A&A575,A45(2015) Astronomy DOI:10.1051/0004-6361/201423972 & (cid:2)c ESO2015 Astrophysics The radio relic in Abell 2256: (cid:2) overall spectrum and implications for electron acceleration M.Trasatti1,H.Akamatsu2,L.Lovisari1,U.Klein1,A.Bonafede3,M.Brüggen3,D.Dallacasa4,5,andT.Clarke6 1 ArgelanderInstitutfürAstronomie,UniversitätBonn,AufdemHügel71,53121Bonn,Germany e-mail:[email protected] 2 SRONNetherlandsInstituteforSpaceResearch,Sorbonnelaan2,3584CAUtrecht,TheNetherlands 3 HamburgerSternwarte,UniversitätHamburg,Gojenbergsweg112,21029Hamburg,Germany 4 DipartimentodiAstronomia,UniversitàdiBologna,viaRanzani1,40127Bologna,Italy 5 INAF–IstitutodiRadioastronomia,viaGobetti101,40129Bologna,Italy 6 NavalResearchLaboratory,4555OverlookAve,SW,DC20375Washington,USA Received10April2014/Accepted1November2014 ABSTRACT Context.Radiorelicsareextendedsynchrotronsourcesthoughttobeproducedbyshocksintheoutskirtsofmerginggalaxyclusters. TheclusterAbell2256hostsoneofthemostintriguingexamplesinthisclassofsources.Ithasbeenfoundthatthisradiorelichas aratherflatintegratedspectrumatlowfrequenciesthatwouldimplyaninjectionspectralindexfortheelectronsthatisinconsistent withtheflattestallowedbythetestparticlediffusiveshockacceleration(DSA). Aims.WeaimattestingtheoriginsoftheradiorelicinAbell2256. Methods.Weperformednewhigh-frequencyobservations at2273, 2640, and4850MHz.Combiningthesenewobservationswith imagesavailableintheliterature,weconstraintheradio-integratedspectrumoftheradiorelicinAbell2256overthewidestsampled frequencyrangecollectedsofarforthisclassofobjects(63−10450MHz).Moreover,weusedX-rayobservationsoftheclusterto checkthetemperaturestructureintheregionsaroundtheradiorelic. Results.Wefindthattherelickeepsanunusually flatbehavioruptohighfrequencies. Althoughtherelicintegratedspectrumbe- tween 63 and 10450 MHz is not inconsistent with a single power law with α10450 = 0.92±0.02, we find hints of a steepening atfrequencies>1400MHz.Thetwofrequencyranges63−1369MHzand13696−310450MHzare,indeed, bestrepresentedbytwo differentpowerlaws,withα1369 = 0.85±0.01andα10450 = 1.00±0.02.Thisbrokenpowerlawwouldrequirespecialconditions 63 1369 tobeexplainedintermsoftest-particleDSA,e.g.,non-stationarityofthespectrum,whichwouldmaketherelicinA2256arather youngsystem,and/ornon-stationarityoftheshock.Ontheotherhand,thesinglepowerlawwouldmakeofthisrelictheonewiththe flattestintegratedspectrumknownsofar,evenflatterthanwhatisallowedinthetest-particleapproachtoDSA.Wefindaratherlow temperatureratioofT /T ∼1.7acrosstheGregionoftheradiorelicandnotemperaturejumpacrosstheHregion.However,inboth 2 1 regionsprojectioneffectsmighthaveaffectedthemeasurements,therebyreducingthecontrast. Keywords.galaxies:clusters:general–galaxies:clusters:individual:Abell2256–accelerationofparticles 1. Introduction lifetimeoftheemittingelectronsimpliestheneedforsomeform of in situ productionor (re-)accelerationof the electrons in all A fraction of galaxy clusters exhibit diffuse Mpc-scale syn- thesesources,eventhoughtheunderlyingphysicalmechanisms chrotronemission(referredtoasradiohalosandradiorelics)not are thought to be different for the different classes of sources. related to any particular cluster galaxy (for reviews see Feretti Moreover,thesediffuseradioemittingregionsaremostlyfound etal.2012;Brüggenetal.2012).Thisemissionmanifestsitself inunrelaxedclusters,suggestingthatclustermergersplayakey bythepresenceofrelativisticelectrons(∼GeV)andweakmag- roleinproducingthem. netic fields (∼μG) in the intracluster medium (ICM), together Radio gischt are large, extended arc-like sources, believed with the hot thermal plasma emitting X-rays. Radio halos per- to be synchrotron emission from electrons accelerated or re- meate the central Mpc3 of galaxy clusters and the radio emis- accelerated in merger or accretion shocks through diffusive sionusuallyfollowstheroundishX-rayemissionfromthether- shock acceleration (DSA, Fermi-I process; see Ensslin et al. mal gas. Radio relics are more irregularly shaped and are lo- 1998; Kang & Ryu 2011). A textbook example of such gi- cated at the clusters periphery. They are usually further subdi- ant radio relic has been observed in the galaxy cluster CIZA videdintothreeclasses:radiogischt,radiophoenices,andAGN J2242.8+5301(vanWeerenetal.2010;Stroeetal.2013).Radio relics(seeKempneretal.2004),dependingontheircharacteris- phoenices are believed to be the result of the re-energization ticsandproposedorigin(asdescribedbelow).Thecombination viaadiabaticcompression,triggeredbyshocks,offossilplasma oftheMpcsizeofsuchsourcesandtherelativelyshortradiative fromswitched-offAGNradiogalaxies(Enßlin&Gopal-Krishna 2001;Enßlin&Brüggen2002).TherelativisticplasmaofAGN (cid:3) AppendixAisavailableinelectronicformat origin had the time to age and without the re-energization http://www.aanda.org would not longer be visible at the currently observable radio ArticlepublishedbyEDPSciences A45,page1of18 A&A575,A45(2015) frequencies. An example of a radio phoenixhas been found in spectrumathighenergiesupward,withoutmodifyingthespec- the galaxy cluster A2443 (Cohen & Clarke 2011). AGN relics tral slope. Even steeper spectra are expected in case of AGN are indeed such fossil radio galaxies where the AGN switched relics.TheaveragespectralindicesreportedinTable4ofFeretti off more recently and no re-energizationoccurred.The plasma et al. (2012) for integrated spectra with measured steepening, is still emitting at observable radio frequencies, and it simply rangefrom1.7to2.9. evolvespassivelyuntilitbecomesinvisibleintheradiowindow With these ingredients, detailed studies of the integrated (Komissarov&Gubanov1994). spectrum and of the spectral-index distribution across the sources,allowustotestthecurrentmodelsandstudytheshock