Table Of Content

Supernova remnants: the X-ray perspective Jacco Vink Abstract Supernovaremnantsarebeautifulastronomicalobjectsthatarealsoofhighscientificinterest, because they provide insights intosupernova explosion mechanisms, and because they are the 2 likelysourcesofGalacticcosmicrays.X-rayobservationsareanimportantmeanstostudythese 1 objects. Andinparticular the advances madeinX-rayimaging spectroscopy over thelast two 0 decadeshasgreatlyincreasedourknowledgeaboutsupernovaremnants.Ithasmadeitpossible 2 tomap theproducts of freshnucleosynthesis, andresulted intheidentificationof regions near n shock frontsthatemitX-raysynchrotron radiation. SinceX-raysynchrotron radiationrequires a 10-100TeVelectrons,whichlosetheirenergiesrapidly,thestudyofX-raysynchrotronradiation J hasrevealedthoseregionswhereactiveandrapidparticleaccelerationistakingplace. 3 InthistextalltherelevantaspectsofX-rayemissionfromsupernovaremnantsarereviewed andputintothecontextofsupernovaexplosionpropertiesandthephysicsandevolutionofsu- ] E pernovaremnants.Thefirsthalfofthisreviewhasamoretutorialstyleanddiscussesthebasics H ofsupernovaremnantphysicsandX-rayspectroscopyofthehotplasmastheycontain.Thisin- cludeshydrodynamics,shockheating,thermalconduction,radiationprocesses,non-equilibrium . h ionization,He-likeiontripletlines,andcosmicrayacceleration.Thesecondhalfoffersareview p oftheadvancesmadeinfieldofX-rayspectroscopyofsupernovaremnantsduringthelast15year. - ThisperiodcoincideswiththeavailabilityofX-rayimagingspectrometers.Inaddition,Idiscuss o theresultsofhighresolutionX-rayspectroscopywiththeChandraandXMM-Newtongratings. r t Although these instruments arenot ideal for studying extended sources, they nevertheless pro- s a vided interesting results for alimited number of remnants. These results provide a glimpse of [ whatmaybeachievedwithfuturemicrocalorimetersthatwillbeavailableonboardfutureX-ray observatories. 2 Indiscussing theresultsof the lastfifteenyears I havechosen todiscuss afew topics that v are of particular interest. These include the properties of Type Ia supernova remnants, which 6 7 appear toberegularly shaped andhavestratifiedejecta,incontrast tocorecollapse supernova 5 remnants,whichhavepatchyejectadistributions.ForcorecollapsesupernovaremnantsIdiscuss 0 thespatialdistributionoffreshnucleosynthesisproducts,butalsotheirpropertiesinconnection . totheneutronstarstheycontain. 2 ForthematuresupernovaremnantsIfocusontheprototypalsupernovaremnantsVelaandthe 1 1 CygnusLoop.AndIdiscusstheinterestingclassofmixed-morphologyremnants.Manyofthese 1 maturesupernovaremnantscontainstillplasmawithenhancedejectaabundances.Overthelast : fiveyearsithasalsobecomeclearthatmanymixed-morphology remnantscontainplasmathat v isoverionized.Thisisincontrasttomostothersupernovaremnants,whichcontainunderionized i X plasmas. ThistextendswithareviewofX-raysynchrotronradiationfromshockregions,whichhas r a madeitclearthatsomeformofmagnetic-fieldamplificationisoperatingnearshocks,andisan indicationofefficientcosmic-rayacceleration. 1 CONTENTS J.Vink Supernovaremants Contents 1 Introduction 3 2 Supernovae 4 2.1 Corecollapsesupernovae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.2 Thermonuclearsupernovae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3 Theclassificationofsupernovaremnants 7 4 Thehydrodynamicstructureandevolutionofsupernovaremnants 9 4.1 Evolutionaryphases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 4.2 Analyticalmodels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 4.3 Supernovaremnantsevolvinginsidewind-blownbubbles . . . . . . . . . . . . . . . . . . . . . . . . . . 11 5 Collisionlessshockheatingandparticleacceleration 12 5.1 Shockheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 5.2 Collisionlessshocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 5.3 Temperatureequilibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 5.4 Thermalconduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 5.5 Cosmic-rayaccelerationbysupernovaremnantshocks . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 6 X-rayradiationfromsupernovaremnants 19 6.1 ThermalX-rayemission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 6.2 Lineemissionassociatedwithradioactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 6.3 Non-thermalemission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 7 X-rayspectroscopywithChandra,XMM-Newton,andSuzaku 27 8 TypeIasupernovaremnants 30 8.1 X-rayspectroscopyofTypeIasupernovaremnants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 8.2 EvidenceforTypeIaprogenitorimprintsonthecircumstellarmedium . . . . . . . . . . . . . . . . . . . 35 9 Corecollapsesupernovaremnants 36 9.1 Oxygen-richsupernovaremnants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 9.2 TheX-rayemissionfromSN1987A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 9.3 Corecollapsesupernovaremnantsandneutronstars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 10 Supernovaremnantsinorapproachingtheradiativephase 46 10.1 Fromnon-radiativetoradiativeshocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 10.2 EnhancedmetalabundancesmatureSNRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 10.3 Mixed-morphologysupernovaremnants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 11 Shockheatingandparticleacceleration:observations 53 11.1 Electron-iontemperatureequilibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 11.2 X-raysynchrotronemittingfilaments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 11.3 X-raybasedevidenceforefficientcosmic-rayacceleration . . . . . . . . . . . . . . . . . . . . . . . . . 57 11.4 TherelationbetweenX-raysynchrotronandg -rayemission. . . . . . . . . . . . . . . . . . . . . . . . . 60 12 Concludingremarks 62 2 INTRODUCTION J.Vink Supernovaremants 1 Introduction pectsofSNRshockphysics.Firstofallthelineemission providesinformationaboutthetemperatureandionization state oftheplasma,buttheabsenceoflinesorweakline Supernovae play a central role in modern astrophysics. emission in young SNRs usually points toward the con- Theyareof primeimportanceforthechemicalevolution tribution of X-ray synchrotronradiation. Studying X-ray of the Universe, and they are one of the most important synchrotronradiationoffersapowerfuldiagnostictoolto sourcesofenergyfortheinterstellarmedium(ISM).Part study electron cosmic-ray acceleration, as the presence of that energyis in the form of cosmic rays, which have anenergydensityintheGalaxyof1-2eVcm 3,thuscon- ofX-raysynchrotronradiationdependssensitivelyonthe − shock acceleration properties. Moreover, the size of the stitutingaboutonethirdofthetotalenergydensityofthe X-ray synchrotron emitting regions can be used to infer ISM.Finally,supernovae,inparticularTypeIasupernovae magnetic-fieldstrengths. (seeSect.2),playacentralroleinpresentdaycosmology One ofthe majoradvancesin X-rayspectroscopyover astheirbrightnessallowsthemtobedetectedathighred- thelast15-20yearshasbeentheemergenceofX-rayimag- shifts(uptoz 1.7,Riessetal.,2007).Theirusehasled ∼ ing spectroscopy.Although presently the spectral resolu- totherecognitionthattheexpansionoftheUniverseisac- tionofferedislimited,itprovidesadirectwaytomapthe celeratinginsteadofdecelerating(Perlmutteretal.,1998; spatialdistributionofelementsandtemperaturesinSNRs. Garnavichetal.,1998). For young SNRs it also helps to separate thermal from Supernova remnants (SNRs) are an important means non-thermal(synchrotron)X-rayemission. tostudysupernovae.Sincesupernovaearerelativelyrare ThepowerofX-rayimagingspectroscopyisillustrated (2-3 per century in a typical spiral galaxy like our own), in Fig. 1,whichshowsTycho’sSNR (theremnantofthe SNRs providethe bestway to study the localpopulation historical SN 1572) as observed by Chandra. The im- ofsupernovae.Inaddition,SNRscanrevealdetailsabout ageshowsbeautifullythedistributionofsiliconandiron, theexplosionsthataredifficulttoobtainfromstudyingsu- butalsorevealsthesynchrotrondominatedemissionfrom pernovaedirectly. For example,youngSNRs can inform neartheshockfront. us about nucleosynthesis yields of individual SNRs and Over the last ten years high spectral resolution X-ray abouttheinherentasymmetriesoftheexplosionitself,as spectroscopy has gained in importance, due to the avail- revealedby the spatialandvelocitydistributionof heavy ability of grating spectrometers on board Chandra and elements.Moreover,SNRsprobetheimmediatesurround- XMM-Newton.Thesespectrometersarealsousedtostudy ingsofsupernovae,whichareshapedbytheirprogenitors. AnotheraspectofSNRsconcernstheinterestingphysi- cal process that shape their properties. SNR shocks provide the best Galactic examples of high Mach number, collisionless shocks. The physics of these shocks is not wellunderstood,asheatingoftheatomsoccurscollision- less, i.e. shockheatingdoesnotoperatethroughparticle- particle (Coulomb) interactions, but through electro- magneticfluctuationsandplasmawaves.TheSNRshocks are also thoughtto be the locationswhere partof the ex- plosionenergyisconvertedtocosmic-rayenergy.Thisis supportedbyobservationsthatindicatethatGeVandTeV particles are present in SNRs through their radiative sig- natures: radio to X-ray synchrotron emission from relativistic electrons, and g -ray emission caused by acceler- atedelectronsandions. X-rayobservationsareoneofthemostimportantmeans to study the many aspects of SNRs. X-ray spectroscopy isessentialto obtainabundancesoftheprimenucleosyn- thesis products of supernovae, which are the so-called alpha-elements(O,Ne,Mg,Si,S,Ar,Ca)andiron-group elements (chiefly Fe, Ni, and some trace elements with 20<Z <28). All these elements have prominent emission lines in the 0.5-10 keV band for temperatures be- Figure 1: Three-color Chandra image of Tycho’s SNR tween0.2-5keV,whichhappenstobethetypicalelectron (SN1572/G120.1+1.4). The color red shows Fe L-shell temperatures of plasma heated by SNR shocks. The hot emission, green Si XIII, and blue continuum (4-6 keV). plasmasofSNRsarealsoopticallythin,soinferringabun- Note the very narrow continuum rims near the shock dancesisrelativelystraightforward.X-rayspectroscopyof front(blueish/purpleinthisimage),causedbyX-raysyn- SNRs, therefore, provide us with a record of alpha- and chrotron radiation from electrons with energies up to iron-groupelementproductionbysupernovae. 100TeV. (Imagemadeby theauthorfromChandradata, X-rayspectroscopycanalsobeusedtostudyseveralas- seealsoHwangetal.,2002;Warrenetal.,2005). 3 2.1 Corecollapsesupernovae J.Vink Supernovaremants 2003), or through binary interaction (e.g. Podsiadlowski etal.,1992).ForTypeIcthemasslossseemstohavere- movedeventhehelium-richlayersoftheprogenitor.The factthattheyarepreferentiallyfoundinthemostluminous regions of galaxies suggests that they are the explosions ofthemostmassivestars(e.g.Kellyetal.,2008).Interest- ingly,alsolongdurationgamma-rayburstsareassociated withTypeIcsupernovae(e.g.Galamaetal.,1998;Stanek etal.,2003). The sub-division of the Type II class in Type IIP (plateau), Type IIL (linear light curve) and Type IIb is basedona combinationoftwoobservationalcriteria:optical spectroscopy and light-curve shape ( Fig. 2). Type Figure2: Theclassificationofsupernovae,basedonopti- IIP are the most common type of core collapse super- calspectroscopyandlight-curveshape. novae, and optical studies of potential progenitor stars confirm that their progenitors have initial masses in the 8 17 M range, and that they explodein the red su- SNRs,butthecontributionofgratingspectrometerstoour ∼ − ⊙ pergiant phase, while still having a substantial hydrogen understanding of SNRs has been limited. The reason is envelope(see the discussionsin Smartt, 2009;Chevalier, thatthe slitless gratingspectrometersare notparticularly 2005).TypeIILprogenitorsprobablyhaveasubstantially wellsuitedforextendedsourcessuchasSNRs.Neverthe- lessmassiveenvelope,eitherduetostellarwindmassloss, less,someimportantstudiesofSNRswithpropertiesthat or due to binary interaction. Type IIb supernovae are a make them suitable for grating spectroscopy have been class intermediate between Type Ib and Type II, in that published.Withtheemergenceofhighspectralresolution their spectra would initially identify them as Type II ex- imagingspectroscopy,aswillbepossiblewiththecalori- plosions,butatlatetimestheirspectraevolveintoTypeIb metric spectrometers on board Astro-H (Takahashi et al., spectra. Also this can be understoodas the result of sub- 2008) and IXO/ATHENA (Bookbinder, 2010), high reso- stantial, but not complete, removal of the hydrogen-rich lutionX-rayspectroscopywillbecomemuchmorepromi- envelope due to stellar wind mass loss, or binary inter- nent. action. The prototypal Type IIb supernova is SN 1993J Here I review the currentstatus of X-ray spectroscopy (Podsiadlowskietal.,1993;Woosleyetal.,1994).Interest- of SNRs. Since there is a lack of good text books or ex- tendedreviewsonSNRsin general1,thefirstpartofthis ingly,recentidentificationandsubsequentspectroscopyof thelightechoofthesupernovathatcausedtheextensively reviewhasatutorialcharacter,describingourcurrentun- studied SNR Cassiopeia A (Cas A) shows that it is the derstanding of the evolution and physics of SNRs. The remnantof a Type IIb supernova,as the spectrumshows secondhalf of the review focusseson the X-rayobserva- both hydrogenand weak helium line absorption (Krause tions themselves, with an emphasis on the achievements etal.,2008a). obtained with data from the three most recent X-ray ob- Not listed in Fig. 2 is the Type IIn class. Type IIn su- servatories:Chandra,XMM-Newton,andSuzaku. pernovaearecharacterizedbynarrowhydrogenemission lines, which are thought to come from a dense circum- 2 Supernovae stellar environment,probablycausedbysubstantialmass lostbytheprogenitor.Itsplaceinthediagramisnotquite Supernovaearedividedintotwobroadcategories,reflect- clear,asatleastoneTypeIInsupernova,SN2001ic,was ingourunderstandingoftheexplosionprocesses:corecol- observedtobeaTypeIasupernovawhosespectrumsubse- lapsesupernovaeandthermonuclearsupernovae.Inaddi- quentlyevolvedintoaTypeIInsupernova(Hamuyetal., tionanobservationalclassificationschemeisused(Fig.2) 2003). that goes back to Minkowski (1941), who observed that some supernovae do not show hydrogen absorption in 2.1 Core collapsesupernovae theirspectra(TypeI)andothersdo(TypeII). Type II supernovae are invariably core collapse super- Core collapse supernovae mark the end of the lives of novae,butTypeIsupernovaecanbeeithercorecollapseor massive stars; that is, those stars with main sequence thermonuclearsupernovae.Thethermonuclearexplosions masses M & 8 M (see Woosley and Janka, 2005, for are associated with spectroscopic class Type Ia (Elias ⊙ a review). Just prior to collapse the star consists of dif- etal.,1985),whichhaveSiabsorptionlinesintheirspec- ferent layers with the products of the different consecu- tra. Type Ib and Ic supernovaeare now understoodto be tive burning stages. From the core to the outside one ex- explosionsofstarsthathavelosttheirhydrogen-richenve- pects: iron-group elements in the core (silicon-burning lopeasaresultofstellarwindmassloss(e.g.Hegeretal., products), then silicon-group elements (oxygen-burning 1ButseeVink(2006);Reynolds(2008);Badenes(2010)forrecent, products), oxygen (a neon burning product), neon and moretopicalreviews. magnesium(carbon-burningproducts),carbon(ahelium- 4 2.1 Corecollapsesupernovae J.Vink Supernovaremants Figure 3: Left: Supernovayields for the most abundant X-ray emitting elements. The squares/black line indicates the mean yield for core collapse supernovae, whereas the circles indicate thermonuclear supernovae (the W7 deflagration modelin red and the WDD2 delayed detonationmodelin magenta).The modelyields were taken from Iwamoto et al. (1999).Right:Oxygenyieldofcorecollapsesupernovaeasafunctionofmainsequencemass.Thecirclesandsquaresare thepredictionsofWoosleyandWeaver(1995),thetrianglesarepredictionsofChieffiandLimongi(2004),andthecrosses ofThielemannetal.(1996).Ingeneraltheoxygenyieldsobtainedbythevariousgroupsareverysimilar,butabove30M oneseesthatcertainmodels(WoosleyandWeaver,1995,using1051ergexplosionenergies)predictadiminishingoxyge⊙n yield.Thereasonisthatabove30M stellar coresmaycollapseinto blackholes,andpartoftheoxygenfallsontothe blackhole.Theamountoffall-backis⊙governedbytheexplosionenergyandtheamountofpre-supernovamassloss,but itisalsosensitivetothenumericaltreatmentoftheexplosion. burning product), helium (a hydrogen-burning product), tionsbyBurrowsetal.(2007)suggestthatacousticpower, and,finally,unprocessedhydrogen-richmaterial. generatedintheproto-neutronstarduetog-modeoscilla- Thecreationoftheiron-groupcore,whichlastsabouta tions, ultimately leads to a successful explosion (but see day, is the beginningof the end of the star, as no energy Weinberg and Quataert, 2008). Finally, there have been can be gained from nuclear fusion of iron. The core col- suggestionsthatneutrinodepositionisnotthemostimpor- lapsesintoaproto-neutronstar,andforthemostmassive tantingredientfora successfulexplosion,butthatampli- stars into a black hole. Most of the gravitational energy ficationofthestellarmagneticfield,duetodifferentialro- liberated(E GM2/R 1053erg,withR theneutron tationandcompression,mayleadtomagneto-centrifugal ns ns ∼ ∼ starradius)isintheformofneutrinos.Thishasbeencon- jetformation,whichdrivestheexplosion(Wheeleretal., firmedwiththedetectionofneutrinosfromSN1987Aby 2002). theKamiokande(Hirataetal.,1987)andIrvine-Michigan- SASI, acoustic power, and magneto-centrifugal mod- Brookhaven (Bionta et al., 1987) water Cherenkov neu- els all predict deviations from spherical symmetry. In trinodetectors. themagneto-rotationalmodelsoneevenexpectsabipolar The supernova explosion mechanism itself, which re- symmetry.Anotherreasonwhydeviationsfromspherical quiresthat&1051 ergofenergyisdepositedintheouter symmetryhasreceivedmoreattentionisthatlongduration layers, is not well understood.The formation of a proto- gamma-rayburstsareassociatedwithveryenergeticType neutronstar suddenlyterminatesthe collapse, and drives Ic supernovae(hypernovae).Giventhe natureof gamma- ashockwavethroughtheinfallingmaterial.However,nu- raybursttheseexplosionsarelikelyjetdriven.Thisraises mericalsimulationsshowthattheshockwavestalls. Itis thepossibilitythatalsomorenormalcorecollapsesuper- thoughtthattheshockwavemaybe energizedbytheab- novaehavejet components(Wheeleret al.,2002).There sorptionofafractionoftheneutrinosescapingtheproto- isindeedevidence,basedonopticalpolarimetry,thatcore neutron star, but most numerical models involving neu- collapse supernovae,especially Type Ib/c, are aspherical trinoabsorptionarestillunsuccessfulinreproducingasu- (e.g. Wang et al., 2001). In Sect. 9.1 I will discuss evi- pernova explosion (Janka et al., 2007). The most recent dencethatalsotheremnantsofcorecollapsesupernovae research,therefore,focussesontheroleofaccretioninsta- showdeviationsfromsphericalsymmetry. bilities for the supernovaexplosion,as these instabilities Theejectaofcorecollapsesupernovaeconsistprimarily helptoenhancetheneutrinoabsorptionincertainregions ofstellarmaterial,exceptfortheinnermostejecta,which just outside the proto-neutronstar. A promising instabil- consist of explosivenucleosynthesisproducts,mostly Fe ity is the so called, non-spherically symmetric standing and Si-group elements. These elements are synthesized accretion shock instability (SASI, Blondin et al., 2003), from protons and alpha-particles, which are the remains whichmay also help to explainpulsar kicksandrotation of the heavy elements that have disintegrated in the in- (Blondin and Mezzacappa, 2007). Alternatively, simula- tenseheatinthetheinnermostregionssurroundingtothe 5 2.2 Thermonuclearsupernovae J.Vink Supernovaremants collapsingcore(Arnett,1996).Someoftheexplosivenu- themostlikelyprogenitorsystemsare.ItisclearthatC/O cleosynthesisproductsare radioactive,such as 56Ni, and white dwarfshaveto accrete matter in orderto reachthe 44Ti. In particular the energy generated by the decay of Chandrasekhar limit, so thermonuclear supernovae must 56Ni into 56Co, and finally 56Fe, heats the ejecta, which occurinabinarysystem.However,thisstillleavesseveral leavesa majorimprinton the evolutionof the supernova possibilities for the progenitor systems: double degener- light curve. The yields of these elements depend sensi- ate systems (two white dwarfs), or white dwarfs with ei- tively on the details of the explosion, such as the mass ther a main sequence star, or an evolved companion (i.e. cut (the boundary between material that accretes on the single degenerate). Only a limited range of mass trans- neutron star and material that is ejected), explosion en- fer rates, 4 10 8 7 10 7 M yr 1, lead to stable − − − ∼ × − × ⊙ ergy,andexplosionasymmetry.Sincethemassoftheneu- growthofthewhitedwarf(Nomoto,1982;ShenandBild- tronstar/blackhole,thelocationofenergydepositionand sten,2007).Novae,forinstance,arerelativelyslowaccre- thepresenceofasymmetriesarenotwellconstrained,the tors, which are thoughtto blow off more mass than they expectedyieldsoftheseelementsareuncertain,andvary accrete.Theonlystablewhitedwarfaccretorsseemtobe substantially from one set of modelsto the other (Fig. 3, supersoftsources2,buttheirpopulationseemstoosmallto WoosleyandWeaver,1995;Thielemannetal.,1996;Chi- accountfortheobservedsupernovarate(e.g.Ruiteretal., effiandLimongi,2004). 2009).Dependingon the progenitorevolutionand accre- Overalltheyieldsofcorecollapsesupernovaearedom- tionscenarios,theprogenitorbinarysystemmayaffectits inated by carbon, oxygen, neon and magnesium, which circumstellar medium. For example, in the case of wind are products of the various stellar burning phases (e.g. accretion,notallthemasslostfromthedonorwillendup WoosleyandWeaver,1995;Thielemannetal.,1996;Chi- onthewhitedwarf.Hachisuetal.(1996,1999)proposed effiandLimongi,2004).Theseyieldsareafunctionofthe a scenarioin whichRoche-lobeoverflowfromthe donor initialmassoftheprogenitor(Fig.3).Itisforthisreason starisstabilizedthroughafast,opticallythick,windfrom thatoxygen-richSNRs(Sect.9.1)areconsideredtobethe thewhitedwarf,whichwillalsoaffecttheimmediatesur- remnantsofthemostmassivestars. roundingsoftheSNR. For some time it was thought that Type Ia supernovae were associated with population II stars, in which case 2.2 Thermonuclear supernovae they must evolve on time scales &109 yr, but it has re- centlybeenestablishedthatalsoashort“channel”exists, TypeIasupernovaearegenerallythoughttobethermonu- clear explosions of C/O white dwarfs, i.e. the explosion withanevolutionarytimescaleof 108yr(e.g.Mannucci ∼ etal.,2006;Aubourgetal.,2008). energy originates from explosive nuclear burning, rather than from gravitational energy liberated during the col- Models for thermonuclear explosions come in three classes: detonation, deflagration, and delayed detonation lapseofastellarcore. models.Indetonationmodelstheexplosivenucleosynthe- Althoughthere is some variationin peak brightnessof TypeIasupernovae,thevariationismuchlessthanthatof sisoccursduethecompressionandheatingoftheplasma byashockwavemovingthroughthestar.Indeflagration Type II supernovae. This is in line with the idea that all models the burning front proceeds slower than the local TypeIasupernovaeareexplosionsofsimilarobjects:C/O whitedwarfswithmassesclosetotheChandrasekharlimit sound speed. The nuclear fusion in the burning front is sustainedbyconvectivemotionsthatmixesunburntmate- (1.38M ).Moreover,anempiricalrelationexistsbetween theirpea⊙kbrightnessandthepostpeakdeclinerateofthe rial into the hot burning zone. The classical deflagration modelistheW7ofNomotoetal.(1984). light curve (Phillips et al., 1992), which can be used to Pure detonation models predict that almost all of the calibratetheabsolutepeakbrightnessofeachevent.This makes Type Ia supernovae excellent distance indicators, white dwarf matter will be transformed into iron-group elements,whereasopticalspectroscopyofType Ia super- onwhichmuchoftheevidencereststhattheexpansionof novae show that the ejecta contain significant amountof theUniverseisaccelerating(Perlmutteretal.,1998;Gar- navichetal.,1998;Riessetal.,2007). intermediatemasselements(Branchetal.,1982).Onthe other hand, the pure deflagration models overpredictthe There is no direct observational evidence that Type Ia progenitorsarewhitedwarfs,butthefactthatonlyTypeIa production of 54Fe with respect to 56Fe and predict too narrow a velocity range for the intermediate mass ele- supernovaecanoccuramongoldstellar populationsindi- ments compared to observations. For these reasons, the catesthatmassivestarscannotbetheirprogenitors.Their relative uniformity can be best explained by assuming a currentlymostpopularmodelforTypeIasupernovaeare the delayeddetonation(DDT) models(Khokhlov,1991), progenitortypewithanarrowmassrange(Mazzalietal., inwhichtheexplosionstartsasadeflagration,butchanges 2007). Moreover, C/O white dwarfs close to the Chan- drasekharmass limit are very likely Type Ia progenitors, to a detonationwave burning the remainderof the white dwarfintointermediatemasselements(IMEs),suchassil- since their high density makes for an ideal “nuclear fu- sionbomb”(Arnett,1996):Onceanuclearreactioninthe icon. There is observationalevidencethat the fractionof coreistriggered,itwillresultinanexplosion.Thetrigger 2SoftX-raypoint sources that are thought tobewhite dwarfs that mechanismitselfis,however,notwellunderstood. stablyaccretematerialfromacompanion(KahabkaandvandenHeuvel, A more serious problem is that we do not know what 1997). 6 THECLASSIFICATIONOFSUPERNOVAREMNANTS J.Vink Supernovaremants C/O that is burnedis roughlyconstantM=1.1 M , but the momentof inertia. Thisenergyloss producesa wind ⊙ that the variation in peak brightness is either caused by ofrelativisticelectronsandpositrons,whichterminatesin theratioofiron-groupelementsoverIMEproducts(Maz- a shock, where the electrons/positronsare accelerated to zali et al., 2007),or by the ratio of stable iron over 56Ni ultra-relativistic energies. These particles advect and dif- (Woosleyetal.,2007).56Nidecaysinto56Fe,andtheheat fuseawayfromtheshockcreatinganebulaofrelativistic thatthisgeneratesdeterminesthebrightnessofaTypeIa electron/positronswhichemitsynchrotronradiationfrom supernova(seeSect.6.2). the radio to the soft g -ray bands, and inverse Compton Figure 3 shows the predicted overall abundance pat- scatteringfromsoftg -raytotheTeVband(Gaenslerand tern of deflagrationand DDT models. Compared to core Slane, 2006,fora review).Sucha nebulais aptlynamed collapse supernovae, thermonuclear supernovae produce a pulsarwindnebula.Themostfamouspulsarwindneb- much more iron-groupelements ( 0.6 M ). The explo- ula is the Crab Nebula (also known as M1 or Taurus A), ∼ ⊙ sionenergy,E,ofthermonuclearexplosionsisdetermined whichispoweredbythepulsarB0531+21.Thisnebulais by the mass of their burning products (Woosley et al., associatedwiththehistoricalsupernovaof1054(Stephen- 2007): son and Green, 2002). As a result the Crab Nebula, but also similar objects like 3C58, have the status of a SNR. E =1.56M +1.74M +1.24M 0.46, (1) 51 Ni Fe IME− Sincethenebulahasamorphologythatisbrightinthecen- with the E the final kinetic energyin units of 1051 erg, ter,anddoesnotshowashell,theyarecalledfilledcenter 51 andthemassesofstableFe,56NiandIMEinsolarunits. SNRsorplerions,withthenameplerionderivedfromthe Greekwordfor”full”,pleres(WeilerandPanagia,1978). It is arbitrarywhether one should call the Crab-nebula 3 The classification of supernova and related objects plerions or pulsar wind nebula. The remnants radio and X-ray emission from plerions is powered by the pulsar wind, not by the supernova explosion, and as suchthenebulashouldnotreallybereferredtoasaSNR. Giventhefactthatsupernovaecanbebroadlyclassifiedin To complicate matters, the Crab nebula does show opti- corecollapseandthermonuclear/TypeIasupernovae,one wouldhopethatSNRswouldbeclassifiedascorecollapse cal line emission from supernovaejecta, which could be referred to as a SNR. Pulsar wind nebulae can be found orTypeIaSNRs.Althoughitispossiblenowtodetermine around both young and old pulsars, even a few millisec- the supernova origin of a given SNR using X-ray spectroscopy (see Sect. 7 and 8), for many old SNRs, whose ond pulsars have small nebulae aroundthem (Kargaltsev andPavlov,2008).Since these olderpulsarsdonothave emissionismainlycomingfromswept-upgas,onehasto rely on secondary indicators for determining their super- a connection with recent supernova activity, one should notcallthemplerions.Sothenamepulsarwindnebulais nova origin. The most reliable indicator is of course the more generic and informative than plerion, and is, there- presence of a neutron star inside the SNR, which makes it clear that the SNR must have a core collapse origin. fore,preferable. But even here one has to be aware of chance alignments Energetic pulsars with ages less than 20,000 yr are ∼ expectedtohaveblownapulsarnebulawhiletheyarestill (Kaspi, 1998), in particular for SNRs with large angular diameters. A secondary indicator for a SNR to originate surroundedbytheSNRshell.Oneexpectsthenaradioand X-ray morphology that consists of a pulsar wind nebula fromacorecollapseeventiswhetheraSNRislocatedina starformingregionorinsideanOBassociation(e.gWest- surroundedbyashell.IndeedafewSNRshavethischar- acteristic(Fig.4c)andareclassifiedascompositeSNRs.In erlund,1969).Butthisdoesnotconstituteproofforsuch fact, it is still puzzlingwhya youngobjectlike the Crab anorigin.Incontrast,apositionofaSNRhighabovethe Galactic plane can be taken as supportingevidencefora NebuladoesnotshowaSNRshell(seeHester,2008,for adiscussion). TypeIaorigin.Suchisthecasefor,forexample,SN1006 Since the 1980s the SNR classification has become (StephensonandGreen,2002). Because the supernova origin of SNRs is often diffi- broader. Due to the imaging capabilities of X-ray satel- liteslikeEinsteinandROSATitbecameclearthatthereare culttoestablish,SNRs havea classificationoftheirown, manySNRsthatdisplayashell-typemorphologyinradio which is mostly based on their morphology. Tradition- emission, but whose X-ray emission mainly comes from ally,thisclassificationrecognizedthreeclasses:shelltype thecenteroftheSNR(WhiteandLong,1991;RhoandPe- SNRs, plerions, and composite SNRs. As the blast wave tre,1998),asillustratedinFig.4d.Ingeneral,theseSNRs sweepsthroughthe interstellar medium(Sect. 4 ) a shell arerelativelyold,andareassociatedwithdenseinterstel- of shock heated plasma is created. Therefore, in many larmedium.TheX-rayemissionfromthecenterofthese casesthemorphologyofaSNRischaracterizedbyalimb SNRswasnotpoweredbyapulsar,butconsistedofther- brightenedshell,whichclassifiestheSNRasashell-type malemissionfromahotplasma.TheseSNRsarereferred SNR. to as mixed-morphology SNRs (Rho and Petre, 1998) or However,incaseofacorecollapseonemayexpectthe as thermal-composite SNRs (Shelton et al., 1999). I will presenceofarapidlyrotatingneutronstar.Itlosesenergy with a rate of E˙ =IW W˙ =4p 2IP˙/P3, with W , the angu- discusstheirpropertiesinSect.10.3. larfrequency,PtherotationalperiodandI 1045 gcm2 X-ray spectroscopy has greatly enhanced our ability ≈ 7 THECLASSIFICATIONOFSUPERNOVAREMNANTS J.Vink Supernovaremants Figure4: TheSNRmorphologicalclassificationillustratedwithexamples.Fromtoplefttobottomright:a)TheCygnus Loop,ashell-typeSNRwithadiameterof3 ,asobservedbytheROSATPSPCinstrument(Levensonetal.,1998),red ◦ is very soft emission from 0.1-0.4 keV, green 0.5 1.2 keV, and blue 1.2 2.2 keV. b) 3C58, a plerion/pulsar ∼ ∼ − ∼ − windnebula,asobservedbyChandra(Slaneetal.,2004).Thelongaxisofthisobjectis 7.c)ThecompositeSNRKes ′ ∼ 75asobservedby Chandra(Helfandet al.,2003)with the innerpulsarwindnebula,whichhasa hardX-rayspectrum, poweredbythepulsarJ1846-0258.Thepartialshellhasaradiusof 1.4.Thecolorsindicate1-1.7keV(red,NeandMg ′ ∼ emission),1.7-2.5keV(green,Si/S), and2.5-5keV(blue,mostlycontinuumemissionfromthepulsarwindnebula).d) The“thermal-composite”SNRW28asobservedinX-raysbytheROSATPSPC(blue)andinradiobytheVLA(Dubner etal.,2000).(Imagecredit:Chandrapressoffice,http://chandra.harvard.edu/photo/2008/w28/more.html) 8 4.2 Analyticalmodels J.Vink Supernovaremants to derive the origin of SNRs (Sect. 7 and 8). So one Also escaping cosmic rays may add to the energylosses now also encountersclassifications such as Type Ia SNR. in phase I andII (see Sect. 5.1).Despite these shortcom- Other SNRs show optical and X-ray evidence for en- ings, the labels haveprovento be of some value,as they hanced oxygen abundances; these so-called oxygen-rich providesomeframeworktocharacterizetheevolutionary SNRs(Sect.9.1)arelikelytheremnantsofthemostmas- phaseofagivenSNR. sivestars.Dependingonthecontextthesedesignationsare In the literature one also often finds designations for oftenmorehelpfulthanthemorphologicalclassifications. SNRs like “young”, “mature” and “old” (Jones et al., Notethatthesenewclassificationsarenotmutuallyex- 1998). These designations do not have a very precise clusive with the traditionalmorphologicalclassifications. meaning,buta generalguidelineis thatyoungSNRs are For example oxygen-richSNRs such as G292.0+1.8and lessthan 1000 2000yrold,andareinphaseIorearly ∼ − theLargeMagellanicCloudSNRB0540-69.3,alsoharbor in phase II, mature SNRs are in late phase II, or early a pulsar wind nebulae. They can therefore be classified phase III, whereas the label old SNRs is usually given bothascompositeSNRsandasoxygen-richSNRs. to the very extended structures associated with SNRs in phase IV. These old SNRs hardly produce X-ray emis- sions,sointhisreviewwewillonlyencounteryoungand 4 The hydrodynamic structure and matureSNRs. evolution of supernova remnants 4.2 Analytical models 4.1 Evolutionary phases SeveralanalyticalmodelsfortheshockevolutionofSNRs exist. The most widely used is the Sedov-Taylor self- The evolution of SNRs is usually divided in four phases similar solution (Sedov, 1959; Taylor, 1950). It assumes (Woltjer, 1972): I) the ejecta dominated phase, in which that the explosion energy E is instantaneously injected the mass of the supernova ejecta, M , dominates over ej into a uniform medium with uniform density r (i.e. a the swept-up mass, M ; II) the Sedov-Taylor phase, for 0 sw pointexplosion),andthatnoenergylosses occur.Inthat which M >M , but for which radiative losses are not sw ej casetheshockradiusR andvelocityV willdevelopas energeticallyimportant;III)thepressure-driven,or“snow- s s plough”phase,inwhichradiativecoolinghasbecomeen- Et2 1/5 ergeticallyimportant;theevolutionoftheshockradiusis R = x , (2) s r nowbestdescribedbyusingmomentumconservation;IV) 0 (cid:16) (cid:17) the mergingphase, in which the shock velocity and tem- peraturebehindtheshockbecomecomparableto,respec- V = dRs = 2 x E 1/5t 3/5= 2Rs. (3) s dt 5 r − 5 t tively,theturbulentvelocityandtemperatureoftheinter- 0 (cid:16) (cid:17) stellarmedium. The dimensionless constant x depends on the adiabatic Althoughthese discrete phasesprovidea usefulframe- index; x = 2.026 for a non-relativistic, monatomic gas worktothinkofSNRs,itshouldbekeptinmindthatitis (g = 5/3). An analytical solution exists for the density, anoversimplification,andthephaseofanindividualSNR pressure,andvelocityprofilesinsidetheshockedmedium, isnotalwaysthateasilylabeled.Moreover,differentparts which is shown in Fig. 5. The Sedov-Taylor solution ofaSNRmaybeindifferentphases.Forexample,arem- can be generalized to a gas medium with a power-law nant like RCW 86 has radiative shocks in the southwest density profile r (r) (cid:181) r s: R (cid:181) tb , V = b R /t, with − s s s (phaseIII),whereasinthenortheastithasveryfast,non- b = 2/(5 s), the expansion parameter. An astrophys- − radiative, shocks (phase I). This is the result of the com- ically relevant case is s = 2, corresponding to a SNR plexityofthemediumitisinvolvingin,whichisprobably shock moving through the progenitor’s stellar wind (see shapedbythestellarwindoftheprogenitorthatcreateda below);asituationthatlikelyappliestotheyoungGalac- cavity,surroundedbyadenseshell,withwhichpartofthe tic SNR Cas A (e.g. van Veelen et al., 2009). This gives shockwave is interacting(e.g. Rosadoet al., 1994;Vink b =2/3,whichshouldbe comparedto theexperimental etal.,1997). value found for Cas A based on the expansion found in Phase I is sometimes called the free-expansion phase, X-rays:b =0.63 0.02(Vink et al., 1998;Delaney and ± and phase II the adiabatic phase. Both names are some- Rudnick,2003;PatnaudeandFesen,2009). what misleading. Free expansion suggests that the shock TheSedov-Taylorsolutionsdonottakeintotheaccount velocityis describedbyV =R /t,with R , the radiusof structureofthesupernovaejectaitself.Thisisagoodap- s s s theoutershock,andttheageoftheSNR.However,asde- proximation once the swept-up mass exceeds the ejecta scribed below, even in phase I one hasV <R /t. Since mass. However, in the early phase, only the outer layers s s energy losses are according to standard models not im- of the supernovatransfertheir energyto the surrounding portant in both phase I and II, it is also misleading to medium. As time progresses, more energy is transferred callexclusivelyphaseIItheadiabaticphase.Moreover,in from the freely expanding ejecta to the SNR shell. This the very early evolution of SNRs (almost the supernova takesplaceatashockseparatinghotejectafromfreelyex- phase)SNRsmaygothroughashortradiativelossphase panding(cold)ejecta,theso-calledreverseshock(McKee, (e.g. Truelove and McKee, 1999; Sorokina et al., 2004). 1974). 9 4.2 Analyticalmodels J.Vink Supernovaremants Figure5: ThestructureofaSNRasgivenbytheself-similarmodelsofSedov(1959)(left)andChevalier(1982)(n= 7,s=2).Thevaluesfortheparametershavebeennormalizedto thevaluesimmediatelybehindtheforwardshock.For the Sedov model the radius is expressed in units of the shock radius, for the Chevalier model in units of the contact discontinuityR ,theborderbetweenswept-upandejectamaterial.Thesolidlinesshowthedensity,thedottedlinesthe c velocity, and the short-dashed lines the pressure profiles. For the Sedov model also the temperature is indicated (long- dashedline).Forthismodelthetemperaturegoestoinfinitytowardthecenter.FortheChevaliermodelthedensitygoes toinfinityforr=R .Notethatintheobserverframethevelocitydropsatthereverseshock. c Twoanalyticalmodelsexisttodescribethestructureand eter will evolve toward the Sedov solution (thus the ex- evolutionofSNRstakingintoaccounttheinitialvelocity pansion parameter evolves from b =(n 3)/(n s) to structureoftheejecta.Thefirstone,byChevalier(1982), b =2/(5 s).ItisclearfromtheChevalie−rsolutio−nsthat describestheearlyevolutionofSNRs,inwhichthefreely from a ve−ry early stage on b <1. There is also observa- expandingejectahaveapproximatelyapower-lawdensity tionalevidenceforthis.Forexample,theinitialexpansion distributionr (cid:181) v n.Thisisareasonableapproximation parameter of SN 1993J in the galaxy M81 has recently ej −ej for the outer ejecta structure as foundin numericalmod- been determinedto be b =0.85 (Marcaide et al., 2009); els of supernova explosions, with n=7 a situation that assuminganinteractionoftheshockwithacircumstellar describes reasonably well the ejecta structure of Type Ia wind(s=2)thisimpliesn 8.5. ≈ supernovae,whereasn=9 12isavalidapproximation Ananalyticalmodelthattakesintoaccountthesmooth − forthedensitystructureofcorecollapsesupernovae. transition from phase I to phase II was obtained by Tru- As shown by Chevalier (1982) the SNR evolution can elove and McKee (1999). Their models employ the fol- bedescribedbyaself-similarsolutionoftheform: lowingcharacteristiclength,timeandmassscales: Rs(cid:181) tb , (4) Rch≡Me1j/3r 0−1/3, withb theexpansionparametergivenby3 tch≡E−1/2Me5j/6r 0−1/3, (6) M M , n 3 ch≡ ej b = − . (5) n s with M the ejected mass, E the explosion energy, and − ej Fors=0,n=7thisgivesb =0.57,andforn=9,s=2 r 0 thecircumstellarmediumdensity.Thesenumberscan thisgivesb =0.86.Themodelsdescribeaself-similarve- be used to construct a set of solutions, which now only locity,pressure,anddensitystructure.Iwillnotreproduce dependonnands,andthedimensionlessvariablesR∗= them here, butshow as an examplethe density,pressure, R/Rchandt∗=t/tch.Themodelsarecontinuous,butcon- andvelocitystructureofan=7,s=2 model(Fig.5).It sistoftwoparts;onefortheevolutionintheejectadomi- importantto realize that the Chevalier solutionsdescribe natedphaseandonefortheSedov-Taylorphase,withtST the earlyevolutionof SNRs, whenthe reverseshock has the dimensionless transition age. For example, the blast notyetreachedtheinnermostejecta.Oncetheinnerejecta wavetrajectoryinthen=7,s=0modelisR∗s1.06=t∗4/7 are reached by the reverse shock the expansion param- fort∗<tST andR∗s =(1.42t∗−0.312)2/5fort∗>tST,with t =0.732.Thesesolutionsshowthatinitiallytheexpan- ST 3Theexpansionparameterisoften,butnotuniformly,indicatedwith sion parameter is identical to the one derived by Cheva- thesymbolm,butsincethecharactersi,...,nhaveanintegerconnotation, lier(1982)(Eq.5),whileitasymptoticallyapproachesthe Iopthereforb .NotethatTrueloveandMcKee(1999)useh (butthis Sedov-Taylorsolution,b =2/5. conflicts with its use in acceleration theory, Sect. 5.5), and Chevalier (1982)uses1/l . Fig. 6 illustratesthe evolutionof SNR shocksasgiven 10