The reasons which make the detection and identification of
rays from SNRs that are located in ordinary (low-density) regions of the ISM rather difficult, are twofold: (i) slow interaction rate
allowing only limited efficiency of conversion of energy of relativistic protons to Y-rays,
(ii) significant "contamination" due to the IC component of radiation, especially at TeV energies. Therefore, the best hope to obtain solid evidence of Y-rays of hadronic origin is associated with SNRs in dense environments.In this regard two scenarios seem to be relevant – "SN explosion inside the GMC" and "GMCs overtaken by the supernova shell". The features of y-radiation of both cases are discussed in Sec.4.2 without specifying the type of particle accelerator. In the particular case of SNR shocks operating as particle accelerators, the scenario of "SN explosion inside the GMC" cannot guarantee effective particle acceleration to very high energies , because of rapid cooling of the shock. Moreover, even if particles are somehow accelerated to very high energies, they may escape the clouds quickly, leading to formation of steep particle spectra inside the cloud. The second scenario gives a more or less passive role to giant molecular clouds (GMCs) which act as targets for protons, significantly enhancing the production rate of n0-decay Y-rays. This scenario seems the most natural way to amplify the Y-radiation to a level comparable with the sensitivity of EGRET. Therefore, if the the reported associations of several EGRET sources with SNRs (Sturner and Dermer, 1995; Esposito et al., 1996; Torres et al., 2003) are not result of accidental geometrical coincidence, dense GMCs overtaken by SNRs seem the most natural sites of Y-ray production. A SNR in cloudy environments may in fact initiate a cluster consisting of several Y-ray sources associated with it.
Montmerle (1979) was perhaps the first to suggest that SN explosions occurring in OB associations which are rich in massive molecular clouds may lead to observable high energy Y-ray fluxes. Such evidence has been found by Pollock (1985) who argued that the interaction of the SNR G78.2+2.1 (y Cygni) with a nearby cloud should be responsible for Y-rays detected by COS B in that direction.
The discovery by EGRET of approximately 80 Y-ray sources at low galactic latitudes, the most of them being still unidentified, initiated new interest in SNRs interacting with nearby clouds. In the last several years the possible correlation of sources from the 3rd EGRET catalog (Hart-man et al., 1999) with relatively young galactic SNRs has been extensively explored. These efforts, motivated by a search for counterparts for the unidentified EGRET sources at low galactic latitudes, revealed new evidence of SNRs interacting with molecular clouds. An interesting outcome of these studies was, for example, the discovery of a large (8° x 8°), low brightness shell type SNR in the Capricornus region, the radio structure of which spatially correlates with three unidentified EGRET sources (Combi et al., 2001). Moreover, the 21 cm line observations revealed that two of these sources coincide with HI clouds. Although the possibility of a chance association cannot be ruled out, this cluster of Y-ray sources can be naturally interpreted as radiation of dense clouds overtaken by the remnant. Despite large observational and theoretical uncertainties (e.g. in the ambient gas density and strength of the magnetic field, the age of the remnant, the energy spectra and e/p ratio of accelerated particles, etc.), the interpretation of Y-ray data as the result of interactions of accelerated particles in dense regions
favours two radiation mechanisms – electron bremsstrahlung and n0-decay Y-rays . On the other hand it is difficult to estimate the contributions from each of these channels based merely on the low-energy Y-ray observations. Detailed multiwavelength studies of a large sample of SNR-GMC interacting systems are needed to understand the acceleration and radiation processes in these objects.
X-ray observations are of special interest. Hard X-rays of synchrotron origin detected from several shell type SNRs are an unambiguous indicator of multi-TeV electrons accelerated, most likely, by strong SNR shocks. Since the cutoff energy of this component determined by Eq.(5.14) is always expected around or below 0.1 keV, the synchrotron X-radiation of SNRs can be easily recognised by its steep spectrum above 1 keV, as well as by its spatial localisation in the shell, where the acceleration takes place. Besides the X-rays of synchrotron origin, we may expect another component of hard X-radiation due to the nonthermal bremsstrahlung of subrelativistic electrons or protons. Because of the Coulomb-loss-flattened distribution of electrons in high density environments, this mechanism predicts an extremely hard X-ray spectrum with power-law photon index r ~ 1 (Uchiyama et al., 2002a). Therefore, the high energy Y-radiation of either bremsstrahlung or n0-decay decay origin should be accompanied of subrelativistic X-radiation with a characteristic e-1 type spectrum, detection of which may serve as an indicator of the existence of a large amount of subrelativistic particles in clouds. For a given spectrum of the electron population, the ratio of the X- and Y-ray fluxes depends on the position of the Coulomb break in the electron spectrum, which in its turn depends on the product of the ambient density and the age of the source. It is likely that such radiation has been indeed detected from y Cygni (Uchiyama et al., 2002a) and RX J1713.7-3946 (Uchiyama et al., 2002b).
