Table Of Content

Mon.Not.R.Astron.Soc.000,1–26(0000) Printed5February2008 (MNLaTEXstylefilev2.2) X–ray Scaling Properties of Early–type Galaxies Ewan O’Sullivan⋆, Trevor J. Ponman, Ross S. Collins SchoolofPhysicsandAstronomy,UniversityofBirmingham,Edgbaston,BirminghamB152TT Accepted2003??.Received2002??;inoriginalform2002?? 3 ABSTRACT 0 We presentananalysisof39X–rayluminousearly–typegalaxiesobservedwiththeROSAT 0 PSPC.Usingmulti–componentspectralandspatialfitstothesedatawehavemeasuredhalo 2 abundance,temperature, luminosity and surface brightnessprofile. We compare these mea- n surements to similar results from galaxy groups and clusters, fitting a number of relations a commonlyused in the study of these larger objects. In particular,we find that the σ:TX re- J lation for oursample is similar to that reportedfor clusters, consistent with β = 1, and spec 9 thattheLX:TX relationhasasteepslope(gradient4.8±0.7)comparablewiththatfoundfor galaxygroups.Assumingisothermality,weconstruct3-dimensionalmodelsofourgalaxies, 1 allowingustomeasuregasentropy.Wefindnocorrelationbetweengasentropyandsystem v mass,butdofindatrendforlowtemperaturesystemstohavereducedgasfractions.Wecon- 3 cludethatthegalaxiesinoursamplearelikelytohavedevelopedtheirhalosthroughgalaxy 5 winds,influencedbytheirsurroundingenvironment. 1 1 Keywords: galaxies:ellipticalandlenticular–galaxies:halos–X-rays:galaxies 0 3 0 / h p 1 INTRODUCTION in more massive systems gas entropy scales with the total mass, - in groups it appears to reach a roughly constant minimum level. o Early-typegalaxieshavebeenknowntopossesslargehalosofhot A number of models have been put forward to explain this be- r gassincethedetectionbyEinsteinofX–rayemissionfromtheel- t haviour,includingraisingtheentropythroughtheinjectionofen- s lipticalpopulation intheVirgocluster (Formanetal. 1979). Suc- ergy by AGN (Wuetal. 2000) or star formation (Ponmanetal. a cessivegenerationsofX–rayobservatorieshavebeenusedtoob- 1999),orthroughtheradiativecoolingandremovaloflowentropy v: serve these galaxies, and the advent of XMM-Newton and Chan- gas(Muanwongetal.2001).Itisnotablethatforanyofthesepro- i drahasallowedthecomplexnatureoftheiremissiontobestudied X cesses,theoriginoftheentropyrisewouldberelatedtothegalax- in detail. Most of the work in this area has focused on the vari- iesinthesystem.Asearly–typegalaxiespossesstheirownhalos, r oussourcesofemissionwithinearly–typegalaxies(hotgas,X–ray a wemightexpecttoseeevidenceoftheseprocessesintheirX–ray binaries, AGN) and on the surprisingly complicated relation be- properties,andfortheeffecttobestrongestinthesesystems,owing tweenopticalandX–rayluminosity.However,ataquitefundamen- totheirpositionatthebottomofthemassscale. tal level, early–type galaxies resemble the groups and clusters in There is also evidence from previous studies of early–type whichtheytypicallyreside.Simulationsofdarkmatterhalossug- galaxies(Helsdonetal.2001;O’Sullivanetal.2001a)that galax- gestthattheyhavesimilarprofilesatallmassscales(Navarroetal. iesinthecentresofX–raybrightgroupsareaffectedbytheiren- 1997). If we consider clusters, groups and galaxies as potentials vironment.TheyhaveasignificantlysteeperL :L relation,and containinghotgas,wemightexpectthepropertiesofthehalosto X B areonaverageconsiderablymoreluminousthannormalellipticals. besimilaracrossawiderangeofmasses. A large fraction of group dominant galaxies in one sample have A comparison between halos on different scales becomes beenshowntohavetemperatureprofilesindicativeofcentralcool- increasingly interesting considering the importance of entropy ing(Helsdonetal.2001),leadingtothesuggestionthattheirhalos changes in governing the behaviour of group and cluster halos. areactually the product of cooling flowsassociated withthe sur- Observations of galaxy groups have demonstrated that these sys- roundinggroup.Consideringthedifferencesbetweenthesedomi- tems do not behave as might be expected from scaling clusters, nantgalaxiesandtheirmorenormalcounterparts,andthebiasing but instead require non-gravitational processes. One of the clear- effect their inclusion in samples of early–type galaxies seems to estsignsofthisistheentropyfloor(Ponmanetal.1999).Whereas have,furtherinvestigationoftheprocesseswhichhaveshapedtheir halosseemswarranted. ⋆ Presentaddress:Harvard-SmithsonianCenterforAstrophysics,60Gar- We have compiled a sample of 39 large, X–ray luminous denStreet,Cambridge,MA02138,USA early–typegalaxiesforwhichthereisgoodqualityROSAT PSPC Email:[email protected] dataavailable.Wehaveanalysedthesedata,andfittedtwodimen- (cid:13)c 0000RAS 2 Ewan O’Sullivanetal. sional,twocomponentsurfacebrightnessprofilestothem.Wehave 3 DATAREDUCTION alsofittedtwocomponentspectralmodels,temperature,abundance DatareductionandanalysisoftheX–raydatasetswerecarriedout and hardness profiles and produced three dimensional models of thegalaxies.Theseallowustomodeloutcontaminationfromsur- usingtheASTERIXsoftwarepackage.Beforethedatasetscouldbe used,varioussourcesofcontaminationhadtoberemoved.Possible rounding cluster or groupemission andthediscretesourcepopu- sources include charged particles and solar X–rays scattered into lationwithinthegalaxy.Wecanthereforeexaminetheproperties thetelescopefromtheEarth’satmosphere. Onboard instrumenta- ofthehaloindetail,forthefirsttimeinasampleofthissize.We tionprovidesinformationwhichallowsperiodsofhighbackground canalsocomparethebehaviour of thissampletothatof samples tobeidentified.Themastervetocounterrecordsthechargedparti- ofgroupsandclustersthroughrelationsbetweenparameterssuch cleflux,andweexcludedalltimeperiodsduringwhichthemaster astemperature,opticalandX–rayluminosity,velocitydispersion, veto rate exceeded 170 count s−1. Solar contamination causes a surfacebrightnessslopeandgasentropy.Inmostcases,thisisthe significantoverallincreaseintheX–rayeventrate.Toremovethis first timethese relations have been studied for halos at thismass contamination we excluded all times during which the event rate scale. The paper is organised as follows. In Section 2 we describe deviatedfromthemeanbymorethan2σ.Thisgenerallyremoved nomorethanafewpercentofeachdataset. our sample and the selection criteria used to create it. Section 3 gives details of the techniques used in reduction of the ROSAT Afterthiscleaningprocesseachdatasetwasbinnedintoa3– PSPC data, and Section 4 describes the spectral and spatial fit- dimensional(x,y,energy)datacube.Spectraorimagescanbeex- ting processes. Our results are presented in Section 5, with data tractedfromsuchacubebycollapsingitalongtheaxes.Amodel fromclustersandgroupsofgalaxiesincludedforcomparison.We ofthebackground wasthengenerated based onanannulus taken discuss the results and their implications in Section 6, and give fromthiscube.Weusedannuliofwidth0.1◦,andinnerradius0.4◦ our conclusions in Section 7. Throughout the paper we assume wherepossible.Incaseswherethiswouldplacetheannulusclose H =50 kms−1Mpc−1,inordertosimplifycomparisonwithpre- tothesourcewemovedtheannulus,generallytor=0.55◦.Toen- 0 viousstudiesofgroupsandclusters.Opticalluminositiesarenor- surethatthebackground modelwasnotbiasedbysourceswithin malisedusingthesolarluminosityintheBband,LB⊙=5.2 1032 theannulus,aniterativeprocesswasusedtoremovepointsources ergs−1. × of>4.5σsignificance.Anumberofourgalaxiesarefoundwithin groupsandclustersofgalaxies,manyofwhichhavetheirownX– rayhalos.Ourintentionwastomodelthesespectrallyandspatially in order to accurately remove the effects of their contamination of our target galaxies. We therefore moved the annulus outward 2 SAMPLESELECTION to avoid the emission, where possible. In cases where the emis- Our sample was selected from the Lyon-Meudon Extragalactic sionappearedtoextendtotheedgeofthefieldofview,weuseda DataArchive(LEDA),specificallythePGC-ROM1996(2nd edi- backgroundannulusatr=0.9◦.Thisoccurredforasmallnumberof tion). This contains information on 100,000 galaxies, of which galaxieswhichlieinthecentresofclusters(e.g.NGC1399).The ∼ 40,000 have the necessary redshift and morphological data. useofabackgroundannuluswhichlieswithintheclusteremission ∼ Galaxieswereselectedtomatchthefollowingselectioncriteria: meansthatwearelikelytooverestimatethetruebackground and henceover-correctforit.However,asweareusingthelargestan- (1) AbsolutemagnitudeM <–19 B nuluspossible,weshouldmimimizethedegreeofoverestimation. (2) MorphologicalT-type<–2 Wecanalsoexpectthecentralgalaxycomponent oftheemission (3) Virgocentric flow corrected recession velocity V < rec tohaveamuchhighersurfacebrightnessthantheclusteremission, 10,500kms−1 sothatoversubtractionwillhaveanegligibleeffectonit.Surface brightnessfitsshouldthereforebeaccurateforthecentralcompo- Thesecriteriawerechosentoproduceaselectionofopticallylumi- nent,whichisourmaininterest,andasgoodasispossibleforthe nousnearbyearly–typegalaxies. clustercomponent. Alistofgalaxiesmatchingthesecriteriawasthencomparedto acatalogueofROSATPSPCpointings,toproduceaninitialsample Theresultingbackground modelwasthenusedtoproducea ofgalaxieswithX–raydata.OnlygalaxieslyingwithinthePSPC background-subtractedcube.RegionsnearthePSPCwindowsup- supportstructure(i.e.within 30′ ofthepointing)wereaccepted, portstructurewereremovedfromtheseimages,asobjectsinthose soastoensure that theX–ra∼ydatawerenot stronglyaffected by areas would have been partially obscured during the observation. vignettingeffectsoroff–axisresolutionproblems.Pointingsofless The cube was further corrected for dead time and vignetting ef- than10ksecwerealsoignored,asthesewereunlikelytoprovide fects,andpointsourceswereremoved. X–ray data of sufficient quality. This initial sample contained 47 Examinationofbackground subtractedimagesallowedusto galaxies. locateeachgalaxyandproducearadialprofileofthesurrounding We then examined images of the raw X–ray data for each region.Fromtheseprofiles,regionsofinterestwereselected,from galaxy,tolookforpotentialproblems.Insomecaseswefoundthat whichimagesorspectraforuseinfittingcouldthenbeextracted. thegalaxiesappearedtobeextremelycompactorpoint–like,sug- Themajorityofourgalaxiesareknowntobemembersofgroups gestingthatsurfacebrightnessfittingwouldbedifficultorimpossi- or clusters, and as such weexpected to see emission from an in- ble.Theseobjectswereremovedfromthesample,asweregalaxies tergalactic medium (IGM) surrounding them. In cases where the inwhichanAGNornearbyquasardominatedtheX-rayemission, galaxydidnotappeartobecontaminatedwithotheremission,we toproduceafinalsampleof39X–rayluminousearly-typegalaxies. defined the region of interest (RoI) as being within the radius at Itisworthnotingthatasthefractionofgalaxiesinwhichthehalo