Table Of Content

Mon.Not.R.Astron.Soc.000,1–??(2008) Printed21January2009 (MNLATEXstylefilev2.2) A spectacular giant arc in the massive cluster lens − ⋆ MACS J1206.2 0847 9 H. Ebeling1, C.J. Ma1, J.-P. Kneib2, E. Jullo2, N.J.D. Courtney3, E. Barrett1, A.C. Edge3, J.-F. 0 0 1 Institute for Astronomy, Universityof Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822, USA 2 2 Laboratoire d’Astrophysique de Marseille, OAMP, CNRS-Université Aix-Marseille, 38 rue Frédéric Joliot-Curie, 13388 Marseille Cedex 13, France 3 Department of Physics, Universityof Durham, South Road, Durham, DH1 3LE, UK n 4 Laboratoire d’Astrophysique de Toulouse-Tarbes, Université de Toulouse, CNRS, 14 Avenue Edouard Belin, F-31400 Toulouse, France a J 4 1 Accepted 2008TBD.Received2008TBD;inoriginalform2008TBD ] E ABSTRACT H We discuss the X-ray and optical properties of the massive galaxy cluster . MACSJ1206.2−0847(z =0.4385),discoveredintheMassiveClusterSurvey(MACS). h Our Chandra observation of the system yields a total X-ray luminosity of 2.4×1045 p erg s−1 (0.1–2.4 keV) and a global gas temperature of 11.6±0.7 keV, very high val- o- ues typical of MACS clusters. In both optical and X-ray images MACSJ1206.2−0847 r appears close to relaxed in projection, with a pronounced X-ray peak at the location t s of the brightest cluster galaxy (BCG); we interpret this feature as the remnant of a a cold core. A spectacular giant gravitational arc, 15′′in length, bright (V∼ 21) and [ unusuallyred(R−K =4.3),isseen20arcsecwestoftheBCG;wemeasurearedshift 1 of z = 1.036 for the lensed galaxy. From our Hubble Space Telescope image of the v cluster we identify the giantarc andits counter image as a seven-foldimagedsystem. 4 An excess of X-ray emission in the direction of the arc coincides with a mild galaxy 4 overdensityandcouldbe the remnantofa minormergerwith agroupof galaxies.We 1 deriveestimatesofthetotalclustermassaswellasofthemassoftheclustercoreusing 2 X-ray, dynamical, and gravitational-lensing techniques. For the mass enclosed by the 1. giant arc (r <119 kpc) our strong-lensing analysis based on Hubble Space Telescope 0 imaging yields a very high value of 1.1×1014 M⊙, inconsistent with the much lower 9 X-ray estimate of 0.5×1014 M⊙. Similarly, the virial estimate of 4×1015 M⊙ for the 0 total cluster mass, derived from multi-object spectroscopy with CFHT and the VLT : of 38 cluster members, is significantly higher than the corresponding X-ray estimate v i of 1.7×1015 M⊙. We take the discrepancy between X-ray and other mass estimates X to be indicative of pronounced substructure along the line of sight during an ongoing r merger event, an interpretation that is supported by the system’s very high velocity a dispersion of 1580 km s−1. Key words: gravitational lensing — X-rays: galaxies: clusters — galaxies: clusters: general — galaxies: clusters: individual (MACSJ1206.2−0847 1 INTRODUCTION andshapeofsucharcs,aswellasoflens-generatedmultiple images,canbeusedtomodeltheprojectedmassintheclus- The concentration of both dark and baryonicmatter in the ter, and in-depth follow-up of the brightest lensed features cores of clusters of galaxies has many profound implica- often yields valuable insights into the properties of distant, tions for our understanding of cluster growth and cosmol- faint galaxies (e.g. Kneib et al. 2004; Smail et al. 2007). ogy.Firstly,thestructureandevolutionofthegravitational Thirdly,thehigh density and temperatureof thegas in the potential of a cluster of galaxies depends on the nature of core of clusters leads to intense X-ray emission that cur- dark matter and thus allows direct comparison with pre- rent instrumentation can detect out to redshifts well above dictions from numerical simulations (e.g. Navarro, Frenk unity.Theselection ofmassiveclustersthroughX-rayemis- & White 1997). Secondly, the surface density of mass in- sion has proved very successful at providing cosmological tegrated through the core of a cluster is often sufficiently constraints (Henry 2000; Borgani et al. 2001; Allen et al. hightostrongly lensbackgroundgalaxies intogravitational 2003; Pierpaoli et al. 2003; Allen et al. 2008; Mantz et al. arcs (Soucail et al. 1987; Mellier et al. 1991; Kneib et al. 2008)andfollow-upobservationsofX-rayluminousclusters 1996; Smith et al. 2001). Detailed analysis of the location 2 H. Ebeling et al. have revealed many spectacular cases of gravitational lens- mask was designed, covering the ∼7’ field of FORS1 with ing (Gioia & Luppino 1994; Smith et al. 2001; Dahle et al. 19 slitlets of fixed length (22”). Credible redshifts could be 2002; Covone et al. 2006). measuredfor14objects.Duringtheexposuretheseeingwas In this paper we present a comprehensive multiwave- 0.6 arcsec. Spectra of thespectrophotometric standard star length study of a complex gravitational arc and its host EG274 were obtained for calibration. cluster MACSJ1206.2−0847, an X-ray selected system at Additionalmulti-object spectroscopy of colour-selected intermediate redshift found by the Massive Cluster Survey, galaxies in the field of MACSJ1206.2−0847 was performed MACS (Ebeling, Edge & Henry 2001, 2007). We describe onMay8,2003usingtheMOSspectrographontheCanada- our optical, NIR and X-ray observations in Sections 2 and