PASJ:Publ.Astron.Soc.Japan,1–??, — (cid:13)c 2008.AstronomicalSocietyofJapan. X-Ray Study of Temperature and Abundance Profiles of the Cluster of Galaxies Abell 1060 with Suzaku Kosuke Sato,1 Noriko Y. Yamasaki,2 Manabu Ishida,2 Yoshitaka Ishisaki,1 Takaya Ohashi,1 Hajime Kawahara,3 Takao Kitaguchi,3 Madoka Kawaharada,3 Motohide Kokubun,3 Kazuo Makishima,3 Naomi Ota,4 Kazuhiro Nakazawa,2 Takayuki Tamura,2 Kyoko Matsushita,5 Naomi Kawano,6 Yasushi Fukazawa,6 and John P. Hughes7 1 Department of Physics, Tokyo Metropolitan University, 1-1 Minami-Osawa, Hachioji, Tokyo 192-0397 7 [email protected] 0 2 Institute of Space and Astronautical Science (ISAS), Japan Aerospace Exploration Agency, 0 3-1-1 Yoshinodai, Sagamihara, Kanagawa 229-8510 2 3 Department of Physics, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033 n 4 Cosmic Radiation Laboratory, The Institute of Physical and Chemical Research (RIKEN), a 2-1 Hirosawa, Wako, Saitama 351-0198 J 5 Department of Physics, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku, Tokyo 162-8601 1 6 Department of Physical Science, School of Science, Hiroshima University, 1 1-3-1 Kagamiyama, Higashi-Hiroshima, Hiroshima 739-8526 7 Department of Physics and Astronomy, Rutgers University, Piscataway, NJ 08854-8019, USA 1 v 8 (Received —;accepted —) 2 Abstract 3 1 Wecarriedoutobservationsofthecentraland20′eastoffsetregionsoftheclusterofgalaxiesAbell1060 0 with Suzaku. Spatially resolved X-ray spectral analysis has revealed temperature and abundance profiles 7 of Abell 1060 out to 27′≃380h−1 kpc, which corresponded to ∼0.25r . Temperature decrease of the 0 70 180 intra cluster medium from 3.4 keV at the center to 2.2 keV in the outskirt region are clearly observed. / h Abundances of Si, S and Fe also decrease by more than 50%from the center to the outer, while Mg shows p fairly constant abundance distribution at ∼0.7 solar within r<17′. O shows lower abundance of ∼0.3 ∼ - o solar in the central region (r <∼6′), and indicates a similar feature with Mg, however it is sensitive to r the estimated contribution of the Galactic components of kT ∼0.15 keV and kT ∼0.7 keV in the outer st annuli(r>13′). Systematiceffectsduetothepointspreadfun1ctiontails,contamin2ationontheXISfilters, ∼ a instrumental background, cosmic and/or Galactic X-ray background, and the assumed solar abundance : v tables arecarefullyexamined. Results ontemperatureandabundancesofSi, S,andFe areconsistentwith i those derived by XMM-Newton at r <13′. Formation and metal enrichment process of the cluster are X ∼ discussed based on the present results. r a Key words: galaxies: clusters: individual (Abell 1060) — X-rays: galaxies — X-rays: ISM 1. Introduction withinitialmassabove∼10M . Themetalsproducedin ⊙ the galaxies are transferedinto the ICM by galactic wind Clusters of galaxies, being the largest virialized system and/or ram pressure stripping. in the universe, are filled with the intracluster medium ASCA firstly revealed the distribution of Si and Fe (ICM), which consists of X-ray emitting hot plasma with in the ICM (Fukazawa et al. 1998; Fukazawa et al. typical temperatures of a few times 107 K. X-ray spec- 2000; Finoguenov et al. 2000; Finoguenov et al. 2001). troscopy of the ICM can immediately determine its tem- The derived iron-mass-to-light ratios (IMLR; Ciotti et perature and metal abundances. The metal abundances al. 1991; Renzini et al. 1993; Renzini 1997) are nearly ofICMhavealotofinformationtounderstandthechem- constant in rich clusters and decrease toward poorer sys- ical history and evolution of clusters. A large amount tems (Makishima et al. 2001). Recent observations with of metals of the ICM are mainly produced by supernovae Chandra and XMM-Newton allowed detailed studies of (SN)inearly-typegalaxies(Arnaudetal.1992;Renziniet the metals in the ICM. These observations, however, al. 1993), which are classified roughly as Type Ia (SN Ia) showed abundance profiles of O, Mg, Si and Fe only for andType II (SN II). Si, S andFe aresynthesizedin both the central regions of very bright clusters or groups of SNIaandSNII,whileαelementssuchasO,Ne, andMg galaxies dominated by cD galaxies in a reliable manner aremainly inSNII, whichareexplosionsofmassivestars (Finoguenovetal.2002;Fukazawaetal.2004;Matsushita 2 K. Sato et al. [Vol. , et al. 2003; Tamura et al. 2003). The abundance pro- Table 2. Estimated column density of the contaminant for files of O and Mg, in particular for the cluster outer re- eachsensoratthecenter ofCCDinunitof1018 cm−2. gions, are still poorly determined, because these satel- lites are characterized by relatively high intrinsic back- XIS0 XIS1 XIS2 XIS3 ground levels. Tamura et al. (2004) derived IMLR for Carbon ......... 1.37 1.83 2.52 4.04 five clusters within 250 h−1010 kpc to be ∼0.01 M⊙/L⊙, Oxygen ......... 0.228 0.305 0.420 0.673 and the oxygen mass within 50 h−1 kpc for several clus- 100 ters. However, oxygen-mass-to-light ratios (OMLR) for richclustersarenotreliableduetotheloweremissivityof illuminated sensors (FI = XIS0, XIS2, XIS3). The BI OVII and OVIII lines in higher temperatures. De Grandi and FI sensors have different advantages. The former &Molendi(2001);Hayakawaetal.