whichtheemissiondropped tothebackground level. Thisregion wastoocompactorfaintforanalysiswassmall( 4%),itappears was suitable for both spatial and spectral fitting. In cases where ∼ thatthemajorityofmassiveearly–typegalaxiesdopossessbright, contaminating intergalactic emission was seen, we defined sepa- extendedX–rayhalos.Table1listsourtargets. rateregionsofinterest,oneforspectralandoneforspatialfits.For (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 3 Name RA DEC σ Vrec D Re T Environment (2000) (2000) (kms−1) (kms−1) (Mpc) (′) ESO443-24 130101.6 -322620 279.9 4970.1 99.4 0.388 -3.1 BGG IC1459 225709.5 -362737 321.4 1522.0 28.3 0.644 -4.7 BGG IC4296 133638.8 -335759 341.2 3587.9 71.8 0.953 -4.8 BGG IC4765 184719.0 -631949 288.4 4344.9 86.9 0.239 -3.9 BGG NGC499 012311.5 +332736 264.2 4482.7 82.8 0.346 -2.9 BGG NGC507 012340.0 +331522 295.8 5015.7 100.3 1.285 -3.3 BGG NGC533 012531.4 +014535 250.0 5411.4 95.5 0.792 -4.7 BGG NGC720 015300.4 -134421 237.7 1622.0 31.2 0.659 -4.7 BGG NGC741 015620.9 +053744 288.4 5529.9 91.6 0.869 -4.8 BGG NGC1332 032617.3 -212009 328.1 1355.8 29.5 0.467 -2.9 Group NGC1380 033626.9 -345833 240.4 1617.6 27.2 0.659 -2.3 Cluster NGC1395 033829.6 -230140 241.0 1516.1 30.8 0.757 -4.8 BGG NGC1399 033828.9 -352658 329.6 1211.2 27.2 0.706 -4.5 BCG NGC1404 033851.7 -353536 212.3 1701.9 27.2 0.446 -4.7 Cluster NGC1407 034012.3 -183452 279.3 1612.0 30.9 1.199 -4.5 BGG NGC1549 041545.0 -553531 203.2 932.5 21.7 0.792 -4.3 Group NGC1553 041610.3 -554651 167.5 805.8 21.7 1.094 -2.3 BGG NGC2300 073219.6 +854232 263.0 2249.7 41.5 0.524 -3.4 BGG† NGC2832 091946.5 +334502 341.2 6992.2 128.9 0.426 -4.3 BGG NGC3091 100013.8 -193814 303.4 3670.2 76.2 0.512 -4.7 BGG NGC3607 111654.1 +180312 216.8 999.6 29.7 1.094 -3.1 BGG NGC3923 115102.1 -284823 269.8 1468.0 26.8 0.889 -4.6 BGG NGC4073 120426.5 +015348 267.9 5970.6 119.1 0.931 -4.1 BGG NGC4125 120807.1 +651022 239.9 1618.6 38.9 0.998 -4.8 BGG NGC4261 121922.7 +054936 316.2 2244.0 47.2 0.644 -4.8 Cluster NGC4291 122018.1 +752221 287.7 2043.9 36.8 0.245 -4.8 Group NGC4365 122427.9 +071906 268.5 1290.4 23.9 0.830 -4.8 Cluster NGC4472 122946.5 +075958 304.8 931.8 23.9 1.734 -4.7 BCG NGC4552 123539.9 +123325 264.2 372.3 23.9 0.500 -4.6 Cluster/AGN NGC4636 124249.8 +024117 211.3 1125.2 23.9 1.694 -4.8 Cluster/BGG NGC4649 124340.2 +113258 342.8 1221.9 23.9 1.227 -4.6 Cluster NGC4697 124835.9 -054802 173.4 1232.2 22.7 1.256 -4.7 BGG NGC5128 132529.0 -430100 142.6 385.6 5.8 0.708 -2.1 BGG/AGN NGC5322 134915.5 +601129 239.9 2035.5 41.7 0.587 -4.8 BGG NGC5419 140338.6 -335841 329.6 4027.5 80.5 0.723 -4.2 BGG† NGC5846 150629.3 +013625 250.0 1890.0 34.4 1.377 -4.7 BGG NGC6269 165758.4 +275119 224.4 10435.0 208.7 0.574 -4.8 BGG NGC6482 175149.0 +230420 302.0 4102.0 82.0 0.132 -4.8 Field? NGC7619 232014.7 +081223 310.5 3825.5 60.0 0.536 -4.7 BCG Table1. Alistofgalaxiesincludedinoursample.RAandDECaretakenfromtheLEDAcatalogue,asaremorphologicaltype(T),Recessionvelocity(which iscorrectedforVirgocentricflowandmovementwithinthelocalgroup),andvelocitydispersion,σ.DistancesaretakenfromPrugniel&Simien(1996)where possible,orcalculatedfromtherecessionvelocity.H0=50isassumedinbothcases,andformostobjects,thedistanceusedisthatofthegrouporclusterin whichisresides.Opticaleffectiveradii,Re,aretakenfromFaberetal.(1989)andPrugniel&Heraudeau(1998).Galaxyenvironmentistakenfromthegroup cataloguesofGarcia(1993),Zabludoff&Mulchaey(1998)andWhiteetal.(1999).BGGandBCGstandforBrightestGroupGalaxyandBrightestCluster Galaxy,indicatingthatthegalaxyisthebrightest(andpresumablydominant)objectinthesurroundingsystem.†indicatesgalaxieswhicharenotidentifiedas brightestgroupgalaxiesinthegroupcatalogues,butlieatthecentreoftheX-rayhalooftheirgroup. spatial fitting,weagaindefinetheRoI asbeing withintheradius radius0.05◦ largerthanthenewregionofinterest,andusethisto atwhichemissiondropstothebackgroundlevel.Thiswillcontain generatealocalbackgroundmodel.Thislocalbackgroundshould boththegalaxyandthesurroundinggrouporcluster,allowingus account for both the cosmic X–ray background in the region of to fit models to both and thereby accurately remove contaminat- interest,andforthegroup/cluster contaminationalongthelineof ingemission. Abackground annulusfor usewiththisregionwas sight.Dependingontheformofthegroup/cluster halo,wemight selectedasdescribedabove. expecttounder-subtractthiscontaminationtosomedegree.Forex- ample,iftheclusterhaloissteeplydeclininginsurfacebrightness For spectral fitting in cases where the galaxy is surrounded outside our region of interest, the local background annulus will by contaminating group/cluster emission, a smaller RoI was de- contain fewer counts and we would expect to underestimate the fined,usingtheradiusatwhichthegalaxyemissiondroppedtothe contaminationalongthelineofsight.Ideallywewouldhopethat levelofthesurrounding group orclusterhalo.Withinthisradius, anyextendedgrouporclusterhalowouldhaveacoreradiussome- emissionshouldbedominatedbycomponentsassociatedwiththe whatlargerthantheregionofinterest,sothatitssurfacebrightness galaxy,thoughitmaystillbecontaminatedbygroup/clusteremis- isrelativelyconstantoverthewholeareaweareconsidering.Any sionalongthelineofsight.Theemissionoutsidethisradiusshould seriousunder-subtractionofgrouporclustercomponentwillhave beprimarilyproduced by thegroup or cluster halo. Wetherefore an effect on the spectral fits we obtain, particularly in the more take a local background spectrum from an annulus with an inner (cid:13)c 0000RAS,MNRAS000,1–26 4 Ewan O’Sullivanetal. massiveclusterswherecontaminationbythesurroundinghotICM parameters,providinguswithcrudemetallicityprofilesforafrac- wouldproducefitswithhigher thanexpectedtemperatures.Simi- tion of our sample. Given the limited spectral range of ROSAT, larly,ifwehavemisjudgedtheradiusatwhichemissionassociated theinabilityof the PSPCtoresolve individual spectral lines,and withthegalaxy becomes lessimportant thanthat associated with thesmallnumberofcountsineachannulus,theabundancesfitted thesurrounding structure, wemightexpect tosubtract partofthe shouldnotbetakenasaccuratemeasurements.However,insome galaxyemission.Again,wewouldexpecttoseeevidenceofthisin casestheydoshowinterestingtrendswhenconsideredinconjunc- theresultsofspectralfitstothedata.Wereturntothisquestionin tionwiththetemperatureprofiles. Section5. Hardnessprofileswerecalculatedinasomewhatsimilarman- ner.Againthelargerregionofinterestwassplitintoanumberof annuli.Fromeachofthese,countsinsoft(0.3–1.3keV)andhard (1.3–2.4 keV) bands were extracted and divided to produce a ra- 4 SPECTRALANDSPATIALANALYSIS tioofhard/softemission.Simulatedspectraindicatethata0.5keV Spectra for each galaxy were obtained by removing all data out- MEKALspectrumproduces avalueof 0.5,whileapowerlaw ∼ side the region of interest and collapsing the data cube along its ofΓ=1.7anda7keVbremsstrahlungspectrumproducevaluesof x and y axes. As with the background annulus, an iterative pro- 1.1and1.2respectively.Theseprofilescanbeusedtogiveaba- ∼ cess was used to remove point sources of >4.5 σ significance in sicideaofchangesinemissionacrossthegalaxyandinparticular the region of interest, although any point sources within the D toidentifyAGN. 25 diameterwereassumedtobeassociatedwiththegalaxyitselfand In order to study the spatial properties of the galaxy X–ray thereforenotremoved.Thespectracouldthenbefittedwithavari- emission, we also performed fits to the 2–dimensional surface etyofmodels.Toprovideabaselineforlaterfitsandtomeasurethe brightnessprofileofeachgalaxy.Followingtheinitialdatareduc- basicpropertiesofthegalaxyhalo,eachspectrumwasfittedwitha tion described in Section 3, we extracted an image in the 0.5–2 MEKALhotplasmamodel(Kaastra&Mewe1993;Liedahletal. keVbandandcorrecteditforvignetting.Thiswasdoneusingan 1995). Initially, only normalisation was fitted. Hydrogen absorp- energy–dependentexposuremap(seeSnowdenetal.1994forafull tioncolumndensitieswerefixedatvalues determinedfromradio description). Point sources were removed as in the spectral anal- surveys(Starketal.1992),and temperatureandmetalabundance ysis, and unrelated extended sources identified and excluded by werefixedat1keVand1solarrespectively.Parameterswerethen hand.Useoftheenergydependentexposuremapresultsinacon- freed in order (temperature, hydrogen column, metallicity), and stantbackgroundlevelacrosstheimage,soaflatbackgroundwas onlyre-frozenattheirstartingvaluesiftheybecamepoorlydefined alsodeterminedandsubtractedfromthedata. ortendedtoextremevalues.Thebasictemperatureandmetallicity Asinthecaseofspectralanalysis,wecanchoosetofitavari- valuesarelikelytoberepresentativeofthemajorityofearly–type etyofsurfacebrightnessmodelstoourdata.Themostcommonly galaxies(Matsushitaetal.2000;Matsushita2001),butclearlyfit- usedinthisworkwasamodifiedKingfunction(or“β–profile”)of tedvaluesarepreferable. theform: We then attempted to fit two component spectral mod- S(r)=S (1+(r/r )2)−3βfit+0.5 (1) 0 core els for each galaxy. These generally included a power–law + MEKAL model and bremsstrahlung + MEKAL model in which whereS(r)isthesurfacebrightnessatagivenradius, S0 is thebremsstrahlungtemperaturewasfixedat7keV.Thefirstcom- thecentralsurfacebrightnessrcoreisthecoreradiusandβfit isa ponent was intended to represent a hard component produced measure of theslope of the surfacebrightness profile.At various by the population of X–ray binaries and other unresolved stellar stages of the analysis we also fitted point source models and de sourceswithineachgalaxy.RecentChandrastudiesofextragalac- Vaucouleursr1/4lawmodels,usingtheform: ticX–raybinarypopulationssuggest thatthisemissionisreason- S(r)=S exp 7.67[(r/r )0.25 1] (2) ablymodeledbyapower–lawofindex 1.2(Sarazinetal.2001; e· {− e − } ∼ Blantonetal. 2001), while ASCA studies show good fits using a whereS isthesurfacebrightness atr , theeffectiveradius e e high temperature bremsstrahlung model (Matsushitaetal. 