propertiesincaseofDSA.Thisisaccomplishedbyobservations Modelsanddiagnostics made over a broad range of frequencies. However,an accurate The proposed formation mechanisms differ in the predictions measurementoftheintegratedspectraofradiorelicsisadifficult of the morphological and spectral characteristics of the differ- task.Thesesourcesusuallycontainanumberofdiscretesources, ent classes of relics. DSA of both thermal and pre-accelerated whosefluxdensityneedstobecarefullysubtractedfromthetotal electrons,shouldproducelargerandmoreperipheralstructures, diffuseemission.Thisrequireshigh-resolutionimagingatmany with strong polarization, and pure power-law integrated syn- frequenciesusingradiointerferometers.However,increasingthe chrotron spectra (Brüggen et al. 2012). Fermi processes natu- observing frequencies, interferometers encounter the technical rally predict an injection power-law energydistribution for the problemofthemissingshortspacingsthatmakesthem“blind” acceleratedelectronpopulationoftheform1 f(E)∝ E−δinj.From toveryextendedstructures.Ontheotherhand,singledishesare synchrotrontheory,theemissionproducedbythispopulationof optimalto catch all the emission from a field but they lack an- electronsisalsodescribedbyapowerlaw2, gularresolution.Indeed,integratedspectraoverawiderangeof frequenciesare available in the literature only for few of these δ −1 objects(seeFerettietal.2012). S(ν)∝ν−αinj with αinj = inj2 · (1) An independentmeasureofthepropertiesofshocksispro- videdbydeepX-rayobservations.Throughthemeasurementsof Emitting particles are naturally subject to energylosses. These temperatureand/orpressurejumpsatthelocationoftheshock, lossesaregovernedbymanyphysicalfactorssuchastheprop- properties such as the shock Mach number M and the shock ertiesofthemagneticfield(seeKardashev1962;Jaffe&Perola compression ratio C can be inferred (see review by Brüggen 1973;Komissarov&Gubanov1994,foradescriptionofthedif- etal. 2012).Inthe testparticlesapproximation3ofDSA, if the ferentmodelsofelectronaging).Theabsenceofanyconstantin- particlediffusionisspecified,theshockMachnumberisthepri- jectionofnewelectronswouldleadtoacutoffinthehigh-energy maryparameterthatdeterminestheefficiencyoftheacceleration regionoftheintegratedspectrum,movingtowardlowerfrequen- mechanismandtheenergydistributionoftheparticlesatinjec- cies in the course of time. The presence of constant injection tion (Kang & Ryu 2010). In this case, a simple direct relation of particles with the same energyspectrum, on the other hand, betweentheshockMachnumber M andtheinjectionindexδ wouldeventuallymaskthecutoff,leadinginsteadtoabreakwith oftheenergyelectronsdistributionexists: inj a changeof0.5inthe spectralindexoftheintegratedemission (continuousinjectionmodel, Kardashev1962): 2(M2+1) δ = · (3) inj (M2−1) α =α +0.5. (2) obs inj However, radio relics are usually observed in the outskirts Thisconditiontranslates,incaseofDSA,intheassumptionthat of clusters where the very low density of electrons (ne < the properties of the shock remain unchanged (stationarity for 10−4 cm−3) make the detection of shocks in the X-ray very theshock).IftheshockhasbeenpresentintheICMfora time challenging(Akamatsu&Kawahara2013).Indeed,afewclear exceeding the electron cooling time, a single power law with X-ray shock detections are known in the literature (see review spectral index α is expected for the integrated radio spec- by Brüggen et al. 2012). In conclusion, multifrequency radio obs trum (stationarity for the spectrum). The observed spectral in- measurements, combined with deep X-ray observations, allow dicesreportedin Table 4of Ferettietal.(2012)forstraightin- asearchforandaproperstudyoftheseshockstotesttheshock- tegratedspectrarangefrom1.1to1.6.Agradientisexpectedin originmodelforrelics. thespectral-indexdistributionacrossthesource,withtheflattest valuesmarkingthepositionoftheshockfrontwheretheparticle ThecaseofAbell2256 getaccelerated,andthesteepeningshowingtheradiativelosses astheelectronsareadvectedawayfromtheshock.Suchagradi- One of the most intriguing clusters hosting both a radio relic entisclearlyobservedintheradiorelicinCIZAJ2242.8+5301 and a radio halo is the galaxy cluster A2256 (z = 0.058). (vanWeerenetal.2010;Stroeetal.2013). The radio relic emission in this cluster differs in many aspects In case of revival via adiabatic compression of old radio from the textbook examples of radio gischt in merging clus- plasma left behind by radio galaxies and pushed towards the ters, e.g.,in CIZA J2242.8+5301(vanWeeren etal. 2010)and cluster outskirtsby buoyancy,we expectinsteadmore filamen- in A3376 (Bagchi et al. 2006). The A2256 relic emission is, tary and smaller radio structures(<50 kpc), again strongly po- indeed, dominated by a complex filamentary structure as con- larized,butwithsteeperandcurvedintegratedspectraduetothe firmed by new wide-band VLA observations published during already aged populationof electrons that are re-accelerated.In the reviewing process of the present paper (Owen et al. 2014). facttheadiabaticcompressionwouldjustshiftthealreadyaged Itis moreovercharacterizedby an unusuallylargeaspectratio, being nearly as wide as it is long, and by an unusual proxim- 1 The distribution is truncated at high energy by the existence of a ity to the cluster center respect to the majority of giant relics maximumenergytowhichelectronscanbeaccelerated. 