(2006)foundthatclus- has higher quantum efficiency in the soft energy band ters associated with cD galaxies and central cool compo- (E <∼ 1 keV), while the latter shows lower instrumental nents showedabundanceconcentrationinthe clustercen- non X-ray background (NXB). The XIS was operated in ter,whileclusterswithoutcDgalaxiesshowedflatterpro- the Normal clocking mode (no window nor burst option, files. Thecentralmetallicityenhancementinthecoolcore so8sexposureperframe),withthestandard5×5or3×3 clusters were further studied and the excess metals were editing mode (Koyama et al. 2006). shown to be supplied from the cD galaxies (De Grandi et Theopticalblockingfilters(OBF)oftheXIShavebeen al. 2004). graduallycontaminatedintimebyout-gasfromthesatel- Abell 1060 (hereafter A 1060) is a nearby cluster of lite, and the degradation of the low energy transmission galaxies(z=0.0114)characterizedby a smoothandsym- was already significant in November 2005. The thickness metricdistributionofintraclustermedium(ICM),andhas of the contaminant is different among sensors,and is also nocD-galaxyatthecenter. Chandraobservationdetected dependent on the location on the CCD. The estimated very compact X-ray emissions from the central two ellip- column density (C/O=6 in number ratio is assumed) at tical galaxies (Yamasaki et al. 2002), and the tempera- the center of the CCD is listed in table 2, and the cal- tureandabundancedistributionswereshowntobe some- culated X-ray transmissionfor BI (XIS1) at each annular what inhomogeneous at the cluster center (Hayakawa et regionis plottedinfigure2.1 Thoughthe thicknessofthe al. 2004). Hayakawaet al. (2006) detected a temperature OBF contaminantis different on each annulus, this effect drop by ∼30% from the central region to r ∼13′ with is consideredin the calculation of the Ancillary Response XMM-Newton,whileithadpreviouslybeenconsideredas File (ARF) by the “xissimarfgen”Ftools task (Ishisaki et flat on the basis of the ASCA and ROSAT observations al.2007). Theenergyresolutionwasalsodegradedslightly (Tamura et al. 1996,Furusho et al. 2001). (FWHM∼150eVat5.9keV)afterthelaunch,duetothe This paper reports results from Suzaku observations radiation damage of the CCD. of A 1060. Owing to the low-background nature of the 2.2. Data Reduction Suzaku XIS, we are able to measure the temperature and abundanceprofilestoamuchouterregionthantheprevi- We used the version 0.7 processing data (Mitsuda et ousXMM-Newtonstudy. WeuseH =70kms−1 Mpc−1, al. 2007), and the analysis was performed with HEAsoft 0 Ω = 1−Ω = 0.73 in this paper. At a redshift of version 6.0.6 and XSPEC 11.3.2t. We started the event Λ M z =0.0114, 1′ corresponds to 14 kpc, and the virial ra- screening from the cleaned event file, in which selection dius, r =1.95h−1 khTi/10 keV Mpc (Markevitch et ofthe eventgradeandbadCCDcolumn,disposalofnon- 180 100 al. 1998), is 1.53 Mppc for the average temperature of observational intervals (during maneuver, data-rate low, hTi = 3 keV (Hayakawa et al. 2006). Throughout this South Atlantic Anomaly, Earth occultation, bright Earth paper we adopt the Galactic hydrogen column density of rim to avoid scattered solar X-ray), and removal of hot N =4.9×1020cm−2 (Dickey&Lockman1990)inthedi- andflickeringpixelsbythe“cleansis”Ftools,werealready H rection of A 1060. Otherwise noted, the solar abundance conducted. The exposure time given in table 1 is for the tableisgivenbyAnders&Grevesse(1989),anderrorsare cleaned event file. We further applied the Good-Time 90% confidence region for a single interesting parameter. Intervals (GTI) given for excluding the telemetry satura- tion by the XIS team. The light curve of each sensor in 2. Suzaku Observation and Data Reduction the0.3–10keVrangewith16stimebinwasalsoexamined to reject periods of anomalous event rate greater or less 2.1. Observation than ±3σ around the mean. After the above screenings, SuzakucarriedouttwopointingobservationsforA1060 remainingexposureofthe centralobservationwas40.2ks in November 2005, the central region and 20′ east offset and that of the offset observation was 52.9 or 54.6 ks (FI region, with exposure of 40.5 and 55.5 ks, respectively. orBI). These exposuresarenotsodifferentfromthose in The observation log is summarized in table 1, and the table 1, which represent that the NXB was almost stable combined X-ray Imaging Spectrometers (XIS; Koyama during the both observations. The event screening with et al. 2006) image in the 0.5–7 keV range is shown in the cut-offrigidity(COR)wasnotperformedinourdata. figure 1. We utilize only the XIS data in this paper. 