2000). (theisophotalradiuscontaininghalfthetotalluminosity.) All models were fitted using the Cash statistic (Cash 1979). The ModelswereconvolvedwiththePSPCpointspreadfunction Cashstatisticisdefined as-2lnL whereL isthe likelihood func- atanenergydeterminedfromthemeanphotonenergyoftheemis- tion.ThismeansthatthemostlikelymodelhasaminimumCash sionintheregionofinterestandthenfittedtothedata.Bothspheri- statisticandthatdifferencesinthestatisticarechi-squared(χ2)dis- calandellipticalfitswerepossiblewhenusingtheKinganddeVau- tributed. Thus confidence intervals can be calculated in the same couleursmodels,withthepositionangleandmajortominor axis was as for a conventional χ2 fit. By comparing the best fit Cash ratio measuring the shape and orientation of elliptical fits. When statisticforeachmodel,andvisuallyexaminingthespectralfit,we using King models, all parameters (core radius, β normalisa- fit selectedthebestfitmodelforeachgalaxy.Fromthiswecouldex- tion,xandypositionandtheellipticityparameters)wereusually tract(inmostcases)theX–raytemperatureandmetallicityofthe allowedtovaryfreely,asweretheparametersinpointsourcemod- galaxyhalo,aswellastheX–rayfluxfromthegalaxyhaloandthe els(xandyposition,normalisation).ThedeVaucouleursmodelis stellarcontribution. intendedtorepresenttheunresolveddiscretesourcepopulationof For each galaxy in the sample, we also derived simple pro- thegalaxies,whichweassumewilltakethesameformasthestel- jected temperature and hardness profiles. Temperature profiles lar population. An alternative approach would be to assume that were produced by splitting the larger, surface brightness region thediscretesourcepopulationfollowsthedistributionofglobular ofinterestintoseveralannuli,fromwhichspectrawereextracted. clusters,butwelackaccuratespatialmodelsofthisdistributionfor Thesespectrawerethenfittedusingthebestfittingspectralmodel manyofourgalaxies.Wethereforeinitiallysettheeffectiveradius forthegalaxyasawhole.Initiallythemodelswerefittedwiththe parameterofthedeVaucouleursmodelstothevalueoftheoptical metallicityandhydrogencolumndensityfrozenattheirglobalbest effectiveradius,andhelditfrozeninmostcases.Wedidallowthe fit values. However, if the data quality permitted, we freed these effectiveradiustovaryforasmallnumber ofgalaxies,wherethe (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 5 Figure1.Measuredprojectedtemperatureprofilesforoursampleofgalaxies (cid:13)c 0000RAS,MNRAS000,1–26 6 Ewan O’Sullivanetal. Figure1–continued (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 7 Figure1–continued Figure2. FittedmetallicityprofilesofNGC1399(leftpanel)andNGC4636(rightpanel) fitwaswellconstrainedbythedata,inordertoinvestigatediffer- determined, and a Gaussian fitted to the resulting spread of val- encesbetweentheopticalandX–raystellarprofiles.Howeveronly ues. Bycomparing theactual Cashstatistictothisdistributionof onegalaxy,NGC4697,wasbestfitbyamodelincludingadeVau- values, we were able to determine the probability that the model couleurscomponent,demonstratingthatinmostofthegalaxiesin could haveproduced thedata. Weweretherefore abletoidentify thesample,eitheremissionfromhotgasdominatesortheX-raybi- caseswherethe2-component fitwasnomorelikelytoreproduce narypopulationdoesnotfollowthestellarpopulation.AChandra thedatathanthe1-componentfit,anddiscardthe2-componentfits observation of NGC 4697 hasshown ittohavearelativelysmall forthesegalaxies. gas halo, with much of its emission contributed by point sources (Sarazinetal.2001),sothesuccessofthedeVaucouleurscompo- nentinthiscasecouldisperhapsunsurprising.TheChandradata 5 RESULTS for this galaxy are best fit by a surface brightness model which includesadeVaucouleurscomponentandaKingmodelwhosepa- 5.1 Spectralandspatialfits rametersaresuchthatitisflat,providingafairlyconstantcontri- Table2showstheresultsofourspectralfits.Asmentionedprevi- butionovertheareastudied. ously,metalabundancesfromROSATPSPCspectraareinherently The use of 2–dimensional datasets to fit the surface bright- unreliable,duetotherelativelypoorspectralresolutionofthein- nessdistributioncanresultinalownumber ofcountsinmanyof strument.Thisisreflectedbythelargeerrorsonsomeofourfitted the data bins. Under these conditions χ2 fitting performs poorly values,andbythefactthatinsomecaseswehadtoholdmetallic- (Nousek&Shue 1989) so, as in the spectral analysis, maximum ityfrozeninordertosecureastablefit.Thetemperaturevaluesare likelihood fitting based on the Cash statistic was used. However, morereliable,andgiveameantemperatureof0.67 0.29keV. ± theCashstatisticgivesnoindicationoftheabsolutequalityofthe AsdiscussedinSection3,apoorchoiceoflocalbackground fit,onlythequalityrelativetootherfits.Inordertogainsomees- for our targets could result in spectral fits biased by inclusion of timateofthetruefitquality,weusedaMonteCarloapproach, in group or cluster emission, or accidental subtraction of some of which the best fit 1- and 2-component model was used to gener- the galaxy emission. The clearest sign of this bias would be un- ate1000imagesofthegroups,towhichPoissonnoisewasadded. usuallyhighorlowfittedtemperature,significantlydifferentfrom Thesewerethencomparedtotheoriginalimage,theCashstatistic thosefoundinotherstudies.Fourofourgalaxieshavetemperatures (cid:13)c 0000RAS,MNRAS000,1–26 8 Ewan O’Sullivanetal. above 1 keV, and one (NGC 4697) has a temperature lower than cores(e.g.NGC1399),hotcores(e.g.NGC720),isothermal(e.g. 0.3keV.Theseareoutsidetherangecommonlyconsideredtypical NGC4697)andthosewherethedataqualitypreventsajudgment forellipticalgalaxies,sowecomparetheresultsforthesegalaxies (e.g. NGC 4552, where the errors on the outermost bin are large to those in the literature. NGC 507 has atemperature marginally enoughtomakeitsuspect).Coolandhotcoregalaxiesareselected above1keV.PreviousROSAT andASCAstudieshavefoundsim- undertherequirementthattheircentralbinmustbehotterorcooler ilar temperatures (Kim&Fabbiano 1995; Matsumotoetal. 1997) thananaverageoutertemperaturebyatleast20%,andthattheer- andmetalabundances(Buote2000),andmorerecentChandradata rorsinT mustbesmallerthanthisamount.Itisalsopossibleto X alsosupportsatemperatureof 1keV(Formanetal.2001).Buote seeevidenceofAGNactivityinsomeoftheprofiles,particularly ∼ (2002) fits a two temperature model to XMM-Newton EPIC data in the case of NGC 5128, where the central bin has a tempera- forNGC1399,andrecoverstemperaturesof 1.5and 0.9keV tureof 5keVandtherestofthegalaxy<1keV.Aninteresting within1′,withbothcomponentsapproachinga∼temperatu∼reof1.3- feature∼of some of thebetter defined profileswhich show central 1.5keVat3-10′.Ourvalueof 1.2keVisquitecomparabletothe cooling(e.g.NGC507,NGC1399,NGC4636)isthatthetemper- ∼ coolercomponent,consideringtheregionfromwhichourspectrum ature rises with radius to a value above the apparent outer mean wasextracted.NGC4073hasthemostextremetemperatureinTa- temperature,andthefallsbacktothatmean,producingatempera- ble2,kT=1.6keV.AnalysisofXMM-NewtonEPICdataforNGC turepeakatmoderateradii.Asallofthesegalaxiesareembedded 4073 and its surrounding group (O’Sullivanetal. 2003, in prep.) inlargergrouporclusterhalos,thispeakmaymarktheboundary suggestatemperaturegradientwithinthestellarbodyofthegalaxy, ofagrouporclusterscalecoolingflow,orthepointofinteraction withprojectedtemperaturesrisingfrom1.4to 2keV.Wealsofind between the galaxy and its environment. The observations avail- ∼ a high temperature for NGC 6269, kT=1.4 0.2 keV. No XMM- ableforNGC1399andNGC4636containverylargenumbersof ± NewtonorChandraresultsareavailableintheliteratureforNGC counts,allowingustoincludemetallicityasafreeparameterinthe 6269,butpreviousanalysesoftheROSATdatahavefoundtemper- profilefitting.Metallicityprofilesofthesetwogalaxiesareshown aturesof1.3 0.15keV(Dahlem&Thiering2000)and1.36 0.07 inFigure2.Despitethelargeerrorsinsomebins,itisnotablethat ± ± keV (Mulchaeyetal. 1996). These are identical within the errors bothmetallicityprofilesfollowthesamestructureasseenintem- with our result. Lastly, NGC 4697 was one of the first elliptical perature;acentraltrough,risingtoapeakatmoderateradius,with galaxiestobeobservedwithChandra,whichshowedittopossess an outer region of relatively low abundance. The peak tempera- arelativelycoolgascomponentwithkT 0.29keV(Sarazinetal. tureandmetallicityoccuratapproximatelythesameradiusinboth ∼ 2001).Thisisafairlygoodmatchtoourmeasuredtemperatureof cases.Inordertocheckthatcorrelationsbetweentemperatureand kT=0.24 0.2keV, anditshouldbenoted thatat leastpart ofthe abundanceinthefitswerenotbiasingtheresultswemodeledthe ± differencebetweentheseresultsmayarisefromChandracalibra- fitspaceforthetwobinsoneithersideoftheapparentbreakinthe tionissues.Ingeneral,thesecomparisonssuggestthatourmethod profileofNGC4636,calculatingfitstatisticsatarangeoftemper- ofbackgroundselectionanddataanalysisproducesfairlyaccurate aturesandmetallicities.Comparingtheconfidenceregionsforthe fitstothedata. twopointsshowsthattheyaredissimilartoatleast9σsignificance, Whencalculatingtheluminositiesofourtargets,wearefaced stronglysuggestingthebreakintheprofileisreal.Confidencere- with the difficulty that while we have surface brightness models gionsforthetwobinsareplottedinFigure3. which should provide an accurate estimate of the total number Table3liststheresultsofthesurfacebrightnessfittingforthe of counts from the galaxy halo, we only have spectral informa- galaxyhalosofourtargetgalaxies.Inthemajorityofgalaxieswe tion for a smaller central region. Ideally we would be able to si- obtaingoodqualityfits,withrelativelysmallerrorsonthecorera- multaneouslyfitspectralandspatialdata,givingatrueluminosity diusandslope.Asweareabletofitellipticalmodels,wealsolist foreachcomponent(e.g.Lloyd-Daviesetal.2000).Inpracticewe thepositionangleofthemajoraxisandaxisratioofthemodelfits. havechosentoscaleuptheMEKALcomponentofthespectralfit Forfivegalaxies,wewereunabletodetermineareliableposition to the number of counts found for the surface brightness model. angle,asthemodelwasconsistent(withinerrors)withbeingspher- Thisallowsustocalculateagasluminosityforthegalaxycompo- ical.Thebestfitpositionanglesofthesegalaxiesarelistedwithout nent,butignoresthecontributionfromdiscretesources.Thislumi- errors,andshouldonlybeconsideredasroughestimates. nositywillthereforebeanoverestimateofthetruegasluminosity associatedwiththegalaxy.However,weexpectlargeopticallylu- 5.2 TheL :T relation minousgalaxiessuchasourtargetstobealmostentirelydominated X X bygasemission. Withasmallnumber ofexceptions, thespectral Figure4showsL plottedagainsttemperatureforoursample.As X fitsconfirmthis,suggesting thatinmost casestheoverestimation can be seen, we find a fairly tight relation between L and T , X X issmall.Itisalsonotablethatbecausethebremsstrahlungcompo- with only a small number of outlier points. Using Kendall’s K- nentpeaksatahigherenergythantheMEKAL,agivenluminosity statistic(Ponman1982)tomeasurethestrengthofcorrelation,we correspondstoasmallernumberofbremsstrahlungcounts(inthe find a significance of 4.6σ. We fitted the relation using ODR- ∼ ROSAT band)thanitwouldforaMEKALmodel.Thismeansthat PACK (Boggsetal. 