . keV cool component is also suggested in thenorth ′ < r < ′ spectra. As already mentioned, the 1kT modelfit to the data gave a marginal NHP of 9.9% and the residualspectra has a soft excess. With the 2kT model, the fitting im-proved as shown in table 3, and f -test shows significant NHP of0.04%. The normalization of the cool component here is ∼ %of the hotter one. Thus, the minor cool component is also sug-gested to be mixed in the ICM at the region between the northrelic candidate and the cluster center.Another temperature jump candidate is at the outer rim ofthe NW relic candidate at r ∼ ′ , from 3.3 keV at ′ < r < ′ to 1.8 keV at ′ < r < ′ annulus. With only statisticalerror, the ratio is . ± . , and with the CXB fluctuation itbecomes . ± . . Again corrected for the “ICM temperaturegradient”, it becomes . ± . . This is ∼ . σ , meaning thatthe possibility the temperature is “higher” in the inner annulus is 89%. This result marginally prefers the temperature jumpat the outer rim of the NW relic candidate, but not significantenough to conclude on it. Note that we see no symptom ofsimilar temperature jump at the north regions. With the ICM temperature and density obtained, we then cal-culated the astrophysical entropy profile, given as K = kT ( n e ) / (e.g. Ponman et al. 1999). Here, n e is the electron density.Since our Suzaku observation only covers the northern portionof the cluster, here we assume the β model gas distribution witha core radius r c = 2 ′ . and β = 0 . , provided by Neumann& Arnaud (1999) obtained from the ROSAT PSPC analysis. Asshown in figure 5, the fitted apec normalization profile matcheswell with the β model with residuals less than ∼ %. By scal-ing the β -model to the apec normalization profile, the central n e was derived as n e ( r = 0) = 1 . × − cm − .The obtained entropy profile is shown in figure 6. Error bars ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 (cid:38)(cid:68)(cid:81)(cid:71)(cid:76)(cid:71)(cid:68)(cid:87)(cid:72)(cid:53)(cid:72)(cid:79)(cid:76)(cid:70)(cid:3)(cid:85)(cid:72)(cid:74)(cid:76)(cid:82)(cid:81) r arcmin no r m Fig. 5.
Normalization of the spectral fitting towards the NW (red) and thenorth (green) relic candidates. Error bars are in the same format as in figure4. Black dash line is the scaled β model. See text for detail. r kpc K = k T n k e V c m - / -
100 1000 10000 10 100 1000 (cid:38)(cid:68)(cid:81)(cid:71)(cid:76)(cid:71)(cid:68)(cid:87)(cid:72)(cid:53)(cid:72)(cid:79)(cid:76)(cid:70)(cid:3)(cid:85)(cid:72)(cid:74)(cid:76)(cid:82)(cid:81)
Fig. 6.
Entropy profile estimated by assuming the β model distribution of theICM. Profile to the NW (red) and north (green) are shown. Error bars are inthe same format as in figure 4. For comparison, average profile of the 109high central entropy ( K > keV cm ) clusters by Cavagnolo et al. (2009)is shown in solid black line. The profile of so called gravitational accretionheating model from Voit et al. (2002) is also shown (dashed black). include both of them from the n e and kT estimations. Centralentropy of K = 400 keV cm − at around 50 kpc from the clus-ter center is one of the highest among the clusters with thistemperature. Actually, there is only one object with entropyhigher than this in the T cluster < keV panel of the entropy pro-files (figure 5) of Cavagnolo et al. (2009), compiled from theChandra data of 239 clusters with various temperature. Becausethe center spectra are fitted with 2kT model and we only em-ployed the hotter one as the ICM component, the central en-tropy would be a little overestimated. However, when applyingthe 1kT fit results and perform the same calculation, we get acentral entropy of K = 320 keV cm − , which is still high.In the profile, apparent “dip” in the NW direction (red lines)at around r ∼ kpc and candidate jump at ∼ kpc both reflects the temperature structure discussed in the last subsec-tion. Over all, the entropy is high and flat, with no significantstructure. This is consistent with the X-ray image being rela-tively circular, as well as its general lack of strong temperaturestructure. The diffuse radio sources are presumably synchrotron emissionby GeV electrons interacting with ∼ µ G magnetic field in theICM. The same electrons scatter the Cosmic microwave back-ground up to the X-ray energy band, i.e. so called inverse-Compton (IC) emission. Since the X-ray spectra around theradio sources are well modeled with thermal emission, here weestimate the upper limit on the emission.Again, we focus on the NW relic and select a region witha radius of ′ . around it. Region ′ around source 4 is alsomasked out. When fitted with the 1kT model, we obtain anacceptable result with kT ∼ . +0 . − . . − . ± . keV and χ / dof = 142 . / . However, as already suggested in theannulus spectra, the residual around 0.8–1 keV exists, and2kT model gives significantly better fit with kT cool = 0 . ± . ± . ± . keV, kT hot = 3 . +0 . − . . − . . − . keV and χ /dof = 118 . / . Spectra of the 1kT and 2kT model fit isshown in figure 7.The IC component will have a power-law like spectra. Herewe assume its photon index to be Γ = 2 . (fixed), i.e. flat in νF ν plot, for simplicity. Although the value observed in 1.4 GHzradio (Feretti et al. 2006) is a bit softer, they are still consis-tent within the error. Assuming 1 µ G magnetic field, electronsscattering 8 keV X-rays corresponds to those emitting 38 MHzradio, well below the observed 1.4 GHz band. Thus, assuminga little harder spectral index there is natural. Unfortunately, thehotter component of the 2kT fit has very similar shape to the
Γ = 2 . power-law. Actually, if we replace it by a power-lawwith Γ free, it was derived as Γ = 2 . +0 . − . (errors are mostlyfrom the CXB fluctuation). Nonetheless, we fitted the spectrawith a fixed Γ ( = 2 . ) power-law in addition to the 2kT modeland estimated the upper-limit flux of the former component as . × − erg s − cm − at 2–10 keV.By integrating the power-law energy distribution of elec-trons in Lorentz factor of < γ < × , and assumingthe relic has a spherical shape with a radius of 190 kpc ( ′ . ),the electron energy density becomes U e < . eV cm − . Thisis not well constrained, compared to the thermal energy den-sity of ∼ . eV cm − (calculated assuming n e ∼ × − cm − and electron-ion number ratio of 1.2). Combining the1.4 GHz radio flux ( ± Jy, Feretti et al. 2006) and the hardX-ray flux, we obtain the lower limit magnetic field strengthof > . µ G, which is consistent with the equipartition field of0.9 µ G (Solovyeva et al. 2008).
Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 −5 −4 −3 χ −4−2024 Energy (keV) c oun t s s − k e V − χ (cid:16084)(cid:16143)(cid:16085)(cid:16076)(cid:16093)(cid:16076)(cid:16151)(cid:16128)(cid:16084)(cid:16142)(cid:16085)(cid:16076)(cid:16094)(cid:16076)(cid:16151)(cid:16128)(cid:16084)(cid:16141)(cid:16085)(cid:16076)(cid:16094)(cid:16076)(cid:16151)(cid:16128) Fig. 7. (a) Spectra obtained around the NW relic candidate region, fitted withthe 2kT model, and (b) its residual. Panel (c) is the residual with the 1kTmodel.
We analyzed the Suzaku deep (84.4 ks) observation data ofAbell 548W and measured the ICM properties out to r = 16 ′ (or 880 kpc) from its center, well beyond the two relic candi-dates. When estimated from the “average” ICM temperatureof 3.6 keV, r and r become 1.3 Mpc and 860 kpc, re-spectively. Here we used the data provided in Arnaud et al.(2005) and estimate r = 704 × p kT / keV kpc and r =452 × p kT / keV kpc. Thus, our observation range reaches 2/3of r and slightly exceeds r derived under hydrostatic as-sumption. Out to this radius, the ICM morphology, temperatureand entropy do not show strong structure, with marginal evi-dence for small temperature variation.The ICM temperature is ∼ keV at its center, and ∼ keVat the outermost regions along the two relic candidates. Wealso observe a temperature “dip” around the NW relic candi-date, which is understood as a ∼ keV cold gas mixed withthe ∼ keV ICM emission. Outer rim of the NW relic showsmarginally higher temperature than those of the ICM outside,consistent with the relic candidate being located at the shockedge. Its significance is marginal, i.e. only 89% confidence,not strong enough to conclude its existence. Astrophysical en-tropy calculated from the ICM density and temperature reaches400 keV cm − at around 50 kpc from the cluster center, whichis among the highest of clusters. This value directly reflects thelow surface brightness nature of the cluster. k T e q = . k e V r ( k T = . k e V ) r ( k T = . k e V ) ’ ( ob s e r v ed ou t e r m o s t ) k T e q = . k e V r ( k T = . k e V ) r ( k T = . k e V ) r (kpc) G a s m a ss f r a c t i on Fig. 8.