2 The synchrotron spectrum has an exponential cutoff at high fre- 3 WhenthedynamicalfeedbackoftheCosmicRayselectronspressure quency,reflectingthetruncationintheparticledistribution. isignored. A45,page2of18 M.Trasattietal.:TheradiorelicinAbell2256:overallspectrumandimplicationsforelectronacceleration known in the literature. It also shows all typical signatures of arepotentiallysensitivetoemissionsonscaleupto∼13(cid:6)witha a merging cluster system although its dynamical state is not fullresolutionof∼9(cid:6)(cid:6). yet completely understood. This cluster has been the first ob- The main limiting factor of the field of view is the effect served with LOFAR at very low frequencies (20−63 MHz) by of the primary beam attenuation. For the WSRT this can be vanWeeren etal. (2012).Theycollecteddata upto 1400MHz described by the function cos6(c · ν · r) where r is the dis- and found a radio-integratedspectrum for the relic that can be tance from the pointing center in degrees, ν is the observing described by a power law with an unusual flat spectral index frequencyinGHzandtheconstantc=68is,tofirstorder,wave- α1369 =0.81±0.03.Theoccurrenceofsimilarflatspectralinde- length independentat GHz frequencies(decliningto c = 66 at 63 ceshavebeenreportedbyKale&Dwarakanath(2010)inthefre- 325MHzandc = 63at4995MHz).Theresultingfieldofview quencyrange 150−1369MHz. Assuming stationaryconditions at 2273 MHz is 0.37◦. In order to image a field big enough to inthetest-particlecaseofDSA,thiswouldrequireaninjection recovertheextendedemissioninA2256,theobservationswere spectral index which is not consistent with the flattest possible carried out in the mosaic mode. Three different pointing cen- injectionspectralindexfromDSA.Indeed,adirectconsequence ters were chosen (details in Table 1). In order to have a good ofthetest-particleapproachtoDSAisthatinthelimitofstrong uvcoverageforeachpointing,theobservationswereperformed shocks(M (cid:5)1)theparticleindexδ approachesanasymptotic switchingthetelescopefromonepointingtoanothereveryfive inj valueof2.Thismeansthatparticleenergydistributionproduced minutes,havingfourhoursofobservationsforeachpointingfor bytest-particleDSAcannotbeflatterthan2(itmustbeδ >∼2). a total of twelve hours for the entire cluster. The observations inj Asaconsequencethesynchrotronspectraatinjectioncannotbe werecarriedoutonthe25thJanuary2003.Theexcellentphase flatter than 0.5 (α >∼ 0.5). So, we should not observe relics stability of the system allow us to observe primary calibrators inj with spectra α <∼ 1. The flat spectrum could be reconciled onlyatthebeginningandtheendofanobservationtocalibrate obs withshockaccelerationiftheshockhasbeenproducedveryre- WSRTdata.3C286and3C48wereobservedforthispurpose. cently(∼0.1Gyrago)andstationarityhasnotbeenreachedyet. Flagging,calibration,imagingandself-calibrationwereper- Inthiscasea steepeningoftheintegratedspectrumisexpected formedwiththeAIPS(AstronomicalImageProcessingSystem) atfrequencies(cid:2)2000MHz. package,withstandardproceduresfollowingtheguidelinepro- Inthispaperwepresentnewhigh-frequencyradioobserva- vided on the ASTRON web-page5. All the antennas were suc- tions(Sect.2)ofA2256at2273,2640and4850MHzMHzper- cessful,withsomeoccasionalRFI,flaggedoutintheearlystages formed both with an interferometer (the Westerbork Synthesis ofdatacalibration.3C286wasusedasthemainfluxdensitycal- RadioTelescope,WSRT)andasingledish(theEffelsberg100m ibratorusingtheBaarsetal.(1977)scale(taskSETJYinAIPS), Telescope),complementedbyX-rayobservations(Sect.3)per- whichprovidesfluxdensitiesrangingfrom11.74Jyinthefirst formedwiththe SuzakuandXMM-Newtonsatellites. InSect. 4 IFto11.39JyintheeighthIF.Thethreepointingswereimaged wepresentanewdeterminationoftherelicradiospectrumover and self-calibrated separately. For each pointing we performed thewidestsampledfrequencyrangecollectedsofarforthiskind three phase-only cycles of self-calibration, followed by a final ofobject(63MHz−10450MHz)4.InSect.5weshowtheICM amplitude and phase self-calibration cycle. The diffuse emis- temperatureinregionsacrosstheradiorelicemission.InSect.6 sion flux was included in the model for the self-calibration. A we consider the effect of the thermal Sunyaev-Zeldovich(SZ) multiresolutioncleanwasperformedwithintheIMAGRtaskin decrement on our flux density measurements at high frequen- AIPStobetterreconstructthecomplexdiffuseemissionpresent cies.DiscussionandconclusionsarepresentedinSects.7and8. intheclusterinthefinalimagesofthepointings.Imagesofthe We adopted the cosmological parameters H0 = Stokes parameter I, U and Q were obtained for each pointing 71 kms−1 Mpc−1, ΩΛ = 0.73 and Ωm = 0.27 (Bennett and were then combined together (separately for I, U and Q) etal.2003),whichprovidealinearscaleof1.13kpcarcsec−1at andcorrectedfortheprimarybeamattenuationwiththeFLATN theredshiftofA2256. taskinAIPSprovidingacentralregionwithauniformσnoise distributionof∼0.027mJy/beam.Theprimarybeamcorrection determinesanincreaseofthenoiseintheouterregions. 2. Radioobservationsanddatareduction A2256wasobservedwiththeWSRTat2273MHzandwiththe 2.2.EFFELSBERGobservations Effelsberg100mTelescopeat2604and4850MHz. Inthissectionwepresenttheobservationsandthemainsteps Part of the observations were performed with the Effelsberg of the calibration and image-making process. All the observa- 100 m Telescope. We used the 11 cm (=2640 MHz) and 6 cm tionsincludefullpolarizationinformation.Inthispaperwefo- (=4850MHz) receivers. Single-dish observationsdo not suffer cusonthetotalintensitypropertiesofthecluster.Wepostponea from the zero-spacing problem, and can trace large scale fea- detailedlocalanalysisbasedonpolarizationpropertiesandspec- tures,althoughwithmodestresolution. tralindexmapstoaforthcomingpaper(Trasattietal.,inprep.). The data reduction of Effelsberg data was performed with theNOD2softwarepackage,followingthestandardprocedures provided on the MPIfR web-page6. The raw images of both 2.1.WSRTobservations A2256 and the calibrators were partly processed using dedi- For this project we choose for the WSRT the maxi-short con- catedpipelinesavailableforeach receiver.The