1989) to perform orthogonal least squares re- ouroverestimateofluminosityisreduced,asthenumberofcounts gression, and found a slope of 4.8 0.7. This fittingmethod uses ± associatedwiththebremsstrahlungcomponent,andassumedinthe theerrorsinxandyforeachpoint,butisunabletotakeofaccount scalingtobepartoftheMEKALcomponent,willproduceonlya ofthefactthattheerrorsareasymmetric. Themeanof theupper smallincreaseingasluminosity. andlowererrorisusedfortheerrorineachaxis. Temperatureprofilesforoursampleofgalaxiesareshownin This fittingmethod should be accurate as long as the points figure1.Theextentoftheprofilesvaries,owingtotherelativequal- deviate from the mean relation only on account of the statistical ity of data, length of exposure and galaxy distance. Comparison errors.Incaseswherethereisalargeintrinsicscatteraboutthere- of these profiles with those shown in Helsdon&Ponman (2000) lation,thestatisticalerrorsdonotprovideanappropriatebasisfor shows them to be similar in most cases where the samples over- theweightingofthedatapoints.Analternateapproachistoweight lap.Wehaveclassifiedthegalaxyprofilesintofourgroups;cooling allthepointsequally,ignoringthestatisticalerrorsasmisleading. (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 9 Name nH TX Z LogLX Model Profile (×10−21cm−2) (keV) (Z⊙) (ergs−1) ESO443-24 0.50+5.3 0.69+0.07 1.0+1.0 41.78+1. MK+BR I −0.19 −0.39 −0.7 −0.06 IC1459 0.069±0.012 0.51±0.05 1.0 40.3±0.2 MK+BR H IC4296 0.63+0.33 0.72±0.07 0.23±0.92 41.7+0.3 MK+BR C −0.24 −0.8 IC4765 2.6+3.8 0.64+0.09 0.17±0.17 42.4+0.7 MK+BR C −0.9 −0.37 −0.1 NGC499 0.79+0.32 0.70+0.02 1.0 42.6±0.3 MK+BR I −0.14 −0.03 NGC507 0.59±0.09 1.03±0.05 0.95+2.2 42.9+0.4 MK+BR C −0.33 −0.7 NGC533 0.29±0.03 0.84+0.05 1.1±1.1 42.2±0.2 MK+BR C −0.04 NGC720 0.15±0.15 0.50±0.04 0.28+0.06 41.1±0.3 MK+BR H −0.04 NGC741 0.54+0.11 0.71±0.07 0.26±0.26 41.9+0.3 MK+BR C −0.08 −0.4 NGC1332 0.16+0.04 0.41+0.06 0.59±0.59 40.6±0.2 MK+BR H −0.03 −0.05 NGC1380 0.18±0.10 0.30+0.08 0.11+0.44 40.4+0.4 MK+BR ? −0.05 −0.07 −1.0 NGC1395 0.069±0.016 0.65+0.04 1.0 40.6±0.2 MK+BR C −0.05 NGC1399 0.12±0.01 1.21±0.03 1.1+0.3 41.8±0.3 MK+BR C −0.2 NGC1404 0.18±0.02 0.60±0.01 0.35+0.05 41.66+0.04 MK C −0.04 −0.05 NGC1407 0.72+0.22 0.79+0.08 0.14+0.14 41.5+0.2 MK+BR C −0.16 −0.07 −0.06 −0.3 NGC1549 0.046±0.046 0.25+0.08 0.14+0.27 39.7+0.2 MK+BR I −0.05 −0.08 −0.4 NGC1553 0.045±0.033 0.53±0.15 0.10+0.16 40.2+0.3 MK+BR ? −0.05 −0.4 NGC2300 0.97+0.97 0.62+0.07 0.23±0.23 41.43+0.2 MK+BR ? −0.30 −0.14 −0.07 NGC2832 0.13±0.02 0.82±0.05 1.0 41.9±0.2 MK+BR ? NGC3091 0.38+0.08 0.64+0.04 1.0 41.7±0.2 MK+BR C −0.06 −0.05 NGC3607 0.015±0.015 0.45±0.06 0.71+0.57 40.7+0.3 MK+BR I −0.21 −0.4 NGC3923 0.58±0.22 0.46+0.04 1.0 40.8±0.2 MK+BR H −0.05 NGC4073 0.16+0.02 1.6±0.2 1.6+1.1 43.1+0.1 MK+BR C −0.03 −0.4 −0.3 NGC4125 0.10±0.04 0.34+0.05 0.34±0.34 41.27±0.07 MK ? −0.04 NGC4261 0.087±0.017 0.67+0.04 1.3±1.3 41.0±0.2 MK+BR C −0.05 NGC4291 0.22+0.18 0.59+0.06 0.63+4.8 41.20+0.4 MK I −0.09 −0.07 −0.13 −0.02 NGC4365 0.11+0.06 1.0+0.3 0.064+0.11 40.5±0.1 MK I −0.05 −0.2 −0.053 NGC4472 0.162±0.007 0.88±0.01 1.0 41.5±0.2 MK+BR C NGC4552 0.18±0.03 0.54±0.06 1.0 40.7±0.2 MK+BR ? NGC4636 0.25+0.05 0.55±0.03 0.40+0.32 42.0+0.1 MK+BR C −0.06 −0.13 −0.2 NGC4649 0.24+0.06 0.78±0.02 0.80+1.0 39.3+0.3 MK+BR C −0.07 −0.36 −0.5 NGC4697 0.16±0.03 0.24±0.02 0.40±0.40 39.0±0.2 MK+BR I NGC5128 0.43+0.07 0.35+0.04 1.0 40.2±0.2 MK+BR H −0.05 −0.03 NGC5322 0.045±0.037 0.33+0.100 0.19±0.19 40.3±0.3 MK+BR ? −0.06 NGC5419 0.29+0.07 0.69±0.26 0.32±0.32 42.0+0.3 MK+BR ? −0.05 −0.4 NGC5846 0.30±0.03 0.66±0.02 1.0 40.5+0.2 MK+BR C −0.3 NGC6269 0.53±0.12 1.4±0.2 0.33+0.36 43.4±0.1 MK C −0.17 NGC6482 0.68+0.72 0.55+0.04 1.0+1.0 41.2+0.5 MK+BR I −0.19 −0.07 −0.5 −0.4 NGC7619 0.34+0.08 0.81±0.03 1.9±1.9 42.0+0.3 MK+BR C −0.06 −0.8 Table2. Resultsofthespectralfitstooursamplegalaxies.Wherepossible,aabsorbedMEKAL+bremsstrahlung(MK+BR)modelwasfitted,butincases wherethebremsstrahlungnormalisationalwaystendedtozero,thiscomponentwasremovedfromthefit.Allupperandlower(1σ)errorsonfittedparameters werecalculatedindividually,butareshownasasingle±errorwhentheyareidenticaltotwosignificantfigures.Thosegalaxiesforwhichmetallicitycould notbesuccessfullyfittedarelistedwithafixedsolarmetallicitywithnoerrors.Temperatureprofilesareclassifiedasisothermal(I),coolcore(C),hotcore (H),oruncertain(?). To check our result we also fitted our L :T relation using the plottedalongsidethoseforsamplesofgroups(Helsdon&Ponman X X SLOPESpackage(Isobeetal.1990)toperformanOLSbisectorfit 2000)and clusters(Davidetal.1993;Mushotzky&Scharf1997; andignoringstatisticalerrors.Thebestfitslopeforthistechnique Fairleyetal. 2000).Our datafollow arelationof similarslopeto was5.05 0.44,verysimilartotheslopefoundwhenusingtheer- that of the groups (Helsdon & Ponman find a best fit slope of ± rors. 4.9 0.8), but offset to a lower luminosity or higher temperature. ± Foragiventemperature,ourgalaxiesareafactorof 3lesslumi- From the temperature profiles for each galaxy, we are able ∼ nous. Because of the scatter inbothsamples, thereissomeover- to identify which of our targets have strong temperature gradi- lapbetweenthegroupsandgalaxies,andsomeofourgalaxydata entswhichcouldaffectthemeasuredmeantemperature.Excluding pointslieabovethegroupbestfitline.Conversely,thebestfitre- these20galaxiesweakenstherelationto 2.8σ significance,and ∼ lationforclustersissignificantlyshallowerthanthatforgroupsor givesabestfitslope(fittedusingstatisticalerrors)of5.9 1.3.We ± galaxies,though againthereisasmallregion ofoverlapbetween alsofitthesampleofgalaxieswithknown temperaturegradients, themostluminousgalaxiesandthefaintestclusters. andfounda 2.7σcorrelation,withaslopeof 3.7. ∼ ∼ The L :T relation has been used extensively in the study X X ofgroupsandclustersofgalaxies.Figure5showsourdatapoints (cid:13)c 0000RAS,MNRAS000,1–26 10 Ewan O’Sullivanet al. Name CoreRadius βfit AxisRatio PositionAngle RoIRadius Model (arcmin) (degrees) (degrees) ESO443-24 0.106±0.004 0.55±0.03 1.7+0.3 306±7 0.045 KI −0.2 IC1459 0.17±0.02 0.89+0.36 1.6+1.4 276±7 0.065 KI+KI −0.09 −0.2 IC4296 0.080±0.004 0.66+0.18 2.2+0.8 29+10 0.12 KI+KI −0.02 −0.4 −8 IC4765 0.39±0.02 1.24+0.6 1.1+0.1 190 0.17 KI+KI −0.06 −0.7 NGC499 0.85±0.04 0.70±0.03 1.05+0.03 290+30 0.25 KI+KI −0.04 −20 NGC507 0.82±0.03 0.54±0.03 1.05+0.04 270 0.14 KI+KI −0.8 NGC533 0.058±0.002 0.53+0.04 1.24±0.06 240±10 0.16 KI+KI −0.02 NGC720 0.15±0.02 0.483+0.010 1.34+0.09 294+20 0.11 KI+KI −0.009 −0.08 −7 NGC741 0.172±0.006 0.65+0.64 1.41+0.08 128+10 0.12 KI+KI −0.06 −0.1 −9 NGC1332 0.010±0.001 0.548+0.009 1.8+0.3 295±4 0.080 KI+KI −0.008 −0.2 NGC1380 0.090±0.011 0.51+0.04 1.1+0.2 210 0.045 KI −0.03 −0.3 NGC1395 0.23±0.03 0.52+0.06 1.3±0.1 41+9 0.10 KI+KI −0.03 −10 NGC1399 0.11±0.01 0.59±0.02 1.23+0.04 176±3 0.12 KI+KI −0.03 NGC1404 0.32±0.04 0.770+0.005 1.05+0.02 294+10 0.20 KI+KI −0.004 −0.01 −6 NGC1407 0.18±0.02 0.56+0.02 1.20±0.06 91+7 0.20 KI+KI −0.01 −8 NGC1549 0.021±0.003 0.509±0.003 1.56+0.10 174+7 0.090 KI −0.09 −8 NGC1553 0.43±0.07 0.66+0.06 1.4+0.2 304+10 0.17 KI+KI −0.09 −0.1 −9 NGC2300 0.23±0.02 0.69+0.12 1.1+0.2 330 0.16 KI+KI −0.06 −0.1 NGC2832 0.0100±0.0003 0.314±0.007 1.4±0.1 357±6 0.15 KI+PS NGC3091 1.17±0.05 1.60+0.03 1.31±0.04 325+7 0.15 KI+KI −0.04 −6 NGC3607 0.63±0.07 0.48±0.02 1.17+0.07 350±10 0.14 KI −0.06 NGC3923 0.010±0.001 0.55+0.05 1.8+0.4 54+6 0.080 KI+KI −0.01 −0.2 −5 NGC4073 0.072±0.002 0.46+0.02 1.20+0.06 265+7 0.10 KI+KI −0.01 −0.05 −8 NGC4125 0.017±0.001 0.48+0.01 1.629+0.3 269+8 0.070 KI+PS −0.08 −0.003 −9 NGC4261 0.37±0.03 1.2+1.4 1.8+0.3 37±5 0.10 KI+KI −0.2 −0.2 NGC4291 0.38±0.04 0.57+0.04 1.3±0.1 106+9 0.065 KI −0.03 −10 NGC4365 0.54±0.08 0.60+0.04 2.0±0.2 31±4 0.090 KI −0.03 NGC4472 0.25±0.04 0.597+0.009 1.08±0.02 83+10 0.24 KI+KI −0.008 −9 NGC4552 0.098±0.014 0.60+0.02 1.7+0.2 164±5 0.045 KI −0.03 −0.1 NGC4636 0.40±0.06 0.535+0.007 1.02+0.02 24 0.24 KI+KI −0.006 −0.09 NGC4649 0.13±0.02 0.567±0.008 1.18±0.03 26±6 0.080 KI+PS NGC4697 0.99±0.15 0.46+0.13 1.8±0.3 28+6 0.045 KI+DV −0.07 −12 NGC5128 0.90±0.53 0.55±0.01 2.21±0.07 52.0±0.9 0.25 KI+KI NGC5322 0.0100±0.0008 0.49±0.01 1.6+0.3 57+8 0.057 KI −0.2 −9 NGC5419 4.5±0.2 0.50+0.18 1.44±0.06 44±6 0.20 KI+PS −0.07 NGC5846 1.3±0.1 0.80+0.05 1.15±0.03 45+7 0.080 KI+PS −0.04 −6 NGC6269 1.21±0.02 0.40+0.04 1.19+0.1 38±15 0.10 KI+PS −0.02 −0.08 NGC6482 0.162±0.007 0.524+0.009 1.13+0.07 220±10 0.10 KI+PS −0.001 −0.05 NGC7619 0.031±0.002 0.447+0.006 1.18+0.07 308±10 0.20 KI+KI −0.005 −0.06 Table3. Resultsofthesurfacebrightnessfitstoourgalaxies,forthecomponentsassociatedwiththegalaxyhalo.Bestfitpositionanglevalueswithouterrors aregivenforthosegalaxieswheretheanglewasessentiallyunconstrained.Allerrorsarequotedatthe1σ. 5.3 L :L andL :T relations ieshavingX–rayluminositiesafactorof 9higherthanthoseof X B B X ∼ groupswithequalopticalluminosity,orconverselyL values 2.3 B ∼ One of the more common relations used in studies of early-type timeslowerthangroupsofsimilarLX.Thisresultcanbecompared galaxiesistheLX:LBrelation.NumerousstudiesbasedonROSAT, totheLX:TX relationshowninSection5.2,inwhichgalaxiesare ASCAorEinsteindatahavebeenpublished(e.g.Beuingetal.1999; offset to lower luminosities at a given temperature, compared to Matsushita2001;Brown&Bregman1998;Fabbianoetal.1992), groups. and we have previously examined this relation in some detail in O’Sullivanetal.(2001a),towhichwedirectreadersforafulldis- The L :T relation for our galaxies is shown in Figure 7. B X cussion of the relation and the effects of galaxy environment on Once again, for this relation we find a fairly strong correlation it. Figure 6 shows the L :L relation for our galaxies. For the ( 4σ significance). The slope of the relation is comparable to X B ∼ sample as a whole, there is a 3.9σ correlation, with a slope of thatfoundforgroupsandclusters;1.91 0.33forourgalaxysam- ± 2.7. Thisisquite asteeprelation, comparable tothat found for ple,1.64 0.23forgalaxygroups(Helsdon&Ponman2002),and ∼ ± asampleofBGGsinpreviouswork(O’Sullivanetal.2001a).We 1.5 for galaxy clusters (Lloyd-Davies&Ponman 2002). How- ∼ havealsoplottedthebestfitrelationfoundforX-raybrightgalaxy ever,wheretherelationsforgroupsandclustersareessentiallythe groups,fromHelsdon&Ponman(2002).Theslopeofthisrelation same(Helsdon&Ponman2002),ourrelationforgalaxiesissignif- (2.6 0.4)isverysimilartothatfoundforoursampleofgalaxies. icantlyoffsettohighertemperatures(byafactorof 2,compared ± ∼ However, our relation is offset from that for groups, with galax- togroups)orlowerL (byafactorof 3). B ∼ (cid:13)c 0000RAS,MNRAS000,1–26

ebook img

X-ray Scaling Properties of Early-type Galaxies PDF

26 Pages·0.54 MB·English
by  Ewan O'Sullivan
#additional_collections #journals #arxiv
Save to my drive
Quick download
Download
Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview X-ray Scaling Properties of Early-type Galaxies