France-Hawaii Telescope (CFHT) on Mauna Kea. We used 3,investigatethepropertiesofthegiantarc,theclusterlens, the B300 grism and the EEV1 CCD, which provides a res- andthebrightestclustergalaxyinSection4to7,andderive olution of 3.3˚A/pixel, and observed through a broadband massestimates for theclustercore andtheentiresystemin filter (#4611, General Purpose set) to produce truncated Section 7. We present a discussion of our results as well as spectra, coveringabout 1800˚A centred on 6150˚A, suchthat conclusions in Section 8. spectra could be stacked in three tiers along the dispersion Throughout we use a ΛCDM cosmology (Ω = 0.3, direction. This choice of filter and grism ensured that, for M Ωλ =0.7) and adopt H0 = 70 km s−1 Mpc−1. allclustermembers,redshiftscouldbeobtainedfromtheCa H+K lines which fall at 5660˚A and 5710˚A at the approximate cluster redshift of z = 0.44. Weather conditions were poor though, and only 48 of the 67 objects observed (total 2 OBSERVATIONS integration time: onehour) yielded reliable redshifts. The galaxy cluster MACSJ1206.2−0847 was originally dis- Finally, MACSJ1206.2−0847 was observed on Decem- covered ina short two-minuteR-bandimage taken on June ber 6, 2005 with the Advanced Camera for Surveys (ACS) 15, 1999 with the University of Hawaii’s 2.2m telescope aboard the Hubble Space Telescope as part of program (UH2.2m) on Mauna Kea. The observation, performed as SNAP-10491 (PI Ebeling), for a total of 1200 seconds in part of the MACS project, was triggered by the presence theF606Wfilter,resultinginahigh-resolutionimageofthe oftheX-raysource1RXSJ120613.0−084743 intheROSAT cluster core, including the giant arc. Bright SourceCatalogue which had noobvious counterpart in the standard astronomical databases and could also not trivially be identified by inspection of the respective Digi- 2.2 Near-infrared tized SkySurveyimage. Near-infrared observations of the core of Following the initial, tentative identification of the MACSJ1206.2−0847 were performed in the J and K RASSX-ray source as a potentially massive galaxy cluster, bands using the United Kingdom Infra-Red telescope weconductedarangeoffollow-upobservationstofirmlyes- (UKIRT) on April, 5 2001 using the UFTI imager during tablish the cluster nature of this source and to characterize a period of good seeing. The observations consisted of two its physical properties. iterations of a nine-point dither pattern, each of 60 second exposures for a total integration time of 1080 seconds. The 2.1 Optical seeing measured from these observations was 0.47 and 0.59 arc-seconds in theJ and K bands,respectively. SpectraoftwogalaxiesinMACSJ1206.2−0847,oneofthem the BCG, were taken with the Wide-Field Prism Spectro- graph on the UH2.2m on July 4, 1999, using a 420 l/mm 2.3 X-ray grism, a Tektronix 20482 CCD yielding 0.355 arcsec/pixel, MACSJ1206.2−0847 was observed on December 18, 2002 and a 1.6 arcsec slit. The two redshifts were found to be withtheACIS-IdetectoraboardtheChandraX-rayObser- concordant, establishing an approximate cluster redshift of vatoryforanominaldurationof23.5ks,aspartofaChan- z≈0.434. dra Large Programme awarded to the MACS team. The Moderately deep,multi-passband imaging observations target was placed about two arcmin off the standard aim- (3×240s, dithered by 10 arcsec, in each of the V, R, and pointtoavoidfluxbeinglost inthechipgapsoftheACIS-I I filters) of the cluster were obtained with the UH2.2m on detector, while still maintaining good (sub-arcsec) angular January 29, 2001, using again the Tektronix 20482 CCD resolution across the cluster core. VFAINT mode was used whichprovidesascaleof0.22arcsecperpixelanda7.3×7.3 inordertomaximizetheefficiencyofparticleeventrejection arcminute2fieldofview.Theseeingwasvariablethroughout in the post-observation processing. thenight;wemeasureseeingvaluesof0.85,1.05,0.90arcsec in the V, R, and I passbands, respectively, from the final, co-added images. Spectroscopicobservationsofpresumedclustergalaxies 3 DATA REDUCTION aswellasofthegiantarcinMACSJ1206.2−0847 wereper- 3.1 Optical imaging formed with the FORS1spectrograph in multi-object spectroscopy mode at the UT3 Melipal telescope of the VLT Standard data-reduction techniques (bias-subtracting, flat- on April11, 2002. TheG300V grism, an ordersorting filter fielding, image combination and registration) were applied (GG375), and a 1 arcsec slit were used, yielding a wave- to the V, R and I band data using the relevant iraf pack- length coverage from ∼4000˚A to ∼8600˚A at a resolution of ages. The data were photometrically calibrated via the ob- R=500. The total exposure time was 38 minutes. A single servation of Landolt standard-star fields. The massive cluster lens MACSJ1206.2−0847 3 In order to measure aperture magnitudes, seeing- matched frames were produced by applying a Gaussian 4466′′ 3300″″ smoothing totheV,I,Jand Kimages such thattheseeing in these frames was degraded to match the 1.1 arc-seconds measured in the R band. Using SExtractor, the seeing- 0000″″ matched (undegraded) images were then used to measure aperture (total) magnitudes. 