Ratio of gas mass to total mass calculated out r , estimated byusing the ICM temperature. Solid line indicates that derived from the totalmass assuming hydrostatic equilibrium with kT = 3 . keV. Dotted line in-dicates those obtained by assuming the hydrostatic equilibrium temperatureto be 2.1 keV, i.e. heated by a factor of 1.7. For reference, r is alsoshown in each plot. Vertical dot dashed line stands for r = 880 kpc, whichcorresponds to the outer bounds r = 16 ′ of our data analysis. Because the X-ray image is relatively circular, we here assumethat the cluster itself is spherically symmetric for simplicity.The ICM distribution is well modeled with the β model, and gasmass integrated within a radius r can be estimated. Assuminghydrostatic equilibrium, we can also calculate the total mass ofthe cluster. We then derived the “gas-mass fraction” ( f gas ) us-ing the ratio of these two, calculated out to r . As shownin figure 8, f gas is derived as . % at 1.3 Mpc ( ∼ r for kT = 3 . keV). Gas mass integrated out to r is derived as M gas ( r ) ∼ . × M ⊙ , while the hydro-statically esti-mated total mass is M tot ( r ) ∼ × M ⊙ . The derived f gas is less than a half of the value generally reported in otherclusters. For example, Walker et al. (2013) derived f gas ∼ . at r in Centaurus cluster, which has a temperature of ∼ keV,similar to that of Abell 548W. Allen et al. (2008) also showed f gas ∼ . in the analysis of 42 clusters with temperatureshigher than 5 keV.To compare f gas with many other clusters, we also derivedthe value at r . Then we have M gas ( r ) ∼ . × M ⊙ and M tot ( r ) ∼ × M ⊙ , resulting in f gas = 0 . .According to Pratt et al. (2007), who analyzed 31 nearby clus-ters, the averaged gas mass fraction ¯ f gas ( r ) is . for a clus-ter with M tot ( r ) = 19 × M ⊙ (see figure 8 of the paper).Thus, also at r , Abell 548W shows slightly more than a halfof f gas of that of the ordinary cluster. Note that there are a fewclusters at around M tot ( r ) ∼ × M ⊙ which showed f gas ( r ) as small as 0.05 in their plot. Because these plots ac-tually include the three LSB clusters presented at the beginning(Abell 2399, Abell 3771 and Abell 2328), it means that Abell548W has a typical f gas ( r ) of LSB clusters. While the mor- ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 phologies of the three LSB clusters are disturbed, that of Abell548W is rather circular, which makes this cluster peculiar.In this scenario we assumed that the ICM of this cluster is inhydrostatic equilibrium and simply its f gas is small. As alreadynoted in, e.g. Ota et al. (2014), the ICM cannot be radiativelycooled down to the “ordinary” entropy within years, andthus this cluster remains LSB for a long period.The fact that the XMM image has ∼ X-ray peaks possiblyassociated with the 2 elliptical galaxies in its center, suggeststhat the cluster is dynamically young. In addition, as alreadynoted, its galaxy velocity dispersion σ V = 1300 km s − is toohigh as a kT ∼ . keV cluster. Based on the redshift distri-bution of 193 galaxies in the Abell 548W region, Solovyeva etal. (2008) interpreted it as a mixture of 2 clusters each with σ V = 700 and km s − , merging with a relative velocity of ∼ km s − .We then consider a scenario that the ICM temperature isheated up by cluster merger. In this case, hydrostatic equilib-rium is not taking place, and the relic candidates can be inter-preted as the shock front propagating outward. Let’s here as-sume that the ICM is heated up by a factor of 1.7, i.e. it willsettle down to 2.1 keV after final relaxation (dynamically, notby cooling). In this case, total mass derived in the last para-graph is overestimated by the same factor. As shown in figure8, real r becomes as small as 650 kpc with M tot ( r ) ∼ . × M ⊙ . Then, f gas ( r ) reaches 0.08, which is thetypical value shown in figure 8 of Pratt et al. (2007). In themerger scenario, the cool component seen in a few regions canbe understood as remnants of the pre-shock low entropy gas. The simplest toy-model is a line-of-sight, 1:1 major merger,with the pre-shock temperature of ∼ . keV. Pre-shock soundvelocity becomes v s ∼ km s − and with a colliding veloc-ity of 1500 km s − , it can generate a shock with Mach ∼ ,and hence post-shock temperature of 3.6 keV from Rankine-Hugoniot conditions. Specifically, defined inward to the cen-ter of gravity of the system, pre-shock bulk velocity is as-sumed to be ( = 1500 / km s − , that of post-shock0 km s − , and shock plane velocity outward km s − .With these parameters, the Mach number becomes (750 +600 km s − ) /