1 Thecalibrationdatabasefileofae xiN contami 20060525.fits The XIS is an X-ray CCD camera, which consists of was used for the estimation of the XIS contamination (N = one back-illuminatedsensor(BI = XIS1) andthree front- 0,1,2,3correspondingtotheXISsensor). No. ] Temperature and Abundance Profiles of Abell 1060 with Suzaku 3 Fig. 1. (a)CombinedXISimageofthecentralandoffsetobservationsinthe0.5–7.0keVenergyrange. TheobservedXIS0-3images are added on the sky coordinate after removing each calibration source region, and smoothed with σ=16 pixel ≃17′′ Gaussian. EstimatedcomponentsofextragalacticX-raybackground(CXB)andinstrumentalbackground(NXB)aresubtracted,andexposure and vignetting are corrected. (b) Exposure mapinks, inwhichcoordinates of the image arethe samewith (a). Four XIS sensors aretreatedseparately. SmallexposureattheCCDcornerscorrespondtothe55Fecalibrationsourcelocations. Theannularregions utilizedinthespectralanalysisareindicatedbygreencircles. Table 1. SuzakuobservationofA1060andNGC2992 Target name Sequence number Date Exposure time (RA, Dec) in J2000 ∗ (l, b) A 1060 center 800003010 2005-Nov-22 40.5 ks (10h36m42.s8, −27◦31′42′′) (269.◦60, 26.◦49) A 1060 offset 800004010 2005-Nov-20 55.5 ks (10h38m03.s8, −27◦31′42′′) (269.◦88, 26.◦65) NGC 2992 † 700005010 2005-Nov-06 38.8 ks (09h45m45.s2, −14◦16′10′′) (249.◦69, 28.◦83) NGC 2992 † 700005020 2005-Nov-19 38.7 ks (09h45m41.s7, −14◦16′05′′) (249.◦66, 28.◦82) NGC 2992 † 700005030 2005-Dec-13 48.3 ks (09h45m51.s0, −14◦16′51′′) (249.◦70, 28.◦84) ∗ AveragepointingdirectionoftheXIS,writtenintheRA NOMandDEC NOMkeywordsoftheeventFITSfiles. † WeusedNGC2992datatoestimatetheGalacticcomponent insubsection4.5. (a) (b) 1 1 XIS1 0.8 0-2′ n o 2-4′ missi 0.6 4-6′ ssion 0.8 Trans 0.4 96--193′′ Transmi 1.0 keV 13-17′ 0.7 keV 0.2 17-27′ 0.6 0.6 keV 0.5 keV 0 0.2 0.5 1 2 5 10 16 Energy (keV) 0−2’ 2−4’ 4−6’ 6−9’ 9−13’ 13−17’ 17−27’ Fig. 2. (a)EstimatedtransmissionofthecontaminantontheBI(XIS1)sensorforeachannularregionusedinthespectralanalysis plottedagainsttheX-rayenergy. Transmissionsinthecentralobservationsaredrawnbysolidlines,andtheoffsetobservationsare bydashedlines. Thesetransmissionsarecalculatedby“xissimarfgen”withacalibrationfileofae xi1 contami 20060525.fits, and writteninthecontami transmiscolumnoftheARFresponse. (b)Thecalculatedtransmissionplottedagainsteachannularregion intheenergiesof0.5,0.6,0.7,and1.0keV. 4 K. Sato et al. [Vol. , 2.3. NXB & CXB Subtraction InordertosubtracttheNXBandtheextra-galacticcos- micX-raybackground(CXB;seeBrandt&Hasinger2005 2) for review), we used the night earth database of 770 ks -c e s exposure provided by the XIS team for the NXB, and es- c ar 1 timated the CXB component using the ASCA results. 1 -s The night-earth spectra were extracted from the same s nt detectorregionasthe A1060observationinordertocan- u o cel positional variation of the NXB. Furthermore, we di- 7 c -0 videdthenight-earthdatareferringtoCORintheranges 1 of <4 GV, 4–13 GV in 1 GV step, and > 13 GV, be- (eV cause it is known that the intensity and energy spectrum 8 - 3 k of the NXB are primarily correlated with COR at the S0. orbital location of the satellite. The NXB spectra to be -1 subtractedinthespectralfittingwereestimatedbyadding 10 theseCOR-sortednight-earthspectraweightedwithexpo- 4 suretimesofthe A1060observationinthe corresponding σ) 2 COTRheraCnXgBe.componentwasestimatedbythe“fake”com- ual ( 0 d mandofXSPECusinguniform-skyARFs,whichisgener- esi -2 R atedby“xissimarfgen”assumingthatthewholesky(prac- -4 tically,r<20′)hastheconstantintensityandthesameX- ray spectrum. We assumed a power-law spectrum for the 1 10 CXBwiththevaluesbyKushinoetal.(2002),Γ=1.4and Projected radius (’) S =5.97×10−8 ergcm−2 s−1 sr−1 (2–10keV),absorbed X withtheneutralhydrogencolumnofNH=4.9×1020cm−2 Fig. 4. In the upper panel, a radial profile of the sur- (Dickey & Lockman 1990). The above CXB intensity is face brightness of A 1060 in the 0.8–3 keV band are plot- taken from table 3 of Kushino et al. (2002), for the in- ted for XMM-Newton MOS1+2 (r < 13′). The best-fit double-β model is shown by the solidgray line, and the two tegrated spectrum with source elimination brighter than β-components are indicated by dashed lines. In the bottom S =2×10−13ergcm−2 s−1 (2–10keV)intheGISfiledof 0 panel,fitresidualsareshowninunitofσ. view with Γ=1.4 (fix) and the nominal NXB level (0%). Itisconfirmedthatthisgivesareasonableestimateofthe Table 3. Thebest-fitparametersoffigure4. CXBcontributionfortheXISinsubsection6.2ofIshisaki et al. (2007) and Fujimoto et al. (2007) β r S ∗ Figures3 (a)–(c)showthe backgroundlevelforSuzaku c 0.8–3keV BI, FI sensors,and XMM-Newton MOS1, at the same 6– narrower .... 1.3±0.1 3.7′±0.3′ 1.26±0.14 9′ annulus of the cluster. The observed spectrum after wider † ..... 0.69 (fix) 7.3′ (fix) 0.96±0.05 the CXB and NXB subtraction are compared with the estimated CXB and NXB spectra separately for Suzaku, * S0.8–3keV atthecenterinunitof10−7 countss−1 arcsec−2. † β andrc isfixedtothevaluesbyHayakawaetal.(2006). while sum of the CXB and NXB are indicated for the XMM-Newtonbecausetheblank-skydatawasusedasthe background. These figures show the background level of sis were performed with SAS version 6.0 and HEAsoft Suzaku XIS is ∼50% lower than that of XMM-Newton. version 6.0.6. As for the XMM-Newton data, data re- It is also notable that there is a strong Al-K peak at α duction and spectral analysis were based on Sato et al. 1.49keVforthe XMM-Newtonbackground,whichmakes (2005)andHayakawaetal.