4477′′ 3300″″ Photometry of the giant arc was performed by man- ually defining an aperture mask fitted to the arc profile. Tthheescaireenacedefrfianmedeabnydtahibsaacpkegrrtouurnedwiamsatgheepnromdausckeeddboyumt oe-f DeclinationDeclination 0000″″ diansmoothingovertheabsentarc.Thisbackgroundimage 4488′′ 3300″″ wassubtractedfromthescienceframeandVRIJKphotom- etryobtainedbyapplyingtheaperturetotheresultingsky- subtracted images. 0000″″ 3.2 Optical spectroscopy --0088°° 4499′′ 3300″″ 200 kpc 37.7" For the reduction of our spectroscopic data we applied the same standard techniquesas for the imaging data, followed 1188ss 1166ss 1144ss 1122ss 1100ss 0088ss 1122hh 0066mm 0066ss by straightening of the skylines, extraction of spectra, and RRiigghhtt AAsscceennssiioonn wavelength calibration using therelevant iraf packages. Preliminaryredshiftsweredeterminedviavisualinspec- Figure 1. Colour image (IVR mapped to RGB) of tion,typicallyfromthecalciumHandKlines.Finalrefined MACSJ1206.2−0847 from 3×240s observations (per filter) obtained with the UH 2.2m telescope (see text for observational redshiftswerefoundwithamulti-templatecross-correlation method using the iraf task fxcor. details). Redshiftswithcross-correlation peakheightsexceeding 0.70 are typically accurate to 0.0002; for peak heights between 0.5 and 0.7 theerroris typically 0.0005. All redshifts were converted to the heliocentric reference frame. Table 2 lists the coordinates of all galaxies observed, as well as the measured redshifts with their uncertainties. 4466 3.3 Near-infrared imaging The J and K imaging data were reduced using the ORAC- DseRrvadtaitoan-roefdoubcsteiornvaptiopreylisnteananddarwdesrteacrsa.libratedthroughob- DeclinationDeclination 4488 3.4 X-ray We use Ciao (version 3.3), the standard suite of software --0088°° 5500′′ tools developed for the analysis of Chandra data at the Chandra Science Center, as well as the most recent calibration information, toreprocess theraw ACIS-Idata.Our 250 kpc 47.2" inspection of the lightcurve of the event count rate in the source-free regions of the ACIS-I detector finds no signif- 2255ss 2200ss 1155ss 1100ss 0055ss 1122hh 0066mm 0000ss RRiigghhtt AAsscceennssiioonn icant flaring, leading to an effective (dead-time corrected) total exposure time of 23.2 ks. Toinvestigatepotentialspatialvariationsinthecluster Figure 2. Locations of all galaxies for which spectroscopic redshifts were measured with the VLT (circles) and CFHT (dia- gas temperature we define various source and background monds), overlaid on the UH2.2m R-band image. Bold symbols regions. For each of these regions we generate auxiliary re- markgalaxiesfoundtobeclustermembers.SeeTable2forcoor- sponse files (ARF) and response matrix files (RMF) which dinatesandredshifts. weigh the position-dependent instrument characteristics by the observed count distribution in the respective area. Fol- lowingMarkevitch&Vikhlinin(2001),onlythe0.5–2.0keV andback-illuminatedACISchips(Vikhlininetal.2002).Fi- band data are used to create these maps of spatial weights, nally, we use the acisabs package to correct all ARFs for sincetheeffectsofvignettingaresmallinthisenergyrange. theeffectsofthe(time-dependent)buildupofacontaminat- Wealsoapplyacorrectionofafactorof0.93totheeffective ing deposit on theoptical detector window which results in areadistributionatenergiesbelow2keVinallARFsassug- a reduction of theeffective area at low energies. gested by a comparison of calibration results for the front- Background regions are defined by copying the respec- 4 H. Ebeling et al. 1155 20 1100 nn 55 −1) 15 [OII] Å 0000..3300 00..4400 00..5500 00..6600 00..7700 −2−1 s 88 zz −18 erg cm 10 Mg II (?) 66 flux (10 5 nn 44 22 00 0 00..442200 00..443300 00..444400 00..445500 00..446600 4000 5000 6000 7000 8000 zz λ (Å) Figure 3. Histogram of galaxy redshifts in the field of Figure 5. Spectrum of the giant gravitational arc in MACSJ1206.2−0847 as observed with multi-object spectro- MACSJ1206.2−0847 as observed with FORS1 on the VLT. We graphsonCFHTandtheVLT(cf.Table2).TheoverlaidGaus- interpret the singleemissionlineobserved at 7589˚A as being [O sian curve is characterized by the best-fit values for the sys- II],whichtranslatesintoaredshiftof1.036forthelensedgalaxy. temic redshift and cluster velocity dispersion of z = 0.4385 and Ourmeasurementisconfirmedbyindependentobservationscon- σ=1575+191 km/s,respectively. ductedalmostsimultaneouslybySandetal.(2003)usingtheESI −190 echellespectrographonKeck-IIwhichresolvesthe[OII]doublet therebymakingtheidentificationunambiguous. tivesource regions tothesame chip-ylocation on theother three ACIS-I CCDs. This strategy minimizes the impact of any residual chip-y dependence of the background on the dataanalysis,aneffectthatisunavoidableifthebackground 5 ARC PROPERTIES is selected as a source-free region on the same CCD as the The bright (V =21.0) giant arc is centered at α (J2000) = cluster. 