Table Of Content

VHE Gamma-ray supernova remnants S. Funk1 Kavli Institute for Particle Astrophysics and Cosmology, SLAC, 2575 Sand Hill Road, PO Box 0029, Stanford, CA-94025, USA 7 0 0 2 Abstract n a J Increasing observational evidence gathered especially in X-rays and γ-rays during 6 thecourseofthelastfewyears supportthenotion thatSupernovaremnants(SNRs) 1 are Galactic particle accelerators up to energies close to the “knee” in the energy spectrum of Cosmic rays. This review summarises the current status of γ-ray ob- 1 v servations of SNRs. Shell-type as well as plerionic type SNRs are addressed and 1 prospect for observations of these two source classes with the upcoming GLAST 7 satellite in the energy regime above 100 MeV are given. 4 1 0 7 Key words: , Supernova remnant, Pulsar Wind Nebula, Gamma-ray astronomy 0 PACS: 98.38.Mz, Supernova remnants / h PACS: 98.70.Rz, γ-ray sources p PACS: 98.70.Sa, Cosmic rays (origin, acceleration, and interactions) - o r t s a : v i X 1 Introduction r a Supernova remnants (i.e. remnants of Supernova explosions) are commonly considered to be Cosmic particle accelerators. In this review paper I will sum- marise experimental evidence, gathered through γ-ray observations mainly in the VHE regime supporting this notion (please note, that in the following γ-ray will be used to stand for VHE γ-rays). Typically, SNRs were detected through radio observations (Green, 2004). Recent advances in the understanding of these objects has been made through X-ray observations with instruments such as ASCA, BeppoSax, XMM-Newton and Chandra and through γ-ray observations with instruments such as HEGRA (High Energy Gamma Email address: [email protected](S. Funk). 1 previously at: Max-Planck-Institut fu¨r Kernphysik, P.O. Box 103980, D 69029, Heidelberg, Germany Preprint submitted to Elsevier 3 February 2008 Ray Astronomy), and H.E.S.S. (High energy stereoscopic system). Based on morphological properties, SNRs can be classified into three broad categories: shell-like, plerionic (also called Pulsar Wind Nebulae or Crab-like) connected toaPulsarandcomposite (inwhichboth,ashellandaPlerionarepresent),the later often showing markedly different radio and X-ray morphologies. For an excellent detailed review on plerionic SNRs, see e.g. Gaensler & Slane (2006). SNRs are thought to be responsible for the acceleration of Cosmic rays up to ∼ 15 energies around the “knee” ( 10 eV) at which the spectrum of Cosmic rays significantly hardens from ∼ 2.7 to ∼ 3.2. This statement is backed by experimental facts, as well as by theoretical considerations. Experimental evidence is lent mainly by (1) X-ray observations of young shell-type SNRs such as SN1006 (Koyama et al., 1995), and Cas A (Vink et al., 2001), in which sites dominated by hard non-thermal X-ray synchrotron emission were found, indi- ∼ cating an electron population extending up to 100 TeV, far beyond thermal energies.(2)Veryhigh energy(VHE) γ-ray observations(> 100GeV)revealed sites of non-thermal particle populations (Aharonian et al., 2004, 2006a,b). Through theoretical considerations is has been known for a long time that Supernova explosions release just about the right amount of energy into their surrounding to account for the energy budget of the Cosmic rays (assuming ∼ that they convert 10% of their energy into kinetic energy of the Cosmic rays) (Ginzburg & Syrovatskii, 1964). Furthermore well-established theoretical models exists explaining how particles can be accelerated in Supernova shock waves to energies approaching the knee.In shell-type SNRs particles are accelerated in the expanding shock waves through diffusive shock (also called first order Fermi) acceleration (Bell, 1978; Blandford & Ostriker, 1978; Drury, 1983;Blandford & Eichler,1987;Jones & Ellison,1991;Malkov & Drury,2001). In plerionic SNRs particles are accelerated to non-thermal energies in the ter- mination shock between the relativistic outflow of electrons from the pulsar surface and the outer nebula. Predictions on the γ-ray visibility of SNRs (later confirmed by H.E.S.S. γ-ray observations although there is still an experimental ambiguity in the underlying particle population resonsible for the γ-ray emission) were given by Drury et al. (1994). 