Bremsstrahlung X-rays from j Cygni
The broad-band spectral energy distribution of the so-called Hard X-ray Clump (HXC) detected by ASCA in the supernova remnant Y Cygni is shown in Fig. 5.7. Nonthermal bremsstrahlung from the accelerated electrons is a natural source of the HXC flux, because the shocked dense clouds act as an effective target for energetic electrons. Due to Coulomb interactions the high density gas results in a significant hardening of low-energy electrons below the "Coulomb break", giving rise to the standard e-1 type bremsstrahlung spectrum in the X-ray band, which agrees with the ASCA data. The bremsstrahlung spectrum above the "Coulomb break" essentially repeats the acceleration spectrum of electrons.
Fig. 5.7 presents the results of numerical calculations for 2 sets of parameters which describe the gas density and the acceleration spectrum of electrons, assuming that electron bremsstrahlung is responsible for both the ASCA hard X-ray and the EGRET Y-ray fluxes. For the electron spectrum with the acceleration index re = 2.1, the best fit is achieved for a gas density of n =34 cm-3. A steeper acceleration spectrum with re = 2.3 requires larger gas density, n = 130 cm-3. Note that the adopted acceleration spectra are consistent with the reported radio spectral index ar = 0.5 ± 0.15. Also, an exponential cutoff in the electron spectrum at 10 TeV is assumed. If the electron distribution with re = 2.3(2.1) extends beyond GeV energies, for the magnetic field 10-5 G the calculated radio flux density amounts to about 10%(60%) of the measured radio flux density integrated over the whole remnant. Furthermore, if the electron distribution extends beyond TeV energies, the bremsstrahlung spectrum with re = 2.1 exceeds the Whipple upper-limit, whereas the spectral index re = 2.3 is still in agreement with the Whipple result. Both combinations of model parameters quite satisfactorily fit the spectral shape and the absolute flux of hard X-rays.
Ignoring the energy losses of electrons would lead to significantly steeper X-ray spectra, and would also result in overproduction of absolute X-ray fluxes (dotted curves in Fig. 5.7). Note that the main contribution to X-rays comes from relatively high energy electrons with energies close to 10 MeV for the acceleration index re = 2.3.
Because of its poor angular resolution, EGRET measurements do not provide clear information about the site(s) of production of high energy Y-rays. Nevertheless, it is likely that only a part (perhaps, even only a small part) of the reported high energy Y-ray fluxes originates in the HXC region.
The Y-ray fluxes could be easily suppressed by assuming lower gas densities. Indeed, such an assumption would lead to a shift of the "Coulomb break" energy in the electron spectrum to lower energies, and the predicted high energy Y-ray spectra would appear significantly below the reported EGRET fluxes (the solid curve in Fig. 5.7).
A more likely candidate site for the production of the bulk of high energy Y-rays is the region called DR4 from which most of the radio emission emerges. A massive cloud with a density of ~ 300 cm-3 occupying ~ 5% of the SNR volume has been suggested to exist in the vicinity of DR4 to explain the Y-ray flux . Actually the reported EGRET error circle is somewhat removed from the HXC, but closer to the DR4. A density of ~ 300 cm-3 is higher than the upper limit density of the HXC. Such high gas density implies a high (about 50 MeV) "Coulomb break" energy in the electron spectrum, and considerable suppression of the X-ray flux. This would naturally explain the lack of noticeable hard X-ray flux from the DR4 region, which is bright in radio and perhaps also in Y-rays.