12h 06m 10.75s,δ (J2000) =−08◦ 48’04.5”, about 20′′west of the BCG. Its unusually red colour is apparent in Fig. 4 which shows a composite V, I, K image of the cluster core. Also prominent, and marked by the arrow in Fig. 4, is the 4 CLUSTER GALAXY DISTRIBUTION counterimage,clearlyidentifiablebyitscolour,atα(J2000) Figure 1 shows a colour image of the cluster generated = 12h 06m 11.27s, δ (J2000) = −08◦ 47’ 43.0”. Enlarged from the V, R, I images obtained with the UH2.2m. high-resolution views of arc and counterimage provided by MACSJ1206.2−0847 is found to be an optically very rich HST are shown in the margins of Fig. 4. The photomet- system with a single dominant central galaxy and no obvi- ric properties of arc and counter image are summarized in ous subclustering in the apparent, projected cluster galaxy Table 1; note howeverthat at theresolution of our ground- distribution. A giant arc is clearly visible about 20 arcsec based images both the arc and its counter image is blend to the West of the BCG. The astrometric solution used is ofseveralobjects.Figure5showsthespectrumofthegiant based on eight stars within the field of view of the V band arcinMACSJ1206.2−0847 asobservedwithFORS1onthe imagethathaveaccuratecelestialcoordinatesintheHubble VLT. Guide StarCatalogue (GSC2). AtR−K =4.3thegiantarcinMACSJ1206.2−0847 is Ourspectroscopicobservationsof85galaxiesinthefield amongthereddeststronglylensedfeaturescurrentlyknown. of MACSJ1206.2−0847 (Fig. 2) yield redshifts as listed in ItscolouriscomparabletothatofthegiantarcinAbell370 Table2.TwogalaxieswereobservedwithboththeVLTand (R−K = 4.1, Aragón-Salamanca & Ellis 1990) and only CFHT;theirspectroscopicredshiftsagreewithintheerrors. slightly bluer than the red arc in Abell 2390 (R−K =4.6, Usingonlythemostaccurateredshiftswithcorrelationpeak Smail et al. 1993). heights exceeding 0.7, we apply iterative 3σ clipping to the Our multi-band photometry allows the classification of redshift histogram to obtain a systemic cluster redshift of the background galaxy lensed into the giant arc. Standard z=0.4384andaveryhighvelocitydispersioninthecluster templatespectralenergydistributions(SEDs)offivegalaxy rest frameofσ=1581 km/sbased on38redshifts.Entirely types were redshifted to match that of the giant arc. The consistent values of z =0.4385 and σ =1575+191 km/s are filter response curves of each of theV, R, I, J and K bands −190 foundusingtheROSTATstatisticspackage(Beers,Flynn& were then convolved with these SEDs and normalized to Gebhardt 1990). The resulting redshift histogram is shown theRbandtoproducepredictedcoloursforeachofthefive in Fig. 3. Despite the extremely high velocity dispersion of galaxytypes,allrelativetotheRband.Theresultingmodel thesystemwefindnoobvioussignsofsubstructurealongthe m –m colours,aswellastheequivalentobservedcoloursof λ R lineofsight;aone-sidedKolmogorov-Smirnovtestfindsthe thearc,areshowninFig.6.Wefindthecolourdistribution observedredshiftdistributiontobeonlymildlyinconsistent ofthearcin MACSJ1206.2−0847 tobeconsistent with the with a Gaussian (2.05σ significance). background galaxy being a normal spiral of class Sbc. The massive cluster lens MACSJ1206.2−0847 5 4400″″ 4477′′ 5500″″ [h] 0000″″ nn oo atiati nn clicli ee DD 1100″″ 2200″″ 30 kpc −−0088°° 4488′′ 3300″″ 5.7" 1144ss 1133ss 1122ss 1122hh 0066mm 1111ss RRiigghhtt AAsscceennssiioonn Figure4.Colourimage(KIVmappedtoRGB)ofMACSJ1206.2−0847from3×240sobservations(I,V)and18×60sobservations(K) obtainedwiththeUH2.2mandUKIRT,respectively(seetextforobservationaldetails).Anarrowpointstothecounterimageoftwoof thefourbackground galaxiesdistorted bythecluster’sgravitational fieldto createthe prominentgiantarc20arcsecwestoftheBCG. Smallerpanelsontopandtotherightshowenlargedhigh-resolutionviewsofthearcanditscounterimageasobservedwithHST/ACS ina1200s snapshotwiththeF606Wfilter. Table 1.Extinction-corrected magnitudes andcoloursofthegiantarcinMACSJ1206.2−0847 anditscounter image(cf.Fig.4). object I R V J K R−I V−I V−R R−J R giantarc 19.27±0.07 20.27±0.11 20.99±0.08 17.82±0.06 15.95±0.06 1.00±0.12 1.72±0.10 0.72±0.13 2.45±0.12 4.32 counter image 20.41±0.06 21.34±0.07 21.94±0.07 18.88±0.06 17.14±0.06 0.93±0.09 1.52±0.09 0.59±0.10 2.46±0.09 4.20 6 H. Ebeling et al. 4466 4477 0 -2 DeclinationDeclination 4488 4499 Irr Scd -4 −−0088°° 5500′′ Sbc 200 kpc 37.7" Sab E 2200ss 1155ss 1100ss 1122hh 0066mm 55ss RRiigghhtt AAsscceennssiioonn 8000 12000 16000 20000 24000 Figure 7. Iso-intensity contours of the adaptively smoothed X- ray emission from MACSJ1206.2−0847 in the 0.5–7 keV range Figure 6. Broad-band colours of the giant gravitational arc in as observed with Chandra’s ACIS-I detector, overlaid on the MACSJ1206.2−0847, relative to the R band, compared to pre- UH2.2mR-bandimage. Thealgorithm used, Asmooth(Ebeling, dictedcoloursforstandardgalaxytypesatthesameredshift.The White & Rangarajan 2005), adjusts the size of the smoothing bestagreementisfoundforaspiraloftypeSbctoScd. kernelacrosstheimagesuchthatthesignalunderthekernelhas constant significance (here 3σ) with respect to the local background. The lowest contour is placed at twice the value of the 6 INTRA-CLUSTER GAS PROPERTIES X-ray background in this observation; all contours are logarith- Figure 7 shows contours of the adaptively smoothed X- micallyspacedbyfactorsof1.5. ray emission from MACSJ1206.2−0847, as observed with Chandra/ACIS-I in the 0.5–7 keV band, overlaid on the UH2.2m R-bandimage1. (where r = r(φ) is the variable radius of an ellipse with ellipticity ǫand orientation angle Θ),an additional circular At large distances from the center, the system’s X-ray Gaussian component to account for the compact core, and appearanceisclosetosphericalinprojection.Thecentralre- gionatr<250kpc,however,showsapronouncedellipticity