15 Since charged particles below the knee at 10 eV are deflected in ubiquitous magnetic fields on their way from the origin to us, we have to turn to neutral messengers to reveal the acceleration sites (the gyroradius of 1 TeV cosmic rays in a magnetic field of µG-scale is of the order of 0.1 pc, much smaller than the thickness of the Galaxy of 200–300pc). Since neutrino detectors have not yet proved to be sensitive enough to detect neutrinos from astrophysical sources (apart from the Sun and the direct Supernova explosion SN 1987A), observations in the radio, X-rays and γ-ray wavebands are so far our best access to non-thermal acceleration processes in SNRs. γ-rays (and neutrinos) are produced in hadronic interactions with subsequent pionic decay and can reveal the acceleration sites since they travel un-deflected from their origin. 2 However, γ-rays not only reveal the sites of hadronic acceleration; they also act as a tracer for energetic electrons that produce γ-rays via IC scattering off background photon fields (such as star-light or the Cosmic microwave background (CMBR)). An ambiguity or duality therefore exists in the responsible radiating particle population in most cases when detecting γ-rays from astrophysical objects. In spite of this ambiguity the detection of γ-rays above ∼ 1 GeV from SNRs gives us direct access to particle acceleration processes and the advantage of γ-rays in comparison to other wavebands is that these are not affected by dust obscuration, which is particularly important for the population of SNRs located within the Galactic plane. A large volumne of the Galaxy can thus be probed for γ-ray emission from SNRs by observations through the Galactic disk. If SNRs are indeed sites of particle acceleration, γ-ray emission is expected and one of the puzzling aspects of previous γ-ray observations of SNRs was the rather low level of emission compared to model predictions (Buckley et al., 1998; Hillas, 2005). The history of soft γ-ray (or hard X-ray) detection of SNRs started with the detection of the Crab Nebula in 1964 with a scintillation counter detector flown on a balloon launched from Palestine, Texas (Clark, 1965). To- day hard X-rays up to 100 keV have been detected from various SNRs both young shell-types such as Cas A (Vink et al., 2001; Vink & Laming, 2003) and SN 1006 (Allen et al., 2001; Vink et al., 2000) and plerionic-types such as the Vela-X PWN (Mangano et al., 2005) and MSH 15-52 (Tamura et al., 1996; Marsden et al., 1997; Mineo et al., 2001; Forot et al., 2006). Thin X- ray filaments in young shell-type SNRs detected with high-angular resolution instruments such as XMM-Newton and Chandra point to regions with high- magnetic fields (up to 0.5 mG) in which electrons rapidly lose energy through synchrotron emission (Koyama et al., 1997; Slane et al., 2001; Bamba et al., 2003; Cassam-Chena¨ı et al., 2004). In higher energies γ-rays COMPTEL de- 44 tected the radioactive Ti-line at 1.157 MeV from the two shell-type SNRs Cas A (Iyudin et al., 1994) and RXJ0852.0–4622 (Vela Junior) (Iyudin et al., 1998). However, it should be noted, that higher sensitivity INTEGRAL observations provided a confirmation for this detection for Cas A, but could 44 not detect this Ti-line in RXJ0852.0–4622. Therefore, these claims are still somewhat controversial. EGRET at energies above 100 MeV did not detect prominent young shell-type SNRs such as Tycho, Kepler, Cas A or SN 1006, noticed however several in- triguing spatial coincidences of unidentified sources in the Galactic plane with individualprominentradioSNRs,suchasW28,andγ-Cygni(Sturner & Dermer , 1995; Esposito et al., 1996; Romero, Benaglia & Torres, 1999). The combina- tion of source confusion especially in the Galactic