defaultstrategy figuration which has optimized imaging performance for very tocalibrateEffelsbergdataistoobserveprimarycalibratorsdur- extendedsources.Thereceivercoversthefrequencyrangefrom ingthesessionandthenuseautomatic2DGaussfitpipelinesto 2193MHzto2353MHzwitheightcontiguousintermediatefre- quencies(IFs)of20MHzwidtheach;theresultingcentralfre- 5 http://www.astron.nl/radio-observatory/ quencyis 2273 MHz and the total bandwidthis 160 MHz. We astronomers/analysis-wsrt-data/ analysis-wsrt-dzb-data-classic-aips/analysis-wsrt-d 4 A very recent paper reports the first observation of a radio relicat 6 https://eff100mwiki.mpifr-bonn.mpg.de/doku.php? 16GHz(Stroeetal.2014). id=information_for_astronomers:user_guide:reduc_maps A45,page3of18 A&A575,A45(2015) Table1.WSRTobservationalparameters. Pointingcenter(J2000) Frequency Bandwidth Exposuretime Telescopeconfiguration RA Dec (MHz) (MHz) (h) 170107.998 +784503.701 2273 160 4 maxi-short 170425.000 +784503.701 2273 160 4 maxi-short 170242.300 +783459.988 2273 160 4 maxi-short Table2. Effelsbergobservationalparameters. Mapcenter(J2000) Frequency Bandwidth Mapsize RA Dec (MHz) (MHz) (‘×’) 170400 +780400 2640 80 48×48 170400 +780400 4850 500 40×35 calculate the factor to scale the final image converting it from the range 4600−5100 MHz. The resulting central frequency is mapunit/beamtoJy/beam(taskRESCALEinAIPS). 4850MHzandthetotalbandwidthis500MHz.Theresolution oftheobservationis2.(cid:6)43 ×2.(cid:6)43. Multihornsystemsuseadifferenttechniquetoovercomethe 2.2.1. Observationsat2640MHz scanning effect problem. The scanning is done in an azimuth- TheEffelsberg11cmreceiverisasingle-hornsystemequipped elevation coordinate system, and must be done only in az- with a polarimeter with eight small-band frequency chan- imuth direction so that all horns will cover the same sky area nels, each 10 MHz wide, covering the frequency range subsequently. At any instant each feed receives the emission 2600−2680MHz,plusonebroad-bandchannel,80MHzwide, fromadifferentpartoftheskybuttheyareaffectedbythesame overthesamefrequencyrange.Theresultingcentralfrequency atmosphericeffects,whichthencancelouttakingthedifference is2640MHzandthetotalbandwidthis80MHz.Theresolution signalbetweenthetwofeeds(Emersonetal.1979).Similarlyto oftheobservationis4.(cid:6)5 ×4.(cid:6)5. the11cmreceiver,datain(R,L,U,Q)areprovidedforeachof TomaptheA2256fieldweusedthemappingmode,which thetwohorn. consist in rastering the field of interest by moving the tele- We performed a total of 25 coverages of the A2256 field, scope, e.g., along longitude (l), back and forth, each subscan 15duringthenightbetweenthe22ndandthe23rdofJuneand shifted in latitude (b) with respect to the other. At centime- 10onthe26thofJune2011.DuetoRFIproblemsonly22cov- terwavelengthsatmosphericeffects(e.g.,passingclouds)intro- eragescouldbeused. duce additional emission/absorption while scanning, leaving a Forthecalibrationweobserved3C286andNGC7027dur- stripy pattern along the scanning direction (the so-called scan- ing the session. The flux densities used for the two calibrators ning effects). Rastering the same field along two perpendicular are7.44Jy(fromBaarsetal.1977)and5.48Jy(fromPengetal. directions(bothalonglongitudeandlatitude)helpsinefficiently 2000)respectively. suppressing these patterns, leading to a sensitive image of the region(Emerson&Graeve1988).Thistechnique,calledbasket- 3. X-rayobservationsanddatareduction weavingtechnique,helpsalsoinsettingthezero-baselevel.The detailsoftheobservationsaresummarizedinTable2.Foreach InthissectionwepresenttheX-rayobservationsofA2256per- coverage of the field, the receiver provides four images (R, L, formed with Suzaku and XMM-Newton and the main steps of U, Q) foreach of the nine channels.As circular polarizationis datareduction. generallyveryweak,theimagesinRandLareverysimilarand Suzaku observed the radio relic region in A2256 (OBSID: canbeaveragedinthelaterstepsofdatareductionprovidingthe 801061010, Tamura et al. 2011) with an exposure time totalintensityimage. of 95.2 ks. The satellite X-ray Imaging Spectrometer (XIS: We performed a total of 14 coverages in the longitude di- Koyamaetal.2007)hasaverylowdetectorbackground,which rection and 15 coverages in the latitude direction; due to RFI allows us to investigate weak X-ray emission targets such as (Radio Frequency Interference) and pointing problems we had clusteroutskirts(seeReiprichetal.2013,forareview).TheXIS todiscardasmallportionofthedata.Thetimerequiredtocom- wasoperatedinthenormalclocking,3×3and5×5mode.To plete one coverage is ∼17 min in both direction, so we have a increasethesignal-to-noiseratiowe filteredthedatasetusinga totalobservingtime on sourceof 8.2 h. The observationswere geomagneticcosmic-raycut-offrigidity(COR)>8GV.Thefil- carried out in the nightbetween the 15th and 16 August2012. tered exposure time is 89.2 ks. The data were processed using 3C286and3C48wereusedasabsolutefluxdensitycalibrators standard Suzaku pipelines (see Akamatsu et al. 2012, for more using the Baars et al. (1977) scale that provide flux densities details). at 2640 MHz of 10.65 Jy and 9.38 Jy respectively for the two We complementedSuzakuobservationswith XMM-Newton calibrators. observations retrieved from the archive (OBSID: 0141380101 and OBSID:0141380201) and reprocessed with SAS v11.0.1. The data were heavly affected by soft proton flares. The data 2.2.2. Observationsat4850MHz werecleanedforperiodsofhighbackgroundduetosoftproton TheEffelsberg6cm receiverisa double-hornsystem,with the flareswithatwostagefilteringprocess(seeLovisarietal.2009, twofeedsfixedinthesecondaryfocuswithaseparationof6cm, 2011,formoredetailsonthecleaningprocess).Inthisscreening each with one broad-band (500 MHz) frequency channel in process