Mon.Not.R.Astron.Soc.000,1–26(0000) Printed5February2008 (MNLaTEXstylefilev2.2) X–ray Scaling Properties of Early–type Galaxies Ewan O’Sullivan⋆, Trevor J. Ponman, Ross S. Collins SchoolofPhysicsandAstronomy,UniversityofBirmingham,Edgbaston,BirminghamB152TT Accepted2003??.Received2002??;inoriginalform2002?? 3 ABSTRACT 0 We presentananalysisof39X–rayluminousearly–typegalaxiesobservedwiththeROSAT 0 PSPC.Usingmulti–componentspectralandspatialfitstothesedatawehavemeasuredhalo 2 abundance,temperature, luminosity and surface brightnessprofile. We compare these mea- n surements to similar results from galaxy groups and clusters, fitting a number of relations a commonlyused in the study of these larger objects. In particular,we find that the σ:TX re- J lation for oursample is similar to that reportedfor clusters, consistent with β = 1, and spec 9 thattheLX:TX relationhasasteepslope(gradient4.8±0.7)comparablewiththatfoundfor galaxygroups.Assumingisothermality,weconstruct3-dimensionalmodelsofourgalaxies, 1 allowingustomeasuregasentropy.Wefindnocorrelationbetweengasentropyandsystem v mass,butdofindatrendforlowtemperaturesystemstohavereducedgasfractions.Wecon- 3 cludethatthegalaxiesinoursamplearelikelytohavedevelopedtheirhalosthroughgalaxy 5 winds,influencedbytheirsurroundingenvironment. 1 1 Keywords: galaxies:ellipticalandlenticular–galaxies:halos–X-rays:galaxies 0 3 0 / h p 1 INTRODUCTION in more massive systems gas entropy scales with the total mass, - in groups it appears to reach a roughly constant minimum level. o Early-typegalaxieshavebeenknowntopossesslargehalosofhot A number of models have been put forward to explain this be- r gassincethedetectionbyEinsteinofX–rayemissionfromtheel- t haviour,includingraisingtheentropythroughtheinjectionofen- s lipticalpopulation intheVirgocluster (Formanetal. 1979). Suc- ergy by AGN (Wuetal. 2000) or star formation (Ponmanetal. a cessivegenerationsofX–rayobservatorieshavebeenusedtoob- 1999),orthroughtheradiativecoolingandremovaloflowentropy v: serve these galaxies, and the advent of XMM-Newton and Chan- gas(Muanwongetal.2001).Itisnotablethatforanyofthesepro- i drahasallowedthecomplexnatureoftheiremissiontobestudied X cesses,theoriginoftheentropyrisewouldberelatedtothegalax- in detail. Most of the work in this area has focused on the vari- iesinthesystem.Asearly–typegalaxiespossesstheirownhalos, r oussourcesofemissionwithinearly–typegalaxies(hotgas,X–ray a wemightexpecttoseeevidenceoftheseprocessesintheirX–ray binaries, AGN) and on the surprisingly complicated relation be- properties,andfortheeffecttobestrongestinthesesystems,owing tweenopticalandX–rayluminosity.However,ataquitefundamen- totheirpositionatthebottomofthemassscale. tal level, early–type galaxies resemble the groups and clusters in There is also evidence from previous studies of early–type whichtheytypicallyreside.Simulationsofdarkmatterhalossug- galaxies(Helsdonetal.2001;O’Sullivanetal.2001a)that galax- gestthattheyhavesimilarprofilesatallmassscales(Navarroetal. iesinthecentresofX–raybrightgroupsareaffectedbytheiren- 1997). If we consider clusters, groups and galaxies as potentials vironment.TheyhaveasignificantlysteeperL :L relation,and containinghotgas,wemightexpectthepropertiesofthehalosto X B areonaverageconsiderablymoreluminousthannormalellipticals. besimilaracrossawiderangeofmasses. A large fraction of group dominant galaxies in one sample have A comparison between halos on different scales becomes beenshowntohavetemperatureprofilesindicativeofcentralcool- increasingly interesting considering the importance of entropy ing(Helsdonetal.2001),leadingtothesuggestionthattheirhalos changes in governing the behaviour of group and cluster halos. areactually the product of cooling flowsassociated withthe sur- Observations of galaxy groups have demonstrated that these sys- roundinggroup.Consideringthedifferencesbetweenthesedomi- tems do not behave as might be expected from scaling clusters, nantgalaxiesandtheirmorenormalcounterparts,andthebiasing but instead require non-gravitational processes. One of the clear- effect their inclusion in samples of early–type galaxies seems to estsignsofthisistheentropyfloor(Ponmanetal.1999).Whereas have,furtherinvestigationoftheprocesseswhichhaveshapedtheir halosseemswarranted. ⋆ Presentaddress:Harvard-SmithsonianCenterforAstrophysics,60Gar- We have compiled a sample of 39 large, X–ray luminous denStreet,Cambridge,MA02138,USA early–typegalaxiesforwhichthereisgoodqualityROSAT PSPC Email:[email protected] dataavailable.Wehaveanalysedthesedata,andfittedtwodimen- (cid:13)c 0000RAS 2 Ewan O’Sullivanetal. sional,twocomponentsurfacebrightnessprofilestothem.Wehave 3 DATAREDUCTION alsofittedtwocomponentspectralmodels,temperature,abundance DatareductionandanalysisoftheX–raydatasetswerecarriedout and hardness profiles and produced three dimensional models of thegalaxies.Theseallowustomodeloutcontaminationfromsur- usingtheASTERIXsoftwarepackage.Beforethedatasetscouldbe used,varioussourcesofcontaminationhadtoberemoved.Possible rounding cluster or groupemission andthediscretesourcepopu- sources include charged particles and solar X–rays scattered into lationwithinthegalaxy.Wecanthereforeexaminetheproperties thetelescopefromtheEarth’satmosphere. Onboard instrumenta- ofthehaloindetail,forthefirsttimeinasampleofthissize.We tionprovidesinformationwhichallowsperiodsofhighbackground canalsocomparethebehaviour of thissampletothatof samples tobeidentified.Themastervetocounterrecordsthechargedparti- ofgroupsandclustersthroughrelationsbetweenparameterssuch cleflux,andweexcludedalltimeperiodsduringwhichthemaster astemperature,opticalandX–rayluminosity,velocitydispersion, veto rate exceeded 170 count s−1. Solar contamination causes a surfacebrightnessslopeandgasentropy.Inmostcases,thisisthe significantoverallincreaseintheX–rayeventrate.Toremovethis first timethese relations have been studied for halos at thismass contamination we excluded all times during which the event rate scale. The paper is organised as follows. In Section 2 we describe deviatedfromthemeanbymorethan2σ.Thisgenerallyremoved nomorethanafewpercentofeachdataset. our sample and the selection criteria used to create it. Section 3 gives details of the techniques used in reduction of the ROSAT Afterthiscleaningprocesseachdatasetwasbinnedintoa3– PSPC data, and Section 4 describes the spectral and spatial fit- dimensional(x,y,energy)datacube.Spectraorimagescanbeex- ting processes. Our results are presented in Section 5, with data tractedfromsuchacubebycollapsingitalongtheaxes.Amodel fromclustersandgroupsofgalaxiesincludedforcomparison.We ofthebackground wasthengenerated based onanannulus taken discuss the results and their implications in Section 6, and give fromthiscube.Weusedannuliofwidth0.1◦,andinnerradius0.4◦ our conclusions in Section 7. Throughout the paper we assume wherepossible.Incaseswherethiswouldplacetheannulusclose H =50 kms−1Mpc−1,inordertosimplifycomparisonwithpre- tothesourcewemovedtheannulus,generallytor=0.55◦.Toen- 0 viousstudiesofgroupsandclusters.Opticalluminositiesarenor- surethatthebackground modelwasnotbiasedbysourceswithin malisedusingthesolarluminosityintheBband,LB⊙=5.2 1032 theannulus,aniterativeprocesswasusedtoremovepointsources ergs−1. × of>4.5σsignificance.Anumberofourgalaxiesarefoundwithin groupsandclustersofgalaxies,manyofwhichhavetheirownX– rayhalos.Ourintentionwastomodelthesespectrallyandspatially in order to accurately remove the effects of their contamination of our target galaxies. We therefore moved the annulus outward 2 SAMPLESELECTION to avoid the emission, where possible. In cases where the emis- Our sample was selected from the Lyon-Meudon Extragalactic sionappearedtoextendtotheedgeofthefieldofview,weuseda DataArchive(LEDA),specificallythePGC-ROM1996(2nd edi- backgroundannulusatr=0.9◦.Thisoccurredforasmallnumberof tion). This contains information on 100,000 galaxies, of which galaxieswhichlieinthecentresofclusters(e.g.NGC1399).The ∼ 40,000 have the necessary redshift and morphological data. useofabackgroundannuluswhichlieswithintheclusteremission ∼ Galaxieswereselectedtomatchthefollowingselectioncriteria: meansthatwearelikelytooverestimatethetruebackground and henceover-correctforit.However,asweareusingthelargestan- (1) AbsolutemagnitudeM <–19 B nuluspossible,weshouldmimimizethedegreeofoverestimation. (2) MorphologicalT-type<–2 Wecanalsoexpectthecentralgalaxycomponent oftheemission (3) Virgocentric flow corrected recession velocity V < rec tohaveamuchhighersurfacebrightnessthantheclusteremission, 10,500kms−1 sothatoversubtractionwillhaveanegligibleeffectonit.Surface brightnessfitsshouldthereforebeaccurateforthecentralcompo- Thesecriteriawerechosentoproduceaselectionofopticallylumi- nent,whichisourmaininterest,andasgoodasispossibleforthe nousnearbyearly–typegalaxies. clustercomponent. Alistofgalaxiesmatchingthesecriteriawasthencomparedto acatalogueofROSATPSPCpointings,toproduceaninitialsample Theresultingbackground modelwasthenusedtoproducea ofgalaxieswithX–raydata.OnlygalaxieslyingwithinthePSPC background-subtractedcube.RegionsnearthePSPCwindowsup- supportstructure(i.e.within 30′ ofthepointing)wereaccepted, portstructurewereremovedfromtheseimages,asobjectsinthose soastoensure that theX–ra∼ydatawerenot stronglyaffected by areas would have been partially obscured during the observation. vignettingeffectsoroff–axisresolutionproblems.Pointingsofless The cube was further corrected for dead time and vignetting ef- than10ksecwerealsoignored,asthesewereunlikelytoprovide fects,andpointsourceswereremoved. X–ray data of sufficient quality. This initial sample contained 47 Examinationofbackground subtractedimagesallowedusto galaxies. locateeachgalaxyandproducearadialprofileofthesurrounding We then examined images of the raw X–ray data for each region.Fromtheseprofiles,regionsofinterestwereselected,from galaxy,tolookforpotentialproblems.Insomecaseswefoundthat whichimagesorspectraforuseinfittingcouldthenbeextracted. thegalaxiesappearedtobeextremelycompactorpoint–like,sug- Themajorityofourgalaxiesareknowntobemembersofgroups gestingthatsurfacebrightnessfittingwouldbedifficultorimpossi- or clusters, and as such weexpected to see emission from an in- ble.Theseobjectswereremovedfromthesample,asweregalaxies tergalactic medium (IGM) surrounding them. In cases where the inwhichanAGNornearbyquasardominatedtheX-rayemission, galaxydidnotappeartobecontaminatedwithotheremission,we toproduceafinalsampleof39X–rayluminousearly-typegalaxies. defined the region of interest (RoI) as being within the radius at Itisworthnotingthatasthefractionofgalaxiesinwhichthehalo whichtheemissiondropped tothebackground level. Thisregion