1988ApJ. . .335. .215P similarity isprobablydueto unstableradiativecoolingand ing resemblanceonallscales totheopticalappearance.This der Laan1975;Green1986; Mufson etal1986)bearsastrik- uum radiosurfacebrightness at 6cmand20(Duinvan medium. ThisenhancementisprobablyanHicloud(DeNoyer located 2°tothenortheastof IC443(Fesen1984).Thecontin- and alarge-scaledensityenhancementinthelocalinterstellar which hasarisenfromtheencounterbetweenshockfront network ofbrightfilaments,fillingthenortheasternquadrant, optical shell,itsappearanceisdominatedbyacomplex bands. Althoughtheremnantpossessesanearlycomplete with structuresintheinterstellarmediumarepresentthrough- out thereportedobservationsinX-ray,radio,andoptical distinctive characterarisingasaconsequenceofitsinteraction interstellar medium. pictures ofasupernova’sinteractionwiththeinhomogeneous remnant. Nevertheless,itisstillpossibletosynthesizeoverall preclude simplemodelingofanybutthesmallestregionsa 1978), whichisprobablyassociated withtheHnregionS249 by itsimmediateenvironment,whileatthesametimetendsto to giveeachsupernovaremnantadistinctivecharactershaped clumps ofdiversedensityandcomposition.Thisvarietytends neither homogeneousnorsmoothlyvarying,butarefilledwith ger 1983).Theconcretecircumstellarandinterstellarmediaare not faithfullyreproducethetheorist’ssphericallysymmetric ground-based observations,showthatphysicalremnantsdo imaging andspectralresolutioninwavebandsaccessiblefrom Sedov (1959)hydrodynamics(see,e.g.,GorensteinandDanzi- wealth ofimagesandspectraavailablefromthelastdecade’s mathematical abstractionsconstructedtodescribethem.The series ofX-rayobservatories,combinedwithadvancesin have revealedthatfew,ifany,closelyresembletheelegant The AstrophysicalJournal,335:215-238,1988December1 © 1988.TheAmericanAstronomicalSociety.AUrightsreserved.PrintedinU.S.A. In thecaseofsupernovaremnantIC443,indications a Recent detailedstudiesofindividualsupernovaremnants (0.1 pc)shell. Subject headings:nebulae:individual(IC443)—supernovaeX-rays:sources gests thatdespitetheatypicalX-rayappearance,emissioninIC443isprobablyconfinedtoathin remnant areaccountedforbyvariationsinabsorptionaninterveningmolecularcloud.Ouranalysissug- librium orthattheX-raysariseinamultiphasemedium.PronouncedsoftX-rayspectraldifferencesacross correlation betweenX-rayandopticalorradiofeatures.Thebest-fitmodelsofthelow-energyspectrum of thebrightestarearemnantsuggesteitherthathasnotyetattainedionizationequi- observe asoftX-rayappearancehighlyatypicalofsupernovaremnantintheadiabaticphase,withlittle using theSSS,IPC,andHRIofEinsteinObservatoryMEDHEAO1A-2experiment.We © American Astronomical Society • Provided by the NASA Astrophysics Data System We presenttheresultsofacomprehensivestudyX-rayemissionfromsupernovaremnantIC443, A COMPREHENSIVESTUDYOFTHEX-RAYSTRUCTUREANDSPECTRUMIC443 I. INTRODUCTION Laboratory forHighEnergyAstrophysics,NASA/GoddardSpaceFlightCenter Received 1986August26;accepted1988May31 Harvard/Smithsonian CenterforAstrophysics R. PetreandA.E.Szymkowiak Leicester University R. WlLLINGALE F. D.Seward ABSTRACT 215 AND -1 4 _1 6 _1 -1 -1 -3 443 hasencounteredeither the molecularcloudorHi Haynes (1979)indicatesthatvirtually theentirefrontfaceofIC of shockchemistry.Amap shockedHibyGiovanelliand (DeNoyer 1979a,b;DeNoyer andFrerking1981;Treffers front ofIC443andinteractingwithitisdemonstratedby the H, CO,andOH,allwitha typicalvelocityof—30kms detection of10differentshocked molecularspeciesincluding extinction inthecentralregionsofremnant.Thatitis in western partsofIC443(Cornett,Chin,andKnapp1977;Sco- ville etal1977).Thiscloudisresponsibleforsignificantvisual molecular cloudcoveringvirtuallytheentiresouthern and revealed thepresenceofalarge(M%10M),thin(3±1 pc) filaments. ObservationsofCOlineradiationat2.6cmhave 1979). IC443hasthusbecome arichlaboratoryforthestudy stellar featuresisnotisolatedtothelocaleofbrightoptical shock velocityof~400kms(Woodgate,Lucke,andSocker around thesefilamentswithTæ10K,corresponding to a ing mediummuchfasterthantheopticalfilamentsaremoving. optical filamentsconfirmsthepresenceofsomegasinor The detectionof[Fex]26374coronalemissionnearthebright suggests thattheshockfrontispropagatingintosurround- having radialvelocitiesof400-500kms(Lozinskaya1979), east ofHaemissionfromgasunassociatedwiththefilaments, for thefilamentsof65kms(Lozinskaya1969).However, some filaments(Lozinskaya1979).Thepresenceinthenorth- radial velocitiesashigh220kmshavebeenobservedin 2 1979). Kirshner 1980).Hainterferometryhasyieldedameanvelocity gas areontheorderof15,000Kand500cm(Fesen typical temperaturesanddensitiesinsidetheopticallyemitting compression ofshockedgasandtheconsequent to thedirectionofshock. of ambientmagneticfieldsandrelativisticparticlestransverse 0 The interactionbetweentheshockfrontandlargeinter- Spectrophotometry ofthebrightfilamentshasindicatedthat 1988ApJ. . .335. .215P 1-2 _1 216 PETRE (Mufson eia/.1986). (5.8 x10"ergscmsinthe2-6keVband—Forman et comprises theremainderofoptical/radioshell.Thisinter- containing thebrightnortheastopticalfilaments,andthird confirms theexistenceofthisbreakout.Theirinterpretation into alarge,low-densityinterstellarcavitytothesoutheastof work demonstratestheexistenceofa“breakout”shock probably fromatomiclineemissionshockedclouds der iscurrentlyundetermined,althoughapproximately20% gas-to-dust ratio(Dweketa/.1987).Theoriginoftheremain- cloud. Approximatelyhalfoftheinfraredemissionarisesas infrared emissionregionsarebestspatiallycorrelatedwiththe primary reasonforthis,thelowobservedfluxfromIC 443 hinted attheunusualnatureofIC443,butmostwerecarried et al(1986). pretation hasbeendisputedbybothGreen(1986)andMufson nected shells.Oneisthebreakoutregion,asecondregion their dataisthatIC443consistsofthreedistinct,intercon- of aphysicalconnectionbetweenIC443andtheHnregion the optical/radionebulosity.Fesenalsofoundstrongevidence shell doesnotrepresenttheentiresupernovaremnant.His that IC443asdemarcatedbytheextentofoptical/radio thermal emissionfromshock-heateddust,assuminganormal luminosity tobe360timestheX-rayluminosity.Thebrightest IRAS bandluminosityisbyfarthedominantsourceofradi- fim bands(BraunandStrom1986;Mufsonetal1986).The into accounttwoimportantaspects.First,observationsusing expansion, itseemsasthoughinteractionwithitsdenseneigh- cloud andthatapproximately1000Mhasbeensweptupby cidence toarisefromthepropagation