Fig. 5.7 Broadband spectral energy distribution of the HXC region of y Cygni. The range of the power-law fit of the hard X-ray component is shown together with Y-ray data (> 100 MeV) of 2EG J2020+4026 taken from Esposito et al. (1996) and the Whipple TeV upper limit from Buckley et al. (1998). The bremsstrahlung photon spectra from the loss-flattened electron distribution are calculated for the electron index re = 2.1 and the gas density n = 34 cm-3 (long-dashed line), re = 2.3 and n = 130 cm-3 (dashed line), and re = 2.3 and n =10 cm-3 (solid line). The dotted lines show the bremsstrahlung spectra corresponding to the acceleration spectra of electrons, i.e. ignoring the Coulomb losses of electrons.
For a gas density of about 100 cm-3 in the HXC, the X-ray flux is produced predominantly by electrons with energies of about 10 MeV. The X-ray flux is roughly proportional to the product of the gas density and the number of relativistic electrons, because the relativistic bremsstrahlung cross-section depends only logarithmically on the electron energy. On the other hand, the Coulomb energy loss rate of relativistic electrons, 
is proportional to the gas density and almost independent of the electron energy. Therefore the energy loss rate of the bulk of the electrons can be uniquely determined by the hard X-ray luminosity, independent of the density and the shape of the electron energy distribution. The estimated hard X-ray luminosity,
can be converted to an energy loss rate of
The energy released in relativistic electrons then would be estimated as
This enormous energy deposition due to Coulomb collisions would heat the emission region of the HXC. Subsequently the heat would be radiated away in the far-infrared band by molecular line emission, if the gas is comprised of molecules. The observed infrared luminosity of Y Cygni is indeed comparable to the above estimate of the Coulomb energy loss rate.
The case of RX J1713.7-3946
This shell type supernova remnant is of great interest, being one of three SNRs so far detected both in synchrotron X-rays (Koyama et al., 1997; Slane et al., 1999, Uchiyama et al., 2003) and TeV Y-rays (Muraishi et al., 2000, Enomoto et al., 2002). Two very massive
and dense
clouds are found in the vicinity of this SNR, one of which (cloud A) probably borders with the shock-wave region of the remnant (Slane et al., 1999). The results based on observation in ASCA’s large FoV revealed an extended X-ray source (Uchiyama et al., 2002b) coincident with cloud A (see Fig.5.8a). The cloud has a positional association also with the unidentified EGRET source 3EG J1714-3857. Butt et al. (2001) claimed that the shock front of RX J1713.7-3946 is interacting with the cloud A, and argued that this would be a natural site of production of > 100 MeV Y-rays, presumably of hadronic origin.
The X-rays from cloud A have, most likely, nonthermal origin. As discussed in the previous section, the unusually flat spectrum (the pho-ton index r = 1 ± 0.3) can be best interpreted in terms of characteristic bremsstrahlung emission from the ionization-loss-flattened distribution of either subrelativistic electrons or protons.
Fig. 5.8 (a) Overall ASCA view of RX J1713.7-3946 in the energy band 5-10 keV (left; from Uchiyama, Takahashi, 2002) and (b) the X-ray image of the northwest rim in the Chandra’ s 17′ X 17/ field-of-view in the 3-5 keV band.
Although this model explains the observed spectral features of the X-ray emission, it requires huge (marginally acceptable) kinetic energy of about 1050 erg in the form of subrelativistic particles, if the estimate for the distance to the source of about 6 kpc (Slane et al., 1999) is correct. The energy requirement is reduced to a quite reasonable level if one assumes that the SNR is located much closer, e.g., at d — 1 kpc (Koyama et al., 1997). In any case, the relation of this radiation to RX J1713.7-394 is not quite clear, and in fact is a subject of further studies.
In this regard, the X-ray emission detected from the northwest rim of RX J1713.7-39, as well as the TeV radiation detected from the same direction have more fundamental implications for this supernova remnant. The synchrotron origin of the X-radiation implies the existence of multi-TeV electrons, and contains information about the parameter B1/2E0. In order to disentangle B and E0, one needs an additional piece of information, and so long the 2.7 K CMBR remains the main reservoir for target photons and the magnetic field does not significantly exceed 10 ^G, such information is contained in IC Y-rays. But these observations would have much more fundamental implications, in particular for the origin of galactic CRs, if the detected TeV-radiation was of hadronic origin.