bad pixels have been excludedand only eventpatterns A45,page4of18 M.Trasattietal.:TheradiorelicinAbell2256:overallspectrumandimplicationsforelectronacceleration 0−12fortheMOSdetectorsand0forthepndetectorwerecon- whichlabeledinthispaperasI2,I3,K2,G2,J2.Theradiohalo sidered.Thefilteredexposuretimeis∼19ksforMOS1,∼20ks emission present in the center of the cluster around source D forMOS2and∼9ksforpn. (Clarke&Ensslin2006)iscompletelyfilteredoutduetoacom- Forbothsatellitesthebackgroundemissioncanbedescribed binationof effects: its low surface brightnessat this frequency, asthesumofaparticlebackgroundcomponentandaskyback- combinedwiththelackofsampledshortspacingsintheWSRT groundcomponent.Theformerisproducedbytheinteractionof observations,thatdeterminesthelossoftheveryextendedweak high-energyparticleswith the detectors. The latter can be sub- emission. divided into at least two thermal components, one unabsorbed Therelicemissionexhibitstworegionsofenhancedsurface duetotheLocalHotBubble(LHB:kT∼0.08keV)andoneab- brightness:awell-definedarc-likeregion(Gregion)inthenorth- sorbedduetotheMilkyWayHalo(MWH:kT ∼ 0.3keV),and ernpart,andalessdefinedregion(Hregion)inthewesternpart a power-law componentdue to the Cosmic X-ray Background (seeFig.1).Thetworegionsareconnectedbyabridgeoflower (CXB:Γ=1.41). brightnessemission.Atfullresolutionthe entirerelic emission The particle background has been modeled and subtracted coversanareaofabout10.(cid:6)6 ×5(cid:6). ThesizereportedbyClarke fromthedataofbothsatellitesbeforethespectralfitspresented &Ensslin(2006)at1.4GHzandataresolutionof52(cid:6)(cid:6)×45(cid:6)(cid:6)is in Sect. 5.1. For the Suzaku observations its contribution 16.(cid:6)9 ×7.(cid:6)8. ConvolvingourWSRT image tothe same resolu- has been estimated from the Night-Earth database with the tionweobtainasimilarangularsize(notshown).Nevertheless, xisnxbgen FTOOLS (Tawa et al. 2008). For XMM-Newton the as we might be anyway missing some of the flux on the most particle component spectra have been extracted from the filter extendedscales,wedonotusethisimageforthecomputationof wheel closed (FWC) observations and renormalized by using therelicintegratedspectrum. the out-of-field-of-viewevents. These spectra were supplied as backgroundspectratotheXSPECfittingroutine. 4.1.2. EFFELSBERGimages Unliketheparticlebackground,theskybackgroundwasnot In Fig. 2 we show the 2640 MHz (left panel) and 4850 MHz subtractedfromthedatabutitsdifferentcomponentsweremod- (right panel) Effelsberg images of A2256. In both images, the eledtogetherwiththeICMemissionduringthespectralfitting. diffuse emission from the relic is mixed up with the emission To fix the model parameters for the different components in from the more compact sources present in the field due to the Suzakuobservations,weusedspectraextractedfroma1degree lowresolutionoftheobservations.Inthe2640MHzimagethe offsetobservationperformedwiththesatellite(PI:Kawaharada, relic emission is blended with the emission from the A+B+C OBSID: 807025010).For XMM-Newton data, we followed the complexandfromsourceF.At4850MHzitiseasiertoseparate methodpresentedinSnowdenetal. (2008)inwhichthediffer- the relic emission from the emission of the A+B complex, but ent componentsare estimated using a spectrum extracted from still part of the tail of source C is inevitably superimposed on ROSATdatainanannulusbeyondthevirialradiusofthecluster. thewesternpartoftherelic.Moreover,Brentjens(2008)derived Theoffsetspectrawerefittedwithaskybackgroundmodelcon- a spectrum α = 1.5± 0.2 for the radio halo between 351 and sidering the LHB, MWH and CXB components. In the fitting, 1369MHz.Ifthereisnosteepeninginthespectrumofthehalo, we fixed the temperature of the LHB component to 0.08 keV. we expect a flux density of ∼37 mJy at 2640 MHz and a flux Abundance(Anders&Grevesse1989)andredshiftofLHBand densityof∼15mJyat4850MHz.Sinceasingledishissensitive MWHwerefixedto1and0,respectively.Thetemperatureofthe toalltheemissioninthefield,itsemissionissmoothedwiththe MWHdeterminedinthefitis0.21±0.03keV.Wealsochecked othersourcesinourEffelsbergimagesifthehalospectrumkeeps forthepossibilityofanadditional“hotforeground”component straightat these frequencies.Constrainingthehalo spectrumat withkT ∼0.6−0.8keV(Simionescuetal.2010)addinganother highfrequencieswouldrequireadedicatedcarefulanalysisthat thermal component to the background model described above. isbeyondtheaimsofthispaper. However,theintensityofthisadditionalcomponentresultednot significantin the offset field and was notincludedin the back- 4.2.Spectralanalysis groundmodeling. For the spectral analysis of the radio relic in A2256 we com- binedourhigh-frequencyobservationswith imagesobtainedat 4. Radioanalysisandresults otherfrequenciesprovidedbytheauthors:the351MHzWSRT image (Brentjens 2008), the 1369 MHz VLA C and D con- 4.1.Radioimages figuration images7 (Clarke & Ensslin 2006) and the Effelsberg 4.1.1. WSRTimage 10450 MHz image (Thierbach 2000). We moreover got infor- mation on the flux densities in the LOFAR image at 63 MHz In Fig. 1 we present the 2273 MHz total intensity WSRT im- published by van Weeren et al. (2012) by the author. All the age of the central region of A2256. The image has been pro- different images were calibrated according to the flux scale of duced with natural weighting of the visibilities in the range [u −u ] = [260−21035λ].Theshortestspatialfrequency Baars et al. (1977) or to its extension to lower frequencies min max (below 408 MHz; Perley & Taylor 1999)8. In this way sampled u determines the largest spatial scale recovered by min this observationsLSS (cid:8) 1/u = 13.