oftheshockfrontintoa eastern filaments.Parkeset al (1977)interpretedthiscoin- luminous inX-rays. pared withoptical/radiomorphologicalcounterpartssuch as al 1978),alsomadespectralanalysisquitedifficult.Whencom- resolution andsensitivitytoperformdetailedmapping. The subsequent revelations,andalllackedsufficientspatial out priortothediscoveryofshockedmolecularcloudand of 1.5kpc.SensitiveHimappingbyBraunandStrom(1986) shocked Hiemissionfromthenortheastandmolecular ative cooling;Dweketal(1987)estimatetheIRASband bright sourceoffar-infraredemissioninthe12,25,60,and100 the IRASsatellitehaverevealedthatIC443isanextremely now eventhemostdetailedworkonremnanthasnottaken bor hasforcedIC443intoasortofpremature“midlifecrisis.” s orhigherelsewhereintheremnantsuggestingadiabatic ative, momentum-conservingexpansionphase(Cox1972; that atleastsomeportionsofIC443haveenteredtheradi- the shockfront.ThepresenceofthisHishellstronglysuggests formed themostinformative study todateoftheX-raystruc- X-ray emissionwassupplied by Levineetal(1979)whoper- dense cloud.Acontraryinterpretation oftheorigin all, theX-rayemissionoriginatesbehindbrightnorth- al counteronAriel5,claimedthatatleasthalf,andpossibly and Parkesetal(1977),observingwiththeMSSLproportion- the grazing-incidencesoftX-rayconcentratorsonCopernicus, S249, therebyestablishingafirmdistanceestimateforIC443 Straka 1974;Falle1975).Withtheshockvelocitiesof400km ture byproducinga0.5-1.5keV imagefroma60ssounding Puppis AandtheCygnusLoop,IC443appearshighlyunder- 0 In theX-rayband,previousmorphologicalstudieshave Charles, Culhane,andRapley(1975),observingIC443with Second, opticalspectroscopybyFesen(1984)hasrevealed Recent discoveriesaboutIC443haverevealedthatuntil © American Astronomical Society • Provided by the NASA Astrophysics Data System ET AL.Vol.335 721-2 mosaic maps:anIPCmapoftheentiresupernovaremnant, marized above,theseobservationspaintaninteresting and fields withintheportionofremnantwithhighest The spectroscopicdataconsistoffiveuseful6'diameterSSS prehensive viewtodateofthespatialandspectralmorphology compilation oftheobservationsusingthreefourfocal ments intheopticalandradiobandswasEinsteinObserva- inextricably boundwiththemolecularandHiclouds.As pre- keV. Takentogetherwiththeobservationsinotherbandssum- HE AO1A-2experimentareusedtoprovideaglimpseof the hr pointedobservationbytheMediumEnergyDetectorof the the entireremnantfromIPC.Inaddition,dataa6 surface brightness,aswellaspectralhardnessratiomapof and anHRImapoftheregionhighestsurfacebrightness. of IC443.Theimagingdatahavebeenusedtocreatephoto- Counter (IPC),theHighResolutionImager(HRI),and plane instrumentsonEinstein,theImagingProportional ness overlap,butarenotidenticalwith,thebrightnortheastern viously reported,theregionsofhighestX-raysurfacebright- self-consistent portraitofIC443,whoseoddmorphology is spatially averagedspectrumoftheentireremnantoutto ~ 8 measurements ofIC443comparableinqualitytomeasure- image, theX-ray-emittinggas inIC443isprobablyconfined properties bythespectralobservations, takentogetherwith duction ofX-rayemittinggas. Theconstraintsplacedongas surface ofIC443andtosome degreeterminatingthepro- suggest thatthecloudisabsorbingX-raysfrommuchof the of sighttoIC443andespeciallywherethetwoareinteracting spectral hardeningwherethemolecularcloudcrosses line optical filaments.AdecreaseinX-raysurfacebrightnessand a Solid StateSpectrometer(SSS),toyieldthemostcom- tory (Giacconietal1979).Inthispaperwepresenta X-ray bandtoperformspatialmappinganddetailedspectral mates ofsomeaveragetemperatureandcolumndensity. strongly suggestthatdespitethe impressiongivenbytheX-ray observational resultsinother bands,particularly[Fex], with highlyspatiallydependentspectradominatedbyline ( =1.2x10K),withacolumndensityof3-4cm.As rather thancontinuumemission,simplemodelsofanentire supernova remnantsareusuallyrelativelycomplicatedentities, The typicalbest-fittemperatureofthesefitsis~1keV (Galas, VenkatesanandGarmire1981;Charlesetal1981). with multicomponentthermalspectra,includinglineemission SNR arelimitedintheinformationtheycanprovidetoesti- bremsstrahlung models(e.g.,Malina,Lampton,andBowyer restricted portionsofIC443(i.e.,thenortheast)withthermal have beenmadetofitproportionalcounterspectraof perform detailedspectralmeasurementsofIC443.Attempts from theinteriororsouthernportionsofIC443. ments lackedthesensitivitytospatiallyresolveX-rayemission sensitivity andthespatialspectralresolutionneededto adjacent toeachotheralongtheshock.Allofaboveexperi- shock densityemittingopticallyarespatiallydistinct,lying responsible fortheX-radiationandregionsofhigherpre- gested thattheshockedregionsoflowpreshockdensity filaments, withnostrongspatialcorrelation.Levineetalsug- but alongthenorthernsegmentofshell,overlapping consisted ofonly~30events,itlocatedthebrightestregion 1976; Parkesetal1977)ortofitspectraoftheentireremnant soft X-rayemissionnotbehindnortheasternopticalfilaments rocket observation.Thoughthisimagehad~6'resolutionand The firstX-rayinstrumentsufficientlysensitiveinthesoft Previous X-rayspectralobservationshavealsolackedthe 1988ApJ. . .335. .215P hms No. 1,1988 imately 0.2-3.1keV.Noattempthasbeenmadeinthemosaic collected (inIPCchannels4-13). ing atthreedifferentlocationswithintheremnant.Thedetails image ofIC443.Thedarknessapixelvariesapproximately observation (allthreepointings),atotalof53,700photonswere of thesepointingsarelistedinTable1.Duringthe5130s shadowing bythewindowsupportstrongbackrequiredpoint- central pointsource. molecular cloud,andplacealimitontheX-rayfluxfrom produced viaWienerfiltering(Wiener1949;BraultandWhite worst residualmodulationduetothestrongbackribsappears map, onlythoseeventsoccurringinpulseheightchannels4-13 context ofthecollisionbetweensupernovashockand discuss theimplicationsofourresults,particularlyin as inferredfromthebest-fitspectralparameters.In§V,we devoted toadiscussionaboutthephysicalconditionsinIC443 in §IIandthespectralresultsIII.SectionIVwillbe to athin(~0.1pc)shell.WepresenttheHRIandIPCimages around ô=22°20'andisatworst~10%.Thefinalimagewas rib shadowsfromdifferentpointingsoverlaponeanother.The distortions intheIC443imagebecausenoinstancedo procedure tocorrectforshadowingintheimagedue mirror yieldsaneffectiveenergyrangeforthemapofapprox- interstellar mediumandthehigh-energycutoffofEinstein have beenincluded.Takingintoaccountabsorptionbythe map