wastoocompactorfaintforanalysiswassmall( 4%),itappears was suitable for both spatial and spectral fitting. In cases where ∼ thatthemajorityofmassiveearly–typegalaxiesdopossessbright, contaminating intergalactic emission was seen, we defined sepa- extendedX–rayhalos.Table1listsourtargets. rateregionsofinterest,oneforspectralandoneforspatialfits.For (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 3 Name RA DEC σ Vrec D Re T Environment (2000) (2000) (kms−1) (kms−1) (Mpc) (′) ESO443-24 130101.6 -322620 279.9 4970.1 99.4 0.388 -3.1 BGG IC1459 225709.5 -362737 321.4 1522.0 28.3 0.644 -4.7 BGG IC4296 133638.8 -335759 341.2 3587.9 71.8 0.953 -4.8 BGG IC4765 184719.0 -631949 288.4 4344.9 86.9 0.239 -3.9 BGG NGC499 012311.5 +332736 264.2 4482.7 82.8 0.346 -2.9 BGG NGC507 012340.0 +331522 295.8 5015.7 100.3 1.285 -3.3 BGG NGC533 012531.4 +014535 250.0 5411.4 95.5 0.792 -4.7 BGG NGC720 015300.4 -134421 237.7 1622.0 31.2 0.659 -4.7 BGG NGC741 015620.9 +053744 288.4 5529.9 91.6 0.869 -4.8 BGG NGC1332 032617.3 -212009 328.1 1355.8 29.5 0.467 -2.9 Group NGC1380 033626.9 -345833 240.4 1617.6 27.2 0.659 -2.3 Cluster NGC1395 033829.6 -230140 241.0 1516.1 30.8 0.757 -4.8 BGG NGC1399 033828.9 -352658 329.6 1211.2 27.2 0.706 -4.5 BCG NGC1404 033851.7 -353536 212.3 1701.9 27.2 0.446 -4.7 Cluster NGC1407 034012.3 -183452 279.3 1612.0 30.9 1.199 -4.5 BGG NGC1549 041545.0 -553531 203.2 932.5 21.7 0.792 -4.3 Group NGC1553 041610.3 -554651 167.5 805.8 21.7 1.094 -2.3 BGG NGC2300 073219.6 +854232 263.0 2249.7 41.5 0.524 -3.4 BGG† NGC2832 091946.5 +334502 341.2 6992.2 128.9 0.426 -4.3 BGG NGC3091 100013.8 -193814 303.4 3670.2 76.2 0.512 -4.7 BGG NGC3607 111654.1 +180312 216.8 999.6 29.7 1.094 -3.1 BGG NGC3923 115102.1 -284823 269.8 1468.0 26.8 0.889 -4.6 BGG NGC4073 120426.5 +015348 267.9 5970.6 119.1 0.931 -4.1 BGG NGC4125 120807.1 +651022 239.9 1618.6 38.9 0.998 -4.8 BGG NGC4261 121922.7 +054936 316.2 2244.0 47.2 0.644 -4.8 Cluster NGC4291 122018.1 +752221 287.7 2043.9 36.8 0.245 -4.8 Group NGC4365 122427.9 +071906 268.5 1290.4 23.9 0.830 -4.8 Cluster NGC4472 122946.5 +075958 304.8 931.8 23.9 1.734 -4.7 BCG NGC4552 123539.9 +123325 264.2 372.3 23.9 0.500 -4.6 Cluster/AGN NGC4636 124249.8 +024117 211.3 1125.2 23.9 1.694 -4.8 Cluster/BGG NGC4649 124340.2 +113258 342.8 1221.9 23.9 1.227 -4.6 Cluster NGC4697 124835.9 -054802 173.4 1232.2 22.7 1.256 -4.7 BGG NGC5128 132529.0 -430100 142.6 385.6 5.8 0.708 -2.1 BGG/AGN NGC5322 134915.5 +601129 239.9 2035.5 41.7 0.587 -4.8 BGG NGC5419 140338.6 -335841 329.6 4027.5 80.5 0.723 -4.2 BGG† NGC5846 150629.3 +013625 250.0 1890.0 34.4 1.377 -4.7 BGG NGC6269 165758.4 +275119 224.4 10435.0 208.7 0.574 -4.8 BGG NGC6482 175149.0 +230420 302.0 4102.0 82.0 0.132 -4.8 Field? NGC7619 232014.7 +081223 310.5 3825.5 60.0 0.536 -4.7 BCG Table1. Alistofgalaxiesincludedinoursample.RAandDECaretakenfromtheLEDAcatalogue,asaremorphologicaltype(T),Recessionvelocity(which iscorrectedforVirgocentricflowandmovementwithinthelocalgroup),andvelocitydispersion,σ.DistancesaretakenfromPrugniel&Simien(1996)where possible,orcalculatedfromtherecessionvelocity.H0=50isassumedinbothcases,andformostobjects,thedistanceusedisthatofthegrouporclusterin whichisresides.Opticaleffectiveradii,Re,aretakenfromFaberetal.(1989)andPrugniel&Heraudeau(1998).Galaxyenvironmentistakenfromthegroup cataloguesofGarcia(1993),Zabludoff&Mulchaey(1998)andWhiteetal.(1999).BGGandBCGstandforBrightestGroupGalaxyandBrightestCluster Galaxy,indicatingthatthegalaxyisthebrightest(andpresumablydominant)objectinthesurroundingsystem.†indicatesgalaxieswhicharenotidentifiedas brightestgroupgalaxiesinthegroupcatalogues,butlieatthecentreoftheX-rayhalooftheirgroup. spatial fitting,weagaindefinetheRoI asbeing withintheradius radius0.05◦ largerthanthenewregionofinterest,andusethisto atwhichemissiondropstothebackgroundlevel.Thiswillcontain generatealocalbackgroundmodel.Thislocalbackgroundshould boththegalaxyandthesurroundinggrouporcluster,allowingus account for both the cosmic X–ray background in the region of to fit models to both and thereby accurately remove contaminat- interest,andforthegroup/cluster contaminationalongthelineof ingemission. Abackground annulusfor usewiththisregionwas sight.Dependingontheformofthegroup/cluster halo,wemight selectedasdescribedabove. expecttounder-subtractthiscontaminationtosomedegree.Forex- ample,iftheclusterhaloissteeplydeclininginsurfacebrightness For spectral fitting in cases where the galaxy is surrounded outside our region of interest, the local background annulus will by contaminating group/cluster emission, a smaller RoI was de- contain fewer counts and we would expect to underestimate the fined,usingtheradiusatwhichthegalaxyemissiondroppedtothe contaminationalongthelineofsight.Ideallywewouldhopethat levelofthesurrounding group orclusterhalo.Withinthisradius, anyextendedgrouporclusterhalowouldhaveacoreradiussome- emissionshouldbedominatedbycomponentsassociatedwiththe whatlargerthantheregionofinterest,sothatitssurfacebrightness galaxy,thoughitmaystillbecontaminatedbygroup/clusteremis- isrelativelyconstantoverthewholeareaweareconsidering.Any sionalongthelineofsight.Theemissionoutsidethisradiusshould seriousunder-subtractionofgrouporclustercomponentwillhave beprimarilyproduced by thegroup or cluster halo. Wetherefore an effect on the spectral fits we obtain, particularly in the more take a local background spectrum from an annulus with an inner (cid:13)c 0000RAS,MNRAS000,1–26 4 Ewan O’Sullivanetal. massiveclusterswherecontaminationbythesurroundinghotICM parameters,providinguswithcrudemetallicityprofilesforafrac- wouldproducefitswithhigher thanexpectedtemperatures.Simi- tion of our sample. Given the limited spectral range of ROSAT, larly,ifwehavemisjudgedtheradiusatwhichemissionassociated theinabilityof the PSPCtoresolve individual spectral lines,and withthegalaxy becomes lessimportant thanthat associated with thesmallnumberofcountsineachannulus,theabundancesfitted thesurrounding structure, wemightexpect tosubtract partofthe shouldnotbetakenasaccuratemeasurements.However,insome galaxyemission.Again,wewouldexpecttoseeevidenceofthisin casestheydoshowinterestingtrendswhenconsideredinconjunc- theresultsofspectralfitstothedata.Wereturntothisquestionin tionwiththetemperatureprofiles. Section5. Hardnessprofileswerecalculatedinasomewhatsimilarman- ner.Againthelargerregionofinterestwassplitintoanumberof annuli.Fromeachofthese,countsinsoft(0.3–1.3keV)andhard (1.3–2.4 keV) bands were extracted and divided to produce a ra- 4 SPECTRALANDSPATIALANALYSIS tioofhard/softemission.Simulatedspectraindicatethata0.5keV Spectra for each galaxy were obtained by removing all data out- MEKALspectrumproduces avalueof 0.5,whileapowerlaw ∼ side the region of interest and collapsing the data cube along its ofΓ=1.7anda7keVbremsstrahlungspectrumproducevaluesof x and y axes. As with the background annulus, an iterative pro- 1.1and1.2respectively.Theseprofilescanbeusedtogiveaba- ∼ cess was used to remove point sources of >4.5 σ significance in sicideaofchangesinemissionacrossthegalaxyandinparticular the region of interest, although any point sources within the D toidentifyAGN. 25 diameterwereassumedtobeassociatedwiththegalaxyitselfand In order to study the spatial properties of the galaxy X–ray thereforenotremoved.Thespectracouldthenbefittedwithavari- emission, we also performed fits to the 2–dimensional surface etyofmodels.Toprovideabaselineforlaterfitsandtomeasurethe brightnessprofileofeachgalaxy.Followingtheinitialdatareduc- basicpropertiesofthegalaxyhalo,eachspectrumwasfittedwitha tion described in Section 3, we extracted an image in the 0.5–2 MEKALhotplasmamodel(Kaastra&Mewe1993;Liedahletal. keVbandandcorrecteditforvignetting.Thiswasdoneusingan 1995). Initially, only normalisation was fitted. Hydrogen absorp- energy–dependentexposuremap(seeSnowdenetal.1994forafull tioncolumndensitieswerefixedatvalues determinedfromradio description). Point sources were removed as in the spectral anal- surveys(Starketal.1992),and temperatureandmetalabundance ysis, and unrelated extended sources identified and excluded by werefixedat1keVand1solarrespectively.Parameterswerethen hand.Useoftheenergydependentexposuremapresultsinacon- freed in order (temperature, hydrogen column, metallicity), and stantbackgroundlevelacrosstheimage,soaflatbackgroundwas onlyre-frozenattheirstartingvaluesiftheybecamepoorlydefined alsodeterminedandsubtractedfromthedata. ortendedtoextremevalues.Thebasictemperatureandmetallicity Asinthecaseofspectralanalysis,wecanchoosetofitavari- valuesarelikelytoberepresentativeofthemajorityofearly–type etyofsurfacebrightnessmodelstoourdata.Themostcommonly galaxies(Matsushitaetal.2000;Matsushita2001),butclearlyfit- usedinthisworkwasamodifiedKingfunction(or“β–profile”)of tedvaluesarepreferable. theform: We then attempted to fit two component spectral mod- S(r)=S (1+(r/r )2)−3βfit+0.5 (1) 0 core els for each galaxy. These generally included a power–law + MEKAL model and bremsstrahlung + MEKAL model in which whereS(r)isthesurfacebrightnessatagivenradius, S0 is thebremsstrahlungtemperaturewasfixedat7keV.Thefirstcom- thecentralsurfacebrightnessrcoreisthecoreradiusandβfit isa ponent was intended to represent a hard component produced measure of theslope of the surfacebrightness profile.At various by the population of X–ray binaries and other unresolved stellar stages of the analysis we also fitted point source models and de sourceswithineachgalaxy.RecentChandrastudiesofextragalac- Vaucouleursr1/4lawmodels,usingtheform: ticX–raybinarypopulationssuggest thatthisemissionisreason- S(r)=S exp 7.67[(r/r )0.25 1] (2) ablymodeledbyapower–lawofindex 1.2(Sarazinetal.2001; e· {− e − } ∼ Blantonetal. 2001), while ASCA studies show good fits using a whereS isthesurfacebrightness atr , theeffectiveradius e e high temperature bremsstrahlung model (Matsushitaetal. 