(2006). Hayakawaetal.(2006) the determination of the Mg abundance quite difficult. have reported the temperature and abundance profiles The S/N ratio of the Suzaku XIS BI/FI sensors are by with XMM-Newton observation, however, we reanalyzed about1.8/2.0timeshigherthanthatoftheXMM-Newton the data in order to compare with the Suzaku results on MOS at 1 keV, and 1.3/1.8 times higher at 4 keV. The the same criterion. The differences form Hayakawa et al. particlebackgroundismorestableforSuzakuthanXMM- (2006) are: (1) the event extraction region, (2) using the Newton due to its low-earth orbit. vapec thin thermal plasma model (Smith et al. 2001) in- steadofvmekal (Meweetal.1985;Meweetal.1986),and 3. XMM-Newton Observation & Analysis (3) deriving the element abundances separately. Utilizing the XMM-Newton image, we also derived the The XMM-Newton observation of A 1060 was carried surfacebrightnessprofileofA1060neededtogeneratethe out on 2004 June 29 with total exposure of 64 ks. The Suzaku ARFs, because the spatial resolution of XMM- samedatasetwithHayakawaetal.(2006)wasutilized. We Newton is superior to that of Suzaku. Figure 4 shows used only the MOS data, and data reduction and analy- a radial profile of A 1060 in the 0.8–3 keV energy range, No. ] Temperature and Abundance Profiles of Abell 1060 with Suzaku 5 Fig. 3. (a)Theobservedspectrumattheannularregionof6–9′ withtheSuzakuBI(XIS1)sensorisplottedinblackcrossesafter subtracting the estimated CXB and NXB components, which are plotted by gray crosses and a black histogram, respectively. (b) Sameas(a)butfortheFI(XIS0+XIS2+XIS3)sensors. Thecountratedropbelow<∼0.4keVisduetotheeventthresholdoftheFI CCDs(Koyamaetal.2006). Although 0.2–10keV energy rangeisshownhere, only0.4–7.1keV bandwas utilizedforthespectral fittingforbothBIandFI. (c)Sameas(a)butfortheXMM-NewtonMOS1sensor,andtheestimatedCXB+NXBcomponentusing blank-skyobservations areplottedingray. The0.5–8.0keVenergyrangewasusedforthespectralfittingwithXMM-Newton. fittedwithadouble-β model. Theoriginoftheradialpro- Table 4. Area, coverage of whole annulus, source ratio fileisplacedat(RA,Dec)=(10h36m43.s1,−27◦31′46′′)in reg and observed counts for each annular region. source J2000. Thebest-fitparametersaresummarizedintable3, ratio reg represents the flux ratio in the assumed spatial distributiononthesky(double-β model)insidetheaccumu- in which β and r of the wider β-component are fixed to c lation region to the entire model, and written in the header the values by Hayakawa et al. (2006). keywordofthecalculatedARFresponseby“xissimarfgen”. 4. Spectral Analysis and Results Region∗ Area† Coverage† source ‡ Counts§ 4.1. Suzaku XIS Spectra (arcmin2) ratio reg BI FI 0–2′ 12.6 100.0% 7.7% 30,549 64,606 We extracted spectra from seven annular regions of 0– 2–4′ 37.7 100.0% 13.7% 55,186 114,220 2′, 2–4′, 4–6′, 6–9′, 9–13′, 13–17′ and 17–27′, centered 4–6′ 62.8 100.0% 13.8% 49,259 100,080 on (RA, Dec) = (10h36m42.s8, −27◦31′42′′). The first 6–9′ 137.0 96.9% 17.2% 52,719 100,750 four annuli were taken from the central observation, and ....................................................... the rest of thee annuli were from the offset observation. 9–13′ 49.1 17.8% 2.8% 10,696 17,329 Table4listsareasoftheextractionregions(arcmin2),cov- 13–17′ 76.2 20.2% 2.2% 11,032 21,010 erage of the whole annulus (%), the source ratio reg 17–27′ 167.3 12.1% 2.0% 18,055 28,834 values (%; see captionfor its definition) and the observed ∗ Thefirstfourannuliareextracted fromthecentral observa- countsin0.4–7.1keVincludingNXB andCXB forthe BI tion,andothersarefromtheoffsetobservation. and FI sensors. † Thelargestvaluesamongfoursensorsarepresented. AlthoughthethicknessoftheOBFcontaminationisdif- ‡source ratio reg≡Coverage routS(r)rdr/ rmaxS(r)rdr, ferent among sensors as shown in table 2 and figure 5(b), rin 0 where S(r)represents the radialRprofile shownRinfigure 4and we have confirmed that the four sensors give quite con- table3,andthermax issetto30′. sistent fit results after incorporating the contamination § Observedcounts includingNXBandCXBin0.4–7.1keV. effect into the ARFs, as demonstrated in figure 5(a). We therefore add the three FI spectra (XIS0, XIS2, XIS3), hereafter. 1solarabundance,however,itmayvaryfromfieldtofield Eachannularspectrumisshowninfigure6. Theionized bymorethananorderofmagnitude(Kushinoetal.2002). Mg,Si,S,Fe linesareclearlyseenineachring. TheOVII The ICM spectrum (a) can be represented by a variable andOVIIIlinesareprominentintheouterrings,however, abundancethin-thermalplasmamodel,vapec,whosebest- most of the OVII emission is supposed to come from the fit parameters are what we want. An important point is local Galactic emission, which will be examined in detail thatthespatialdistributionsaredifferentbetween(a)and in subsections 4.3–4.5. (b). Theformerfollowsthesurfacebrightnessoftheclus- ter, while the latter is supposedto havealmosta uniform 4.2. Strategy of Spectral Fit distribution in the XIS field of view. Thebasicstrategyofthespectralfitisdescribedinsub- WethereforegeneratedtwodifferentARFsforthespec- section 6.3 of Ishisaki et al. (2007). The observed spec- trum of each annulus, Au and Ab, which respectively trumisassumedtocontain(a)thinthermalplasmaemis- assume the uniform-sky emission and ∼ 1◦×1◦ size of sionfromtheICM,(b)localGalacticemission,(c)CXB, the double-β surface brightness profile obtained with the and(d)NXB. Theestimationof(c)and(d)isdescribed XMM-Newtondata(table3). Auwasusedtoevaluatethe in subsection 2.3, and both are subtracted from the ob- CXB, and the surface brightness of the Galactic compo- servedspectrum. Thespectrumof(b)canberepresented nent(b)ateachannulusincombinationwiththeXSPEC by one or two thin-thermal plasma model(s), apec, with “fakeit”command,andAb wasusedfortheactualfitting. 