Next Contents Previous

3.4. Supernova Remnants: Shell

Supernova remnants (SNRs) are widely believed to be the sources of hadronic cosmic rays up to energies of approximately Z × 1014 eV, where Z is the nuclear charge of the particle (for a review, see Jones et al. 1998). The arguments in support of this statement are twofold. First, supernova blast shocks are some of the few Galactic sites capable of satisfying the energy required for the production of Galactic cosmic rays, although even these must have a high efficiency, ~ 10%-30% (e.g., Drury, Markiewicz, & Völk 1989), for converting the kinetic energy of supernova explosions into high-energy particles. Second, the model of diffusive shock acceleration (e.g., Blandford & Ostriker 1978; Bell 1978; Legage & Cesarsky 1983), which provides a plausible mechanism for efficiently converting the explosion energy into accelerated particles, naturally produces a power-law spectrum of dN / dE propto E-2.1. This is consistent with the inferred spectral index at the source for the observed local cosmic-ray spectrum of dN / dE propto E-2.7, after correcting for the effects of propagation in the Galaxy (e.g., Swordy et al. 1990).

The origins of cosmic rays cannot be studied directly because interstellar magnetic fields isotropize their directions, except perhaps at the highest energies (gtapprox 1018 eV). Thus, we must look for indirect signals of their presence from astrophysical sources. That SNRs accelerate electrons to high energies is well established from observations of synchrotron emission in the shells of SNRs at radio and, more recently, X-ray (e.g., Koyama et al. 1995) wavelengths. However, it is difficult to extrapolate from these data to inferences about the nature of the acceleration of hadrons in these same objects.

Evidence of shock acceleration of hadronic cosmic rays in SNR shells could come from measurements of gamma-ray emission in these objects. Collisions of cosmic-ray nuclei with the interstellar medium result in the production of neutral pions which subsequently decay into gamma-rays. The gamma-ray spectrum would extend from below 10 MeV up to ~ 1/10 of the maximum proton energy (gtapprox 10 TeV), with a distinctive break in the spectrum near 100 MeV due to the resonance in the cross section for pi0 production. As gamma-ray production requires interaction of the hadronic cosmic rays with target nuclei, this emission should be stronger for those SNRs located near, or interacting with, dense targets, such as molecular clouds. The cosmic-ray density, and hence the associated gamma-ray luminosity, will increase with time as the SNR passes through its free expansion phase, will peak when the SNR has swept up as much interstellar material as contained in the supernova ejecta (the Sedov phase) and gradually decline thereafter (Drury, Aharonian, & Völk 1994; Naito & Takahara 1994). Thus, gamma-ray bright SNRs should be "middle-aged."

From the calculations of Drury et al. (1994), the luminosity of gamma-rays from secondary pion production may be detectable by the current generation of satellite-based and ground-based gamma-ray detectors, particularly if the objects are located in a region of relatively high density in the interstellar medium. In support of this hypothesis, EGRET has detected signals from several regions of the sky consistent with the positions of shell-type SNRs (Sturner & Dermer 1995; Esposito et al. 1996; Lamb & Macomb 1997; Jaffe et al. 1997). However, the EGRET detections alone are not sufficient to claim the presence of high-energy hadronic cosmic rays. For instance, the relatively poor angular resolution of EGRET makes it difficult to definitively identify the detected object with the SNR shell. Because of this, embedded pulsars (Brazier et al. 1996; de Jager & Mastichiadis 1997; Harrus, Hughes, & Helfand 1996) and an X-ray binary (Kaaret et al. 1999) consistent with the positions of some of these EGRET sources have been suggested as alternative counterparts. In addition, significant background from the diffuse Galactic gamma-ray emission complicates spectral measurements. To complicate matters further, with the detection of X-ray synchrotron radiation from SNR shells, the possibility that gamma-rays could be produced via inverse Compton scattering of ambient soft photons has been realized (Mastichiadis & de Jager 1996; Mastichiadis 1996). Bremsstrahlung radiation may also be a significant source of gamma-rays at MeV-GeV energies (de Jager & Mastichiadis 1997; Gaisser, Protheroe, & Stanev 1998).