2000). (theisophotalradiuscontaininghalfthetotalluminosity.) All models were fitted using the Cash statistic (Cash 1979). The ModelswereconvolvedwiththePSPCpointspreadfunction Cashstatisticisdefined as-2lnL whereL isthe likelihood func- atanenergydeterminedfromthemeanphotonenergyoftheemis- tion.ThismeansthatthemostlikelymodelhasaminimumCash sionintheregionofinterestandthenfittedtothedata.Bothspheri- statisticandthatdifferencesinthestatisticarechi-squared(χ2)dis- calandellipticalfitswerepossiblewhenusingtheKinganddeVau- tributed. Thus confidence intervals can be calculated in the same couleursmodels,withthepositionangleandmajortominor axis was as for a conventional χ2 fit. By comparing the best fit Cash ratio measuring the shape and orientation of elliptical fits. When statisticforeachmodel,andvisuallyexaminingthespectralfit,we using King models, all parameters (core radius, β normalisa- fit selectedthebestfitmodelforeachgalaxy.Fromthiswecouldex- tion,xandypositionandtheellipticityparameters)wereusually tract(inmostcases)theX–raytemperatureandmetallicityofthe allowedtovaryfreely,asweretheparametersinpointsourcemod- galaxyhalo,aswellastheX–rayfluxfromthegalaxyhaloandthe els(xandyposition,normalisation).ThedeVaucouleursmodelis stellarcontribution. intendedtorepresenttheunresolveddiscretesourcepopulationof For each galaxy in the sample, we also derived simple pro- thegalaxies,whichweassumewilltakethesameformasthestel- jected temperature and hardness profiles. Temperature profiles lar population. An alternative approach would be to assume that were produced by splitting the larger, surface brightness region thediscretesourcepopulationfollowsthedistributionofglobular ofinterestintoseveralannuli,fromwhichspectrawereextracted. clusters,butwelackaccuratespatialmodelsofthisdistributionfor Thesespectrawerethenfittedusingthebestfittingspectralmodel manyofourgalaxies.Wethereforeinitiallysettheeffectiveradius forthegalaxyasawhole.Initiallythemodelswerefittedwiththe parameterofthedeVaucouleursmodelstothevalueoftheoptical metallicityandhydrogencolumndensityfrozenattheirglobalbest effectiveradius,andhelditfrozeninmostcases.Wedidallowthe fit values. However, if the data quality permitted, we freed these effectiveradiustovaryforasmallnumber ofgalaxies,wherethe (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 5 Figure1.Measuredprojectedtemperatureprofilesforoursampleofgalaxies (cid:13)c 0000RAS,MNRAS000,1–26 6 Ewan O’Sullivanetal. Figure1–continued (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 7 Figure1–continued Figure2. FittedmetallicityprofilesofNGC1399(leftpanel)andNGC4636(rightpanel) fitwaswellconstrainedbythedata,inordertoinvestigatediffer- determined, and a Gaussian fitted to the resulting spread of val- encesbetweentheopticalandX–raystellarprofiles.Howeveronly ues. Bycomparing theactual Cashstatistictothisdistributionof onegalaxy,NGC4697,wasbestfitbyamodelincludingadeVau- values, we were able to determine the probability that the model couleurscomponent,demonstratingthatinmostofthegalaxiesin could haveproduced thedata. Weweretherefore abletoidentify thesample,eitheremissionfromhotgasdominatesortheX-raybi- caseswherethe2-component fitwasnomorelikelytoreproduce narypopulationdoesnotfollowthestellarpopulation.AChandra thedatathanthe1-componentfit,anddiscardthe2-componentfits observation of NGC 4697 hasshown ittohavearelativelysmall forthesegalaxies. gas halo, with much of its emission contributed by point sources (Sarazinetal.2001),sothesuccessofthedeVaucouleurscompo- nentinthiscasecouldisperhapsunsurprising.TheChandradata 5 RESULTS for this galaxy are best fit by a surface brightness model which includesadeVaucouleurscomponentandaKingmodelwhosepa- 5.1 Spectralandspatialfits rametersaresuchthatitisflat,providingafairlyconstantcontri- Table2showstheresultsofourspectralfits.Asmentionedprevi- butionovertheareastudied. ously,metalabundancesfromROSATPSPCspectraareinherently The use of 2–dimensional datasets to fit the surface bright- unreliable,duetotherelativelypoorspectralresolutionofthein- nessdistributioncanresultinalownumber ofcountsinmanyof strument.Thisisreflectedbythelargeerrorsonsomeofourfitted the data bins. Under these conditions χ2 fitting performs poorly values,andbythefactthatinsomecaseswehadtoholdmetallic- (Nousek&Shue 1989) so, as in the spectral analysis, maximum ityfrozeninordertosecureastablefit.Thetemperaturevaluesare likelihood fitting based on the Cash statistic was used. However, morereliable,andgiveameantemperatureof0.67 0.29keV. ± theCashstatisticgivesnoindicationoftheabsolutequalityofthe AsdiscussedinSection3,apoorchoiceoflocalbackground fit,onlythequalityrelativetootherfits.Inordertogainsomees- for our targets could result in spectral fits biased by inclusion of timateofthetruefitquality,weusedaMonteCarloapproach, in group or cluster emission, or accidental subtraction of some of which the best fit 1- and 2-component model was used to gener- the galaxy emission. The clearest sign of this bias would be un- ate1000imagesofthegroups,towhichPoissonnoisewasadded. usuallyhighorlowfittedtemperature,significantlydifferentfrom Thesewerethencomparedtotheoriginalimage,theCashstatistic thosefoundinotherstudies.Fourofourgalaxieshavetemperatures (cid:13)c 0000RAS,MNRAS000,1–26 8 Ewan O’Sullivanetal. above 1 keV, and one (NGC 4697) has a temperature lower than cores(e.g.NGC1399),hotcores(e.g.NGC720),isothermal(e.g. 0.3keV.Theseareoutsidetherangecommonlyconsideredtypical NGC4697)andthosewherethedataqualitypreventsajudgment forellipticalgalaxies,sowecomparetheresultsforthesegalaxies (e.g. NGC 4552, where the errors on the outermost bin are large to those in the literature. NGC 507 has atemperature marginally enoughtomakeitsuspect).Coolandhotcoregalaxiesareselected above1keV.PreviousROSAT andASCAstudieshavefoundsim- undertherequirementthattheircentralbinmustbehotterorcooler ilar temperatures (Kim&Fabbiano 1995; Matsumotoetal. 1997) thananaverageoutertemperaturebyatleast20%,andthattheer- andmetalabundances(Buote2000),andmorerecentChandradata rorsinT mustbesmallerthanthisamount.Itisalsopossibleto X alsosupportsatemperatureof 1keV(Formanetal.2001).Buote seeevidenceofAGNactivityinsomeoftheprofiles,particularly ∼ (2002) fits a two temperature model to XMM-Newton EPIC data in the case of NGC 5128, where the central bin has a tempera- forNGC1399,andrecoverstemperaturesof 1.5and 0.9keV tureof 5keVandtherestofthegalaxy<1keV.Aninteresting within1′,withbothcomponentsapproachinga∼temperatu∼reof1.3- feature∼of some of thebetter defined profileswhich show central 1.5keVat3-10′.Ourvalueof 1.2keVisquitecomparabletothe cooling(e.g.NGC507,NGC1399,NGC4636)isthatthetemper- ∼ coolercomponent,consideringtheregionfromwhichourspectrum ature rises with radius to a value above the apparent outer mean wasextracted.NGC4073hasthemostextremetemperatureinTa- temperature,andthefallsbacktothatmean,producingatempera- ble2,kT=1.6keV.AnalysisofXMM-NewtonEPICdataforNGC turepeakatmoderateradii.Asallofthesegalaxiesareembedded 4073 and its surrounding group (O’Sullivanetal. 2003, in prep.) inlargergrouporclusterhalos,thispeakmaymarktheboundary suggestatemperaturegradientwithinthestellarbodyofthegalaxy, ofagrouporclusterscalecoolingflow,orthepointofinteraction withprojectedtemperaturesrisingfrom1.4to 2keV.Wealsofind between the galaxy and its environment. The observations avail- ∼ a high temperature for NGC 6269, kT=1.4 0.2 keV. No XMM- ableforNGC1399andNGC4636containverylargenumbersof ± NewtonorChandraresultsareavailableintheliteratureforNGC counts,allowingustoincludemetallicityasafreeparameterinthe 6269,butpreviousanalysesoftheROSATdatahavefoundtemper- profilefitting.Metallicityprofilesofthesetwogalaxiesareshown aturesof1.3 0.15keV(Dahlem&Thiering2000)and1.36 0.07 inFigure2.Despitethelargeerrorsinsomebins,itisnotablethat ± ± keV (Mulchaeyetal. 1996). These are identical within the errors bothmetallicityprofilesfollowthesamestructureasseenintem- with our result. Lastly, NGC 4697 was one of the first elliptical perature;acentraltrough,risingtoapeakatmoderateradius,with galaxiestobeobservedwithChandra,whichshowedittopossess an outer region of relatively low abundance. The peak tempera- arelativelycoolgascomponentwithkT 0.29keV(Sarazinetal. tureandmetallicityoccuratapproximatelythesameradiusinboth ∼ 2001).Thisisafairlygoodmatchtoourmeasuredtemperatureof cases.Inordertocheckthatcorrelationsbetweentemperatureand kT=0.24 0.2keV, anditshouldbenoted thatat leastpart ofthe abundanceinthefitswerenotbiasingtheresultswemodeledthe ± differencebetweentheseresultsmayarisefromChandracalibra- fitspaceforthetwobinsoneithersideoftheapparentbreakinthe tionissues.Ingeneral,thesecomparisonssuggestthatourmethod profileofNGC4636,calculatingfitstatisticsatarangeoftemper- ofbackgroundselectionanddataanalysisproducesfairlyaccurate aturesandmetallicities.Comparingtheconfidenceregionsforthe fitstothedata. twopointsshowsthattheyaredissimilartoatleast9σsignificance, Whencalculatingtheluminositiesofourtargets,wearefaced stronglysuggestingthebreakintheprofileisreal.Confidencere- with the difficulty that while we have surface brightness models gionsforthetwobinsareplottedinFigure3. which should provide an accurate estimate of the total number Table3liststheresultsofthesurfacebrightnessfittingforthe of counts from the galaxy halo, we only have spectral informa- galaxyhalosofourtargetgalaxies.Inthemajorityofgalaxieswe tion for a smaller central region. Ideally we would be able to si- obtaingoodqualityfits,withrelativelysmallerrorsonthecorera- multaneouslyfitspectralandspatialdata,givingatrueluminosity diusandslope.Asweareabletofitellipticalmodels,wealsolist foreachcomponent(e.g.Lloyd-Daviesetal.2000).Inpracticewe thepositionangleofthemajoraxisandaxisratioofthemodelfits. havechosentoscaleuptheMEKALcomponentofthespectralfit Forfivegalaxies,wewereunabletodetermineareliableposition to the number of counts found for the surface brightness model. angle,asthemodelwasconsistent(withinerrors)withbeingspher- Thisallowsustocalculateagasluminosityforthegalaxycompo- ical.Thebestfitpositionanglesofthesegalaxiesarelistedwithout nent,butignoresthecontributionfromdiscretesources.Thislumi- errors,andshouldonlybeconsideredasroughestimates. nositywillthereforebeanoverestimateofthetruegasluminosity associatedwiththegalaxy.However,weexpectlargeopticallylu- 5.2 TheL :T relation minousgalaxiessuchasourtargetstobealmostentirelydominated X X bygasemission. Withasmallnumber ofexceptions, thespectral Figure4showsL plottedagainsttemperatureforoursample.As X fitsconfirmthis,suggesting thatinmost casestheoverestimation can be seen, we find a fairly tight relation between L and T , X X issmall.Itisalsonotablethatbecausethebremsstrahlungcompo- with only a small number of outlier points. Using Kendall’s K- nentpeaksatahigherenergythantheMEKAL,agivenluminosity statistic(Ponman1982)tomeasurethestrengthofcorrelation,we correspondstoasmallernumberofbremsstrahlungcounts(inthe find a significance of 4.6σ. We fitted the relation using ODR- ∼ ROSAT band)thanitwouldforaMEKALmodel.Thismeansthat PACK (Boggsetal. 