a constant background. ∼ Point sources with detection significances exceeding as well as non-concentric X-ray flux contours in the cluster 99% (as measured with the celldetect algorithm) have core. The observed elongation, as well as the displacement been excised from the image, and a spectrally weighted ex- of the innermost contours toward the very compact X-ray posure map is used as a two-dimensional response function core, are both in the direction of a group of galaxies in the in the fit.We usea composite exposure map to account for vicinity of thegiant arc. the differences between photons of cosmic origin and high- energy particles, the latter being subject neither to off-axis 6.1 Spatial analysis vignetting nor to variable detection rates due to spatial or temporalvariationsintheCCDquantumefficiency(QE).At Using Sherpa, the fitting package provided with Ciao, we energiesbetween0.5and2keVparticlesaccountforroughly fittheobservedX-raysurfacebrightnessdistributionwithin 2/3oftheobservedbackground;athigherenergiesthisfrac- 2.5arcminoftheclustercorewithatwo-dimensionalspatial tionrisestoover90percent.Since,overall, morethan80% model, consisting of an elliptical β model of the background events registered in the 0.5–2 keV band r 2 −3β+21 used here are caused by particles we ignore the sky contri- S(r)=S0 1+ , butionaltogetherandcomputeabackgroundexposuremap (cid:20) (cid:16)r0(cid:17) (cid:21) whichincorporatestheeffectsofbadpixelsandditheringbut not those of a variable CCD QE and vignetting. A second exposuremapiscomputedusingweightsbasedonthespec- 1 For this overlay, as well as for any other comparisons of the trumofthetargetofourobservations.Thespectralweights spatial appearance of the cluster inthe X-rayand optical wave- forthis‘cluster-weighted’ exposuremap are created assum- bands, we have used three X-raypoint sources with obvious op- ing a plasma model with kT =12 keV, a metal abundance tical counterparts to slightly adjust the astrometric solution of theX-rayimage,namelyby−0.03secondsinrightascensionand of 0.3 solar, and the Galactic value of 4.23×1020 cm−2 for −0.4arcsecondsindeclination.Weestimatetheresulting,relative theequivalent Hydrogen column density value in thedirec- astrometrybetween theopticalandX-rayimagestobeaccurate tionofthecluster,consistentwiththeresultsofourspectral tobetter than0.2arcseconds. fits to the global cluster X-ray spectrum (see below). The The massive cluster lens MACSJ1206.2−0847 7 resultingexposuremapshowssignificantvignettingofmore than20%acrosstheACIS-Ifieldofview.Thefinalexposure map used in the following is then the weighted average of thebackgroundandclusterexposuremaps,withtheweights 3300″″ being given by the fraction of counts from the cluster and the background, and with the peak value set to that of the cluster-weighted map. 4477′′ 4455″″ Sincetheimageusedinthefitisrelativecoarselybinned ito(nh2fcet×lhutede2letedawsrcoicon-spdeteicmh2pe)eonmaisnniotodd-nsenapllo,reswaspmedatadfiluaoll-nsnmccotaoitloednceosl(npaPvarSoteilFavfil)et..ctAohBmleelcpmpaoauonrsdeaenemltosefwttaeihtrrhees Declination (2000)Declination (2000) 0000″″ low numberof counts(zero or one) in thelarge majority of image pixels, we use theC statistic (Cash 1979) during the 1155″″ optimization process. Our spatial fit yields best-fit values for the parameters oftheβ modelofr0 =(23.6±0.8) arcsec(correspondingto (134±5) kpc) for the core radius, β = 0.57±0.01 for the --0088°° 4488′′ 3300″″ slopeparameter, ǫ=0.17±0.01 fortheellipticity, andΘ= (56±2) degrees (counted North through West). Our two- 1144ss 1133ss 1122ss 1111ss 1122hh 0066mm 1100ss dimensionalspatialfitalsofindscoordinatesofα(J2000) = RRiigghhtt AAsscceennssiioonn ((22000000)) 12h 06m 12.10s,δ (J2000) =−08◦ 48’01.7”forthecentroid ofthecompactclustercore,offsetby(8.1±1.7)arcsecfrom theposition α(J2000) =12h 06m 12.50s,δ (J2000) =−08◦ 48’07.1”whichmarksthecenteroftheellipticalcomponent 3300″″ that describes the shape of the X-ray emission on larger scales. Figure 8 shows the contours of the best-fitting model overlaidbothontheobservedexposure-correctedX-rayim- 4477′′ 4455″″ atetrarhxgamececeWetasetsnededrosstfforqoenofummttoihhtseeestdihcorelneausb,dsidotcaevoutreraarc.lwesosTerrpreehoe.menNodabroiietntnsaeiginidntgthueoaadwtlawhtbiehmiotneuhabttgtheeh3es0ets0m-hefixooptwcdhmeesoslotsaodirsneeclssgl,upieoabtanor-- Declination (2000)Declination (2000) 0000″″ excluded from the fit. Since the two-dimensional fitting procedure allows no 1155″″ immediate assessment of the goodness of fit other than via visual inspection of the residuals (Fig. 8), we also fit a one- dimensional, spherical β model to the radial X-ray surface brightness profile. In this fit we adopt the center of the el- --0088°° 4488′′ 3300″″ liptical model component as the overall center of theX-ray emission, and exclude a 60-degree-wide azimuthal section 1144ss 1133ss 1122ss 1111ss 1122hh 0066mm 1100ss RRiigghhtt AAsscceennssiioonn ((22000000)) around the compact core and the excess emission to the West-North-West.Againweaccountforvariationsintheex- Figure 8. Iso-intensity contours (logarithmically spaced) of the posuretimeacrossthesourceandbackgroundregions.Since best-fittinganalyticmodel(ellipticalβmodelpluscircularGaus- all