6 K. Sato et al. [Vol. , Fig. 5. (a) The uppermost panel show the observed spectra after subtracting the estimated CXB and NXB components at the annularregionof2–4′ plottedseparatelyfortheXIS0(black),XIS1(red),XIS2(green),andXIS3(blue)sensors. Thecrossmarkers denotetheobservedspectraandthesolidlinesshowthebest-fitmodelwithapec1+apec2+phabs×vapec. Theenergyrangearound theSiK-edge(1.825–1.840keV)isignoredforthespectralfit. LowerthreepanelsshowtheresidualsofthefitinunitofσforXIS1, XIS0,XIS2,andXIS3fromuppertolower,respectively. (b)PlotsofthecalculatedXISeffectivearea(ARF+RMF)includingthe XIS quantum efficiency. These responses are used in the spectral fit in (a) with the same colors. The quantum efficiency is much higherforBI(XIS1;red)thanFIsensors,whiletheOBFcontamination isthickerintheorderofXIS3,XIS2,XIS1,andXIS0. Fig. 6. Theupperpanelsof(a)–(g)showthebackground(NXB+CXB)subtracted spectra at the annular regions which is denoted in the panels, and they are plot- ted by red and black crosses for BI and FI, respectively. The estimated CXB and NXB components are subtracted, and the plotted data are fitted with the apec1+apec2+phabs×vapec modeldrawnbygreenandyellowlinesfortheBIand FIspectra. Theapec1andapec2componentsrepresentingthelocalGalacticemission fortheBIspectraareindicated bycyanandorange lines. The energyrangearound the Si K-edge (1.825–1.840 keV) is ignored for the spectral fit. The energies of the several prominent lines are also indicated in the panels. The lower panels show the fitresidualsinunitofσ. No. ] Temperature and Abundance Profiles of Abell 1060 with Suzaku 7 We have confirmed that the assumed double-β surface brightness profile is consistent with the observed Suzaku image by about±15%in figure 5 of Ishisakiet al. (2007). We further constrained that the surface brightness of the Galactic component (b) is nearly constant among all the extractionannuli,whereasitsspectralshapeisdetermined from the spectral fit of our Suzaku data. Details will be described in the next subsection. We adopted the nominal Redistribution Matrix Files (RMF) of ae xiN 20060213.rmf for spectral fitting, al- thoughslightdegradationinenergyresolutionisexpected (subsection 2.1). The ARFs were generated by “xissi- marfgen”, and were convolvedwith the RMFs and added for three FI sensors, using the “marfrmf” and “addrmf” tasks in Ftools. The spectra from BI and FI are fitted simultaneously in the 0.4–7.1keV band except for energy rangeofanomalousresponsearoundtheSiK-edge(1.825– Fig. 7. (a)Amagnificationof0.49–1.4keVrangeoftheBI 1.840keV). Weignoredbelow0.4keVbecausetheCedge (red)andFI(black)spectrainthe17–27′ annulus. Theyare (0.284keV)seenintheBIspectracouldnotbereproduced fitted with the apec1+phabs×vapec model, and the best– perfectlyinourdata. Energyrangeabove7.1keVwasalso fitmodelaredrawnbygreenandyellowlines. Theestimated ignored because background Ni line (∼7.5 keV) left arti- CXBandNXBcomponentsweresubtracted,andasimultane- ficial structures after the NXB subtraction at large radii. ouslyfitwiththe13–17′ annuluswasconductedintheentire It is also known that the XIS response in E>8 keV are energy range of 0.4–7.1 keV. The apec1 or vapec component ∼ for BI is indicated by a cyan or magenta line, respectively. not fully understood at the presentstage. In the simulta- LowertwopanelsshowthefitresidualsforBIandFIinunit neousfit ofBI andFI, onlythe normalizationareallowed ofσ. Theorangedotted linesinfigure10correspond tothis to be different betweenthem, althoughwe foundthat the modeling of the Galactic component. (b) Same as (a) but derived normalizations are quite consistent between the fittedwiththeapec1+apec2+phabs×vapec model,andthe two. apec2 component for BI is indicated by an orange line. We adopt this model in the spectral fit. See subsection 4.3 for 4.3. Estimation of Galactic Component details. We found that the estimation of the Galactic compo- nent significantly affect the determination of the oxygen abundance, because both the ICM and the Galactic com- ponentcontribute to the OVIII emissionline (0.653keV), and the XIS cannot resolve them by redshift due to the limitedenergyresolution. Ontheotherhand,mostofthe OVII emission is supposed to originate in the Galactic component, because the ICM temperature kT > 2 keV ∼ ≃2.3×107 K is too high to emit the OVII lines (0.561, 0.568, 0.574 