3.4.1. VHE gamma-Ray Observations

Measurements of gamma-rays at very high energies may help resolve the puzzle of the gamma-ray emission from the EGRET-detected sources. VHE gamma-ray telescopes have much better angular resolution than EGRET, reducing the source confusion associated with any detection. Also, because the diffuse Galactic gamma-ray emission has a relatively steep spectrum, propto E-2.4-E-2.7 (Hunter et al. 1997), compared with the expected ~ E-2.1 spectrum of gamma-rays from secondary pion decay, contamination from background gamma-ray emission should be less in the VHE range. Thus, in recent years, searches for emission from shell-type SNRs have been a central part of the observation program of VHE telescopes.

The Whipple Collaboration has published the results of observations of six shell-type SNRs (IC 443, gamma Cygni, W44, W51, W63, and Tycho) selected as strong gamma-ray candidates based on their radio properties, distance, small angular size, and possible association with a molecular cloud (Buckley et al. 1998; Hess et al. 1997). The small angular size was made a requirement because of the limited field of view (3° diameter) of the Whipple Telescope at that time. VHE telescopes can also detect fainter gamma-ray sources if they are more compact, because they can reject more of the cosmic-ray background. IC 443, gamma Cygni, and W44 are also associated with EGRET sources (Esposito et al. 1996). Despite long observations, no significant excesses were observed, and stringent limits were derived on the VHE flux (see Table 4).

Table 4. Observations of shell-type supernova remnants

ObjectName Observation Time
Integral Flux a
(10-11 cm-2 s-1)

Tycho... 867.2 > 0.3 < 0.8 1
IC 443 b... 1076.7 > 0.3 < 2.1 1

678.0 > 0.5 < 1.9 c 2
W44 b... 360.1 > 0.3 < 3.0 1
W51... 468.0 > 0.3 < 3.6 1
gamma Cygni b... 560.0 > 0.3 < 2.2 1

2820.0 > 0.5 < 1.1 c 2
W 63... 140.0 > 0.3 < 6.4 1
SN 1006... 2040.0 > 1.7 0.46 ± 0.6stat ± 1.4sys 3

a Upper limits from Buckley et al. (1998) are at the 99.9% confidence level and those from Hess et al. (1997) are at the 3 sigma confidence level.
b Associated with an EGRET source.
c Upper limits were converted from fractions of > 500 GeV Crab flux using the measured Crab flux of Hillas et al. (1998).
References. - (1) Buckley et al. 1998; (2) Hess et al. 1997; (3) Tanimori et al. 1998a.

In contrast to the upper limits derived by the Northern Hemisphere telescopes, the CANGAROO Collaboration has recently reported evidence for TeV gamma-ray emission from the shell-type SNR SN 1006 (Tanimori et al. 1998a). Observations taken in 1996 and 1997 indicate a statistically significant excess from the northeast rim of the SNR shell (see Fig. 4). The position of the excess is consistent with the location of nonthermal X-rays detected by the Advanced Satellite for Cosmology and Astrophysics (ASCA) experiment (Koyama et al. 1995). If this object is confirmed as a TeV gamma-ray source, it represents the first direct evidence of acceleration of particles to TeV energies in the shocks of SNRs.

Figure 4

Figure 4. Left: Contour map of the statistical significance of the excess emission from SN 1006 as observed with the CANGAROO telescope in 1996. Right: The same plot for data taken in 1997. The solid lines indicate the contour map of nonthermal X-ray emission detected with ASCA. The dashed circles indicate the angular resolution of the CANGAROO telescope. Figure from Tanimori et al. (1998a).