consistsof128x32"pixelsandiscenteredat IPC windowsupportstructure.Thisdoesnotproducelarge height spectrumwouldnotwashouttrueX-rayfeaturesinthe high-energy andlow-energyendsofthe16channelIPCpulse a =61448,ô22°45'.Sothatnoisecontaminationatthe subtracted andexposure-correctedphotomosaicmap.This 1971). C b Imaging theentiresurfaceofIC443withIPCwithout The threeimageswerecombinedintoasinglebackground- Figure lashowsagray-scalemapofthefinalphotomosaic a HEAO 1A-2MED Einstein SSS Einstein IPC... Einstein HRI: a c b SeeFig.2. Pointedobservation. SeeFig.4. Position 6 Position 5 Position 4 Position 3 Position 2 Position 1 Position 3... Position 2... Position 1... Instrument © American Astronomical Society • Provided by the NASA Astrophysics Data System II. IMAGINGOBSERVATIONS IC 443X-RayObservationLog a) IPCObservations TABLE 1 hms 6 1400 6 1300 6 1412 6 1448 6 1455 6 1345 6148 Field Center(1950) 14 48 14 24 14 00 13 59 14 00 13 19 R.A. X-RAY STRUCTUREANDSPECTRUMOFIC443 / 22 3500 22 4500 22 4800 22 4500 22 5100 22 4800 22 5000 22 3600 22 2100 22 4500 22 3200 22 4700 22°4700' Decl. Exposure 3 (10 s) 10.58 4.19 2.94 4.38 8.37 6.05 6.56 3.87 5.10 7.06 1.64 1.81 1.68 -1 o hms -2-1 -1 -2-1 -1 -21 -2_1 (Giovanelli andHaynes1979). TheHimaprevealswherethe shock frontisinteractingwith largestructuresinthelocal brightness, renderedasgray-scale intensity,andacontourmap of shockedHiinthevelocity range—32to—41kms southerly, towardtheradiocenter oftheremnant. HEAO A-2MEDandHED3—thispaper)arelocatedmore higher energyinstruments(OSO7—WinklerandClark1975; Garmire 1981);mostandpossiblyallthefluxdetectedby the cides withthisbrightestfeature(Galas,Venkatesan, and LED arisesthere.Incontrast,theerrorregioncentroids for detected bytheHEAOA-2LowEnergyDetectorsalsocoin- keV byLevineetal(1979).Thecentroidoftheemission maximum surfacebrightnessiswhatwasimagedin0.5-1.5 (Charles, Culhane,andRapley1975;Malina,Lampton, and Bowyer 1976;Parkesetal1977).The12'diameterfeature of observations oftheremnantwithnonimaginginstruments and Strom(1986). that ofonethethreedistinctsubshellsidentifiedbyBraun The extentofthisbrightregioncorrespondsapproximatelyto ness regionwhichfillsthenortheasternthirdofremnant. the firstat<5«2240',splittinghighX-raysurfacebright- extinction bisectingtheremnantroughlyfromnorthtosouth at a«61330,andsecondlaneextendingeastwardfrom rapidly doseemassociatedwithopticalfeatures—alaneof maximum levelwellbehindtheshell. filamentation exists,theX-raysurfacebrightnessreachesits (suggesting limb-brightening),totheeast,wheresimilaroptical northern shellborderinaregionofbrightopticalfilaments is not.AlthoughthedominantX-rayemissionfeatureabuts within theradio-opticalshell;innorthwestandsoutheastit and southwesttheX-rayemissionappearstobeconfined generally overlap,importantdifferencesexist.Inthenortheast bright regionalongthenorthernborderofremnant. morphology sorevealediscuriouslyatypicalofadiabatic south tonorthandfromwesteast,bya12'diameter relatively smoothsurfacebrightnessgradientsrunningfrom of alimb-brightenedshellexists:theimageisdominatedby supernova remnants(e.g.,PuppisA—Petreetal1982;Cygnus Loop—Rappaport etal1979;Ku1984).Littlesuggestion of IC443visibleintheX-raybandforfirsttime.The ness of~18IPCcountsarcminks. ks. Thehighestcontouragainrepresentsasurfacebright- contour levelsincreaselinearlybyincrementsof0.67counts arcmin ksfromthelowestcontourof1.2counts posed onthePalomerSkySurveyredplateofIC443.The counts s. imately theextentofopticalfilaments.Thebackground- increments of1.8countsarcminksfromaminimum for clarity;thelevelsarespacedlinearlyinsurfacebrightness erate surfacebrightnessregionsattheexpenseof“burning enhanced background,andtobringoutstructureinthemod- for thisfigurewaschosensothattheIPCcoverageisvisibleas subtracted IPCcountratecontainedinthemapis12.4±0.2 to amaximumof18.Thedashedcurverepresentsapprox- zero toamaximumof14countsarcminks.Thecontrast logarithmically withsurfacebrightness,fromathresholdof out ”thebrightestregions.Thecontourlevelshavebeenadded Figure 1cdisplaysasuperposition betweentheIPCsurface Emission fromthisbrightregionhasdominatedprevious The twositeswherethesurfacebrightnessvariesmost Although thedistributionsofX-rayandopticalemission In FigureIh,acontourmapoflahasbeensuper- As canbeseeninFigure1,theIPChasrenderedentirety 217 1988ApJ. . .335. .215P -2 _1 h m 4 curve representstheouterboundaryofoptical/radioshell.ThesoftX-raymorphologyishighlyatypicalamiddle-aged,adiabaticsupernova remnant. contours bothrepresent0.2-3.1keVsurfacebrightness.Thearespacedlinearlyinbrightnessincrementsof1.8countsarcmin ks.Thedashed coincide withlowX-raysurfacebrightness,theregionswhere interstellar medium:alargeHicloudtotheeastandmolecu- map isaWiener-filtered,exposure-correctedphotomosaic of two mapsexists:althoughmostoftheshockedHiregions lar cloudinthelineofsight.Nostrongcorrelationbetween IC 443isprovidedbytheHRImapdisplayedinFigure3. This surface brightness. shocked HiisnotdetecteddonecessarilyhavehighX-ray ô =22°42',andconsistsof128xpixelsdimensions played inatangentplaneprojection,centeredat=6 14 structed bythetechniquedescribedinPetre(1982).Itis dis- three separateHRIimages(seeTable1,andFig.2), con- despite bothlongobservation timesandrestrictionoftheHRI brightness. Ofthe9.2x10 countscollectedinthethree the verysmallnumberoftrue X-rayeventsintheimage, HRI resolution)andtheWiener filteringwerenecessitatedby observations toonlythose regions ofthehighestsurface 218 16" x16". Fig. la.