keV). See, e.g., figure 3 of Yoshikawa et al. (2003) for the oxygen line emissivity. It is therefore important to estimate the Galactic com- ponent precisely, which is possible using the offset obser- vation of A 1060 with Suzaku. In order to determine the surface brightness and the spectral shape of the Galactic Fig. 8. A plotof confidence contour between kT1 (temper- component, we performed the simultaneous fit of the 13– ature of the cooler part of the two apec components) and 17′ and 17–27′ annuli. The Galactic component is promi- the O abundance of vapec for the 17–27′ annulus, in the nent in these annuli as shown in figure 6(f) and (g), how- simultaneous fitting of 13–17′ and 17–27′ annuli with the ever the ICM component is still dominant almost all the apec1+apec2+phabs×vapec model. The cross denotes the best-fitlocation,andthetwocontoursrepresent1σ and90% energy range except for the OVII line. We made the si- confidence ranges,frominnertoouter,respectively. multaneous fit in the whole 0.4–7.1 keV range (except 1.825–1.840 keV), assuming one or two apec models for nent is by about twice larger for (a). We examined the the Galactic component, and the fit results are presented improvement of the χ2 (∆χ2 =46) with the F-test, and intable5andfigure7. Theresultantnormalizationofthe addingtheapec componentwasjustifiedwithalargesig- apec model in table 5 is scaledso that it gives the surface nificance (false2probability ∼10−9). We further allowed brightness in the unit solid angle of arcmin2. thenormalizationoftheapec tobecomefreebetweenthe It may appear that the difference between these two 2 twoannuli. Thederivedsurfacebrightnessratioofthe13– modelsarenotlargeinfigure7,however,itisnotablethat 17′annulusto17–27′was0.9±0.2,henceitwasconsistent the derived OVIII line intensity for the Galactic compo- 8 K. Sato et al. [Vol. , Table 5. Thebest-fitparametersoftheapec component(s) forthesimultaneousfitofthespectrain13–17′ and17–27′ annuli. Fit model Norm ∗ kT Norm ∗ kT χ2/dof 1 1 2 2 (keV) (keV) (a) apec +phabs×vapec ................. 1.20±0.07 0.179+0.006 — — 1128/994 1 −0.012 (b) apec +apec +phabs×vapec ......... 1.66±0.35 0.143+0.014 0.21±0.05 0.737+0.063 1082/992 1 2 −0.015 −0.088 ∗ Normalizationofthe apec component dividedbythe solidangle, Ωu=π×(20′)2,assumed inthe uniform-skyARFcalculation (20′ radiusfromtheopticalaxisofeachXISsensor),Norm= nenHdV/(4π(1+z)2DA2)/Ωu×10−20 cm−5 arcmin−2,whereDA istheangulardistancetothesource. R to be constant. We also confirmedthatthe 9–13′ annulus abundances were reasonably determined although the er- gavethesameresult. Wethereforeconcludedthatthetwo rorsare larger than other elements. We therefore decided apecmodelsarerequiredtoaccountforthe Galacticcom- to link the S, Ar, and Ca abundances to be the same, be- ponent. Hereafter,theapec +apec +phabs×vapecmodel cause they showed similar values. The Al abundance be- 1 2 are utilized for the spectral fitting, otherwise stated. came 0.0, which is physically strange, therefore we fixed To demonstrate how sensitive the O abundance of the the Alabundanceto1solar. Theχ2/dofwasincreasedto ICM is to the assumed Galactic component model, we 1898/988withthis treatment,howeverweconfirmedthat present a confidence contour between kT (keV) of the other parametersdidnot changebeyondthe quotederror 1 apec component and the O abundance (solar) of vapec range in table 6. Somehow, the Ne and Ni abundances 1 for the outermost annulus (17–27′) in figure 8. There were larger than other elements. This might be due to appears to exist a negative correlation between the two these element lines could not be resolved from the Fe-L parameters, because higher temperature of the Galactic line complex. Note that the Ni abundance was also de- component produces more OVIII emission line relative to termined by Ni-L lines because we ignored energy range OVII, which contribute to reduce the OVIII line from the above7.1keV,whiletheFeabundancewasdeterminedby ICM (vapec component). Influences on the derived tem- both Fe-L and Fe-K lines. Anyway, we left these Ne and perature and abundance by the modeling of the Galactic Ni abundances to move freely during the spectral fit. We component will be tested in subsection 4.4, too. also present results when the O, Ne, and Mg abundances In order to take into account both existence of the are linked to have the same value in Appendix 2. Galacticcomponentitselfandpropagationofitsstatistical Wealsofoundthattheassumedabundancetable,which error,wesimultaneouslyfittedeachannuluswiththeout- defines the “1 solar” of each element relative to H, sig- ermost annulus of 17–27′. As mentioned in the previous nificantly affects the determination of temperature and subsection,thenormalizationoftheapec +apec compo- abundance. In fact, difference in the abundance table 1 2 nent are constrained to give the same surface brightness affects the fit results in three ways. Firstly, we assume betweenthetwoannuli. Thetemperaturesofthetwoapec the Galactic component to have the apec +apec model 1 2 modelswerealsocommonbetweenthetwo,howevertheir with “1 solar”. Secondly, the Galactic absorption model values (two normalizations and two temperatures) them- ofphabs is alsochangedbythe assumedabundancetable. selvesareleftfree. Itisconfirmedthatthederivednormal- Lastly, the derived abundances are given in unit of “so- izations, Norm and Norm , and temperatures, kT and lar” by the vapec model. The first two effects are rather 1 2 1 kT , are consistent with the values in table 5(b) within complicated, and details are investigated in Appendix 2. 