annuli contain at least 50 photons we are now justified sian component) of the cluster emission in the 0.5–7 keV range, in usingχ2 statistics in the fit. overlaidontheobservedimage(top,logarithmicscaling)andthe The resulting one-dimensional radial profile as well as residualimage(bottom,linearscaling).Thedashedellipseinthe the best-fitting β model are shown in Figure 9. When fit- bottom panel highlights the excess emission in the direction of ting out to a radius of about 150 arcsec (850 kpc), where the arc and marks the region excluded in the spatial fit of the theobservedsurfacebrightnessprofilebeginstodropbelow datatothetwo-dimensionalsurface-brightnessmodel. twice the background level, we find the β model to provide an unacceptablefittothedata at areducedχ2 valueof1.9 (48datapoints,45degreesoffreedom (d.o.f.).However,es- fit, thus allow a credible parametrization of the observed sentially the same model fits the data very well (χ2=1.1 emission out to about 375 kpc, provided the compact core for 19 d.o.f.) within a radius of 66 arcsec (375 kpc). At andtheexcessemissionregiontotheWestareexcluded.At larger radii the observed slope varies such that the best- larger radii, the X-ray morphology of MACSJ1206.2−0847 fittingβ model first systematically underpredicts,then sys- is again too complex to bedescribed by a simple β model. tematically exceedstheobserved values.Thebest-fitvalues Although the one-dimensional model is too simplistic ofS0 =(3.23±0.13)ctarcsec2,r0=(20.1±1.3)arcsec(cor- toadapttothespatialvariationsintheX-rayemissionnear respondingto(114±7)kpc),andβ =0.57±0.02,allofwhich theclustercoreoratverylargeradii,itprovidesanadequate are consistent with the results from our two-dimensional globaldescriptionofthecluster.Extrapolatingthemodelto 8 H. Ebeling et al. r (kpc) r (kpc) 10 100 1000 100 1000 16 -2surface brightness (counts arcsec) 01..1000 kT (keV) 1118024 6 0.01 1 10 100 10 100 r (arcsec) r (arcsec) Figure 9. Radial X-ray surface brightness profile of Figure 10. Radial profile of the ICM temperature in MACSJ1206.2−0847 in the 0.5–7 keV band as observed MACSJ1206.2−0847. The profile is centered on the peak of the with Chandra/ACIS-I. The profile is centered on the peak of ellipticalsurfacebrightnesscomponentasdeterminedinourtwo- the ellipticalsurfacebrightness component as determined inour dimensional fit (see text for details). Horizontal bars mark the two-dimensional fit (see text for details). The solid line shows width of the respective annulus, defined such that each region the best-fitting β model. The dashed horizontal line marks the contains about3,000netphotons. surface brightness corresponding to twice the background level; the vertical dotted line shows the radial limit within which a β modelprovidesanacceptable descriptionofthedata. 450 400 r200 (see Section 8.2) yields values of (4.4±0.07)×10−12 erg s−1 cm−2 and (24.3±0.5)×1044 erg s−1 for the to- 350 talX-rayfluxandX-rayluminosityofMACSJ1206.2–0847, rnpoenorsetlpysaeeconcctutiotviunetnloytviaf(eob0wr.o1auo–ntf2y.1t4shyMeksetpfVaecmc)tf.arTtothimhcaetetrqhXruoe-rorcstaleyuwdsheteimecrrhrisocsareirsnoetndeborio,suahdnnoeddwtettecohvtaebertde,, intensity (a.u.) 235000 [OII] (?) Hδ G band Hγ FeI at larger radii, the β model tends to overpredict the ob- 200 served X-ray surface brightness (see Fig. 9). A more robust measurement can be obtained within r1000 (1.04 Mpc) and 150 CN Ca H+K yields lower limits to thetotal X-rayflux and luminosity of (3.8±0.06)×10−12 erg s−1 cm−2 and (21.0±0.4)×1044 100 5400 5600 5800 6000 6200 6400 erg s−1. λ (Å) Figure11.Low-resolutionspectrumofthecentralclustergalaxy inMACSJ1206.2−0847 asobservedwithCFHT(seetextforde- 6.2 Spectral analysis tails). A tentative detection of OII inemissionishighlighted, as We measure a global temperature for the intra-cluster areaseriesofabsorptionfeaturescharacteristicoftheoldstellar populationdominatingclusterellipticals. medium(ICM)inMACSJ1206.2−0847byextractingtheX- rayspectrumfromr=70kpctor=1Mpc(r1000)andusing Sherpa to fit a MEKAL plasma model (Mewe et al. 1985) core, in agreement with the results of our X-ray imaging with the absorption term frozen at the Galactic value. We analysis. findkT =(11.6±0.66) keV,ahighvalueevenforextremely X-ray luminous clusters (Chen et al. 2007). Although the relatively high reduced-χ2 value of this spectral fit of 1.4 is 7 PROPERTIES OF THE CENTRAL GALAXY statistically acceptable, it could be indicative of systematic effectssuchasspatialtemperaturevariationsorthepresence The spectrum of the central cluster galaxy taken at CFHT ofmultiphasegas.Wefindonlymildevidenceoftheformer (Fig.11,seealsoSection2.1)coversonlyasmallwavelength when fitting absorbed plasma models to the X-ray spectra range mostly redward of the4000˚A break. While thewave- extracted from five concentric annuli. As shown in Fig. 10, lengthcoverageisthusinsufficienttocheckforthepresence the ICM temperature is consistently high (∼12 keV), with ofHβ andOIIIinemission,wedodetectfaintOIIemission the exception of the core region where, at r < 130 kpc, a (λrest =3727˚A),albeitatamuchlowerlevelthantypically significant drop to about 9 keV (still a very high value) is observed in the central cluster galaxies of large cool-core observed. The lower gas