—EinsteinIPCimageofsupernovaremnantIC443.Theisanexposure-corrected,filteredmosaicthreeexposures.gray scaleandsolid A higherresolutionviewofthebrightnortheasterncorner of The largepixeldimensions(compared withthe4"intrinsic © American Astronomical Society •Provided bythe NASAAstrophysics Data System b) TheHRIImage PETRE ETAL. -21 2 4 -21 -21 exists betweenthegeneralappearance ofIC443asseenbythe intensity of~25countsarcminks.Contoursatthelevels red plate. “real” with99%confidence. InFigure3bthecontoursfrom fluctuation abovethebackground, andthusmaybeconsidered clarity. Allthoseareasofthemaplargerthan4arcmin con- taining someamountofshadingrepresentsa3orgreater imately thelevelofnon-X-raybackground,toapeak darkness ofapixelvariesapproximatelylogarithmicallyfrom level diffuseemission. X-ray eventsoccurper16"xcell.Theresultinghigh level images during2.2x10sofobservation,only~2.3 Figure 3ahavebeensuperposed onthePalomarSkySurvey a thresholdof~8countsarcminks,representingapprox- Wiener filtering,inordertohighlightthestructureof low the brightestfeaturesofmap,nomorethan10-15 true counts, one-quarterofthetotal,representtrueX-rayevents; of cell-to-cellstatisticalfluctuationhasbeensmoothed via the remainderrepresentinternaldetectorbackground.Evenin 10, 13,and16countsarcminkshavebeenadded for A comparisonwithFigure 1 revealsthatlittledifference Figure 3aisagrayscalemapoftheHRIphotomosaic.The Vol. 335 1988ApJ. . .335. .215P 2_1 -2_1 -21 feature overlapsbrightoptical filaments.However,thereisa case anabsence)offluxbetween twobrighterregions. X-ray emissionfromtheregion observedbytheHRIiscon- entire complexofopticalfilaments,althoughmuch it is consequence oftheinstruments’differentspectralresponses. optical extinctionat<5ä22°40' coincideswithadrop(inthis tained withintheopticalshell. AsinFigurelb,thefingerof below thelowestcontourlevel.IncontrasttoIPC,all the gests thatanyspectralvariationsoverthisregionarenot sig- nificant enoughtoproducedifferencesinthetwoimages as a that theHRIimagerevealssomewhatmoredetail.This sug- HRI andthecorrespondingportionofIPCimage,except counts arcmin“ks.Thehighestcontourrepresentsasurfacebrightness of18countsarcminks.TheX-rayemissionappearslargelyuncorrelatedwiththe comparison ofthedetailsIC443intwobands.Thecontourlevelsincrease linearlybyincrementsof0.67countsarcminks~fromthelowestcontour1.2 Figure 3aindicatesthattheHRIdoesdetectX-raysfrom the bright opticalshell. No. 1,1988X-RAYSTRUCTUREANDSPECTRUMOFIC443219 ~12' x12'brightnessenhancement. Onlyabouthalfofthis Also, asintheIPCimage, HRImapisdominatedbya Fig. lb—ContourmapoftheIPCimageIC443overlaidonPalomarSkySurveyredplateremnant.MoreX-raycontourshavebeen added toafford 8 © American Astronomical Society •Provided bythe NASAAstrophysics Data System / / / / 22°20 22°40 22°00 23°00 hm 61 e42 (Woodgate, Lucke, andSocker1979). filament; thisisaslikelyaprojection effectasaphysicalassoci- ation. Interestingly,thehighest surfacebrightnessofdiffuse Only themoreeasterlyof twooverlapsabrightoptical surface brightnessoftheregion inwhichtheyareembedded. observed intheseknotsisapproximately 2-3timestheaverage soft X-rayfeaturesinIC443.Thehighestsurfacebrightness ments alongthenorthwesternlimb,representingbrightest minute levelintheformoftwosurfacebrightnessenhance- [Fe x]inIC443occursthe spacebetweenthesetwoknots background levelover—1!5(~0.75pcford=1.5kpc). brightness dropssharplyatthisedge,fromamaximumto the region andtheedgeofopticalshell.TheX-raysurface precise coincidencebetweenthenorthernedgeofthisbright Within thisbrightfeature,substructureisvisibleatthearc- a 1988ApJ. . .335. .215P (for thepurposeofspectralfitting).Atposition1,signal-to- between IC443andadjacentfeaturesintheinterstellarmedium(anHicloudtoeast,amolecularlineofsight). shall henceforthbedisregarded. noise ratioisinsufficientfor spectral analysis,andthesedata the individualspectrahaverelativelylowsignal-to-noiseratios contour map,wherethecircleateachpointinglocation rep- surface brightness.ThebrightestX-rayemissionanticorrelateswiththehighestHibrightnessregions,whichrepresentlocationsofstrongest interaction 220 the SSSobservationswerelimitedtobrighterareasofIC 443, the edgeofremnant(asisposition1).Despitefact that another. Thenorthernmostofthese,position2,isjustinside are atportionsofthebrightestemissionfeature,andthree of resents the6'fieldofviewSSS.Fourthesepointings details. TheyarealsoindicatedinFigure4,atopthe IPC locations arelistedinTable1alongwithotherobservational these liealonganorth-southlinepartiallyoverlapping one observed bytheEinsteinSolidStateSpectrometer(SSS).These mize theeffectofabsorption duetoicecryopumpedonto described elsewhere(Urry1984; Reichertetal.1985).Tomini- the time-dependenticelayerand backgroundcomponents,are 1 Fig. le.—SurfacebrightnesscontoursofHiinthevelocityband—30to—40kms"(GiovanelliandHaynes1979)superposedonagrayscale map oftheIPC Six separatepositionsinthenorthernhalfofIC443were Procedures forfittingtheSSS spectra,includingdealingwith © American Astronomical Society •Provided bythe NASAAstrophysics Data System a) EinsteinSSSObservations Ô III. SPECTRALOBSERVATIONS o# 23°00' 2230 22°15' 22°45' hm 6 161412 PETRE ETAL. (e.g., Clarketal.1982).Thesedifficultiesariseprimarilyfrom (Hamilton, Sarazin,andChevalier1983).Toaccountfor the moderate columndensitytoIC443. have assumedtwoseparatethermal components,andallowed dance determinationscannot be made. limitation ofthesedifficulties isthatabsoluteatomicabun- models ofplasmasinionizationequilibrium(Raymond and cally thin,collisionallyionizedplasma.Wehaveused both has aninconsequentialeffectonourresultsbecauseofthe photon energiesbelow0.75keVfromspectralanalysis.This the detector,weexcludedallpulseheightchannelsrepresenting response toasourcecontaining spatialstructure.Theprimary uncertainties ininputatomicphysicsandthedetector supernova remnantshavebeendescribedindetailelsewhere of MorrisonandMcCammon(1983)wasused. absorption ofX-raysbycoldgasinthelinesight,model that describedinBeckeretal.(1979),usingmodelsofanopti- Smith 1977,1979)andtime-dependentionizationmodels When fittingthespectrawith equilibrium plasmamodels,we The difficultiesofinterpretingthebestfitthermalmodels of Spectral modelfittingwasperformedinamannersimilar to a i) EquilibriumModels Vol. 335 1988ApJ.. .335. .215P 6 7 keV, wherethe emissivityoftheHe-likelines ofsiliconand nova remnants(e.g.,Beckeret al1979,1980;Clarket1982). published two-temperaturefits toSSSspectraofothersuper- In theseotherfits,thelowtemperature isusuallyaround0.7 san, andGarmire1981).The IC443fitsstandincontrastto A-2 LowEnergyDetector(Charles etal1981;Galas,Venkate- temperature fitstospectrafrom thelowerresolutionHEAO1 IC 443(andthusprimarilyofthebrightregion)inmulti- ally consistent.Thesetemperatures(~2.2x10K and peratures ofthetwocomponentsarerelativelylow,andgener- 1.1 x10K)comparefavorablywiththoseinferredforall of magnesium, silicon,sulfur,andirontovaryfreely.Belowwe the temperaturesofbothtobefreeparameters(cf.Beckeretal best-fit values. quoted rangesinTable2represent1aexcursionsfrom the volved best-fitmodeloverlaid,isshowninFigure5. The better qualitypulseheightspectra(position4),withthecon- best fitmodelsforallcasesarecontainedinTable2;oneof the We thenfittedtheindividualspectratolookforstatistically peratures donotvarysubstantiallyfrompositiontoposition). significant parametervariations.Thevaluesof the the bestdefinedparametervalues(assumingthattem- these combineddatasetshavethehighestsignal,theyproduce comprising thefourobservationsofbrightfeature.Since discuss two-temperaturefitsexclusively:foreachoftheindi- of spectra,onecontainingallsixobservations,andthesecond fit ataconfidencelevelofhigherthan95%(viatheF-test). temperature fitsarepreferredoverthebestsingle vidual andsummedspectra,wefoundthatthetwo- 1979, 1980).Wealsoallowedtherelativeabundancesofneon, dashed contourrepresentstheouterboundaryofoptical/radioshell. No. 1,1988 For allspectra,individualandcollective,thebest-fittem- Our approachwastofirstfittwotime-averagedensembles 3 Fig. 2.—ExposuremapshowingthecoverageofnortheasternportionIC443byEinsteinHRI.TheaverageHRIexposurelengthwas7.5 x10s.The © American Astronomical Society • Provided by the NASA Astrophysics Data System X-RAY STRUCTUREANDSPECTRUMOFIC443 2 Ä 21-2 combined spectra. column density,andtemperaturedifferencespresentamong the fiveindividualspectraand betweentheindividualand any hardconclusionsbeingdrawnfromtheslightabundance, parameters inthebest-fitmodel.Thelowsignal-to-noiseratios of thesespectra,asreflectedintheparameterranges,prevent dance, asdeterminedfromtheFeLblend.However,fixing the new crosssectionswastomakeindeterminatetheFeabun- previous results.Oneconsequenceoftheintroduction the yields Nvalueslowerby~20%andmoreconsistentwith ion oftheBrownandGould(1970)absorptioncrosssections Fe abundanceatcosmichaslittleornoeffectontheother (Xv 1);butthey areunsatisfactoryinthat theylendthem- use oftheMorrison-McCammonabsorptionmodel;substitut- description oftheSSSspectra fromastatisticalpointofview 90% confidencelevel.TheslightlyhigheraverageNof 6.5 x10cm,thanwasfoundbypreviousobservers(e.g., bright LMCremnants(Clarketal1982). lower temperaturetomeanthatIC443(andPuppisA)iscloser Malina, Lampton,andBowyer1976),maybeascribedto the to reachingionizationequilibriumthan,forinstance,the prets theneedfortwotemperaturesasanindicationofioniza- tion disequilibrium(Itoh1979),thenonecaninterpretthe which similartwotemperaturefitshavebeenobtainedis Puppis A(Szymkowiak1984),ofcomparableage.Ifoneinter- strained andfixedat4keV.Theonlysupernovaremnantfor the shapeofcontinuumabove2.5keV,isusuallyuncon- sulfur ismaximized.Thehightemperature,requiredtomatch H H The atomicabundancesareallconsistentwithsolaratthe The two-temperature,equilibrium fitsprovideanadequate ii) NonequilibriumModels 221 1988ApJ. . .335. .215P -21 2 (1979). “ natural”interpretationoftheapparentpresencetwo tem- equilibrium componentsmightexist(see§IVh).Amore within theX-rayemittinggas;thattwophysicallydistinct wiak (1984).Thecalculationswhichproducedthesemodels perature componentsisthattheshockedplasmahasnot yet selves toaseeminglyadhocinterpretationoftheconditions boundaries exists. the sameasin(u).Thenorthernboundaryofbrightfeatureisonlylocationremnantwhereastrongcorrelationbetween X-ray andoptical represents theouterboundaryofoptical/radioshell,(b)AcontourmapHRIimageoverlaidonPalomarSkySurveyredplate.The contourlevelsare photomosaic ofthethreeHRIexposuresindicatedinFig.2.Thecontourlevelsrepresent10,13,and16countsarcminks. dashedcontour collisional timescaleparameter rj=nE,whereisthepre- ionization equationsbehind this shock.Themodelsarechar- and Chevalier1983),usingthetechniquedescribedinSzymko- data andagridofnonequilibriummodels(Hamilton,Sarazin, attained ionizationequilibrium,asfirstsuggestedby Itoh assuming instantaneouselectron-ion equipartitionand acterized bytwoparameters, theshocktemperatureT,anda remnant shockandconsistoftime-dependentsolution the assumed Sedovdynamicsfortheexpansionofasupernova For eachvalueofTandrj, twomodelsareavailable:one shock ambientdensityandE istheenergyofexplosion. 222 0 s s Fig. 3.—(a)HRIimageoftheX-ray-bright,northeasternportionIC443.Theisanexposure-corrected,background-subtracted, Wiener-filtered We havethereforeattemptedacomparisonbetweentheSSS © American Astronomical Society •Provided bythe NASAAstrophysics Data System PETRE ETAL. Fig. 3a 7 25 we findthatatthe99%confidence levelfortwoparametersof we attempttodefinestatistically acceptablerangesforTandrj, partition. Thebest-fitmodelparameters arelistedinTable3.If column densityvariationsfrompositiontoposition, we we fixedtheelementalabundancesattheircosmicvalues. models isavailable,weperformedthecomparisonwith the ing locationtoanotherandonlyarelativelycrudegrid of representing energiesabove1 keV. restricted ourfitrangetoonlythosepulseheightchannels able, thetime-averagedspectrumofthosepositions(2,3,4, and nonequilibrium modelswiththehighestsignalspectrumavail- no significantdifferencesinparametervaluesfromonepoint- Coulomb collisions. another assumingtime-dependentelectronheatingfrom Finally, toavoidpotentialfittingproblemsduepossible 5) whosefluxisdominatedbythebrightfeature.Additionally, 10 £j^_1052g -6^ electron-ionequi- s an(erS cm Under theseconstraints,the bestfittingmodelhasT= Since thedetailedfittingtoequilibriummodelsrevealed s Vol. 335 1988ApJ. . .335. .215P 5 3 4 2 523 ably aconsequence ofusingamodelintended torepresentthe and explosionenergy(8x10 ergs).Thisdiscrepancyisprob- yields absurdvaluesfortheage (9x10yr),distance(32kpc), models areallowed.Thisisnotunexpected;inthisrangeof interest (x+9.2),rjisrestrictedtotherange remnant (inferredfromthe HEAO 1A-2MEDspectrum), and anestimateofthecontinuum fluxat2keVfortheentire Chevalier (1983),usingasinputs 7¡,/y,theangulardiameter, tral parameters,aswediscussin§IV.Briefly,astraightforward a simplephysicalinterpretationtothenon-equilibriumspec- cussion in§IVbelow,wewillusethebest-fitTandunder Sedov analysisbasedonFigure 3ofHamilton,Sarazin,and parameter range. band (evenforthesummedspectrum)prohibitsdiscrimination relatively lowqualityoftheIC443pulseheightspectrain this the shapeofcontinuumabove2keV,becomessmall. The partition andnonequipartitionmodels,manifestedprimarily in scale isrelativelyshort,andthedifferencesbetweenequi- relatively lowTandhighrj,theelectron-ionequilibration time grid ofmodels.Also,bothequipartitionandnonequipartition No. 1,1988 the caveatthattheyonlysuggestappropriatemodel between thetwosetsofmodels.Forpurposesour dis- 10 PHOT0NS/(cm2-s-keV) PHA COUNTS/(cm2-s-keV) © American Astronomical Society • Provided by the NASA Astrophysics Data System Fig. 6a Fig. 6b 227 3 7 observed bytheHEAO1A-2MEDduringan8.4x10spointing.The spectrum withsolarabundances.Thetemperaturerequiredbythe solid linerepresentsthebest-fitspectralmodel,a1.2x10Kthermal input best-fitspectralmodelforthe MEDdata,convolvedwiththe used tomodeltheSSSspectraof brightnortheastregion,{b)The MED isconsistentwiththehotteroftwoequilibriumcomponents MED spectralresponse. Fig. 6.—(a)Thecomposite2-10keVspectrumofIC443,as 1988ApJ. . .335. .215P divided bythecountsin0.2-1.3keV band.Symbols,inincreasingorderofprominence,representhardness ratiosofapproximately0.5,0.8,1.0,and1.2.Notethe lack ofcorrelationbetweenspectralhardness andsurfacebrightness. scale hasbeencompressedtoallowdisplayofthecompleteerrorboxes. other withgreaterthan99%confidence.BoththeMEDandHED3boxescontainradiocenterofremnant(markedaplussign).The rightascension locations; thedashedcontourrepresentsouterboundaryofoptical/radioshell.TheLED(0.1-1.5keV)andHED3(2.5-10errorboxes excludeeach confidence contours.TheLEDerrorboxrepresentsthe90%contourandwastakenfromGalasetal.(1981).smallcirclesrepresent theSSSpointing Fig. 8.—TheIPCspectralhardness ratiomapoverlaidontheIPCsurfacebrightnesscontours.Thespectral hardnessisthecountsin1.3-3.1keVband Fig. 7.—ErrorboxesforemissioncentroidsofIC443,asobservedbythreedetectorstheHEAO1A-2experiment.TheMEDandHED3 represent99% © American Astronomical Society • Provided by the NASA Astrophysics Data System hm 620 16H12M6H8M 6 1615 146131211 1 ^ H,MoH.-M^H M H.Mi C itrir 228 a a LOcu CM X-RAY STRUCTURE AND SPECTRUM OF IC 443 229 00LO 00 best fit source centroid. Such a search can be performed only hardness ratio (SHR) map. This map is depicted in Figure 8, ^ along the scan direction since the narrow dimension of the field superposed on the IPC surface brightness contours. The spec- § of view (1?5 FWHM) is in this direction, and the data can be tral hardness ratio is defined here as the ratio between the S binned in scan angle with a resolution much finer than the number of counts in IPC channels 8-13 (nominally 1.3-3.1 2 angular extent of IC 443. keV) and those in channels 3-7 (nominally 0.2-1.3 keV). In In order to define the error box with the highest precision, order to ensure enough counts per bin to determine meaning- we used only the total MED and HED3 count rates. These ful ratios over most of the remnant, the IPC map has been count rates can be characterized as “ colors,” which are func- rebinned into 256" x 256" pixels. Additionally, the SHR data tions of the detector responses and the source spectrum. For a have been binned in four crude levels which we refer to as the 1 keV thermal plasma, the MED has an effective bandpass 0.5, 0.8, 1.0, and 1.2 levels and which span the approximate from 1.5 to 3.5 keV, with a peak of 2.5 keV (since we include all ranges <0.65,0.65-0.9,0.9-1.1, and >1.1. the MED pulse height channels here, the effective bandpass is Figure 8 suggests that pronounced X-ray spectral variations softer than that of the modelled pulse height spectrum). The are present within IC 443. Consistent with the HEAO 1 A-2 HED3 color is slightly “ bluer,” with peak energy of 3.8 keV scanning data, there is a general trend of increasing spectral and an effective bandpass from 2.3 to 5.3 keV. Illustrated in hardness from north to south, but this is largely coincidental, Figure 7 are the 99% confidence error boxes for the MED and due to the spectral softness of the bright northern region. HED3 centroids, superposed on an outline of the IC 443 Excepting this region, there is no strong correlation between optical radio shell. Also shown are the LED 90% confidence spectral hardness and surface brightness. error box (Galas, Venkatesan, and Garmire 1981), and the SSS Figure 9 represents a calibration of this SHR map, con- pointing positions. The LED spectra are clearly biased toward structed by folding a grid of single-temperature Raymond- the northern bright region. There is a definite southward trend Smith equilibrium spectra, plus absorption by various column of error box centroids with increasing energy: the LED and densities, through the IPC response function at the appropri- HED3 error boxes completely exclude each other at the 99% ate gain setting for the IC 443 observations. Although single- level. However, both the MED and HED3 error boxes contain temperature models are not warranted by the SSS results, we the radio center of IC 443 in their small strip of overlap. We may use them to construct the expected hardness ratios for conclude from Figure 7 that the morphology of IC 443 is two-temperature fits by taking weighted sums of the hardness probably much more symmetric above 2 keV than at lower ratios of the two components. We can thereby mimic nonequi- energies, and that, depending upon the preferred interpretation librium spectral models. of the SSS spectra, either the hot equilibrium component or a As a check of the reliability of such an approach, we first single nonequilibrium model applies throughout the entire attempt to reproduce the spectral hardness ratio observed in remnant. the region where the SSS observations were made. For the ratio of IPC count rates inferred from the SSS-derived param- 6 7 c) IPC Spectral Hardness Ratio Map eter values of 7^ = 2.2 x 10 K, T2 — 1.1 x 10 K, 5 < 21 -2 Since the IPC provides pulse height data long with posi- NJN2 < 15, and NH = 5-10 x 10 cm , we find that the tional information, we may use the IPC image to investigate SHR value predicted from Figure 9 is 0.4-0.5, within the range the spectral variations across IC 443 by constructing a spectral of SHR values represented by SHR = 0.5 in Figure 8. Fig. 9.—Calibration of the IPC spectral hardness ratio map. For the IPC gain during the IC 443 observations and an input model of a coronal equilibrium plasma (Raymond-Smith) spectrum, the spectral hardness ratio is plotted as a function of temperature for various column densities. © American Astronomical Society • Provided by the NASA Astrophysics Data System 1988ApJ. . .335. .215P emission locatestheextentofandcolumn densitythroughamolecularcloudinthelineofsighttoIC443.Note thecorrelationbetweenhighCOsurfacebrightness and spectralhardness. 7 22-2 ness decreaseratherthanthoseoflowsurfacebrightness. ness contoursappearatopsitesofrapidX-raysurfacebright- coincides withthecloud,andhighestCOsurfacebright- 67 only notablefeaturesarethatsomeofthebrightX-rayregion brightness mapsdoesnotshowasstrongacorrelation:the temperature. AsuperpositionoftheCOandX-raysurface between theX-rayspectralhardnessandCObrightness the lineofsighttoIC443.Adefinitecorrelationisvisible CO brightnesstemperaturemap(Cornett,Chin,andKnapp esis, wepresentinFigure10anoverlayupontheSHRmapofa varies acrosstheremnant.Asevidencesupportingthishypoth- strong anticorrelationisnotobserved. 1977), whichdemarcatestheextentofmolecularcloudin the IPCsurfacebrightness,assumingconstantpressure.Sucha observe theSHRvaryingasapproximately—3powerof fixed columndensity,theIPCsurfacebrightnessmapisnot.If this werethecorrectdescriptionofremnant,wewould continuous rangeoftemperatures(upto3x10K)witha observed SHRrangeisconsistentwithamodelallowing not covertherangeofobservedSHR.Also,although value lessthan1x10cm,wefindthatmixingTi=2.2 remnant. IfweholdNconstantthroughoutIC443atany with themeasuredspectralpropertiesarerequiredtoproduce the highervaluesofSHRobservedelsewherewithin 230 PETRE H x 10KandT=1.1invaryingproportionsdoes 2 Fig. 10.—TheIPCspectralhardness ratiomap,superposedonasurfacebrightnesscontourmapofCOlineemission (Cornett,Chin,andKnapp1977)TheCO A morereasonablehypothesisisthatthecolumndensity We nextconsiderwhatvariationsinTandNconsistent H © American Astronomical Society • Provided by the NASA Astrophysics Data System ET AL.Vol.335 2 21 21-2 7 cm“. ThisisconsistentwiththeHcolumndensityestimate find thecloudcolumndensityvariesbetween0and6x10 brightness temperature.IfweinterpretthevariationofSHRas column densitywillscaleassomepositivepoweroftheCO a columndensityvariationsacrossthemolecularcloud,we column densities(Cornett,Chin,andKnapp1977),theH temperatures cannotbeconvertedinasimplewayintoH and >1.0,wefindaverageCObrightnesstemperaturesof0.94, strongly withincreasingSHR.FortheSHRlevelsof0.5,0.8, 2 temperature correspondingtoeachSHRlevelincreases in thecoveredareaof>3x10cm,or>1.5cm“ of H.Additionally,wefindthattheaverageCObrightness ference correspondstoanaverageincreaseofcolumndensity 1.24, and1.85(typically±0.5).AlthoughtheseCObrightness (equilibrium) plasmawithtemperatureof1.2x10K,thisdif- 2 the unoccultedregionsis0.60.Forasingletemperature parts ofIC443coveredbythecloud(theareawithin outermost contourinFig.10)is0.75whiletheaverageSHR 2 2 on alargerscale.First,wefindthattheaverageSHRinthose However, itisstillpossibletoquantitativelycomparethemaps fact, apoint-by-pointcorrelationofthetwomapsindicates that theyarecorrelated,butonlyatthe68%confidencelevel. statistics inthelowersurfacebrightnessregionsofIPC.In the SHRmap)willbesomewhatcontaminatedbycounting the provisothatadetailedcorrelation(at256"binsizeof brightness temperaturecanbequantifiedtosomeextent,under This apparentcorrelationbetweentheSHRandCO a 1988ApJ. . . 335 . .215P 7 7,25-6 21-2 is theIPCsurfacebrightness.Fromspectralhardness No. 1,1988X-RAYSTRUCTUREANDSPECTRUMOFIC443231 ratio mapweinferthepresenceofsignificantspectralvaria- whose T=1.2x10Kcorrespondstothehigheroftwo more limitedextent,atime-dependentequilibriummodelwith duced byamolecularcloudinthelineofsight. component ismoreuniformlydistributedintheremnantthan described bysingle-temperatureequilibriumplasmamodel possible tofittheentirerangeofobservedSHRvaluesby strates thatthespatiallyaveragedspectrumabove2keViswell temperatures andapproximatelysolarabundances,or,toa described byeitheranequilibriumplasmamodelwithtwo the SSSspectraofbrightestfeatureIC443arewell tions acrossIC443,aresultofvariablecolumndensitypro- temperatures foundintheSSSfitting,andthatthishotter same two-temperaturemodel. marized above.First,theX-rayemittinggasisessentiallyin ionization equilibriumandthattheemissionarisesfromtwo T =IOKandrj10ergscm.TheMEDdemon- where, usingtheSHRdataalone.Infact,allowingNtovary component representshigherdensitymaterialassociatedwith discrete componentsalongthelineofsight,whichmightbe over therangeallowedbyCOmeasurementsmakesit rule outthepresenceoflow-temperaturecomponentelse- temperature andvaryingcolumndensity.Wecannot,however, (Hamilton andSarazin1984).Werelyonlyontheinferred T medium isclearlynonuniform,itsaferherenottouserjfor librium andtheparametersrjTshouldthereforebeused ofIC 443. ISM andfillsmostoftheremnantvolume,whilecool spatially interspersed.Thehotmaterialisthediffuse,shocked energy EofIC443,becausetheseparametersdependexclu- determining remnantparameterssuchasenergyandmass the entiresupernovaremnantwasnotfittedandambient by thenonequilibriummodel.Sinceintegratedspectrumof taken, however,nottomisinterprettheinformationimparted to characterizethepropertiesofremnant.Caremustbe shocked interstellarfeatures,suchastheHicloudtoeast To estimatevaluesfortheseparameters wesimplyupdatethe case (anassumptionsupportedbythetemperaturesobserved), the sameapproachtobothmodelsforestimatingaget and and emissionmeasure. from theCOmeasurementsof0.5-8.0x10cm(Cornett, not meritamoresophisticated treatment. presupernova conditionsofthis region(Fesen1984;Braunand We recognizethatthisapproach isanoversimplificationofthe analysis ofGulliford(1974), using ourinferredtemperatures. and Sedovdynamicsarebuiltintothenonequilibriummodels. of Sedovdynamicshasbeentacitlyassumedintheequilibrium sively onthedynamicsofsupernovaremnant.Thevalidity low-temperature materialisrestrictedtothebrightfeature, and theremainderofremnantisdescribedbyasingle observed SHRrangecanbeexplainedbyamodelinwhichthe Chin, andKnapp1977). Strom 1986),but,considering theirquality,theX-raydatado s n s s To summarizethemainconclusionsofourspectralanalysis: Two possiblescenariosaccountforthespectralresultssum- The secondscenarioisthatIC443notinionizationequi- Independently ofwhichscenarioispreferred,wecanapply The discussionaboveallowsustoconcludethatthe IV. CONDITIONSINTHEX-RAY-EMITTINGGAS © American Astronomical Society • Provided by the NASA Astrophysics Data System a) AgeandEnergyofIC443 1 51 3 3 7 7 it. ThemultibandcomparisonbyMufsonetal(1986)suggests cating thattherearestrongdensityvariationsalongthelineof ing densityvariationsalongtheremnantshell.Time- this featureislocatedalongtheoncomingedgeofremnant feature partiallyoverlapsthebrightopticalfilamentation,indi- is stronglyreinforcedbyobservationsinotherbands.This resolution imagingobservationsofthe[Fex]line(Brown, models couldbereconcilediftheshockedgasisclumpy,with magnitude. Teskesuggestedthattheobservationsand the failed toaccountforthe[Fex]emissionbytwoorders of dependent collisionalionizationmodelingbyTeske(1984) between twoX-raysurfacebrightnessenhancements,suggest- brightness (Woodgate,Lucke,andSocker1979)occurs also overlapsit.Interestingly,theobserveddiffuse[Fex]peak shell. Thehighest[Fex]26374surfacebrightnessinIC443 sight totheX-ray-emittinggas,andpossiblyinterspersedwith sphere orsurfaceofuniformX-rayemissivity.Thissuggestion that thisfeatureisfairlycomplexandcannotbetreatedasa ness appearsasdiscreteknots,typically0.1pcinextent,with a The [Fex]emissioninregionsofhighestX-raysurfacebright- Woodgate, andPetre1988)haveconfirmedthisprediction. IPC andtheMED. remainder oftheremnant,incorporatingresultsfrom tions withintheX-rayemitting gasheremustincorporatethese small fillingfactor(<1%).Clearly,anydiscussionofthecondi- extrapolate ourinferencesaboutthebrightfeatureto within IC443.Sincetheregionofhighsurfacebrightnesswas data doyieldsimilarestimatesforitsageandenergy. 443, theequilibriumandnonequilibriummodelsofSSS interstellar gasattheexplosioncenter.Wehavethusfound model and0.61n*xl(rergsfortheequilibriummodel, pressure equilibriumandof uniform electrondensity.Fora peratures areinferredfromthe SSSfittingandwhicharein that thebrightfeatureconsists oftwocomponentswhosetem- other aspects. tral resolution(HRIandSSS),weconsideritfirst.Wethen that, despitedifferencesinimplicationsofconditionsinsideIC where n*isthevalueofpreshocknumberdensity equation (14)yieldsl.On*x10ergsforthenonequilibrium estimated bymanypreviousinvestigators.ForE,Gulliford’s [Fe x]arisinginshockedcloudlets.Recenthighspatial observed bytheinstrumentswithhighestspatialandspec- sets ofmodelsaboutgasdensities,pressures,andfillingfactors and 3.4x10yrfortheequilibriumcase,muchyoungerthan then yieldsanagetof2.8x10yrforthenonequilibriumcase ponent, wefind»1.2x10K.Gulliford’sequation(12) librium fittotheSSSdata(whichisalsotemperature ent parametervalueistheshocktemperature.Thenonequilib- in natureandthereforeremainvalidhere;theimportantdiffer- observed bytheMED)torepresentpredominantgascom- the higheroftwotemperaturesinferredfromequi- rium modelyieldeddirectly«1.8x10K.Ifweinterpret dient isintheplaneofsky,fromsouthwesttonortheast. density distribution.Heassumedthatthedirectionofgra- expanding intoamediumwithone-dimensional,exponential the Kompaneets(1960)self-similarsolutionforablastwave Most oftheinputparametersusedbyGullifordaregeometric The arcminutesubstructurerevealedbytheHRIsuggests Next, weexaminewhatinferencescanbemadeusingthetwo We firstconsidertheionization equilibriumcase.Weassume Gulliford’s analysisoftheageandenergyIC443invoked b) TheBrightFeature 1988ApJ. . .335. .215P 4 2-6 -37_1/2 1/2-37_ 1/2-3 -3 n2Ä-6 -6 for theopticalfilamentsat~10K(FesenandKirshner1980) bands. Valuesofthegaspressureinthisregionareavailable equilibrium withthegasatothertemperatures,visiblein emitting gas,weinvoketheassumptionthatitisinpressure equilibrium component. both aresomewhathigherthanthatinferredforthewarmer sion measurenLof33cmpc.Theresultantdensityis inferred abovearesummarizedinTable5. cm, andapressureof2.9x10LK.Thevalues 5.SL~ cm,andthepressureis7.7x10LK; the twocomponents=12L~cmandn2.6L number densitiesincmofthetwocomponents.Weinferfor of theemissionregion,inpc,andnaretotal i^K 6-5cmpc,whereListheline-of-sightthickness the emissionmeasures,findingæ156cmpcand Using adistanceof1.5kpctoIC443(Fesen1984),wesolvefor nominally associatedwithshockedcloudmaterial,is~0.4. ponents, asderivedfromthespectralfitting. where Nandarethenormalizationsoftwocom- inferred spectralparametersinthefollowingway: will beatthelowertemperatureTandremainder at thehighertemperatureT.Thevalueofaisrelatedto volume VofgasobservedbytheSSS,somefractiona K X kX 232 PETRE X2K 2 K l2 l5 2 To infervaluesforthedensityandthicknessofX-ray- For thebest-fitnonequilibriumspectrum,weinferanemis- The inferredvalueofa,thefillingfactorcooler,densergas © American Astronomical Society • Provided by the NASA Astrophysics Data System C Remainder ofSNR: Bright feature: de 1-3 a (one component)...1254 ... 0.6 (two component)2.21281054231.0 Optical* 6-13x10 MED/IPC SSS(Eq) 2.2111006.5156 [Fex] 1.2... case assumesa=0everywhereexcept northeast. Remainder: Bright feature: (NIE) ...185633.4 d e a c b -1/2 -1/2 -1/2 -1/2 -1/2 FixedusingSSSvalue. FixedusingMEDvalue. Brown,Woogate,andPetre1987. Two-componentcaseassumesais constant everywhereinIC443;one-component FesenandKirshner1980. (one component)... 0.8L0.0 (two component)...... 4.8L1.0L0.41 Optical 140-630...0.08-0.8 SSS 12.4L*2.6L~0.412.7L* MED/IPC [Fex] 80....10.0 Ä Ä Ä k 6563 Parameter (10K)cm")(cm“pc) -37 Parameter (cm)a(10cm K) Gas ParametersinIC443 2 Ti TnVn^LL 2il TABLE 5 B. A. ET AL.Vol.335 -3 7-3 3 8 6-3 6 the factorof3uncertaintyassignedtothispressure,wefind n are38cmand7.8,respectively.Whenweconsider K withtheequilibriummodel,wefindL=0.1pc,andn exclusively withthepressureasderivedfrom[Fex]mea- surements. of 2(Sgro1975),wethereforecomparetheX-rayemission to thesurroundingX-rayemittinggasofnomorethanafactor small sizeofthecloudsimpliesanoverpressuringwithrespect lapse. AsthepressureofX-ray-emittinggasismorelikely to besimilarthatofthegasnearerintemperature,and that theopticalfilamentsareinastateofrapidradiativecol- assuming afactorof100shockcompression.Itismorelikely netic fieldswithanuncharacteristicallyhighstrengthof30/¿G, supplying apressureof>3x10cmrequiresambientmag- of magneticfieldandrelativisticparticlepressurecomponents ant pressurecomponentintheopticalfilaments.Anequalmix the magneticfieldsandrelativisticparticlesmustbedomin- sure equilibriumwiththehottergas.Fortoexist, ticles, itseemsunlikelythattheopticalfilamentsareinpres- from compressedambientmagneticfieldsandrelativisticpar- the additionalpressurewithinopticalfilamentsarising -1/2 between thesetwocomponentsmaybereducedbyconsidering cm “K,withinafactorof3.Althoughthepressuredifference the brightestregionsinHRIimagehavenT=1x10 (filaments 1and4;FesenKirschner1980)yieldagas 2 nT æ0.8-8.0x10cmK.The[Fex]gascoincidentwith ments whichliealongthelineofsighttobrightfeature Kx Woodgate, andPetre1988).Thetwoobservedopticalfila- and the[Fex]-emittinggasat1.2x10K(Brown, 8-3 5.8L* ...7.7L k If wecomparethenominal[Fex]pressureof1x10cm «2 ? 1988ApJ. . .335. .215P -3 -23 3- -3 51 51 0.01 pc
1988Apj. . .335. .215P the Astrophysical Journal, 335:215-238