2 the quoted errors. In this section, we treat only the solar abundance ratio of angr (Anders & Grevesse 1989)with the phabs absorp- 4.4. Radial Temperature & Abundance Profiles tion model, simply because they are the default standard Before entering the spectral fit at each annulus, we in- ofXSPEC.Intermsoftheχ2,itappearstogivethe min- vestigate the fit result in the central 0–6′ region to test imum χ2 with the lodd (Lodders 2003) abundance table the capability in the abundance determination with the in combination with the wabs (Morrison & McCammon Suzaku XIS. This region exhibits nearly constant tem- 1983) absorption model which utilizes the abundance ta- perature and metal abundances (figure 10), and the con- ble of aneb (Anders & Ebihara 1982) built-in the code tribution of the Galactic component is almost negligible (table 11). (figure 6). The fit result is presented in table 6, in which We then examined the influence of uncertainty in most of element abundances are allowed to be free in the the OBF contaminant using the spectrum within 6′. spectral fit, except for He, C, and N, which are fixed to Figure9(a)showsamagnificationofthebest-fitspectrum 1 solar with the assumed abundance ratio of angr. in the energy range of 0.4–1.0 keV with the “nominal” Although the fit was not acceptable due mainly to the ARF response, in which transmission of the OBF con- very high photon statistics than the systematic errors in taminant was estimated according to the calibration files the instrumentalresponse,this result wasuseful to assess ae xiN contami 20060525.fits(N=0,1,2,3),asshown whether each element abundance was reasonably (sub- in figures 2 and 5(b). As described in subsection 4.2, the solar to ∼ 2 solar) determined or not. The Ar and Ca spectrumwassimultaneouslyfittedwiththe17–27′annu- No. ] Temperature and Abundance Profiles of Abell 1060 with Suzaku 9 Table 6. Resultofthespectralfitinthecentralregionof0–6′,withtheapec1+apec2+phabs×vapecmodel. Onlytheparameters for the vapec component are presented. The abundance of each element (except for He, C, and N, which are fixed to 1 solar with theassumedabundance ratioofangr)isallowedtobefreeinthespectral fit. Errorsare90%confidence rangeofstatisticalerrors, anddonotincludesystematicerrors. Region kT O Ne Mg Al Si χ2/dof (keV) (solar) (solar) (solar) (solar) (solar) 0–6′ ................... 3.54+0.03 0.31+0.08 1.21+0.10 0.70+0.10 0.00+0.13 0.57+0.05 1871/986 −0.03 −0.08 −0.10 −0.09 −0.00 −0.05 Norm ∗ S Ar Ca Fe Ni (solar) (solar) (solar) (solar) (solar) 302±3 0.57+0.07 0.53+0.16 0.44+0.17 0.43+0.01 0.83+0.23 −0.07 −0.16 −0.17 −0.01 −0.23 ∗ Normalizationofthevapeccomponent scaledwithafactorofsource ratio reg/areaintable4, Norm=sourcearreaatio reg nenHdV/(4π(1+z)2DA2)×10−20 cm−5 arcmin−2,whereDA istheangulardistancetothesource. R Table 7. Resultsofthespectralfitsinthecentralregionof0–6′ bychangingtheamountoftheOBFcontaminant. Contaminant O Ne Mg Al Si S Ar Ca Fe Ni (solar) (solar) (solar) (solar) (solar) (solar) (solar) (solar) (solar) (solar) nominal ......... 0.31 1.21 0.70 0.00 0.57 0.57 0.53 0.44 0.43 0.83 +20% ........... 0.43 1.05 0.48 0.00 0.47 0.50 0.45 0.46 0.41 0.41 free ∗............ 0.38 1.12 0.57 0.00 0.51 0.53 0.47 0.43 0.42 0.54 Contaminant kT Norm† kT Norm ‡ kT Norm ‡ N § χ2/dof 1 1 2 2 CO1/6 (keV) (keV) (keV) BI FI nominal ......... 3.54 302 0.159 1.00 0.752 1.00 1.74 2.32 1878/986 +20% ........... 3.36 298 0.148 1.40 0.721 1.09 2.11 2.72 1746/986 free ∗............ 3.50 301 0.159 0.96 0.765 0.95 2.18 2.38 1681/982 ∗ WefittedtheamountoftheOBFcontaminantasafreeparameter withtheC/Onumberratiofixedto6. † Normalizationofthevapecmodel,calculatedinthesamewaywithtable6 . ‡ Ratiotothenormalizationofthe“nominal”ARF. § ColumndensityoftheOBFcontaminantwithchemicalcompositionofCO1/6 forBIandFIinunitof1018 cm−2. Table 8. Summaryofthebest-fitparametersofthevapeccomponentforeachannularregionwiththeapec1+apec2+phabs×vapec model. Each annulus is simultaneous fitted with the the outermost 17–27′ annulus. Errorsare 90% confidence range of statistical errors,and donot includesystematic errors. The solarabundance ratioof angris assumed. These resultsareplotted infigure 10. TheOabundance intheoutertworegionsbecomes lowerwhentheGalacticcomponent isexpressedbyasingleapecmodel,which isdrawnbyorangedottedlinesinfigure10. ThekT,O,andMgcolumnsatr<6′ areslightlydifferentwhentheOBFcontaminant isincreasedby“+20%”, whichisdrawnbyblackdotted linesinfigure10. TheNeabundance isprobablynotreliablebecause the SuzakuXIScannotresolvetheionizedNelinesfromtheFe-Llinecomplex. Region Norm∗ kT O Ne Mg Si