(cid:6)22 . The image has been we were able to cover the widest range of frequencies min corrected for the primary beam attenuation that determines an 7 Weusedthehigh-resolutionCconfigurationdatatomeasuretheflux increase of the noise in the edge of the image. The high reso- lution (9(cid:6).(cid:6)84 ×9(cid:6).(cid:6)44 ) allowed us to analyze the substructures densityofdiscretesourceswhileforthedeterminationoftherelicflux densityweusedtheDconfigurationimage. of the diffuse relic emission in detail. The map shows several 8 TheoverallfluxscalefortheLOFARobservationsat63MHzwas ofthewell-knownradiofeaturespresentinthecluster(notation obtainedcomparingthemeasuredintegratedfluxdensitiesoffivebright from Bridle et al. 1979;Rottgeringet al. 1994):the radio relic sourcesinthefieldofview withpredictedfluxespartlybasedonflux emission(sourcesGandH),thehead-tailsourcesA,B,CandI, densitiesmeasurementsfromthe1.4GHzNVSS(Condonetal.1998) the complex source F (here resolved in the three components and the 74 MHz VLSS (Cohen et al. 2007), both based on the Baars F1,F2andF3),aswellasmanyotherdiscretesources,someof etal.(1977)scale. A45,page5of18 A&A575,A45(2015) Fig.1.WSRT2273MHztotalintensityradioimage.Contoursaredrawnat[1,2,4,8]×3σ,with3σ=8×10−5Jy/beamandthecolorscalestarts atthesamelevel.Thebeamsizeis9(cid:6).(cid:6)84×9(cid:6).(cid:6)44.Theimageiscorrectedfortheprimarybeamattenuation. (63 MHz−10.45GHz) used so far for the determination of the were finally averaged, separately at the two frequencies. The spectrumofaradiorelic. standarddeviationof these values(at the two frequencies)was We estimated the uncertaintiesσ on the fluxdensitymea- used as the term (cid:8) . For the WSRT data we used, instead, the S cal surementsS withthefollowingformula dispersion of antenna gains. This translates into (cid:8) values of cal (cid:2) 1.6%for the Effelsberg 11 cm data, of 1.2%for the Effelsberg σ = σ 2+σ 2, (4) 6 cm data and of 2% for the WSRT 13 cm data. Combining S rms cal this with the uncertainties on the Baars et al. (1977) scale, we where end up with E (11 cm) = 5.3%, E (6 cm) = 5.2% and cal cal √ E (13 cm) = 5.4%. For the other images from the literature – σrms =σ× Nbeamistheerrorduetotheimagenoise,withσ wcealassumed a similar value E = 6%. It shouldbe noted that cal beingtheimagenoiselevel(quotedintheimage’scaptions) theself-calibrationprocessmightaffectthemeasuredfluxden- andN thenumberofbeamscoveredbythesource; beam sitiesoninterferometricimages.Theuncertaintiesaretherefore – σ = E ×S is theerrordueto calibrationuncertainties, cal cal possiblylargerthanestimatedhere. determinedinturnbytwofactors:theaccuracyoftheabso- lute flux density scale adopted ((cid:8) ) and the uncertainties Where nototherwise specified,we estimated the uncertain- scale related to the applicationof such scale to our data (the cal- tiesσf onthequantities f calculatedfrommeasuredquantities ibration method(cid:3), (cid:8)cal); being these two factors uncorrelated (e.g.,spectralindicesandextrapolatedfluxdensitiesofdiscrete sources) applying the standard error propagation formula. The weusedEcal = (cid:8)s2cale+(cid:8)cal2. spectraofthetotalcluster,relic+sources,relic,GandHregions presentedinthenextsectionshavebeendeterminedcalculating The spectral data provided by Baars et al. (1977) for the flux the flux densities at different frequencieson images convolved density calibratorshave an absoluteuncertaintyof 5%. For the to the same lowest resolution available (4.(cid:6)4 × 4.(cid:6)4) and using calibration of the Effelsberg images we performed four cover- thesameregionsfortheintegration.Theerrorsassociatedtothe agesof3C286andthreecoveragesof3C48at2640MHzand spectralindicesaretheerrorsfromthefitsofthedatatakinginto four coverages of 3C286 and three coveragesof NGC7027 at accounttheuncertaintiesinthefluxdensitymeasurements. 4850MHz.AGaussianfitoftheimageofacalibratorprovides its flux density in map unit. A comparison of this value with All the spectra, including those of the discrete sources theknowncalibratorfluxdensityallowsustocalculatethefac- embedded in the relic emission, are plotted over the same tor to translate map unit in physicalunit. The slightly different fixed x-axis range (frequency range 40−14000 MHz) for easy factors deriving from the different coverages of the calibrators comparison. A45,page6of18 M.Trasattietal.:TheradiorelicinAbell2256:overallspectrumandimplicationsforelectronacceleration Fig.2.Effelsbergtotalintensityradioimages.Leftpanel:Effelsberg2640MHztotalintensityradioimageincolorscaleandblackcontoursdrawn at[–1,1,2,4,8,16]×3σ,with3σ=4×10−3Jy/beam.Thebeamsizeis4.(cid:6)4 ×4.(cid:6)4.Rightpanel:Effelsberg4850MHztotalintensityradioimage incolorscaleandblackcontoursdrawnat[–1,1,2,4,8]×3σ,with3σ=1.86×10−3Jy/beam.Thebeamsizeis2.(cid:6)43 ×2.(cid:6)43.Inbothpanelsthe bluecontoursarefromtheWSRT2273MHzradioimagenotcorrectedfortheprimarybeamattenuationandaredrawnat[1, 2]×8×10−5Jy/beam. 4.2.1. Totalclusteremission Wefirstconsideredtheintegratedradiofluxdensityoftheentire cluster (halo, relic and discrete sources combined) measuring the flux density in the circular region centered at J2000 posi- tion α = 17 03 45 δ = +78 43 00 with 10(cid:6) radius, as de- scribedbyBrentjens(2008).Theclusterradiuswasdetermined by Brentjens (2008) as the one at which the derivative of the integratedfluxwithin thecircle respectto the radiusofthe cir- cle settles to a constant value. The measured flux densities at 351, 1369 (D configuration), 2640, 4850 and 10450 MHz are summarizedinTable4.Thetotalclusterradioemissionbetween 351and10450MHzcanbemodeledbyasinglepowerlawwith spectralindexα=1.01 ± 0.02(Fig.10a). Brentjens(2008)modeledtheclusterfluxdensityasthesum oftwospectralcomponents,oneduetothehaloandtheotherdue to the relic anddiscrete sources combined.He showedthat the secondtermbecomesdominantatfrequenciesabove100MHz. Fig.3.Regionsusedforthespectracomputation.Theredregionmarks Beingtheradiorelicthedominantcontributortothefluxdensity the region considered for the relic spectrum computation. The to- at high frequency,its spectrum at the same frequencies cannot tal region is further subdivided into regions G and H. In color scale be flatter than the total cluster emission spectrum. This shows the WSRT high-resolution image with the green contours from the qualitativelythatathighfrequenciestherelicspectrumdoesnot 2640 MHz Effelsberg image overplotted. The white region mark the Csource. keep the 0.85 slope observed at low frequency by van Weeren etal.