1989) to perform orthogonal least squares re- ouroverestimateofluminosityisreduced,asthenumberofcounts gression, and found a slope of 4.8 0.7. This fittingmethod uses ± associatedwiththebremsstrahlungcomponent,andassumedinthe theerrorsinxandyforeachpoint,butisunabletotakeofaccount scalingtobepartoftheMEKALcomponent,willproduceonlya ofthefactthattheerrorsareasymmetric. Themeanof theupper smallincreaseingasluminosity. andlowererrorisusedfortheerrorineachaxis. Temperatureprofilesforoursampleofgalaxiesareshownin This fittingmethod should be accurate as long as the points figure1.Theextentoftheprofilesvaries,owingtotherelativequal- deviate from the mean relation only on account of the statistical ity of data, length of exposure and galaxy distance. Comparison errors.Incaseswherethereisalargeintrinsicscatteraboutthere- of these profiles with those shown in Helsdon&Ponman (2000) lation,thestatisticalerrorsdonotprovideanappropriatebasisfor shows them to be similar in most cases where the samples over- theweightingofthedatapoints.Analternateapproachistoweight lap.Wehaveclassifiedthegalaxyprofilesintofourgroups;cooling allthepointsequally,ignoringthestatisticalerrorsasmisleading. (cid:13)c 0000RAS,MNRAS000,1–26 X–rayScalingPropertiesof Early–typeGalaxies 9 Name nH TX Z LogLX Model Profile (×10−21cm−2) (keV) (Z⊙) (ergs−1) ESO443-24 0.50+5.3 0.69+0.07 1.0+1.0 41.78+1. MK+BR I −0.19 −0.39 −0.7 −0.06 IC1459 0.069±0.012 0.51±0.05 1.0 40.3±0.2 MK+BR H IC4296 0.63+0.33 0.72±0.07 0.23±0.92 41.7+0.3 MK+BR C −0.24 −0.8 IC4765 2.6+3.8 0.64+0.09 0.17±0.17 42.4+0.7 MK+BR C −0.9 −0.37 −0.1 NGC499 0.79+0.32 0.70+0.02 1.0 42.6±0.3 MK+BR I −0.14 −0.03 NGC507 0.59±0.09 1.03±0.05 0.95+2.2 42.9+0.4 MK+BR C −0.33 −0.7 NGC533 0.29±0.03 0.84+0.05 1.1±1.1 42.2±0.2 MK+BR C −0.04 NGC720 0.15±0.15 0.50±0.04 0.28+0.06 41.1±0.3 MK+BR H −0.04 NGC741 0.54+0.11 0.71±0.07 0.26±0.26 41.9+0.3 MK+BR C −0.08 −0.4 NGC1332 0.16+0.04 0.41+0.06 0.59±0.59 40.6±0.2 MK+BR H −0.03 −0.05 NGC1380 0.18±0.10 0.30+0.08 0.11+0.44 40.4+0.4 MK+BR ? −0.05 −0.07 −1.0 NGC1395 0.069±0.016 0.65+0.04 1.0 40.6±0.2 MK+BR C −0.05 NGC1399 0.12±0.01 1.21±0.03 1.1+0.3 41.8±0.3 MK+BR C −0.2 NGC1404 0.18±0.02 0.60±0.01 0.35+0.05 41.66+0.04 MK C −0.04 −0.05 NGC1407 0.72+0.22 0.79+0.08 0.14+0.14 41.5+0.2 MK+BR C −0.16 −0.07 −0.06 −0.3 NGC1549 0.046±0.046 0.25+0.08 0.14+0.27 39.7+0.2 MK+BR I −0.05 −0.08 −0.4 NGC1553 0.045±0.033 0.53±0.15 0.10+0.16 40.2+0.3 MK+BR ? −0.05 −0.4 NGC2300 0.97+0.97 0.62+0.07 0.23±0.23 41.43+0.2 MK+BR ? −0.30 −0.14 −0.07 NGC2832 0.13±0.02 0.82±0.05 1.0 41.9±0.2 MK+BR ? NGC3091 0.38+0.08 0.64+0.04 1.0 41.7±0.2 MK+BR C −0.06 −0.05 NGC3607 0.015±0.015 0.45±0.06 0.71+0.57 40.7+0.3 MK+BR I −0.21 −0.4 NGC3923 0.58±0.22 0.46+0.04 1.0 40.8±0.2 MK+BR H −0.05 NGC4073 0.16+0.02 1.6±0.2 1.6+1.1 43.1+0.1 MK+BR C −0.03 −0.4 −0.3 NGC4125 0.10±0.04 0.34+0.05 0.34±0.34 41.27±0.07 MK ? −0.04 NGC4261 0.087±0.017 0.67+0.04 1.3±1.3 41.0±0.2 MK+BR C −0.05 NGC4291 0.22+0.18 0.59+0.06 0.63+4.8 41.20+0.4 MK I −0.09 −0.07 −0.13 −0.02 NGC4365 0.11+0.06 1.0+0.3 0.064+0.11 40.5±0.1 MK I −0.05 −0.2 −0.053 NGC4472 0.162±0.007 0.88±0.01 1.0 41.5±0.2 MK+BR C NGC4552 0.18±0.03 0.54±0.06 1.0 40.7±0.2 MK+BR ? NGC4636 0.25+0.05 0.55±0.03 0.40+0.32 42.0+0.1 MK+BR C −0.06 −0.13 −0.2 NGC4649 0.24+0.06 0.78±0.02 0.80+1.0 39.3+0.3 MK+BR C −0.07 −0.36 −0.5 NGC4697 0.16±0.03 0.24±0.02 0.40±0.40 39.0±0.2 MK+BR I NGC5128 0.43+0.07 0.35+0.04 1.0 40.2±0.2 MK+BR H −0.05 −0.03 NGC5322 0.045±0.037 0.33+0.100 0.19±0.19 40.3±0.3 MK+BR ? −0.06 NGC5419 0.29+0.07 0.69±0.26 0.32±0.32 42.0+0.3 MK+BR ? −0.05 −0.4 NGC5846 0.30±0.03 0.66±0.02 1.0 40.5+0.2 MK+BR C −0.3 NGC6269 0.53±0.12 1.4±0.2 0.33+0.36 43.4±0.1 MK C −0.17 NGC6482 0.68+0.72 0.55+0.04 1.0+1.0 41.2+0.5 MK+BR I −0.19 −0.07 −0.5 −0.4 NGC7619 0.34+0.08 0.81±0.03 1.9±1.9 42.0+0.3 MK+BR C −0.06 −0.8 Table2. Resultsofthespectralfitstooursamplegalaxies.Wherepossible,aabsorbedMEKAL+bremsstrahlung(MK+BR)modelwasfitted,butincases wherethebremsstrahlungnormalisationalwaystendedtozero,thiscomponentwasremovedfromthefit.Allupperandlower(1σ)errorsonfittedparameters werecalculatedindividually,butareshownasasingle±errorwhentheyareidenticaltotwosignificantfigures.Thosegalaxiesforwhichmetallicitycould notbesuccessfullyfittedarelistedwithafixedsolarmetallicitywithnoerrors.Temperatureprofilesareclassifiedasisothermal(I),coolcore(C),hotcore (H),oruncertain(?). To check our result we also fitted our L :T relation using the plottedalongsidethoseforsamplesofgroups(Helsdon&Ponman X X SLOPESpackage(Isobeetal.1990)toperformanOLSbisectorfit 2000)and clusters(Davidetal.1993;Mushotzky&Scharf1997; andignoringstatisticalerrors.Thebestfitslopeforthistechnique Fairleyetal. 2000).Our datafollow arelationof similarslopeto was5.05 0.44,verysimilartotheslopefoundwhenusingtheer- that of the groups (Helsdon & Ponman find a best fit slope of ± rors. 4.9 0.8), but offset to a lower luminosity or higher temperature. ± Foragiventemperature,ourgalaxiesareafactorof 3lesslumi- From the temperature profiles for each galaxy, we are able ∼ nous. Because of the scatter inbothsamples, thereissomeover- to identify which of our targets have strong temperature gradi- lapbetweenthegroupsandgalaxies,andsomeofourgalaxydata entswhichcouldaffectthemeasuredmeantemperature.Excluding pointslieabovethegroupbestfitline.Conversely,thebestfitre- these20galaxiesweakenstherelationto 2.8σ significance,and ∼ lationforclustersissignificantlyshallowerthanthatforgroupsor givesabestfitslope(fittedusingstatisticalerrors)of5.9 1.3.We ± galaxies,though againthereisasmallregion ofoverlapbetween alsofitthesampleofgalaxieswithknown temperaturegradients, themostluminousgalaxiesandthefaintestclusters. andfounda 2.7σcorrelation,withaslopeof 3.7. ∼ ∼ The L :T relation has been used extensively in the study X X ofgroupsandclustersofgalaxies.Figure5showsourdatapoints (cid:13)c 0000RAS,MNRAS000,1–26 10 Ewan O’Sullivanet al. Name CoreRadius βfit AxisRatio PositionAngle RoIRadius Model (arcmin) (degrees) (degrees) ESO443-24 0.106±0.004 0.55±0.03 1.7+0.3 306±7 0.045 KI −0.2 IC1459 0.17±0.02 0.89+0.36 1.6+1.4 276±7 0.065 KI+KI −0.09 −0.2 IC4296 0.080±0.004 0.66+0.18 2.2+0.8 29+10 0.12 KI+KI −0.02 −0.4 −8 IC4765 0.39±0.02 1.24+0.6 1.1+0.1 190 0.17 KI+KI −0.06 −0.7 NGC499 0.85±0.04 0.70±0.03 1.05+0.03 290+30 0.25 KI+KI −0.04 −20 NGC507 0.82±0.03 0.54±0.03 1.05+0.04 270 0.14 KI+KI −0.8 NGC533 0.058±0.002 0.53+0.04 1.24±0.06 240±10 0.16 KI+KI −0.02 NGC720 0.15±0.02 0.483+0.010 1.34+0.09 294+20 0.11 KI+KI −0.009 −0.08 −7 NGC741 0.172±0.006 0.65+0.64 1.41+0.08 128+10 0.12 KI+KI −0.06 −0.1 −9 NGC1332 0.010±0.001 0.548+0.009 1.8+0.3 295±4 0.080 KI+KI −0.008 −0.2 NGC1380 0.090±0.011 0.51+0.04 1.1+0.2 210 0.045 KI −0.03 −0.3 NGC1395 0.23±0.03 0.52+0.06 1.3±0.1 41+9 0.10 KI+KI −0.03 −10 NGC1399 0.11±0.01 0.59±0.02 1.23+0.04 176±3 0.12 KI+KI −0.03 NGC1404 0.32±0.04 0.770+0.005 1.05+0.02 294+10 0.20 KI+KI −0.004 −0.01 −6 NGC1407 0.18±0.02 0.56+0.02 1.20±0.06 91+7 0.20 KI+KI −0.01 −8 NGC1549 0.021±0.003 0.509±0.003 1.56+0.10 174+7 0.090 KI −0.09 −8 NGC1553 0.43±0.07 0.66+0.06 1.4+0.2 304+10 0.17 KI+KI −0.09 −0.1 −9 NGC2300 0.23±0.02 0.69+0.12 1.1+0.2 330 0.16 KI+KI −0.06 −0.1 NGC2832 0.0100±0.0003 0.314±0.007 1.4±0.1 357±6 0.15 KI+PS NGC3091 1.17±0.05 1.60+0.03 1.31±0.04 325+7 0.15 KI+KI −0.04 −6 NGC3607 0.63±0.07 0.48±0.02 1.17+0.07 350±10 0.14 KI −0.06 NGC3923 0.010±0.001 0.55+0.05 1.8+0.4 54+6 0.080 KI+KI −0.01 −0.2 −5 NGC4073 0.072±0.002 0.46+0.02 1.20+0.06 265+7 0.10 KI+KI −0.01 −0.05 −8 NGC4125 0.017±0.001 0.48+0.01 1.629+0.3 269+8 0.070 KI+PS −0.08 −0.003 −9 NGC4261 0.37±0.03 1.2+1.4 1.8+0.3 37±5 0.10 KI+KI −0.2 −0.2 NGC4291 0.38±0.04 0.57+0.04 1.3±0.1 106+9 0.065 KI −0.03 −10 NGC4365 0.54±0.08 0.60+0.04 2.0±0.2 31±4 0.090 KI −0.03 NGC4472 0.25±0.04 0.597+0.009 1.08±0.02 83+10 0.24 KI+KI −0.008 −9 NGC4552 0.098±0.014 0.60+0.02 1.7+0.2 164±5 0.045 KI −0.03 −0.1 NGC4636 0.40±0.06 0.535+0.007 1.02+0.02 24 0.24 KI+KI −0.006 −0.09 NGC4649 0.13±0.02 0.567±0.008 1.18±0.03 26±6 0.080 KI+PS NGC4697 0.99±0.15 0.46+0.13 1.8±0.3 28+6 0.045 KI+DV −0.07 −12 NGC5128 0.90±0.53 0.55±0.01 2.21±0.07 52.0±0.9 0.25 KI+KI NGC5322 0.0100±0.0008 0.49±0.01 1.6+0.3 57+8 0.057 KI −0.2 −9 NGC5419 4.5±0.2 0.50+0.18 1.44±0.06 44±6 0.20 KI+PS −0.07 NGC5846 1.3±0.1 0.80+0.05 1.15±0.03 45+7 0.080 KI+PS −0.04 −6 NGC6269 1.21±0.02 0.40+0.04 1.19+0.1 38±15 0.10 KI+PS −0.02 −0.08 NGC6482 0.162±0.007 0.524+0.009 1.13+0.07 220±10 0.10 KI+PS −0.001 −0.05 NGC7619 0.031±0.002 0.447+0.006 1.18+0.07 308±10 0.20 KI+KI −0.005 −0.06 Table3. Resultsofthesurfacebrightnessfitstoourgalaxies,forthecomponentsassociatedwiththegalaxyhalo.Bestfitpositionanglevalueswithouterrors aregivenforthosegalaxieswheretheanglewasessentiallyunconstrained.Allerrorsarequotedatthe1σ. 5.3 L :L andL :T relations ieshavingX–rayluminositiesafactorof 9higherthanthoseof X B B X ∼ groupswithequalopticalluminosity,orconverselyL values 2.3 B ∼ One of the more common relations used in studies of early-type timeslowerthangroupsofsimilarLX.Thisresultcanbecompared galaxiesistheLX:LBrelation.NumerousstudiesbasedonROSAT, totheLX:TX relationshowninSection5.2,inwhichgalaxiesare ASCAorEinsteindatahavebeenpublished(e.g.Beuingetal.1999; offset to lower luminosities at a given temperature, compared to Matsushita2001;Brown&Bregman1998;Fabbianoetal.1992), groups. and we have previously examined this relation in some detail in O’Sullivanetal.(2001a),towhichwedirectreadersforafulldis- The L :T relation for our galaxies is shown in Figure 7. B X cussion of the relation and the effects of galaxy environment on Once again, for this relation we find a fairly strong correlation it. Figure 6 shows the L :L relation for our galaxies. For the ( 4σ significance). The slope of the relation is comparable to X B ∼ sample as a whole, there is a 3.9σ correlation, with a slope of thatfoundforgroupsandclusters;1.91 0.33forourgalaxysam- ± 2.7. Thisisquite asteeprelation, comparable tothat found for ple,1.64 0.23forgalaxygroups(Helsdon&Ponman2002),and ∼ ± asampleofBGGsinpreviouswork(O’Sullivanetal.2001a).We 1.5 for galaxy clusters (Lloyd-Davies&Ponman 2002). How- ∼ havealsoplottedthebestfitrelationfoundforX-raybrightgalaxy ever,wheretherelationsforgroupsandclustersareessentiallythe groups,fromHelsdon&Ponman(2002).Theslopeofthisrelation same(Helsdon&Ponman2002),ourrelationforgalaxiesissignif- (2.6 0.4)isverysimilartothatfoundforoursampleofgalaxies. icantlyoffsettohighertemperatures(byafactorof 2,compared ± ∼ However, our relation is offset from that for groups, with galax- togroups)orlowerL (byafactorof 3). B ∼ (cid:13)c 0000RAS,MNRAS000,1–26

See more

The list of books you might like

Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.
DMCA & Copyright: Dear all, most of the website is community built, users are uploading hundred of books everyday, which makes really hard for us to identify copyrighted material, please contact us if you want any material removed.
Copyright @ PDFdrive.to, 2022 | We love our users