The observations of RX J1713.7-3946 by the CANGAROO collaboration using the new 10 m diameter imaging Cherenkov telescope (Enomoto et al., 2002) not only confirmed the TeV signal reported earlier (Muraishi et al., 2000), but also provided very important information about the shape of the differential spectrum, which between 0.4 TeV and 10 TeV is approximated by a power-law with photon index
If confirmed, this information would be almost sufficient for a definite conclusion about the origin of TeV Y-rays.
Indeed, the energy flux of TeV radiation of about
is quite comparable with the flux in synchrotron X-rays (Enomoto et al., 2002), which implies that the IC origin of TeV Y-rays would require a magnetic field as small as
This requirement is tighter than the case of SN 1006, especially if one takes into account that RX J1713.7-394 is located in the galactic plane. Given the shock compression of the magnetic field in the shell by a factor of 4 or more, this upper limit on the magnetic field seems unrealistic, and consequently the interpretation of TeV data in terms of the IC scattering on the 2.7 K CMBR rather inadequate. Moreover, the observed steep spectrum at relatively low energies does not agree with prediction of the "2.7 CMBR inverse Compton" origin of TeV Y-rays. Indeed, the spectral fit for the synchrotron radio to X-ray data requires
and the electron spectral index ae = 2.08 (Enomoto et al., 2002). For
this gives a lower limit on the cutoff energy of electrons,
Since the target photon is fixed, this allows a robust prediction for the spectral shape of the TeV emission. Namely, it can be described by a hard power-law with a photon index of ^
up to 1 TeV, after which it gradually steepens to
around a few TeV, and
above 10 TeV. Since this is in apparent conflict with the observed single power-law type spectrum with
the IC interpretation can be discarded if the 2.7 K CMBR represents the main target field for electron scattering (see Fig. 2.10).
Formally one may assume that some other seed photon field, different to the 2.7 K CMBR, dominates in the formation of the IC scattering component. In this case the steep y-ray spectrum can be explained by the Klein-Nishina effect. This implies that the mean energy of these seed photons should be 1 eV or larger. In order to make this IC component dominant, one has to assume an unreasonably high density of optical/UV photons in RX J1713.7-39, one that is not supported by observations. The bremsstrahlung component also can be confidently excluded. Although for
the absolute bremsstrahlung flux around 1 TeV can match the observed Y-ray flux, in order to avoid overproduction of Y-rays at higher energies one must assume a cutoff in the electron spectrum around 1 TeV which, however, contradicts the synchrotron-best-fit condition of
After excluding all possible Y-ray components of leptonic origin, one may arrive at conclusion that
Y-radiation remains the only option which can explain the TeV data without being in conflict with the X-ray data. The X-rays can be still explained by synchrotron radiation of electrons. This of course would require suppression of the IC component of Y-radiation assuming
The model calculations presented in Fig. 2.10 demonstrate that the TeV spectral points can be satisfactorily fitted by the n0-decay mechanism assuming an exponential cutoff in the proton spectrum at
For the average hydrogen density in the cloud of about n ~ 300 cm-3, the observed TeV fluxes can be explained if the content of relativistic protons in the cloud is about
The different estimates of the distance to the source vary between 1 kpc (Koyama et al., 1997) and 6 kpc (Slane et al., 1999). The total energy accelerated in protons typically is limited to 1050 erg, a significant part of which is may contained in subrelativistic protons or electrons. Also, for any realistic geometry only a fraction of accelerated protons can be captured/confined in the cloud. Therefore, a relatively small distance to the source, close to its lower limit of about 1 kpc, seems preferable on energy grounds.
The hadronic model of the TeV emission of RX J1713.7-39 has been criticised by Reimer and Pohl (2002) and Butt et al. (2002) who argued that this interpretation violates the Y-ray upper limits set by EGRET at GeV energies. This is, however, a rather shaky argument; the "GeV upper limit" problem can be overcome in several natural ways. For example, (i) adopting a slightly harder proton spectrum, it is possible to avoid the conflict with the EGRET data. Even for a proton spectrum steeper than E-2, it is still possible to suppress the GeV Y-ray flux, if one invokes (iii) the effects of energy-dependent propagation of protons while travelling from the accelerator (SNR shock) to the nearby clouds. Moreover, the lack of the GeV Y-rays can be naturally explained by (iii) confinement of low-energy (< 10 GeV) protons in the shell, in contrast to the effective escape of high-energy (multi-TeV) protons, assuming that the particle acceleration proceeds in the Bohm diffusion regime. And finally, the GeV Y-ray emission perhaps can be additionally suppressed assuming that (iv) the low energy protons do not freely penetrate the densest regions of the cloud due to quasi-static magnetic mirrors or different type of plasma instabilities caused by the same cosmic rays.