680 km s − = 2 . and thus the temperature ra-tio 2.1.Here we consider that this cluster (or two groups) is (are)right at the middle of initial heating phase (see, e.g. figure 5of Ricker and Sarazin 2001). In the following adiabatic ex-pansion, the X-ray luminosity will follow L X ∝ n e n i T . V =( n i V ) n e T . ∝ M gas T . Here, n i is ion number density, M gas is total gas mass, and entropy conservation of K = T / ( n e ) / isapplied. With the cluster total mass to be doubled after merger, its future relaxation temperature will be ∼ M – T relation ( M ∝ T . by Vikhlinin et al. 2006).As the merger moves to later stages, the decreasing tempera-ture will cause the luminosity to get dimmer as ∝ T . Since L X ( r ) ∝ T ( r ) . in Pratt et al. (2007), luminosity deficitwill be slightly relaxed as the cluster settles down. In addition,larger scatter in L X at lower temperature make the peculiarityof this object further relaxed. In other words, this cluster infuture will look like one of many low L X groups of galaxies sometimes seen in the kT – L X plot.Although the X-ray properties of Abell 548W could be un-derstood if it is a major merger of (relatively large) galaxygroups, the merging velocity of km s − itself is ratherhigh, and we need to consider its origin in our future work.What is more, general lack of strong inhomogeneity in both theX-ray morphology and temperature structure requires a finelytuned merger model, e.g. merger axis perfectly aligned to theline of sight, and so on. Other exotic possibilities, such as over-heating by AGN feedback and inherent baryon fraction deficit,still cannot be ruled out with current observational results. Suzaku deep (84.4 ks) observation of Abell 548W detected theICM emission out to r = 16 ′ (or 880 kpc) from its center, wellbeyond the two relic candidates, and measured the ICM temper-ature for the first time out to this radius. The ICM morphology,temperature and entropy do not show strong structure, whilemarginal evidence for small temperature variation is observed.The hot ( ∼ keV) component detected with XMM (Solovyevaet al. 2008) was not confirmed, although the contaminatingpoint source (source 1) makes it difficult for Suzaku to clearlydistinguish the inconsistency. Central entropy of the ICM isamong the highest in a cluster with this temperature, as well.At the NW candidate relic region, symptom of relativelycool ( ∼ keV) component mixed with the ∼ keV ICM emis-sion is detected. In addition, marginal temperature jump at theNW relic rim is suggested. If this is the case, the radio sourcesare consistent with being relics activated with merger shock.When assuming hydro-static equilibrium, the gas-mass frac-tion ( f gas ) of the cluster is estimated to be 0.067 at r and0.052 at r , which both are about a half of the value gener-ally seen. Considering these observational properties, a mergingcluster scenario of two relatively large ( kT ∼ . keV) galaxygroups is discussed. Although these parameters can explain thehigh entropy nature of the cluster, finely tuned model to addressboth the high entropy and featureless and apparently circular X-ray properties at the same time will be needed. In other words,if this cluster is a major merger, the merging axis shall be almostcompletely parallel to the line-of-sight. Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
Acknowledgement
KN, MT and MF are supported in part by JSPS KAKENHIGrant Number 15H03639, 26400218 and 15K05080, respec-tively.