plane, caused by the rather poor angular resolution of the EGRET instrument and the ambiguity in ex- 3 isting counterparts prevented individual identifications. However, a statisti- cal assessment shows a 4-5σ effect when trying to correlate the population of EGRETunidentifiedsourceswiththepopulationofradioSNRs(Sturner & Dermer , 1995).AlsoplerionicSNRshavenotbeenunambiguouslyidentifiedwithEGRET sources, althoughagainintriguing associations of EGRET unidentified sources with prominent plerions such as PSRB1706–44, and the Kookaburra complex exist (Roberts et al., 2005). From a population point-of-view PWN are one of the best candidates to account for low-latitude slowly varying unidentified EGRET sources as proposed by Roberts et al. (2001). Lately, using VHE γ- ray source positions in the Kookaburra region (Aharonian et al., 2006c) the re-analysis of EGRET data provided strong evidence of correlation of the PWN detected in this region with the confused unidentified EGRET source 3EGJ1420–6038 (Reimer & Funk, 2006). This new approach might prove a useful template for connection future GLAST and VHE γ-ray detections. All these possible associations of source classes with unidentified EGRET sources will hopefully be tested following the launch of the upcoming GLAST satellite in late 2007. ThehistoryofVHEγ-ray(E> 100GeV)detectionsofSNRsstartedagainwith the detection of the Crab Nebula, the first object to be reported in this wave- bandbytheWhipplecollaboration(Weekes et al.,1989).Variousclaims ofde- tections of shell-type SNRs have been made before the advent of the H.E.S.S. ∼ telescope system. CasAwasdetectedbyHEGRAinavery deep( 200hours) exposure(Aharonian et al.,2001).DetectionsofSN1006(Tanimori et al.,1997) and RXJ1713.7–3946(Enomoto et al., 2002) have been reported by the CAN- GAROO collaboration. With the advent of the H.E.S.S. telescope system, for the first time a number of Galactic SNRs, both shell-type and plerionic in na- ture could be established. In the following I will describe these populations of cosmic accelerators along with prospects for Supernova remnant observations with the upcoming GLAST satellite. The outline of this paper is as follows: Section 2 provides a description of advances made through the VHE γ-ray detections of shell-type Supernova remnants, Section 3 provides the corre- sponding description for Pulsar Wind Nebulae. Section 4 describes Supernova remnants found in γ-rays and later identified as Supernova remnants, while Section 5 summarises prospects for Supernova remnant observations with the upcoming GLAST satellite. 2 Gamma-ray observations of shell-type SNRs Bymeans ofdata takenwith theH.E.S.S. telescope system during thefirst few years of operation for the first time in VHE γ-ray astronomy resolved images of shell-type SNRs above 100 GeV could be taken. In particular the SNRs ∼ ◦ ∼ ◦ RXJ1713.7–3946 and RXJ0852.0–4622, with diameters of 1 and 2 re- 4 spectively could be resolved with unprecedented detail in this energy band. On the other hand, SN1006, one of the SNRs most expected to emit γ-rays in the energy band (due to the strong non-thermal X-ray emission from the rims) was not detected in deep H.E.S.S. observations (Aharonian et al., 2005a). The upper limit derived by this observations turned out to be an order of magnitude below the previously reported CANGAROO detection. A reanalysis of the CANGAROO data along with newer data from the CANGAROO-III detector is consistent with the H.E.S.S. upper limits (Tanimori et al., 2005). Therefore in the following the H.E.S.S. upper limits will be used in the discus- sionof γ-rayemission fromSN1006.Since several papersonbothRXJ1713.7– 3946(Aharonian et al.,2004,2006a,b)andRXJ0852.0–4622(Aharonian et al., 2005b, 2006d) have been published by the H.E.S.S. collaboration, the main focus of this review will lie on similarities and differences between the two objects with the addition of comparisons to SN1006 as the most prominent non-detected SNR where appropriate. Figure 1 