temperature measured around the clusters (e.g., Allen et al. 1992, Crawford et al. 1995). clustercorecouldbecausedbythepresenceofaminorcool BrightNVSS(162mJyat1.4GHz)andMolonglo (900 The massive cluster lens MACSJ1206.2−0847 9 X-ray peak are not expected in the simplest cool-core sce- nario, although it is worth noting that an offset of the same size as observed by us here (10 kpc) has been seen in thebest-studied cool-core cluster, Perseus (Böhringer et al. 4477′′ 5500″″ 1993), andtherehasbeenawidespread realization thatthe physics of cool cluster cores are more complex and the role of AGN feedback more important than previously thought (Peterson et al. 2001; Soker et al. 2002; Edge 2001; Mittal etal.2008).WeconcludethatMACSJ1206.2−0847 islikely 0000″″ DeclinationDeclination tminoinccloounuststaerirnadcaoiormegsoadalatexrzayt>e(oc0no.e4o)lo,fcwotrhiteeh,matshoewsteolpblosaewsrevarenfdueldxoitsnrteeusmrbkeanlynocwluens- being likely dueto a recent or still ongoing cluster merger. 1100″″ 8 MASS MEASUREMENTS −−0088°° 4488′′ 2200″″ 8.1 Virial mass Using the measured redshifts of cluster members and their spatial distribution as projected on the sky, we determined 1133..55ss 1133..00ss 1122..55ss 1122..00ss 1111..55ss 1122hh 0066mm 1111..00ss RRiigghhtt AAsscceennssiioonn the virial cluster mass based on the method of Limber & Mathews (1960) in which themass is calculated as Figure 12. As Figure 7 but using the HST image and zoomed intoshowonlytheclustercore. M = 3πσP2RH. (1) V 2 G Here σ is the one-dimensional (radial) velocity dis- P mJyat365MHz)radiosourcesarecoincidentwiththecen- persion and R is theprojected mean harmonic point-wise H tralgalaxy.Theradiosourceexhibitsarelativelysteepspec- separation (projected virial radius). R is defined by H trum (α = −1.32±0.05) and is similar to the one found 1 1 in MACSJ1621.3+3810 (Edge et al. 2003) but an order of R−1= , (2) magnitude more powerful (6×1026 W Hz−1 at 1.4 GHz). H N2 Xi<j |ri−rj| The radio source is also detected at 74 MHz in the VLA where N is the number of galaxies, |r −r | is the pro- Low-frequency Sky Survey (VLSS) with a measured flux of i j jected separation ofgalaxies iandj, andtheij sum isover (6.67±0.7)Jythatisinexcellentagreementwiththepower- allpairs.Beingapairwiseestimatorthisquantityissensitive lawpredictionfromthedetectionsat365MHzand1.4GHz. toclosepairsandquitenoisy(Bahcall&Tremaine1981).It Although no clear signs of cavities are detected in the also systematically underestimates the radius for a rectan- clustercore,wecan,fromourChandradataalone,notcon- gular aperture typical of cluster redshift surveys (Carlberg clusively assess the degree of interaction between the BCG etal.1996).Carlbergetal.thereforeintroduceanewradius and the surrounding intra-cluster medium. High-resolution estimator,theringwiseprojectedharmonicmeanradiusR , radio data would be required to map the radio morphology h given by of the BCG. We do observe though a slight, but significant displacement of (1.7±0.4) arcsec – (9.6±2.3) kpc at the 1 1 2π dθ cglaulastxeyr(rαed(sJh2if0t0–0)b=etw1e2ehn0t6hmep12o.s1it4iso,nδo(fJt2h0e00c)en=tra−l0c8lu◦s4te8r’ Rh−1 = N2 Xi<j 2π Z0 ri2+rj2+2rirjcosθ p 03.3”) and the X-ray centroid of the compact cluster core 1 2 = . (3) (Fig. 12). Having carefully aligned the optical and X-ray N2 π(r +r )K(k ) images (seefootnotein Section6)weestimatethatat most Xi<j i j ij 10% of this offset can reasonably be attributed to residual Herer andr aretheprojecteddistancesfromtheclus- i j astrometric uncertainties; the majority of the misaligned is tercenterto galaxies iand j respectively,k2 =4r r /(r + ij i j i thusreal.Similaroffsets(∼10kpc)havebeennotedinpre- r )2, and K(k) is the complete elliptic integral of the first j vious Chandra studies of galaxy clusters (Arabadjis, Bautz kind in Legendre’s notation. This estimator requires an ex- & Garmire 2002) and may also have contributed to minor plicit choice of cluster center and assumes the cluster is optical/X-ray misalignments in thecores ofcooling clusters spherically symmetric with respect to this center. It treats observed with ROSAT(Peres et al. 1998). one of the particles in the pairwise potential |r −r |−1 as i j In addition, the velocity offset of 550 km s−1 between havingitsmassdistributedinaringaroundtheclustercen- the central galaxy and the cluster mean velocity is com- ter. R is less sensitive to close pairs, less noisy, and toler- h parable to the largest peculiar velocities observed in local ates non-circular apertures better than R . If the cluster H clusters(Zabludoffetal.1990;Hill&Oegerle1993;Oegerle is significantly flattened or subclustered, however, R will h & Hill 2001), although we note that the non-Gaussian ve- systematically overestimatethetrueprojectedvirialradius. locity distribution in this cluster (Fig. 3) complicates the Wecalculate both radiusestimators in ourmass deter- measurement of an accurate systemic velocity. minationstoinvestigatetheresultingsystematicuncertainty Spatial and velocity offsets between central galaxy and inthevirialmass.Foroursamplethevirialradiusandmass 10 H. Ebeling et al. derivedusingR are7%largerthanthosebasedonR .We To model the cluster core we used Lenstool2 (Kneib h H choosetouseR asthemorerobustestimatorinouranalysis