On the other hand, the limited energy budget available for cosmic rays requires TeV Y-ray production in very dense regions, in particular in giant molecular clouds which should significantly override the proton acceleration site(s). Unfortunately, the quality of the CANGAROO data is not sufficient for definite conclusions concerning the exact location of the TeV production region(s). The observations of this source by the H.E.S.S. and CANGAROO-III arrays of Cherenkov telescopes should allow detailed morphology of the TeV production region with a bin size as small as several arcminutes. A similar accuracy for study of spatial distribution of GeV Y-rays can be achieved in future observations of RX J1713.7-39 by AGILE and GLAST. Although hadronic models generally predict similar spatial distributions of GeV and TeV Y-rays, some differences nevertheless may arise due to possible energy dependent effects in the propagation of protons from their acceleration site to the nearby clouds.
Fig. 5.9 Multiwavelength synchrotron (solid lines) and IC (dashed lines) radiation spectra from the filaments and the plateau regions of RX J1713.7-39.It is assumed that the electron acceleration takes place in the filaments. The following parameters sets have been used in calculations: The age and the distance to the source: T = 10 000 yr and d =6 kpc. The acceleration rate, the spectral index and the exponential cutoff of the electron spectrum:
a = 1.84, and Eo = 125 TeV, respectively. The magnetic field in the filaments and in the plateau region :
The convective escape time Tcon = 1000 yr, the gyrofactor in the filaments n = 83.
The importance of comprehensive studies of both spatial and spectral properties of nonthermal radiation has been demonstrated by the Chandra observations of the northwest rim of RX J1713.7-39, the brightest region of nonthermal X-rays. The Chandra image revealed (Uchiyama et al., 2003) a complex network of X-ray filaments surrounded by a fainter diffuse plateau, and a dark region of circular shape (see Fig. 5.8b). The examination of the individual spectra from different parts of the rim showed that despite significant brightness variations, the X-ray spectra everywhere in this region are similar, being well fitted with a power-law with photon index r ~ 2.3. Furthermore, the spectra indicate that the synchrotron cutoff is located beyond 10 keV. Both the filamentary structure and the lack of synchrotron cutoff below a few keV may challenge the perceptions of the standard shock acceleration models concerning the production, propagation and radiation of ultrarelativistic electrons.
The X-ray brightness distribution shown in Fig. 5.8b most likely is the result of highly inhomogeneous production and propagation of relativistic electrons.
Fig. 5.10 The same as in Figs. 5.9, but for the following model parameters: T = 1000 yr, 
At first glance, homogeneous distribution of multi-TeV electrons over the rim cannot be ruled out, because in this case the filamentary structure might be referred to a local enhancement of the magnetic field. However, in the regime, when the synchrotron cooling time is shorter than the source age, the synchrotron X-ray luminosity does not depend significantly on the magnetic field. In this regime an equilibrium is quickly established between the electron injection and synchrotron losses. This implies that the synchrotron radiation saturates to the maximum possible rate determined by the electron injection rate.
In this context it seems natural to identify the X-ray filaments as sites of particle acceleration. It is likely that the filaments are bright not only because of the enhanced magnetic field, but also due to high concentration of electrons inside the accelerators. A significant fraction of these electrons may escape from the thin filaments, and fill much larger regions. This will result in synchrotron X-rays from the extended regions (plateau), the brightness of which can be low, but their overall X-ray flux may exceed the total flux from filaments.
The results of a numerical time-dependent treatment, which includes the energy losses and escape of electrons from their accelerators, but do not specify the acceleration mechanism (i.e. assumes an a priori acceleration rate and spectrum of electrons extending to > 100 TeV), are shown in Figs. 5.9 and 5.10, for two different sets of model parameters (Uchiyama et al., 2003). These results show that the hypothesis of electron acceleration in the filaments and hot-spots does explain the fluxes and spectral shapes of the X-ray emission from both the filaments and the plateau region. On the other hand, the calculated IC fluxes are at least one order of magnitude below the TeV fluxes reported by the CANGAROO collaboration.