(2012).InSect.4.2.3wewillquantifythissteepening. differentcomponentsof source C and for the sourcesJ and I2. Whereavailable,wecombinedourdatawithdatacollectedfrom 4.2.2. Discretesources the literature andwe modeledthe integratedspectra of the dis- Figure 3 shows the area selected for measuring the radio relic cretesourcesoverthefrequencyrange63−10450MHz.Allthe flux densities. From the high-resolution image it is possible to measuredandextrapolatedfluxdensities,aswellasthespectral see which are the discrete sources included in such area: the indices of the source C components with references are listed tail of source C C , K, J, I, G2, K2, J2, I2 and I3. To esti- in Table 3. Details on the flux densities derivation are given in tail matethefluxdensityfromthediscretesourcesthatneedstobe Appendix A. The spectra of the sources C, K, J, I and I2, for subtractedfromthe totalemission, we producedtwo imagesat whichmorethantwofluxdensitymeasurementswereavailable, 1369MHz(VLAC configuration)and2273MHz(WSRT) us- are plotted individually in figures from 4 to 8 and all together ingthesameuv-range(262−15460λ),pixel-sizeandrestoring inredinFig.9.ForthesourcesK2,J2,I3andG2onlyourflux beamandwecalculatedthespectralindexofthediscretesources densitymeasurementsat1369and2273MHzwereavailableand embeddedinthe relicemissionbetweenthesetwo frequencies. we simplyassumed straightspectra.Thisassumptionmaylead Moreoverwe measured the flux density at 10450 MHz for the toaslightoverestimateofthefluxdensitiesatbothlowandhigh A45,page7of18 A&A575,A45(2015) sa em c ne 10450MHzS(mJy) (cid:10)a1<d1±a0.80.2<d1(cid:10)a1(cid:10)a1(cid:10)a1(cid:10)a1 onsultrefereRightsite,Int c 10450MHzS(mJy) <±d4.00.9±d2.10.6<±d1.90.7 8500MHzgS(mJy) (...)(...)±1.20.4(...)(...)(...)(...)(...) FormoredetailscedfromFig.5.4 4850MHzS(mJy) ±a11.71.3±a6.00.9±a5.71.5 4900MHzgS(mJy) (...)(...)±2.10.5(...)(...)(...)(...)(...) heliterature.e()ork.Dedu tw 2640MHzS(mJy) ±a27.31.4±a14.00.8±a13.31.6 10450α2273 ±d1.400.14±d1.400.19±d1.420.25 4850MHzaS(mJy) ±0.80.1<1.2±2.40.3<1.4±1.00.1±0.70.1±1.10.1±0.60.1 orpublishedineasuredinthis kM 2273MHzS(mJy) ±d33.71.8±d17.30.9±d16.40.9 2273α1369 ±d1.020.16±d0.720.15±d1.300.16 2640MHzaS(mJy) ±1.60.2<1.5±5.50.3<1.8±1.70.1±1.00.1±1.80.1±1.00.1 herinthisword()al.(1994). Cfluxdensities. 351MHz1369MHzS(mJy)S(mJy) ±±ad228.717.756.53.4±±ad60.65.125.01.5±±ad168.118.431.51.9 (b)SourceCspectralindices. 3271369αα327153 ±±dd0.880.171.030.07±±ef0.500.100.650.10±±aa1.000.221.230.10 (c)Othersourcesfluxdensities. 1446MHz2273MHzcdSS(mJy)(mJy) ±±3.40.31.90.2±±2.10.31.70.2±±8.21.86.80.4±(...)1.90.1±(...)2.00.1±(...)1.10.2±(...)2.00.1±(...)1.10.1 normalfontaremeasuredvalueseitc()2012).MeasuredbyRottgeringetdbyLinetal.(2009). Table3.Propertiesoftheradiosourcesembeddedintheradiorelicemission. (a)Source Component63MHz153MHz327MHzS(mJy)S(mJy)S(mJy) ±±±abc1048256ALL4805024620±±±aaa144.526.992.710.763.45.5HEAD±±±aaa903.5257.0387.351.1182.620.7TAIL 153αComponent63 ±a0.880.17ALL±a0.500.10HEAD±a0.950.35TAIL Source63MHz327MHz351MHz1369MHzaadSSSS(mJy)(mJy)(mJy)(mJy) ±±±±c16.33.06.70.1K7.01.03.30.2<<±±c0.651.00.1J11.20.1±±±±a32.220.317.08.916.48.7I9.30.8(cid:10)(cid:10)(cid:10)±a111I21.00.07±±±56.027.811.32.6K2(...)3.20.19±±±7.963.01.0J2(...)1.40.2±±±35.218.68.92.1I3(...)3.00.2±±±23.913.75.51.4G2(...)1.70.1 Notes.italicValuesinareassumedorextrapolatedinthiswork,whilevaluesinab()()belowandtext.Extrapolatedinthiswork.MeasuredbyvanWeerenetal.(fg()()Rightsite(2009).DeducedfromFig.7,Rottgeringetal.(1994).Measure A45,page8of18 M.Trasattietal.:TheradiorelicinAbell2256:overallspectrumandimplicationsforelectronacceleration 0.01 C all 1 C head ] C tail y J [ y sit n e d x α ≥ 0.35 u α = - 0.25 ± 0.15 Fl 0.001 0.1 ] y J 100 1000 10000 [ y Frequency [MHz] sit n e Fig.6.IntegratedradiospectrumofsourceJ.Filledcirclesaremeasured d x fluxdensitieswhileopencirclesareextrapolatedfluxdensities. u Fl 0.01 α = 0.39 ± 0.21 ] y J [ 0.001 sity 0.01 α = 0.69 ± 0.10 n e d 100 1000 10000 x Frequency [MHz] Flu α = 1.38 ± 0.12 Fig.4. Integrated radio spectrum of source C. Filled circles are mea- suredfluxdensitieswhileopen circlesareextrapolated fluxdensities. Theblacksolidlineisthespectrumoftheentiresource;thereddashed 0.001 line is the source’s head spectrum; the blue dotted-dashed line is the spectrumofthetail.Seetextformoredetails. 