et al. 1996; Jullo et al. 2007) which now uses a Bayesian h anddefinethethree-dimensional(deprojected)virialradius MCMC sampler to optimize the cluster mass model and as generate robust lens results. We have followed the proce- π dure of Limousin et al. (2007) to model the mass distribu- r = R . (4) V 2 h tionusingonecluster-scaledark-matterhalodescribedbya truncated PIEMD (Pseudo Isothermal Elliptical Mass Dis- Our determination of the virial radius estimators R H tribution),aswellasanadditional84truncatedgalaxy-scale andR wasmadeusingall62galaxieswithredshiftswithin h PIEMD potentials to describe the dark-matter halos asso- 3σ of the cluster mean (see Section 4). The resulting pro- ciated with the brightest cluster member galaxies selected jected virial radius is R = 1.176 Mpc (R = 1.096 h H Mpc) with a virial mass of 3.861×1015 M⊙ and a three- from the cluster V–K red sequence. Furthermore, to mini- mize the number of free parameters, we assumed that the dimensional virial radius of r =1.847 Mpc. V mass of galaxy-size halos scales with the K-band luminos- ityoftheassociated galaxy(Natarajan &Kneib1997).The 8.2 X-ray mass obtained critical curves shown in figure 13 display a wind- ing shape in between the galaxies (see close-up in Fig. 14), Toestimate thetotal gravitational mass ofthecluster from which explains the extremeelongation of thegiant arc. its X-ray emission, we need to assume that thecluster is in Usingthemostprobablystrong-lensingmassmodel,we hydrostatic equilibrium. In addition, we need a description estimate the mass enclosed by the giant arc. We find M(< of thedensityaswell as thetemperatureoftheintracluster 21′′) = (112.0 ± 5) × 1012 M⊙ and a mass-to-light ratio gas, often assumed to be isothermal. For a cluster with a interior to thegiant arc of M/L=56±2.5. core region as disturbed as the one of MACSJ1206.2−0847 These numbers are very robust and depend little on such simplifying assumptions are unlikely to be justified; the mass profile assumed for the dark-matter distribution however,anisothermalβ modelshouldprovideanadequate of the cluster. Caution is advised though when extrapolat- description of thecluster outskirtsand allow usto obtain a ing to larger radius, as the slope of the cluster mass profile crudemass estimate. is not well constrained by only one (multiple) arc. Deeper, From the X-ray temperature (see Section 6.2) we esti- high-resolution imaging (with, e.g., ACS or WFC3), how- mate the virial radius R200 using the formula of Arnaud et ever, would likely detect a large numberof multiple images al. (2002), aswasthecaseforAbell1703(Limousinetal.2008),allow- R200 = 3.80 βT1/2∆−z1/2 (1+z)−3/2 ing us to accurately measure the slope of the cluster dark- matter profile. Note that the current best estimate of the kT ×( )1/2h−1 Mpc Einsteinradiusatz∼7isnearly45′′,makingthisclustera 10keV 50 superb cosmological telescope to probe the first galaxies in with theUniverse. ∆z =(200Ω0)/(18π2Ωz). The total magnification of the system (both the giant arc and the part of the counter image that is multiply im- Here βT =1.05 (Evrard et al. 1996) is the normalization of aged)isabout80±10,oneofthelargestamplificationfactors the virial relation, i.e., GMv/(2R200) = βT kT. Then, the known for a giant arc. A detailed model of the arc surface totalmassoftheclusterwithinaradiusr canbecomputed brightnessisbeyondthescopeofthisstudybutwillbepre- withthehelpoftheβ-modelprofilediscussedinSection6.1: sented in a futurepaper (Clement et al., in preparation). T r (r/r )2 M(r) = 1.13×1014βkeVMpc1+(r/cr )2M⊙ (5) c 9 SUMMARY AND CONCLUSIONS (Evrard et al. 1996). Using the above equations we find R200 = (2.3±0.1) Mpc as an approximate value for the We present a comprehensive multi-wavelength analysis of virial radius. TheX-rayestimates forthetotal mass within thepropertiesofthemassivegalaxyclusterMACSJ1206.2– R200 and the mass within the core region (defined as the 0847.Ataredshiftofz=0.4385, thesystemactsasagrav- sphere interior to the giant arc, i.e. r < 119 kpc) are then itationallensforabackgroundgalaxyatz=1.04, resulting (17.1±1.2)×1014 M⊙ and(0.46±0.05)×1014 M⊙,respec- in spectacular gravitational arc of high surface brightness, tively. 15′′in length, a total magnitudeof V=21.0, and of unusual, veryredcolourofR−K =4.3.OurX-rayanalysisbasedon 8.3 Lensing mass ChandradatayieldsglobalX-rayproperties(LX =2.3×1045 erg s−1, 0.1–2.4 keV, and kT = 11.6±0.7 keV) that make Using the high-resolution HST images, we have identified this cluster one of the most extreme systems known at any within thegiant arc two features A& Bthat are replicated redshift. six times. The same two features are also identified in the Belying its relaxed appearance at optical wavelengths, southern part of the counter image as shown in Figure 14. MACSJ1206.2–0847 exhibitsmanysigns ofongoing merger The northern part of the counter image is most likely not activity along thelineof sight when looked at more closely, multiply imaged. Altogether, the giant arc and its counter includingadisturbedX-raymorphologyintheclustercore,a image represent a seven-image multiple system which we use to constrain a strong-lensing model of the cluster mass distribution. 2 publiclyavailableathttp://www.oamp.fr/cosmology/lenstool