shows γ-ray excess maps for RXJ1713.7–3946, RXJ0852.0–4622, and SN1006. Both γ-ray emitting objects show a shell-like structure with a surprising resemblance of their respective X-ray morphology (the correlation coefficients between γ-ray and X-ray counts are ∼ 60% − 80%). For both objects the X-ray emission is completely dominated by non-thermal X-ray emission without traces of line emission, exhibiting small filamentary ∼ structures that are interpreted as zones where the magnetic field is high ( 50µG) such that electrons rapidly lose energy through synchrotron emission in these areas (Cassam-Chena¨ı et al., 2004; Uchiyama et al., 2003; Aschenbach, 1998; Iyudin et al., 2005). Both objects appear rather faint in radio with typical fluxes below or in the several tenth of Jansky-regime for the whole shell, certainly lower than what would be expected from equipartition ar- guments (Lazendic et al., 2004). The distance to both objects is somewhat ∼ uncertain, for RXJ1713.7–3946 it seems that a distance of 1kpc is preferred from the column density inferred from X-ray data. This distance would make RXJ1713.7–3946 most likely the remnant of the historical Supernova event of AD393. For RXJ0852.0–4622 distance estimates range from as close as the ∼ ∼ Vela pulsar ( 250pc) to as far as the Vela Molecular Ridge ( 1kpc). The age ∼ ∼ ranges from 500 years in the close case to 5000 years in the far case. Mor- phologically their γ-ray emission, in particular the width of the shells is rather different. The apparent width of the shell for RXJ1713.7–3946 comprises 45% of the radius of the SNR, while the for RXJ0852.0–4622 it approximates to 20% of the radius. There is no apparent correlation between the dense molecular material surrounding RXJ1713.7–3946 as measured by the NANTEN telescope and the VHE γ-ray emission as measured by H.E.S.S., but in fact, × 50 assuming a typical energy of 1 10 ergs in accelerated protons, the density needed to explain the γ-ray flux through hadronic interactions is only 1 cm−3. SN1006 is somewhat distinct in its multi-frequency picture in that its surface 5 Fig. 1. Acceptance-corrected smoothed excess maps of the 3.5◦ × 3.5◦ fov surrounding the two prominent H.E.S.S. Supernova remnants RXJ1713.7–3946 (2004 and 2005 data) (Aharonian et al., 2006b) and RXJ0852.0–4622 (2005 dataset) (Lemoine-Gourmard et al., 2006; Aharonian et al., 2006d) and non-detected SN1006 (2004 dataset with VLA radio contours in white) (Aharonian et al., ◦ 2005a). The sky-regions shown are of similar size, indicating the large extent (2 diameter) of RXJ0852.0–4622. ∼ brightness is higher in radio ( 100 Jy) (Gardner & Milne, 1965; Roger et al., 1998) showing a pronounced shell-like structure. The X-ray emission, espe- ∼ cially in the shell is dominated by non-thermal emission up to 10keV. SN1006 was not detected in a deep (1000 ksec) INTEGRAL exposure above 20 keV (Kalemci et al., 2006) and was also not detected in sensitive H.E.S.S. observations (Aharonian et al., 2005a). The density surrounding the source was estimated from X-ray as well as optical observations and values as low as n= 0.05 cm−3 have been invoked to explain the apparent absence of γ-ray emission in a hadronic scenario. From the H.E.S.S. non-detection assuming a leptonic γ-ray emission scenario on the CMBR a lower limit on the post-shock magnetic field of B > 25µG can be derived (Aharonian et al., 2005a). Higher values of the magnetic field in excess of 40µG have been derived from X-ray observation and application of diffusive shock acceleration scenarios, so the 6 1) 10-8 -s RXJ1713.7-3946 -2 m RXJ0852.0-4622 c SN1006 g 10-9 r e ( E d N/ 10-10 d 2 E 10-11 10-12 10-13 10-6 10-4 10-2 1 102 104 106 108 1010 1012 1014 Energy (eV) Fig. 2. Spectral energy distribution for the Supernova remnants RXJ1713.7–3946 (lightred),RXJ0852.0–4622 (pink),SN1006(blue).Alsoshownisahadronicmodel (dotted green) and a time-dependent leptonic one-zone model (dashed grey: IC emission, solid grey: synchrotron emission). The parameters for this model are: B-Field: 9µG, age: 1.7 kYears, Electron photon index: 2.1, Electron Cutoff: 80 TeV. lower limit on the magnetic field is not in contradiction to these values. Comparing the energy spectra of the two γ-ray detected SNRs, strong similarities can be made out. The spectral energy distribution (SED) for the three shell-type SNRs discussed here is shown in Figure 2 along with model spectra, showing typical leptonic and hadronic γ-ray emission models. As can be seen from this plot RXJ1713.7–3946 and RXJ0852.0–4622 show a remark- ably similar γ-ray energy spectrum with a rather flat E−2-type distribution at lower energies with a deviation from this power-law at higher energies. The flat spectrum at lower energies has advocated claims that the γ-rays might be generated by pionic decays rather than Inverse Compton scattering (Aharonian et al., 2006a,b). However, Porter et al. (2006) claim that the data can be well fitted in terms of a leptonic model when applying an un- broken electron spectrum along with the Galactic radiation fields. Therefore, at the moment, no strong conclusions can be drawn from the spectral shape on the particle population responsible for the γ-ray emission. The upcoming GLASTsatellite,measuring intheenergyrangebetween 30MeVand300GeV might be able to distinguish between hadronic and leptonic γ-ray production mechanisms. Also interesting to note is that the H.E.S.S. γ-ray upper limit for SN1006 is more than an order of magnitude below these γ-ray detections and therefore starts to be rather constraining for the values of the magnetic field (in a leptonic scenario) or the ambient matter density (in a hadronic scenario). 7 These first unambiguous detections of individual shell-type SNRs allowed for important advances in the understanding γ-ray emission from these objects. However, the open question remains what differentiates non-detected SNRs such as SN1006 from prominent γ-ray emitters such as RXJ1713.7–3946. 3 Gamma-ray observations of Pulsar Wind Nebulae Pulsar wind nebulae (PWN) or Plerions are objects powered by a relativistic particle outflow (electrons and positrons) from a central source – a pulsar. ThispulsarisarapidlyrotatingneutronstargeneratedintheSupernovaevent. The wind of relativistic particles flows freely out until the outflow pressure is balanced by that of the surrounding medium. At that point a standing termi- nation shock is formed at which particles are accelerated (Kennel & Coroniti, 1984; Aharonian, Atoyan & Kifune, 1997). The existence of electrons accelerated to energies > 100 TeV in such PWN has been established by X-ray observations of synchrotron emission, e.g. in the Crab nebula (Weisskopf et al., 2000). VHE γ-rays are generated in PWN from the high-energy electrons by non-thermal bremsstrahlung or inverse Compton (IC) scattering on photon target fields, such as the cosmic microwave background (CMBR) or star-light. Apart from the Crab Nebula, no individual PWNe have been unambigu- ously associated with EGRET sources, although several unidentified EGRET sources are located in close proximity to prominent PWN, such as in the Kookaburra region, or MSH–15–52. PWN are however one candidate for the population of slowly varying low-latitude unidentified sources. GLAST will shed more light on this population and possibly establish PWN as emitters in the MeV to GeV range. In VHE γ-rays PWN make up the majority of the identified Galactic sources detected so far (Funk, 2006; Gallant, 2006). Apart from the Crab Nebula (the brightest steady VHE γ-ray source) several prominent PWN were identified in VHE γ-rays in the last two years. These detec- tionsinclude MSH–15–52 (Aharonian et al.,2005c),Vela X(Aharonian et al., 2006e),thetwosourcesintheKookaburraregion(Aharonian et al.,2006f)and lately HESSJ1825–137 (Aharonian et al., 2005d, 2006g). VHE γ-ray emission from PWN comes in various disguises as shown in Figure 3: These include a) point-like emission such as from the Crab Nebula (Aharonian et al., 2006h) and from the composite SNR G0.9+0.1, where the γ-ray emission was shown to originatefromthecentral PWN(Aharonian et al.,2005e),b) emission tracing the X-ray contours around a central pulsar such as in MSH–15-52 or c) asymmetrically extending to one side and tracing the X-ray contours such as in Vela X and finally d) the emerging new class of offset PWN exemplified by HESSJ1825–137 where the γ-ray emission shows a similar morphology to the X-ray emission but on a much larger scale (Aharonian et al., 2006g). Calculat- ing the efficiency of the energetic pulsars powering the PWN that is necessary 8 Fig. 3. Acceptance-corrected smoothed excess maps of MSH–15–52 (top left) (Aharonian et al., 2005c), the Kookaburra region showing the γ-ray emission coincident with the two non-thermal wings of the Kookaburra (top right) (Aharonian et al., 2006f), HESSJ1825–137 (bottom left) (Aharonian et al., 2005d, 2006g) and Vela X (bottom right) (Aharonian et al., 2006e). Also shown are the energetic pulsars that are thought to power the PWNe. to account for the VHE γ-ray luminosity, values between 0.02% (Crab Neb- ula) and 7.5% (HESSJ1825–137) of the spin-down luminosity are found. The broadband SEDs of the VHE γ-ray PWN can typically be well described by leptonic models, although claims have been made for a hadronic component at the high-energy end of the spectrum (Horns et al., 2006). Vela X is the first VHE γ-ray source in which the peak in the Inverse Compton energy flux has been detected within the H.E.S.S. energy range (Aharonian et al., 2006e). Figure 4 shows, that while the ranges of γ-ray fluxes for the detected PWN is rather small, the differences in the X-ray energy fluxes span a large range. This might be alluded to largely different magnetic fields, to different angular scales on which the X-ray emission has been measured or simply to the fact, that different populations of electrons are responsible for the X-ray and the γ-ray emission. 9 -1)s 10-8 K3/Kookaburra -2 m RMaSbHb i1t5/K-5o2okaburra c 10-9 VELA-X g HESS J1825-137 er SNR + cocoon E ( 10-10 d N/ 2 dE 10-11 cocoon 10-12 Nebula 10-13 Core 10-14 10-6 10-4 10-2 1 102 104 106 108 1010 1012 1014 Energy (eV) Fig. 4. Spectral energy distribution for the PWNe in the Kookaburra region (K3 and Rabbit) (red), MSH–15-52 (turquoise), Vela X (blue), and HESSJ1825–137 (grey). The similar energy flux for the γ-ray emission in comparison to the vastly different energy flux for the X-ray emission is apparent. The most prominent example of the new class of offset PWN is HESSJ1825– 137. This object can serve as a template for a whole new class of γ-ray PWN in which a) the γ-ray emission is shifted away from the pulsar, possibly due to dense material on one side that prevents an isotropic expansion of the PWN and b) the size of the VHE γ-ray PWN is on a much larger scale (∼ 1◦) than ∼ ′ the X-ray PWN ( 1) (Gaensler et al., 2003). Concerning the offset morphology, asymmetric reverse shock interactions were first proposed to explain the offset morphology of the Vela X PWN based on hydro-dynamical simulations by Blondin, Chevalier & Frierson (2001). The different sizes for the γ-ray and X-rayPWNecanbeexplained bythedifferenceinthesynchrotron coolinglife- times of the (higher energy) X-ray emitting and the (lower energy) IC-γ-ray emitting electrons. The γ-ray sources that can be explained in this framework are typically extended, their emission region overlaps with energetic pulsars (energetic enough to explain the γ-ray flux by their spindown power) and very importantly also show evidence for an X-ray PWN. So far only Vela X and HESSJ1825–137 match this picture, several other unidentified VHE γ- ray sources have been proposed to be offset PWN, but all these cases lack the detection of an X-ray PWN. 4 New Supernova remnants found in VHE γ-rays Originally Supernova remnants have been detected by means of radio observation sensitive to synchrotron emission in magnetic fields. Radio observations 10