3.4.2. Implications of the gamma-Ray Observations

If the EGRET detections do indicate the presence of gamma-rays produced by secondary pion decay, the measured flux can be compared with the VHE upper limits. These VHE upper limits and the EGRET measurements are compared with the predicted fluxes from the model of Drury et al. (1994) in Figure 5 (Buckley et al. 1998). The solid curves are normalized to the integral above 100 MeV flux detected by EGRET assuming a source cosmic-ray spectrum of E-2.1, so they assume that the EGRET emission is entirely due to cosmic-ray interactions and that the cosmic-ray spectrum at the source has the canonical spectrum. In the cases of gamma Cygni, IC 443, and W44, the Whipple upper limits lie a factor of ~ 25, 10, and 10, respectively, below the model extrapolations and require either a spectral break or a differential source spectrum steeper than E-2.5 for gamma Cygni and E-2.4 for IC 443 (Buckley 1998). In addition, the measured EGRET spectra, while consistent with a spectral index of about 2, do not show the flattening of the gamma-ray spectrum near 100 MeV which would be expected if the gamma-rays result from secondary pion decay. Gaisser, Protheroe, & Stanev (1998) performed multiwavelength fits to the EGRET and Whipple results and concluded that if the EGRET detections are truly from the shells of the SNRs, the EGRET data must be dominated at low energies by electron bremsstrahlung radiation, but the source spectrum at high energies must still be relatively steep (~ E-2.4) to account for the Whipple upper limits (cf. Buckley et al. 1998).

Figure 5

Figure 5. Whipple Observatory upper limits (W) shown along with EGRET integral fluxes (E) and integral spectra. Also shown are CASA-MIA upper limits (CM) from Borione et al. (1995), Cygnus upper limits (C) from Allen et al. (1995), and the AIROBICC upper limit from Prosch et al. (1996). The solid curves indicate extrapolations from the EGRET integral data points at 100 MeV (triangles). The dashed curves are estimates of the allowable range of fluxes from the model of Drury et al. (1994). Figure from Buckley et al. (1998).

If, on the other hand, the higher than 100 MeV emission results from some other emission mechanism, such as the interactions of high-energy electrons accelerated by embedded pulsars or in the SNR shocks, the EGRET results should not be compared with the Whipple data. Instead, the Whipple data must be considered alone in the context of the secondary pion decay models. This possibility was also investigated by Buckley et al. (1998) and is shown by the dashed curves in Figure 5. The dashed curves represent a conservative estimate of the range of allowable parameter values for the Drury et al. (1994) model, without reference to the EGRET detections. For this comparison, there is still room for the models to work in these objects. However, the upper limits in some (e.g., IC 443) are beginning to strain the limits of the available parameter space. It will take more sensitive measurements with future telescopes to fully span the allowable parameter space and see if we need to reconsider our assumptions about the sources of cosmic rays within our Galaxy.

The TeV emission from SN 1006 detected by the CANGAROO group also does not require the presence of hadronic cosmic rays. In fact, the most common explanation for the detected emission is inverse Compton scattering of electrons with cosmic microwave background photons (Reynolds 1996; Mastichiadis & de Jager 1996). The main arguments for this are that the emission is centered on one of the regions where the synchrotron emission was detected with ASCA and the lack of evidence for a nearby molecular cloud needed to boost the TeV gamma-ray flux to a detectable level. Under the assumption that the emission is from inverse Compton scattering, Tanimori et al. (1998a) have combined their data with the ASCA results to derive an estimate of 6.5 ± 2 µG for the magnetic field within the SNR shell. The observations also provide an upper limit on the acceleration time. Thus, the TeV observations provide previously unknown parameters for models of the shock acceleration in SNRs. Unfortunately, the possibility of inverse Compton emission from the shells of SNRs also confuses the issue for using gamma-rays as a probe of cosmic-ray acceleration in SNRs. Future measurements will need to provide accurate spectra and spatial mapping of the gamma-ray emission from SNRs in order for the source of the gamma-ray emission to be unambiguously resolved.

Next Contents Previous