100 1000 10000 Frequency [MHz] Fig.7.IntegratedradiospectrumofsourceI.Filledcirclesaremeasured fluxdensitieswhileopencirclesareextrapolatedfluxdensities. α = 0.52 ± 0.07 ] y 0.01 J [ y 0.01 sit ] n y e J Flux d α = 1.09 ± 0.02 ensity [ d 0.001 x α ≥ 0.42 u α = -1.27 ± 0.20 Fl 0.001 100 1000 10000 Frequency [MHz] 100 1000 10000 Frequency [MHz] Fig.5. Integrated radio spectrum of source K. Filled circles are mea- suredfluxdensitieswhileopencirclesareextrapolatedfluxdensities. Fig.8. Integrated radio spectrum of source I2. Filledcircles are mea- suredfluxdensitieswhileopencirclesareextrapolatedfluxdensities. frequencies(asforstandardradiosourcesynchrotronspectrawe long at high resolution, ∼410(cid:6)(cid:6) in total in our 2273 MHz im- expect a flattening at low frequenciesand a steepening at high age (see Fig. 3). The published flux densities in the literature frequencies).Ontheotherhand,theyareweaksourcesandtheir refer to the source as a whole. We modeled the source distin- flux densities are not crucial for the relic spectrum determina- guishingbetweenthehead(long76(cid:6)(cid:6))andthetailaswearein- tion.TheirspectraareplottedinblueinFig.9.Thisfigureshows terested only on the contribution from the latter. In Fig. 4 the thespectraofthediscretesourcesincludedintheregionselected, spectrum over the range 63−10450 MHz is plotted for the en- comparedtothetotalfluxdensityintheregion(relic+sources). tiresource,thehead,andthe tail.Ascommonamonghead-tail The main flux density contribution among the discrete sources sources,theheadisflatterthanthetailandthespectrumsteepens in the relic area come from the tail of source C, especially at at high frequency for both components. At lower frequencies, lower frequencies. The source appears noticeably narrow and thetailcontainsalmostalltheemissionofthesources.Athigh A45,page9of18 A&A575,A45(2015) relic+sources Table4.Totalclusterfluxdensities. relic+sources 10 relic+sources C tail ν(MHz) S(mJy) K J 351 3320.0±200.0 I 1369 928.6±57.0 I2 2640 459.0±24.8 G2 K2 4850 246.3±13.3 J 10450 107.8±7.5 1 I3 selected area is shown in Fig. 3. The region is further divided intotworegionsofenhancedradiobrightness(regionsGandH) ] discussedinthenextsection.Asdiscussedintheprevioussec- y y [J tion,thisregionincludetheradiorelicandsourcesCtail,K,J,I, sit G2,K2,J2,I2andI3. n 0.1 We first considered the total emission from the region. e x d Althoughthefluxdensitymeasurementsoftherelic+sourcesin Flu therange63−10450MHzcanbefittedwithasinglepowerlaw withα10450 = 0.93±0.02(Fig.9),hintsofasteepeningatfre- quenci6e3s >1400 MHz are present. A separate fit of the spectra between63and1369MHzandbetween1369and10450MHz shows indeed that these two frequency ranges are best repre- 0.01 sented by two different power laws, with α1369 = 0.86± 0.01 63 and α10450 = 1.02±0.02 (Fig. 9). All the fits are plotted over theen1t3ir6e9range63−10450MHztohighlightdifferences.Since the relic is the major contributor in the region both at low and highfrequency,thissuggeststhatasteepeningmightbepresent in its spectrum as well. Moreover the relic spectrum between 0.001 63and1369MHzmustbeα(relic)1369 ≤ 0.86±0.01asitcan- 63 notbesteeperthantherelic+sourcesspectrum.Similarlyitmust 100 1000 10000 be α(relic)10450 ≥ 1.02 ± 0.02 as it cannot be flatter than the Frequency [MHz] relic+sourc1e3s69spectrum. Indeed, after discrete sources subtrac- Fig.9.Integratedradiospectraofthedifferentcomponentsintherelic tion we find that, although the relic spectrum between 63 and 10450 MHz is not inconsistent with a single power law with regionshowninFig.3.Inblackweshowthefluxdensitiesandspec- traoftheemissionfromtheentireregion(relic+sources).Thedashed α(relic)10450 =0.92±0.02,itisbestrepresentedbytwodifferent 63 (α10450 = 1.02 ± 0.02) and dot-dashed (α1369 = 0.86 ± 0.01) lines powerlaws,withα(relic)1369 = 0.85±0.01andα(relic)10450 = rep13r6e9sent a double power-law fit to the data6.3The solid line (α10450 = 1.00±0.02 (Fig. 10b). T6h3is is supported by a lower va1l3u6e9s of 0.93±0.02)isasinglepower-lawfit.Inredandblueweshowth6e3spec- thereducedχ2inthecaseofthedoublepowerlaw(χ2 =0.19) traofthediscretesourcesincludedintheregion.Redspectraareplotted respecttothesinglepowerlaw(χ2 =0.92).Themearseudredrelic individuallyinfiguresfrom4to8.Bluespectraarestraightpower-law red fluxdensities(beforeandafterdiscretesourcessubtraction)are fitstothemeasuredfluxdensitiesat1369and2273MHz.Seetextfor summarizedinTable5.Thefluxdensitiesmeasuredat63MHz, moredetails. 351MHzand1369MHzareinagreementwithintheerrorbars withthealreadypublishedfluxdensitiesatthesamefrequencies. Thelow-frequencyspectralindexisinagreementwithwhatwas frequenciesthe electrons in the tail are older while the head is foundbyvanWeerenetal.(2012). moreclearlyvisible. Clarke & Ensslin (2006) reporta relic mean spectral index Sources J and I2 have an inverted-spectrum. Their spectra between1369MHzand1703MHzof1.2.However,thisvalue (plottedinFigs.6and8)haveaconvexshapeatGHzfrequen- hasabiguncertainty(notquotedbyauthor)sinceitisderivedas cies,typicalofinverted-spectrumsources,andarelikelyyoung themeanvaluefromthespectralindeximagebetweentwovery radioobjects. closefrequencies.Wecannotthereforeexcludeitisinagreement with our value. Moreover, the very recent JVLA observations by Owen et al. (2014) reports for the relic an overall intensity 4.2.3. Radiorelicspectrum weighted spectral index in the L-band of ∼0.94, in agreement To avoid thatresolutioneffectsat differentfrequenciesmay al- withourfinding. ter the determination of the relic integrated spectrum, we con- volvedalltheimagesusedforthefluxdensitiescomputationto 4.2.4. RegionsGandH thesameresolutionof4.(cid:6)4×4.(cid:6)4.Atsuchlowresolution,itisnot easytoidentifytheregionwheretomeasuretherelicfluxden- Our high-resolution image shows that the relic can be divided sity. We adopted the following strategy: we first computed the into two separate parts: regions G and H in Fig. 3. The cases relic flux density from the 4850 MHz image at full resolution, of double relics in the same cluster are getting more and more whereitiseasiertoseparatetherelicemissionfromthecomplex commonsincethefirstdiscoveryoftwoalmostsymmetricrelics A+Bemission.Wethenchoosetherelicregioninthe4850MHz located on opposite sides in A3667. Since then, several other imageconvolvedtotheresolutionof4.(cid:6)4×4.(cid:6)4,matchingtheflux doublerelicssystemshavebeenfound(seeFerettietal.2012). density measured at full resolution at the same frequency.The We haveinvestigatedthepossibility thatthe twodifferentparts A45,page10of18