S, Ar, Ca Fe Ni χ2/dof (keV) (solar) (solar) (solar) (solar) (solar) (solar) (solar) 0–2′ 605±12 3.34+0.04 0.38+0.16 1.20+0.20 0.76+0.19 0.69+0.11 0.82+0.12 0.49+0.03 1.13+0.46 1240/992 −0.04 −0.15 −0.20 −0.19 −0.11 −0.12 −0.03 −0.45 2–4′ 352±5 3.39+0.03 0.28+0.11 1.11+0.16 0.65+0.14 0.52+0.08 0.52+0.08 0.42+0.02 0.90+0.34 1290/992 −0.03 −0.12 −0.15 −0.14 −0.08 −0.08 −0.02 −0.34 4–6′ 198±3 3.42+0.04 0.29+0.13 1.24+0.18 0.67+0.15 0.58+0.09 0.53+0.09 0.41+0.02 0.87+0.38 1345/992 −0.04 −0.13 −0.16 −0.15 −0.09 −0.10 −0.02 −0.37 6–9′ 109±2 3.25+0.04 0.58+0.14 1.17+0.16 0.59+0.15 0.50+0.08 0.55+0.09 0.40+0.02 1.17+0.36 1329/992 −0.04 −0.14 −0.15 −0.15 −0.08 −0.09 −0.02 −0.35 9–13′ 44±2 2.87+0.10 0.65+0.34 0.89+0.35 0.75+0.33 0.27+0.17 0.19+0.20 0.40+0.05 1.13+0.77 1057/992 −0.10 −0.29 −0.33 −0.32 −0.17 −0.19 −0.05 −0.72 13–17′ 26±1 2.50+0.09 0.72+0.29 0.44+0.29 0.78+0.29 0.30+0.15 0.44+0.18 0.25+0.05 0.16+0.60 1082/992 −0.09 −0.38 −0.28 −0.27 −0.14 −0.17 −0.05 −0.16 17–27′ 13±1 2.22+0.11 0.48+0.41 0.30+0.29 0.21+0.22 0.28+0.12 0.21+0.14 0.21+0.04 0.00+0.41 —† −0.12 −0.37 −0.26 −0.21 −0.12 −0.14 −0.04 −0.00 ∗ Normalizationofthevapec model,calculated inthesamewaywithtable6. † The17–27′ annuluswasfittedsimultaneouslywithotherannuli,andthebest-fitvalueswiththe13–17′ annulusarepresentedhere. 10 K. Sato et al. [Vol. , Fig. 9. (a) A magnification of the best-fit spectrum at 0–6′ annulus in the energy range of 0.4–1.0 keV with the “nominal” ARF response, in which transmission of the OBF contaminant was estimated according to the calibration files ae xiN contami 20060525.fits (N =0,1,2,3). Table 6 and the “nominal” row of table 7 show the best-fit parameters. (b) Thebest-fitspectrumwiththe“+20%”ARF,inwhichamountoftheOBFcontaminantwasincreasedby+20%thanthe“nominal” ARF. The“+20%”rowoftable7showsthebest-fitparameters. (c)Thebest-fitspectrumwiththeno-contaminantARFbutfitted with varabs×(apec1+apec2+phabs×vapec) model to consider the transmission of the OBF contaminant in the XSPEC varabs modelwithchemicalcompositionofCO1/6. The“free”rowoftable7showsthebest-fitparameters. (d), (e),(f)Sameas(a),(b), (c)butforthe17–27′ annulus,respectively. Eachspectrumwassimultaneouslyfittedwith(a),(b),(c). lus shown in figure 9(d). In this figure, it is suggested fromthe “nominal”value forBI isbeyondthe calibration that the absorption in the low energy band is slightly uncertainty, however, almost all of the best-fit values at inconsistent between BI and FI sensors, namely, the BI the “free”rowintable 7 arebetweenvalues at“nominal” spectrum appears to need more absorption in 0–6′ where and “+20%”. We therefore consider the +20% result as the amount of contamination is larger than 17–27′ (see the systematic error range due to the uncertainty in the figure 2). We also found that the contamination mea- OBF contaminant. surementbyRXJ1856.5−3754onOctober25,2005inthe Consideringthesesystematics,resultsofthespectralfit XIS hardware paper (figure 14 of Koyama et al. 2006) at each annulus are summarized in table 8 and figure 10. indicates by about 20% larger amount of the OBF con- We tested the results by changing the background nor- taminant than the “nominal” value for BI (XIS1). malizationby±10%,andtheyareplottedingreendotted We therefore generated ARF responses changing the lines in figure 10. The systematic error due to the back- amount of the OBF contaminant by +20%. The fit re- ground estimation is almost negligible. Difference in the sult is shown in figure 9(b), (e) and table 7 (“+20%” best-fit values by modeling the Galactic component with row). Note that the CXB background to subtract was asingleapecwasinvestigated,andtheywereindicatedby also modified in this fit. The fit residual for 0–6′ was orange dotted lines. The differences are within the sta- improved, and the χ2 was decreased by ∆χ2 =132. We tistical error for the inner five annuli, however, O abun- further testedthe fit byadding anabsorptionofthe OBF dance becomes lower than the 90% confidence error at contaminant with chemical composition of CO to the the outer two annulias indicated in figure 7,and temper- 1/6 fit model using the XSPEC varabs model.2 The fit re- ature and Fe abundance become lower at the outermost sult is shown in figure 9(c) and (f). The best-fit values annulus. This is due mainly to the fact that the XIS can- are summarized in table 7 (“free” row), and the χ2 was not resolve the ICM OVIII line from the Galactic OVIII improved by ∆χ2=197. The derived amount of contam- by redshift. The systematic error range due to the un- inantfor BI wasby 25%largerthan the “nominal” value, certainty in the OBF contaminant is indicated by black and by 3% larger for FI. We think that the deviation dottedlines. Itissometimeslargerthanthestatisticaler- rorsatsmallradii(r<6′),particularlyforkT,O,andMg 2 In this method, we used the same CXB background with the abundances. Though∼Ne abundance is significantly larger “nominal”ARF,andthesurfacebrightnessoftheGalacticcom- than other elements at small radii (r<9′), it is probably ponentwasslightlydifferentbetween 0–6′ and17–27′.
Description: