AnexceptionallyluminousTeVγ-raySNR 1 HESS J1640−465 – an exceptionally luminous TeV γ-ray supernova remnant H.E.S.S.Collaboration,A.Abramowski,1F.Aharonian,2,3,4F.AitBenkhali,2A.G.Akhperjanian,5,4E.Angu¨ner,6G.Anton,7 4 1 S.Balenderan,8A.Balzer,9,10A.Barnacka,11Y.Becherini,12J.BeckerTjus,13K.Bernlo¨hr,2,6E.Birsin,6E.Bissaldi,14J.Biteau,15 0 M.Bo¨ttcher,16C.Boisson,17J.Bolmont,18P.Bordas,19J.Brucker,7F.Brun,2P.Brun,20T.Bulik,21S.Carrigan,2S.Casanova,16,2 2 M.Cerruti,17,22P.M.Chadwick,8R.Chalme-Calvet,18R.C.G.Chaves,20A.Cheesebrough,8M.Chre´tien,18S.Colafrancesco,23 b e G.Cologna,24J.Conrad,25,26C.Couturier,18Y.Cui,19M.Dalton,27,28M.K.Daniel,8I.D.Davids,16,29B.Degrange,15C.Deil,2 F P.deWilt,30H.J.Dickinson,25A.Djannati-Ata¨ı,31W.Domainko,2L.O’C.Drury,3G.Dubus,32K.Dutson,33J.Dyks,11M.Dyrda,34 1 1 T.Edwards,2K.Egberts,14P.Eger,2P.Espigat,31C.Farnier,25S.Fegan,15F.Feinstein,35M.V.Fernandes,1D.Fernandez,35A.Fiasson,36 ] G.Fontaine,15A.Fo¨rster,2M.Fu¨ßling,10M.Gajdus,6Y.A.Gallant,35T.Garrigoux,18G.Giavitto,9B.Giebels,15J.F.Glicenstein,20 E H M.-H.Grondin,2,24M.Grudzin´ska,21S.Ha¨ffner,7J.Hahn,2J. Harris,8G.Heinzelmann,1G.Henri,32G.Hermann,2O.Hervet,17 . A.Hillert,2J.A.Hinton,33W.Hofmann,2P.Hofverberg,2M.Holler,10D.Horns,1A.Jacholkowska,18C.Jahn,7M.Jamrozy,37 h p M.Janiak,11F.Jankowsky,24I.Jung,7M.A.Kastendieck,1K.Katarzyn´ski,38U.Katz,7S.Kaufmann,24B.Khe´lifi,31M.Kieffer,18 - o S.Klepser,9D.Klochkov,19W.Kluz´niak,11T.Kneiske,1D.Kolitzus,14Nu.Komin,36K.Kosack,20S.Krakau,13F.Krayzel,36 r st P.P.Kru¨ger,16,2H.Laffon,27G.Lamanna,36J.Lefaucheur,31A.Lemie`re,31M.Lemoine-Goumard,27J.-P.Lenain,18D.Lennarz,2 a [ T.Lohse,6A.Lopatin,7C.-C.Lu,2V.Marandon,2A.Marcowith,35R.Marx,2G.Maurin,36N.Maxted,30M.Mayer,10T.J.L.McComb,8 2 J.Me´hault,27,28P.J.Meintjes,39U.Menzler,13M.Meyer,25R.Moderski,11M.Mohamed,24E.Moulin,20T.Murach,6C.L.Naumann,18 v M.deNaurois,15J.Niemiec,34S.J.Nolan,8L.Oakes,6S.Ohm,33E.deOn˜aWilhelmi,2B.Opitz,1M.Ostrowski,37I.Oya,6M.Panter,2 8 8 R.D.Parsons,2M.PazArribas,6N.W.Pekeur,16G.Pelletier,32J.Perez,14P.-O.Petrucci,32B.Peyaud,20S.Pita,31H.Poon,2 3 4 G.Pu¨hlhofer,19M.Punch,31A.Quirrenbach,24S.Raab,7M.Raue,1A.Reimer,14O.Reimer,14M.Renaud,35R.delosReyes,2F.Rieger,2 . 1 L.Rob,40C.Romoli,3S.Rosier-Lees,36G.Rowell,30B.Rudak,11C.B.Rulten,17V.Sahakian,5,4D.A.Sanchez,2,36A.Santangelo,19 0 4 R.Schlickeiser,13F.Schu¨ssler,20A.Schulz,9U.Schwanke,6S.Schwarzburg,19S.Schwemmer,24H.Sol,17G.Spengler,6F.Spies,1 1 Ł.Stawarz,37R.Steenkamp,29C.Stegmann,10,9F.Stinzing,7K.Stycz,9I.Sushch,6,16A.Szostek,37J.-P.Tavernet,18T.Tavernier,31 : v A.M.Taylor,3R.Terrier,31M.Tluczykont,1C.Trichard,36K.Valerius,7C.vanEldik,7B.vanSoelen,39G.Vasileiadis,35C.Venter,16 i X A.Viana,2P.Vincent,18J.Vink,41H.J.Vo¨lk,2F.Volpe,2M.Vorster,16T.Vuillaume,32S.J.Wagner,24P.Wagner,6M.Ward,8 r a M.Weidinger,13Q.Weitzel,2R.White,33A.Wierzcholska,37P.Willmann,7A.Wo¨rnlein,7D.Wouters,20V.Zabalza,2M.Zacharias,13 A.Zajczyk,11,35A.A.Zdziarski,11A.Zech,17H.-S.Zechlin1 1Universita¨tHamburg,Institutfu¨rExperimentalphysik,LuruperChaussee149,D22761Hamburg,Germany 2Max-Planck-Institutfu¨rKernphysik,P.O.Box103980,D69029Heidelberg,Germany 3DublinInstituteforAdvancedStudies,31FitzwilliamPlace,Dublin2,Ireland 4NationalAcademyofSciencesoftheRepublicofArmenia,Yerevan 5YerevanPhysicsInstitute,2AlikhanianBrothersSt.,375036Yerevan,Armenia 6Institutfu¨rPhysik,Humboldt-Universita¨tzuBerlin,Newtonstr.15,D12489Berlin,Germany 7Universita¨tErlangen-Nu¨rnberg,PhysikalischesInstitut,Erwin-Rommel-Str.1,D91058Erlangen,Germany 8UniversityofDurham,DepartmentofPhysics,SouthRoad,DurhamDH13LE,U.K. 9DESY,D-15738Zeuthen,Germany 10Institutfu¨rPhysikundAstronomie,Universita¨tPotsdam,Karl-Liebknecht-Strasse24/25,D14476Potsdam,Germany 11NicolausCopernicusAstronomicalCenter,ul.Bartycka18,00-716Warsaw,Poland 12DepartmentofPhysicsandElectricalEngineering,LinnaeusUniversity,35195Va¨xjo¨,Sweden 13Institutfu¨rTheoretischePhysik,LehrstuhlIV:WeltraumundAstrophysik,Ruhr-Universita¨tBochum,D44780Bochum,Germany 14Institutfu¨rAstro-undTeilchenphysik,Leopold-Franzens-Universita¨tInnsbruck,A-6020Innsbruck,Austria 15LaboratoireLeprince-Ringuet,EcolePolytechnique,CNRS/IN2P3,F-91128Palaiseau,France 16CentreforSpaceResearch,North-WestUniversity,Potchefstroom2520,SouthAfrica 17LUTH,ObservatoiredeParis,CNRS,Universite´ParisDiderot,5PlaceJulesJanssen,92190Meudon,France 18LPNHE,Universite´PierreetMarieCurieParis6,Universite´DenisDiderotParis7,CNRS/IN2P3,4PlaceJussieu,F-75252,ParisCedex5,France 19Institutfu¨rAstronomieundAstrophysik,Universita¨tTu¨bingen,Sand1,D72076Tu¨bingen,Germany (cid:13)2c02D0S1M4/RIrAfuS,,CMENARSAacSla0y0,0F,-29–19191Gif-Sur-YvetteCedex,France 21AstronomicalObservatory,TheUniversityofWarsaw,Al.Ujazdowskie4,00-478Warsaw,Poland 22nowatHarvard-SmithsonianCenterforAstrophysics,60gardenStreet,CambridgeMA,02138,USA 23SchoolofPhysics,UniversityoftheWitwatersrand,1JanSmutsAvenue,Braamfontein,Johannesburg,2050SouthAfrica 24Landessternwarte,Universita¨tHeidelberg,Ko¨nigstuhl,D69117Heidelberg,Germany 25OskarKleinCentre,DepartmentofPhysics,StockholmUniversity,AlbanovaUniversityCenter,SE-10691Stockholm,Sweden 26WallenbergAcademyFellow 27Universite´Bordeaux1,CNRS/IN2P3,Centred’E´tudesNucle´airesdeBordeauxGradignan,33175Gradignan,France 28FundedbycontractERC-StG-259391fromtheEuropeanCommunity 29UniversityofNamibia,DepartmentofPhysics,PrivateBag13301,Windhoek,Namibia 30SchoolofChemistry&Physics,UniversityofAdelaide,Adelaide5005,Australia 31APC,AstroParticuleetCosmologie,Universite´ParisDiderot,CNRS/IN2P3,CEA/Irfu,ObservatoiredeParis,SorbonneParisCite´,10,rueAliceDomonetLe´onieDuquet,75205ParisCedex13,France 32UJF-Grenoble1/CNRS-INSU,InstitutdePlane´tologieetd’AstrophysiquedeGrenoble(IPAG)UMR5274,Grenoble,F-38041,France 33DepartmentofPhysicsandAstronomy,TheUniversityofLeicester,UniversityRoad,Leicester,LE17RH,UnitedKingdom 34InstytutFizykiJa¸drowejPAN,ul.Radzikowskiego152,31-342Krako´w,Poland 35LaboratoireUniversetParticulesdeMontpellier,Universite´Montpellier2,CNRS/IN2P3,CC72,PlaceEuge`neBataillon,F-34095MontpellierCedex5,France 36Laboratoired’Annecy-le-VieuxdePhysiquedesParticules,Universite´deSavoie,CNRS/IN2P3,F-74941Annecy-le-Vieux,France 37ObserwatoriumAstronomiczne,UniwersytetJagiellon´ski,ul.Orla171,30-244Krako´w,Poland 38Torun´CentreforAstronomy,NicolausCopernicusUniversity,ul.Gagarina11,87-100Torun´,Poland 39DepartmentofPhysics,UniversityoftheFreeState,POBox339,Bloemfontein9300,SouthAfrica 40CharlesUniversity,FacultyofMathematicsandPhysics,InstituteofParticleandNuclearPhysics,VHolesˇovicˇka´ch2,18000Prague8,CzechRepublic 41AstronomicalInstituteAntonPannekoek,UniversityofAmsterdam,POBox94249,NL-1090GEAmsterdam,theNetherlands Accepted2014January17.Received2014January17;inoriginalform2013October23 Mon.Not.R.Astron.Soc.000,2–9(2014) Printed12February2014 (MNLATEXstylefilev2.2) ABSTRACT Theresultsoffollow-upobservationsoftheTeVγ-raysourceHESSJ1640−465from2004 to 2011 with the High Energy Stereoscopic System (H.E.S.S.) are reported in this work. The spectrum is well described by an exponential cut-off power law with photon index Γ = 2.11±0.09 ±0.10 , and a cut-off energy of E = 6.0+2.0TeV. The TeV emis- stat sys c −1.2 sion is significantly extended and overlaps with the north-western part of the shell of the SNR G338.3−0.0. The new H.E.S.S. results, a re-analysis of archival XMM-Newton data, andmulti-wavelengthobservationssuggestthatasignificantpartoftheγ-rayemissionfrom HESS J1640−465 originates in the SNR shell. In a hadronic scenario, as suggested by the smoothconnectionoftheGeVandTeVspectra,theproductoftotalprotonenergyandmean targetdensitycouldbeashighasW n ∼4×1052(d/10kpc)2ergcm−3. p H Keywords: radiationmechanisms:non-thermal,ISM:supernovaremnants,ISM:individual objects:G338.3−0.0 1 INTRODUCTION designated 2FGL 1640.5−4633 in the two-year Fermi-LAT cata- logue (Nolan et al. 2012). Note that no pulsation has been found Startingin2004theGalacticPlaneSurvey(Aharonianetal.2006b) in any wavelength band so far. Due to the large γ-ray to X-ray performedbytheH.E.S.S.Collaboration,usinganarrayofimaging ratio luminosity (L /L (cid:39) 30; Funk et al. 2007), Slane et al. γ X atmosphericCherenkovtelescopes(IACTs),ledtothediscoveryof (2010)inferredanevolvedPWNwithalowmagneticfieldandan nearly70newsourcesinthevery-high-energy(VHE,E>100GeV) injectionspectrumthatconsistsofaMaxwellianelectronpopula- γ-rayregime(Carriganetal.2013).Thechallengesincethenhas tionwithapower-lawtail(ase.g.proposedbySpitkovsky2008) beentoassociatethesesourceswithastrophysicalobjectsseenin to reproduce the broadband spectral energy distribution (SED) in other wavelengths and to identify the underlying radiation mech- aleptonicPWNscenario.Ahadronicoriginoftheγ-rayemission anisms. A large fraction of the Galactic VHE γ-ray population wasconsideredtobeunlikelyasitwouldrequireratherhighambi- could be associated with regions with recent star-forming activ- entdensities(n (cid:39) 100cm−3),implyingintensethermalradiation ity and to objects at late stages of stellar evolution such as su- inX-raysfromtheSNRshellthathassofarnotbeendetected. pernova remnants (SNRs) and the nebulae produced by powerful Lemiere et al. (2009) performed a detailed study of the young pulsars (for a review, see e.g. Hinton & Hofmann 2009). gaseous environment of G338.3−0.0, and based on the HI ab- InmanycaseswhereanastrophysicalcounterparttotheVHEγ- sorption features, derived a distance of (8 − 13)kpc. A recent ray emission could be identified, however, the nature of the un- study of the nearby stellar cluster Mercer 81 and the giant HII derlying particle population remains unclear. Highly energetic γ- region G338.4+0.1 by Davies et al. (2012) supports this esti- ray emission could be either produced by relativistic electrons or mate,whichimpliesthatHESSJ1640−465isthemostluminous protons(andheaviernuclei).Relativistichadronsundergoinelastic VHE γ-ray source known in the Galaxy. Throughout this work, scatteringwithnucleiintheinterstellarmedium(ISM),producing a distance of 10kpc is assumed. Since the original discovery of π0-decayγ-rayemission.Ultra-relativisticelectrons,ontheother HESS J1640−465, the available H.E.S.S. exposure towards this hand, can up-scatter low-energy photons present at the accelera- source has quadrupled w.r.t the data used in (Aharonian et al. tionsiteviatheInverseCompton(IC)process.Inverydensemedia 2006b),andadvancedanalysismethodsarenowavailablethatal- Bremsstrahlung losses of electrons can significantly contribute to lowforamuchmoredetailedspectralandmorphologicalstudyof thegeneratedγ-rayemission.IACTscanplayakeyroleinidenti- the VHE γ-ray emission. In this work, H.E.S.S. follow-up stud- fyingtheunderlyingparticlepopulationandstudyingnon-thermal iesandare-analysisofXMM-Newtondataarepresented.Boththe processes in γ-ray sources by localising the emission region and broadbandSEDandtheTeVmorphologyrevealevidenceforpro- constrainingtheenergyspectrumatveryhighenergies. tonaccelerationintheSNRshellofG338.3−0.0. The VHE γ-ray source HESS J1640−465 was discovered by H.E.S.S. in the Galactic Plane Survey (Aharonian et al. 2006b) and is positionally coincident with the SNR G338.3−0.0 (Whiteoak&Green1996).UsingXMM-NewtonobservationsFunk 2 H.E.S.S.OBSERVATIONSANDRESULTS et al. (2007) detected a highly absorbed extended X-ray source (XMMUJ164045.4−463131)closetothegeometriccentreofthe H.E.S.S.isanarrayoffiveimagingatmosphericCherenkovtele- SNRandwithintheH.E.S.S.sourceregion.TheX-rayandVHE scopes located in Namibia designed to detect VHE γ-rays. The γ-rayemissioncomponentswereinterpretedassynchrotronandIC fifth telescope started operation in September 2012. All H.E.S.S. emissionfromrelativisticelectronsinapulsarwindnebula(PWN). datausedtoperformthestudiesdescribedbelowhavebeentaken ObservationswithChandraconfirmedthepresenceoftheextended between May 2004 and September 2011 with the four-telescope nebulaandidentifiedapoint-likesourcewhichwassuggestedtobe array(Aharonianetal.2006a).Thetotaldeadtimecorrectedlive the associated pulsar (Lemiere et al. 2009). Recently, Castelletti timeamountsto63.4hr,comparedto14.3hrintheoriginalpubli- et al. (2011) analysed new high-resolution multi-frequency radio cation(Aharonianetal.2006b).Observationshavebeenperformed data of G338.3−0.0 but could only set upper limits on the radio atzenithanglesbetween20◦and65◦withameanvalueof∼33◦. flux from a potential extended radio nebula. Fermi-LAT observa- The data were recorded with pointing offsets between 0.2◦ and tions revealed a high-energy (HE, 100MeV<E<100GeV) γ-ray 1.8◦ with a mean value of 1.1◦ from the HESS J1640−465 po- sourcecoincidentwithHESSJ1640−465(Slaneetal.2010),also sition. Data were analysed using a standard Hillas-type H.E.S.S. (cid:13)c 2014RAS AnexceptionallyluminousTeVγ-raySNR 3 analysis1 fortheeventreconstructionandaboosteddecisiontree basedeventclassificationalgorithmtodiscriminateγ-raysfromthe -46°18'00" H.E.S.S. 120 charged particle background (Ohm et al. 2009). All results were 105 cross-checkedbyanindependentanalysisandcalibrationforcon- sistency(deNaurois&Rolland2009). 24'00" Mercer 81 90 2.1 Morphology 00) 75 0 2 Thahredscouutsrcaendpuossiintigontheanrdingmboarcpkhgorloougnydheasvtiemabteioennmobettahionded(Bwerigthe nation (J 30'00" 60 etal.2007).Inthissetupaminimumintensityinthecameraim- Decli 45 age of 160p.e. is required, resulting in an energy threshold of 36'00" Eth =600GeVandapointspreadfunction(PSF)with68%con- 30 tainment radius of r = 0.09◦ for the morphology studies. The 68 fit of a symmetric two-dimensional Gaussian profile, convolved 15 42'00" with the H.E.S.S. PSF with Sherpa (Freeman et al. 2001) gives a best-fit position of RA 16h40m41.0s ± 1.0sstat ± 1.3ssys and 0 Dec −46◦32(cid:48)31(cid:48)(cid:48) ± 14(cid:48)(cid:48) ± 20(cid:48)(cid:48) (J2000), consistent with 42m00s 30s 41m00s 30s 40m00s 16h39m30s stat sys Right Ascension (J2000) thepreviouslypublishedvalue(Aharonianetal.2006b).Thesys- tematic error on the best-fit position originates from the pointing Figure1.H.E.S.S.excessmapsmoothedwitha2DGaussianwith0.017◦ precisionoftheH.E.S.S.arrayofabout20(cid:48)(cid:48).Thesourceisintrin- varianceandthebest-fitposition(statisticalerrorsonly)andintrinsicGaus- sically extended with a Gaussian width of σ = (4.3 ± 0.2)(cid:48). sian width overlaid as blue solid and dashed lines. 610MHz radio con- S This extension is 1.6(cid:48) (∼2σ) larger than in the original publica- tours are shown in black (Castelletti et al. 2011). The green circle indi- catesthepositionofthecandidatePWNXMMUJ164045.4−463131,and tion, which can be understood as fainter emission belonging to in gray the best-fit position of the Fermi source 2FGL 1640.5−4633 is HESS J1640−465 that can now be revealed with the increased given.ThewhitecircleindicatesthesourceHESSJ1641−463(Oyaetal. dataset.Figure1showstheH.E.S.S.best-fitpositionandextension 2013)andtheregionofhighradioemissionconnectingHESSJ1640−465 overlaidontheVHEγ-rayexcessmap.TheVHEγ-raysourceen- and HESS J1641−463 is the HII region G338.4+0.1. The progenitor of closesthenorthernpartoftheSNRshellofG338.3−0.0,thecandi- G338.3−0.0ispotentiallyassociatedwiththemassiveyoungstellarcluster datePWNXMMUJ164045.4−463131(Funketal.2007)andthe Mercer81(Daviesetal.2012). Fermi-LATsource2FGL1640.5−4633(Slaneetal.2010;Nolan etal.2012).Figure1alsoshowssomeindicationforanasymmetric extensionoftheemissionalongthenorthernpartoftheshelland 2sctoia0owt1inna3rg)wd.tshhTetahhntiestshnueeebxwsttryelaymncstdmiioinsengctroitivshceeGarelsasdououssrsoscieuaeernncmemaoHosddEereleSlsSffirdooJur1maH6l4EthV1Se−HSs4EJk61y3γ6-m4(rO0aa−yypa,4ee6imnt5diasiils--. -1-1-2) s cmN/dE (TeV 111000---111201 d an oversimplification. The residual emission could indicate some 10-13 emission in between HESS J1640−465 and HESS J1641−463. 10-14 This component is however not detected with high significance, makingadiscussionofitsorigindifficultinthiscontext.Morpho- 10-15 logical fits in energy bands do not reveal any significant change 10-16 iancbheasntg-fietipnossoituiorcneamndo/roprheoxlotegnysiwonit,hwehniecrhgywo(auslde.hga.vseeeinndiicnattehde s)Residual ( -022 1 10 Energy (Te1V02) PWNe HESS J1825−137 or HESS J1303−631; Aharonian et al. 1 10 102 Energy (TeV) 2006c;Abramowskietal.2012a). Figure2.VHEγ-rayspectrumofHESSJ1640−465(top)andfluxresid- uals(bottom)extractedwithinthe90%containmentradius(seetext).Also 2.2 Spectrum shownisthebest-fitpowerlaw,plusexponentialcut-offmodeland68% errorband.Allspectralpointshaveaminimumsignificanceof2σ.Thelast TheVHEγ-rayspectrumisshowninFigure2,andhasbeenex- pointisthedifferentialfluxupperlimitinthisenergybandat95%confi- tracted using std cuts (60p.e. minimum image intensity, Eth = dencelevel. 260GeV), using the reflected region background method (Berge etal.2007)andforwardfoldingwithamaximumlikelihoodopti- andacut-offenergyofE = 6.0+2.0TeV.Thesystematicerrors misation (Piron et al. 2001) from the 90% containment radius of c −1.2 theVHEγ-rayemissionofHESSJ1640−465of0.18◦aroundthe onfluxnormandindexforthisdatasetarebasedonthedifference seen between the main and cross-check analysis and are a result best-fit position. The fit of a power law with exponential cut-off: dN/dE =Φ ×(E/1TeV)−Γe−E/Ec resultsinaphotonindex of uncertainties in e.g. atmospheric conditions, simulations, bro- 0 kenpixels,analysiscuts,ortherun-selection.Thefitprobabilityp Γ=2.11±0.09 ±0.10 ,adifferentialfluxnormalisationat stat sys 1TeVofΦ =(3.3±0.1 ±0.6 )× 10−12TeV−1cm−2s−1 foranexponentialcut-offpowerlawmodelisp ∼ 36%,whereas 0 stat sys the fit probability for a pure power law model is p ∼ 1%. The luminosity of HESS J1640−465 above 1TeV at 10kpc distance 1 The software package HAP version 12-03-pl02 with version32 of the is L>1TeV (cid:39) 4.6×1035(d/10kpc)2ergs−1, a factor of ∼ 2.8 lookuptableswasused. higherthanthatoftheCrabnebula. (cid:13)c 2014RAS,MNRAS000,2–9 4 H.E.S.S.Collaboration (1.0−2.0)keV,(2.0−4.5)keV,and(4.5−10.0)keV).Events -1-2) sm 102 HESS J1640-465 RX J1713.7-3946 afrroomunadraelgliosonucrocerrsesdpeotencdtiendgitnoathney9o5f%thceosnetabiannmdsenwterarediuresmoofvtehde c XMM-Newton PSF at the respective source position in the detec- g er tor. The total flux upper limit was derived assuming that the re- -12 0 maining count-rate from a polygon region enclosing the northern 1 E ( 10 part of the shell is due to background. A power-law model with d N / photonindexΓX =−2wasappliedtoconstrainnon-thermallep- 2 dE tonicemission.Twodifferentabsorptioncolumndensitiesasfound in the literature, N = 6.1×1022cm−2 (Funk et al. 2007) and H,1 N =1.4×1023cm−2 (Lemiereetal.2009),havebeenconsid- H,2 1 ered.NodiffuseX-rayemissioncoincidentwiththeSNRshellwas detectedwiththisdataset.Theresulting99%confidenceupperlim- itsfortheunabsorbedflux((2−10)keV)areF (N )=4.4× 99 H,1 10−13ergcm−2s−1andF (N )=8.3×10−13ergcm−2s−1. 10-4 10-3 10-2 10-1 1 10 102 99 H,2 Energy (TeV) Thesevalueshavebeenscaledupby11%toaccountforthemiss- ingareaduetoexcludedpoint-likesources. Figure 3. Comparison of the HE and VHE γ-ray spectra of HESSJ1640−465(filledcircles)andRXJ1713.7−3946(opensquares). Data for RX J1713.7−3946 are from Abdo et al. (2011) and Aharonian etal.(2011),GeVdataofHESSJ1640−465isfromSlaneetal.(2010). 4 DISCUSSION Alsoshownisthebest-fitexponentialcut-offpowerlawmodeltothefull γ-rayspectrum(Table1). The H.E.S.S. source encloses the PWN candidate XMMU J164045.4−463131 as well as the north-western half of the incomplete shell of G338.3−0.0. The comprehensive ThephotonindexasreconstructedwiththenewH.E.S.S.data multi-wavelength data available together with the new H.E.S.S. at TeV energies is compatible with the photon index as recon- andXMM-Newtonresultsallowforamuchmoredetailedinvesti- structedintheGeVdomain(Slaneetal.2010;Nolanetal.2012; gationoftheSEDandhencetheunderlyingnon-thermalprocesses Ackermannetal.2013).Asimultaneousexponentialcut-offpower to be carried out. As the evolutionary state of G338.3−0.0 is law fit to the GeV data points as derived by Slane et al. (2010), essentialforthediscussion,theageoftheSNRisestimated,and andnewTeVdatabetween200 MeVand90 TeV(showninFig- theenvironmentinwhichitlikelyexpandedisinvestigated.These ure3)hasbeenperformed.Theresultofthisfitissummarisedin estimateswillformthebasisforthediscussionoftheoriginofthe Table1andshowsthatthefluxat1TeV,thephotonindexaswell non-thermalemissioninaPWNandSNRscenario. asthecut-offenergyareconsistentwiththefittotheH.E.S.S.-only data.Thefithasaχ2of21for24degreesoffreedom(d.o.f.)with aprobabilityof63%2andimpliesthatnobreakintheγ-rayspec- 4.1 AgeandEnvironmentofG338.3−0.0 trumbetweentheFermiandH.E.S.S.energyrangeisrequiredin The age and environment of the SNR have a large influence on ordertodescribethedata. theinterpretationandmodelingoftheemissionscenarioandthus deservediscussioninthiscontext.Previousestimatesputtheageof theSNRintherangeof(5−8)kyr(Slaneetal.2010),however,as 3 XMM-NEWTONDATAANALYSIS becomesevidentfromthediscussionbelow,itmaybesignificantly youngerthanthat. Funk et al. (2007) reported the detection of the candidate PWN If the X-ray PWN is indeed related to the SNR, then XMMUJ164045.4−463131withXMM-Newtonandintroducedit G338.3−0.0 originated from a core-collapse supernova explo- asapotentialcounterpartofHESSJ1640−465.Asbecomesclear sionofamassivestar.Suchstarsusuallymodifythesurrounding fromFig.1theVHEγ-rayemissionregionalsooverlapswiththe mediumthroughstrongstellarwinds,creatingacavityofrelatively northern part of the shell of SNR G338.3−0.0. To investigate γ- lowdensitysurroundedbyahigh-densityshellofswept-upmate- rayemissionscenariosrelatedtotheSNR,theXMM-Newtondata rial.(seeWeaveretal.1977;Chevalier1999).Suchawind-blown (ObsID:0302560201)werere-analysedtoderiveanupperlimitfor bubblescenariohasneverbeenconsideredforthisobject,butneeds diffuse X-ray emission originating from the northern part of the tobeexploredforadetaileddiscussionoftheγ-rayemissionmech- shell.FortheanalysistheScienceAnalysisSystem(SAS)version anismspossiblyatworkinHESSJ1640−465.Thesecavitieshave 12.0.1 was used, supported by tools from the FTOOLS package significant impact on the evolution of the subsequent supernova andXSPECversion12.5.0(Arnaud1996)forspectralmodelling. shock front, and such scenarios have been evoked to explain the Thedataareaffectedbylongperiodsofstrongbackgroundflaring properties of other SNRs like the Cygnus Loop (e.g. Levenson activityresultinginnetexposuresofonly5.9ks(PN)and13.5ks et al. 1998), RCW86 (Vink et al. 1997), and RX J1713.7−3946 (MOS), following the suggested standard criteria for good-time- (Fukui et al. 2003), all of which have physical diameters simi- interval filtering. To detect and remove point-like X-ray sources lar to G338.3−0.0. Chevalier (1999) estimated the size of wind- the standard XMM-Newton SAS maximum likelihood source de- blowncavitiesbyrequiringapressureequilibriumbetweenthein- tectionalgorithmwasusedinfourenergybands((0.5−1.0)keV, sideofthebubble,whichhasbeenpressurisedbythetotalenergy ofthewind:1/2M˙v2τ,andthesurroundingmedium.Here,M˙ is w 2 ThefithasbeenperformedonthebinnedH.E.S.Sspectrumshownin the mean mass-loss rate, vw is the wind speed and τ is the life- Figure 2 and on the GeV spectrum from Slane et al. (2010) taking into time of the star. With a distance of 10kpc, the radius of the ob- accountstatisticalerrorsonly. servedshellofG338.3−0.0is10pc,whichisassumedheretobe (cid:13)c 2014RAS,MNRAS000,2–9 AnexceptionallyluminousTeVγ-raySNR 5 Table1.Best-fitspectrumresultsofthenewH.E.S.S.dataasshowninFigure2,andincombinationwiththeGeVspectrumfromSlaneetal.(2010). Data Emin Emax Γ Φ0 Ec 10−12cm−2s−1 TeV H.E.S.S. 260GeV 90TeV 2.11±0.09 3.3±0.1 6.0+2.0 −1.2 H.E.S.S.+Fermi-LAT 200MeV 90TeV 2.23±0.01 3.7±0.2 8.8+2.3 −1.5 comparabletothesizeofthewind-blownbubble.Suchsizescan G338.3−0.0,averageneutralgasdensitiesn¯ lowerthanthatare H be achieved by a typical ∼20M O-type star with τ (cid:39)7Myr, alsoplausible.FromtheHIabsorptionmeasurementsandthether- (cid:12) M˙ (cid:39) 10−7M yr−1,andv (cid:39) 2600kms−1,evolvinginanHII malradioemission,thehydrogengas(neutralplusionised)inthe (cid:12) w regionwithtemperature10kK(Osterbrock1989)andaverageden- region is consistent with densities of n¯ (cid:38) (100−150)cm−3. H sityofn∼150cm−3(seebelow,Kudritzki&Puls2000;Muijres Purcelletal.(2012)performedasurveyforhigh-densitygas(n(cid:38) etal.2012).Thiscorrespondstoatotalmasslossinthemainse- 104cm−3)inNH transitionlinesintheGalacticplane.Withthe 3 quencephaseof0.7M .Anextremecasethatmayprovidealower sensitivityofthissurveyandgiventhatnoemissioninthesetran- (cid:12) limittotheageoftheSNRcanbederivedbytheassumptionthat sition lines is seen towards HESS J1640−465 a molecular cloud theremainingmaterialinsidethecavitysolelyoriginatesfromthe moremassivethan∼8000M isnotsupportedbythedata.How- (cid:12) stellar wind. The mean number density then is n ∼ 0.01cm−3 ever,thisdoesnotexcludetheexistenceofsmaller,similarlydense 0 with a total mass swept up by the SNR shock of 0.7M . This clumpsofmaterialintheshellregion(seebelow).Thereisalsono (cid:12) means that the SNR shock would evolve freely expanding up to maseremissiondetectedtowardstheTeVemission,whichwould theradiusofthewind-blownbubble.Assumingaverageshockve- haveindicatedtheinteractionofashockwavewithdensematerial locitiesbetween(5000−10000)kms−1theageoftheSNRwould (e.g.Walshetal.2011). be(1−2)kyr,whichisconsiderablyyoungerthantheestimateof (5−8)kyrbySlaneetal.(2010),owingtothelowerdensity. 4.2 PWNscenario In addition to the SNR age, also the density of the ISM in the immediate vicinity of the shock region has major impact on The positional coincidence of HESS J1640−465 and theinterpretationoftheemissionscenario.Thedensityintheshell 2FGL 1640.5−4633 with the candidate X-ray PWN surroundingthewind-blownbubblecanbeestimatedwithvarious XMMU J164045.4−463131 is seen as evidence for leptonic methods,i.e.viathermalradioemission,thermalX-raymeasure- γ-ray emission from a PWN (Funk et al. 2007; Lemiere et al. mentsandHIabsorptionstudies.Castellettietal.(2011)foundev- 2009; Slane et al. 2010). In these scenarios, electrons are ac- idence for thermal radioemission in the SNR shell indicating the celerated to energies of hundreds of TeV in the PWN, radiate presence of dense material. The authors infer electron densities via synchrotron and IC processes and produce the observed basedonthefree-freeabsorptionfeatureintheradiospectrumof X-ray and HE and/or VHE γ-ray emission. In the following the ne ∼(100−165)cm−3.NodiffuseX-rayemissionfromtheSNR PWN interpretation will be confronted with the new spectral and shellhavebeenreportedinFunketal.(2007),andintheprevious morphologicalH.E.S.S.resultsandtheavailablemulti-wavelength section upper limits have been derived. Slane et al. (2010) argue information. that therefore high gas densities are not supported. However, the Theγ-rayspectrumofmiddle-agedandoldPWNeischarac- lackofobservedthermalX-rayemissionmightbeconsistentwith terisedbyabreakintheSEDof∆Γ=0.5attheenergywherethe theverylargedistanceandhighcolumndensitiesinferredfromthe IC/synchrotronlosstimeoftheparentelectronpopulationissim- XMM-Newton and Chandra spectra (Lemiere et al. 2009) of the ilar to the age of the source (e.g. Hinton & Hofmann 2009). For PWN XMMU J164045.4−463131; especially if the plasma tem- youngPWNe(t (cid:39) 1kyr)theγ-rayspectrumfrominteractionsof peratureisbelow1keV.Onlyforhighertemperatures,ase.g.ob- electronswithmagneticandradiationfieldsiseffectivelyuncooled servedfromKes32(Vink2004),couldobservablethermalX-rays uptothecut-offenergyasICandsynchrotronlosstimesaremuch beexpectedfromthissource.Particularly,SNRsevolvingrapidly longerinatypicalPWNenvironment.Thisleadstoapeakinthe insidelow-densitywind-blowncavitiesarenotexpectedtoproduce ICandsynchrotronspectraatenergiesjustbelowthecut-offenergy significantthermalX-rayemission.OnlywhentheSNRshockhits intheelectronspectrum.AnICpeak(orspectralbreak)isseenfor thesurroundingshell,themediumintheshockregionthermalises alloftheGeVandTeVidentifiedPWNe(e.g.Grondinetal.2011; rapidlyandcoolsextremelyfast,whichmakestheSNRanefficient Aharonianetal.2006c;Abdoetal.2010a;Aharonianetal.2005; emitterofhardthermalX-rays,butonlyduringashorttime.Later, Abramowski et al. 2012b), but not for HESS J1640−465. To re- thetemperaturesareexpectedtodropsignificantlybelow1keVdue producetheobservedγ-rayspectralindexΓ (cid:39) 2.2forayoung γ tothedecreasedshockspeedsofonlyafew100kms−1 (seee.g. object((cid:46) 2.5kyr),theinjectionspectrumhastobeΓ = 3.4,as e Tenorio-Tagleetal.1991).Asoutlinedabove,duetothehighab- Γ =(2Γ −1)–anindexsignificantlysteeperthanpredictedby e γ sorptiontowardsG338.3−0.0suchemissionisnotexpectedtobe Fermi acceleration theory. Slane et al. (2010) suggested an addi- detectable. tionalMaxwellianlow-energyelectroncomponentinordertoex- Finally,theHIabsorptionfeaturecanbeusedtoinferamaxi- plainthesmoothconnectionoftheHEandVHEγ-rayspectra.As mum(neutral)gasdensity.AssumingthatalloftheHIgasasstud- showninSection2.2thenewhigh-qualityH.E.S.S.spectrumcon- iedbyLemiereetal.(2009)between−65kms−1and−55kms−1 nectswiththeGeVspectrumwithoutanydiscernablefeaturesand is associated with G338.3−0.0 and located in a shell with 4pc thusdoesnotrequiresuchacontribution.Infact,aχ2 testofthe thickness (as supported by radio observations) at 10kpc, a max- Slane et al. (2010) model on the binned GeV and TeV spectrum imum density of n (cid:39) 600cm−3 can be derived. How- resultsinaχ2 = 189for25d.o.f.withverylowprobability,not H,max ever,sincesomeoftheabsorbinggasmaynotbeassociatedwith supportingasignificantcontributionofsuchaMaxwelliancompo- (cid:13)c 2014RAS,MNRAS000,2–9 6 H.E.S.S.Collaboration nent. This can be compared to the exponential cut-off power law 4.3 SNRscenario modelasshowninTable1,whichhasaχ2 =21for24d.o.f. Given the spectral and morphological similarity of From a theoretical point of view, the extent of the PWN HESS J1640−465 with other Galactic SNRs interacting with is expected to be smaller than its associated SNR (e.g. Blondin molecular clouds, an SNR origin of the non-thermal emission is etal.2001).Thispredictionissupportedbyobservationsofseveral studiedinthefollowing.Inahadronicγ-rayemissionscenario,a PWNe,includingMSH15−52(Aharonianetal.2005)andVelaX high-densityisrequiredtoprovidesufficienttargetmaterialforthe (Abramowskietal.2012b).TheintrinsicsizeofHESSJ1640−465 relativistic protons to produce neutral pions which subsequently atTeVenergies,however,islargerthanG338.3−0.0andfeatures decay into energetic photons (see e.g. Aharonian et al. 1994). significantoverlapwiththeshelloftheSNR–abehaviourthatis ThishighdensitymaterialoutsidetheSNRshockcouldeitherbe notseenforanyotherPWN. the wind shell surrounding the stellar wind bubble, or the dense At radio wavelengths, Castelletti et al. (2011) derived up- material known to exist in the vicinity of HESS J1640−465. per limits on the possible radio emission from the PWN The relatively low ISM density Inside the wind-blown bubble at various wavelengths, with the most constraining limit of would not be sufficient to account for the bulk of the observed 3.7×10−17ergcm−2s−1at610MHzwithintheX-rayPWN.Due γ-ray emission, and thus the target material must be of different tothedifferentcoolingtimesoftheunderlyingelectronpopulation, origin. In the environment of G338.3−0.0 there could be at thePWNisexpectedtohavealargerextentinradiothaninX-rays least two possibilities for the occurrence of sufficiently dense (e.g.Gaensler&Slane2006).Asnoradioemissionhasbeende- ISM: a) As discussed in Section 4.1 and following Chevalier tectedattheX-rayPWNlocation,itishardtoestimatethesizeand (1999),wind-blownbubblesaresurroundedbyathindenseshell hence total flux from a potential radio PWN. The 610MHz map containing the bulk of the material swept-up by the stellar wind. showsadeficitofemissionattheX-rayPWNlocationandsome IftheexpandingshockofG338.3−0.0isnowclosetothisregion, enhancementinsidetherestoftheSNR.Thiscouldbeassociated accelerated protons might interact with this dense material and withprojectedSNRemission,orwitharelicradioPWN.Foryoung subsequentlyproducetheobservedγrays.b)Asecondpossibility PWNethepeakoftheradioemissionisexpectedtobeclosetothe isthattheSNRshockexpandsintoahighlyinhomogeneousISM pulsarposition.Sincetheradiosurfacebrightnessaroundtheputa- towardsthenearbyHIIregionfeaturingdenseclumpsofmolecular tivepulsarismuchlowercomparedtotherestoftheSNRinterior, gassurroundedbyregionsofcomparativelylowdensity.Here,the thiswouldimplythattheradioexcessisrelatedtoprojectedshell particles could be efficiently accelerated within the inter-clump emission.Foroldersystems,however,theradioPWNcanverywell medium while energetic protons can penetrate into the dense fillthefullinterioroftheSNRshell.Asacompromise,thelimitas clumps and produce the observed γ-ray emission. This scenario givenbyCastellettietal.(2011)isscaledupbyafactorof16to has already been proposed for the young (∼2kyr) VHE γ-ray covertheinterioroftheSNRshell.Inthiscasetheradiolimitisa emitting SNR RX J1713.7−3946 (see Zirakashvili & Aharonian factorof∼fivebelowthemodelcurvesinLemiereetal.(2009)and 2010) where dense molecular cloud cores have been detected in Slaneetal.(2010),andwouldimplyalow-energycut-offoftheun- theshockregion(e.g.Sanoetal.2010).SuchISMconditionsare derlyingelectronspectrumsignificantlyhigherthanthe50GeVas probably also present in the vicinity of G338.3−0.0, due to its usedbyLemiereetal.(2009). vicinitytoamassiveanddenseHIIregion,makingthisemission Insummary,theinterpretationoftheGeVandTeVemission scenarioalsoviableforHESSJ1640−465. as solely originating from a PWN is very difficult as neither the Incontrasttomiddle-agedinteractingSNRslikeIC443(Abdo γ-rayspectrum,northemorphologyortheradiodatasupportsuch etal.2010c)andW44(Abdoetal.2010b)wheretheγ-rayspectra apicture.ApossiblesolutionwouldbethattheGeVemissionhas arestronglypeakedatGeVenergies,RXJ1713.7−3946andother a different origin than the TeV emission. This, however, requires youngSNRsemitalargefractionoftheirhigh-energyemissionin fine-tuningtoexplainthesmoothFermiandH.E.S.S.spectrumand theTeVregime,eitherduetoadifferentradiationprocessortheir thepositionalcoincidenceoftheGeVandTeVsources.Alsothe earlier stage in evolution. Figure 3 shows a comparison between TeVspectrumalonedoesnotshowanysignificantdeviationfroma theGeV–TeVspectraofHESSJ1640−465andRXJ1713.7−3946 purepowerlawbelowthecut-offenergy,whichwouldbeexpected asseenbyFermiandH.E.S.S.Interestingly,theirspectralshapes forayoungPWN.Infact,theradioupperlimitinCastellettietal. in the TeV regime are very similar, which could support an age (2011),theX-raydataandanon-dominantICcomponentintheγ- younger than (10−20)kyr for G338.3−0.0. However, the GeV rayregimewouldbeconsistentwithXMMUJ164045.4−463131 spectrum becomes much harder for RX J1713.7−3946 but keeps being a young PWN (c.f. Fig. 5 in Funk et al. 2007). In general, thesameslopeforHESSJ1640−465.Leptonicmodelsgivingrise the featureless γ-ray spectrum over almost six decades in energy totheobservedshapeoftheγ-rayspectrumofRXJ1713.7−3946 ischallengingforanyleptonicmodelasspectralbreaksandsharp have been discussed in the literature quite extensively (see e.g. cut-offsareexpectedintheresultingSEDduetocoolingandKlein- Abdo et al. 2011; Yuan et al. 2011). However, following Zi- Nishinaeffects,respectively(e.g.Hinton&Hofmann2009). rakashvili&Aharonian(2010),thechangeinslopetowardslower TheTeVemissionalsosignificantlyoverlapswiththenorth- energies for RX J1713.7−3946 could also be explained in a westernpartoftheshellofG338.3−0.0anditishencequitenat- hadronicscenariobythesmallerpenetrationdepthsintothedense uraltoexploreanoriginofthenon-thermalemissionintheSNR molecularcloudcoresforprotonswithlowerenergies(seealsoIn- shell. Especially the spectral characteristics of HESSJ1640−465 oueetal.2012).Theseparticlesthereforecannotinteractwiththe are similar to that of prominent Galactic SNRs interacting with same amount of material as protons with higher energies, giving molecularcloudssuchasW28,W51CorIC443(seeOhm2012, risetoanunder-luminousandharderGeVγ-rayspectrum.Thefact andreferencestherein).Inthefollowingthefocuswillbeonanori- that this feature is not seen for HESS J1640−465 might indicate ginofthenon-thermalemissionintheSNRshell,bearinginmind an older remnant than e.g. RX J1713.7−3946 (i.e. (cid:38) 2.5kyr) or thatsomefractionofthetotalTeVemissioncouldplausiblyorigi- differentdiffusionpropertiesofthelocalISMthatallowalsolow- natefromthePWN. energyprotonstofullypenetratethedensemolecularclumps.An (cid:13)c 2014RAS,MNRAS000,2–9 AnexceptionallyluminousTeVγ-raySNR 7 -46°24'00" 325 -2-1) sm GMRT/ATCA XMM Fermi-LAT H.E.S.S. g c 10-10 Mercer 81 300 E (er d 28'00" 275 dN/ 10-11 2 E 0) 250 00 10-12 2 n (J 32'00" 225 o eclinati 200 10-13 D 175 36'00" 10-14 10-7 10-5 10-3 10-1 10 103 105 107 109 1011 1013 1015 Energy (eV) 150 125 Figure5.HEandVHEγ-rayspectrumofHESSJ1640−465asgivenin 40'00" Slaneetal.(2010)andshowninFigure2,respectively.TheX-raylimit 100 hasbeenderivedinthenorthernpartoftheradioshellandassumingthe 20s 41m00s 40s 20s 16h40m00s Right Ascension (J2000) highercolumndensityasderivedbyLemiereetal.(2009)(seeFigure1and text),andtheradiodataisfromCastellettietal.(2011),scaledbyafac- Figure4.SpitzerMIPS24µmimageinunitsofMJysr−1 withoverlaid torof0.5,assumingthathalfoftheradioemissioncomesfromthenorth- contoursfromthesmoothedH.E.S.S.excessmap(white)andcontoursof ernpartoftheshell.Thelong-dashedblueandreddash-dottedcurvesare thenorth-westernpartoftheSNRshellfromthe610MHzimage,convolved synchrotronandICemissionfromnon-thermalelectrons,respectively.The withtheH.E.S.S.PSF(magenta,c.f.Fig.1). greendashedcurveistheBremsstrahlungcomponentandthesolidblack curveishadronicπ0-decayγ-rayemission. ageof2.5kyrwouldimplysomemixingofthestellar-windmate- rialandtheISMleadingtoaveragedensitiesinthewindbubbleof beconstrainedbytheobservedsynchrotronspectrumfromradioto n ∼0.1cm−3(c.f.Section4.1). X-rays.Inthismodelcalculation,amagneticfieldofB =35µG, 0 WhencomparingtheTeVmorphologyofHESSJ1640−465 maximum electron energy of E = 10TeV and electron spec- c,e toG338.3−0.0(Fig.1)itbecomesclearthatγ-rayemissiononly tralindexofΓ =2.0isrequiredtoreproducetheradiospectrum e showssignificantoverlapwiththenorth-western(NW)partofthe andtonotviolatetheX-raylimit.Thetargetradiationfieldshave radioshell.Thus,inahadronicscenariothelackofemissionfrom been chosen based on Lemiere et al. (2009), with a dust compo- thesouth-eastern(SE)shellneedstobeexplained.Insuchamodel nent that has been increased to account for the five times higher the γ-ray emission is expected to follow the distribution and the radiation field energy density in the northern part of the shell. It densityofavailabletargetmaterialintheshockregion.Indeed,a is clear from Figure 5 that the predicted IC emission is at least correlationbetweenthemolecularandatomicgasandtheVHEγ- two orders of magnitude below the observed γ-ray emission for ray intensity from RX J1713.7−3946 has recently been reported an assumed electron-to-proton (e/p) ratio of 10−2. Furthermore, byFukuietal.(2012).Thus,ifdensetargetmaterialismuchmore the smooth connection of the HE and VHE γ-ray spectrum can- abundant in the northern region of G338.3−0.0 compared to the not be explained. A considerably higher e/p ratio of (cid:39) 0.1 (and south,theobservedTeVmorphologyofHESSJ1640−465iscon- lowermagneticfieldofB (cid:39) 10µG)isrequiredtoreachtheTeV sistentwithahadronicscenario.Figure4showstheSpitzerMIPS flux. Even in this case, the IC spectral shape and maximum en- (Riekeetal.2004)24µmimageofthisregion,whichessentially ergyisnotsupportedbytheVHEγ-rayspectrum.Indenseenvi- tracestheabundanceofinterstellardustanddenseHIIstar-forming ronments,Bremsstrahlungcansignificantlycontributetothenon- regions.Hereitcanbeseenthatthemeaninfraredintensitytowards thermalemission.Densitiesashighas500cm−3 ande/pratiosof theNWpartisafactorof∼5higherthantowardstheSEareaof 0.1are,however,requiredtoreachthefluxobservedbyH.E.S.S. theshell.Therefore,thedifferentdensitiescouldindeedgiverise Inahadronicscenario,atotalenergytransferredintoprotons totheobservedmorphology.Tofurthertestthehypothesisofthe ofW =2.5×1050erg,maximumprotonenergyE =50TeV p c,p NW shell being the origin of the VHE γ-ray emission, only this and spectral index of Γ = 2.2 as well as an average ambient p partoftheradioshellwasusedasatemplateandconvolvedwith density n¯ = 150cm−3, is required to reproduce the GeV – H the H.E.S.S. PSF. The resulting contours are over-plotted on the TeVspectrum.ThemeasuredTeVfluxcoupledwiththelargeesti- SpitzerimageinFig.4andshowagoodagreementwiththeVHE mated distance of ∼ 10kpc would imply that HESS J1640−465 γ-rayexcesscontoursfromH.E.S.S. is the most luminous Galactic VHE γ-ray SNR detected so far Figure5showsthemeasuredSEDofG338.3−0.0alongwith (L (cid:39) 4.6×1035(d/10kpc)2ergs−1). The TeV luminos- >1TeV the new H.E.S.S. data and XMM-Newton limits. Also shown is ityisthereforeaboutoneorderofmagnitudehigherthanthatofthe a single-zone time-dependent model for the continuous injection W51CSNR(Aleksic´etal.2012).Duetotheharderγ-rayspectral of electrons and protons over an assumed age of G338.3−0.0 of index,HESSJ1640−465hasatotalγ-rayluminositycomparable 2.5kyr (e.g. Funk et al. 2007). High-energy electrons produce to W51C. The product of total energy in interacting protons and synchrotron and IC γ-ray emission in interactions with magnetic meanambientdensityofW n¯ (cid:39)4×1052(d/10kpc)2ergcm−3 p H andradiationfields,respectively.High-energyprotonsproduceπ0- requiresaconsiderableamountofSNkineticenergythatistrans- decayγ-rayemissionininteractionswithmaterialintheSNRshell. ferredtohigh-energyprotonsand/orahighaveragedensityofthe ThebroadbandSEDcanbeexplainedinthisscenariowitharea- target material as motivated before. With the gas densities esti- sonable choice of input parameters. The leptonic component can mated above, a very large energy in protons is needed to reach (cid:13)c 2014RAS,MNRAS000,2–9 8 H.E.S.S.Collaboration the measured GeV and TeV flux. This implies that either the SN detailedandconstructivecomments,whichsignificantlyimproved explosion was as energetic as E (cid:39) 4×1051(d/10kpc)2erg thequalityofthepaper.Thisworkisbasedinpartonobservations SN (assuming a canonical 10% of SN explosion energy is channeled madewiththeSpitzerSpaceTelescope,whichisoperatedbytheJet intocosmicrays)and/orthatthefractionofE transferredinto PropulsionLaboratory,CaliforniaInstituteofTechnologyundera SN relativistic protons is significantly larger than the canonical 10%, contractwithNASA. i.e.upto∼40(d/10kpc)2%foratypicalE =1051erg.Note SN thatthisestimatecanbeevenhigher,asonlythenorthernhalfof theSNRshellseemstobeilluminatedbycosmicrays. REFERENCES Abdo A. A. et al. (Fermi-LAT Collaboration), 2010a, ApJ, 714, 927 5 CONCLUSIONSANDOUTLOOK AbdoA.A.etal.(Fermi-LATCollaboration),2011,ApJ,734,28 ThedetailedH.E.S.S.resultspresentedinthisworkshowthatthe Abdo A. A. et al. (Fermi-LAT Collaboration), 2010b, Science, VHE γ-ray emission from HESS J1640−465 significantly over- 327,1103 lapswiththenorth-westernpartoftheSNRshellofG338.3−0.0. Abdo A. A. et al. (Fermi-LAT Collaboration), 2010c, ApJ, 712, Moreover, the VHE γ-ray spectrum smoothly connects with the 459 Fermi spectrum and has a high-energy cut-off that implies that AbramowskiA.etal.(H.E.S.S.Collaboration),2012a,A&A,548, particles with tens of TeV energies are present in the accelera- A46 tion region. The TeV morphology, new radio measurements and Abramowski A. et al. (H.E.S.S. Collaboration), 2012b, A&A, theoverallγ-rayspectrumarehardtoexplaininascenariowhere 548,A38 most of the non-thermal emission is coming from the PWN. The AckermannM.etal.(Fermi-LATCollaboration),2013,ArXive- broadbandSEDandmorphologyofthenon-thermalemissionfrom prints:1306.6772 HESSJ1640−465canbebetterexplainedinascenariowherepro- Aharonian F. et al. (H.E.S.S. Collaboration), 2005, A&A, 435, tonsareacceleratedintheshellofG338.3−0.0andinteractwith L17 dense gas associated with the G338.4+0.1 HII complex. In this Aharonian F. et al. (H.E.S.S. Collaboration), 2006a, A&A, 457, case, the product of total energy in interacting protons and mean 899 ambientdensityWpn¯H ∼4×1052(d/10kpc)2ergcm−3required AharonianF.etal.(H.E.S.S.Collaboration),2006b,ApJ,636,777 toexplainthefluxmeasuredbyFermiandH.E.S.S.iscomparable AharonianF.etal.(H.E.S.S.Collaboration),2011,A&A,531,C1 totheγ-ray-emittingSNRW51C,althoughtheTeVluminosityof Aharonian F. et al. (H.E.S.S. Collaboration), 2006c, A&A, 460, HESSJ1640−465isanorderofmagnitudehigher.Inthispicture, 365 thenon-detectionofthermalX-raysisconsistentwiththelargedis- AharonianF.A.,DruryL.O.,VoelkH.J.,1994,A&A,285,645 tancetoG338.3−0.0andthehighcolumndensityalongthelineof Aleksic´J.etal.(MAGICCollaboration),2012,A&A,541,A13 sight.Highresolutionandhighsensitivitymolecularlineobserva- ArnaudK.A.,1996,inASPConf.Ser.101:AstronomicalData tionsinthisregionarerequiredtolocatethedensegasthatmight AnalysisSoftwareandSystemsV,p.17 actastargetmaterialandtoputlimitsontheexplosionenergyof BergeD.,FunkS.,HintonJ.,2007,A&A,466,1219 G338.3−0.0.ThefutureCherenkovTelescopeArraywithitsmuch BlondinJ.M.,ChevalierR.A.,FriersonD.M.,2001,ApJ,563, betterangularresolutionandsensitivityisneededtofurtherresolve 806 the VHE γ-ray emission region(s) of HESS J1640−465 and to CarriganS.etal.(H.E.S.S.Collaboration),2013,ArXive-prints: distinguish the contribution from the SNR shell and the PWN in 1307.4690 G338.3−0.0. CastellettiG.,GiacaniE.,DubnerG.,JoshiB.C.,RaoA.P.,Ter- rierR.,2011,A&A,536,A98 ChevalierR.A.,1999,ApJ,511,798 Davies B., de La Fuente D., Najarro F., Hinton J. A., Trombley ACKNOWLEDGEMENTS C.,FigerD.F.,PugaE.,2012,MNRAS,419,1860 The support of the Namibian authorities and of the University of deNauroisM.,RollandL.,2009,AstroparticlePhysics,32,231 NamibiainfacilitatingtheconstructionandoperationofH.E.S.S.is FreemanP.,DoeS.,SiemiginowskaA.,2001,inSocietyofPhoto- gratefullyacknowledged,asisthesupportbytheGermanMinistry Optical Instrumentation Engineers (SPIE) Conference Series, forEducationandResearch(BMBF),theMaxPlanckSociety,the Vol.4477,SocietyofPhoto-OpticalInstrumentationEngineers FrenchMinistryforResearch,theCNRS-IN2P3andtheAstropar- (SPIE)ConferenceSeries,StarckJ.-L.,MurtaghF.D.,eds.,pp. ticleInterdisciplinaryProgrammeoftheCNRS,theU.K.Science 76–87 andTechnologyFacilitiesCouncil(STFC),theIPNPoftheCharles FukuiY.etal.,2003,PASJ,55,L61 University, theCzech ScienceFoundation, the Polish Ministry of FukuiY.etal.,2012,ApJ,746,82 Science and Higher Education, the South African Department of FunkS.,HintonJ.A.,Pu¨hlhoferG.,AharonianF.A.,Hofmann Science and Technology and National Research Foundation, and W.,ReimerO.,WagnerS.,2007,ApJ,662,517 bytheUniversityofNamibia.Weappreciatetheexcellentworkof GaenslerB.M.,SlaneP.O.,2006,ARA&A,44,17 the technical support staff in Berlin, Durham, Hamburg, Heidel- GrondinM.-H.etal.,2011,ApJ,738,42 berg,Palaiseau,Paris,Saclay,andinNamibiaintheconstruction HintonJ.A.,HofmannW.,2009,ARA&A,47,523 andoperationoftheequipment.S.O.acknowledgesthesupportof InoueT.,YamazakiR.,InutsukaS.-i.,FukuiY.,2012,ApJ,744, theHumboldtfoundationbyaFeodor-Lynenresearchfellowship. 71 We are also grateful to Gabriela Castelletti, who kindly provided KudritzkiR.-P.,PulsJ.,2000,ARA&A,38,613 the610MHzmapandPatrickSlaneforthePWNmodelcurve.The LemiereA.,SlaneP.,GaenslerB.M.,MurrayS.,2009,ApJ,706, authorswouldalsoliketothanktheanonymousrefereeforher/his 1269 (cid:13)c 2014RAS,MNRAS000,2–9 AnexceptionallyluminousTeVγ-raySNR 9 LevensonN.A.,GrahamJ.R.,KellerL.D.,RichterM.J.,1998, ApJS,118,541 Muijres L. E., Vink J. S., de Koter A., Mu¨ller P. E., Langer N., 2012,A&A,537,A37 Nolan P. L. et al. (Fermi-LAT Collaboration), 2012, ApJS, 199, 31 Ohm S., 2012, in American Institute of Physics Conference Se- ries,Vol.1505,AmericanInstituteofPhysicsConferenceSeries, AharonianF.A.,HofmannW.,RiegerF.M.,eds.,pp.64–71 OhmS.,vanEldikC.,EgbertsK.,2009,AstroparticlePhysics,31, 383 OsterbrockD.E.,1989,Astrophysicsofgaseousnebulaeandac- tivegalacticnuclei Oya I. et al. (H.E.S.S. Collaboration), 2013, ArXiv e-prints: 1303.0979 PironF.etal.,2001,A&A,374,895 RiekeG.H.etal.,2004,ApJS,154,25 PurcellC.R.etal.,2012,MNRAS,426,1972 SanoH.etal.,2010,ApJ,724,59 SlaneP.,CastroD.,FunkS.,UchiyamaY.,LemiereA.,Gelfand J.D.,Lemoine-GoumardM.,2010,ApJ,720,266 SpitkovskyA.,2008,ApJ,682,L5 Tenorio-TagleG.,RozyczkaM.,FrancoJ.,BodenheimerP.,1991, MNRAS,251,318 VinkJ.,2004,ApJ,604,693 VinkJ.,KaastraJ.S.,BleekerJ.A.M.,1997,A&A,328,628 WalshA.J.etal.,2011,MNRAS,416,1764 Weaver R., McCray R., Castor J., Shapiro P., Moore R., 1977, ApJ,218,377 WhiteoakJ.B.Z.,GreenA.J.,1996,A&AS,118,329 YuanQ.,LiuS.,FanZ.,BiX.,FryerC.L.,2011,ApJ,735,120 ZirakashviliV.N.,AharonianF.A.,2010,ApJ,708,965 (cid:13)c 2014RAS,MNRAS000,2–9