Astron. Astrophys. 319, 397-400 (1997)

3. Observations and results

We have used the same set of observations as in the analysis of the 0.75-3 MeV range reported in Paper I to be able to directly compare the results at 3-30 MeV to those in the lower energy bands.

Again, we detect no individual Seyfert galaxy, and no significant flux in the cumulative data which correspond to a net observation time of 70 days. The 2 upper limits for the individual sources are listed in Table 1, together with those for the two low energy bands already presented in Paper I. The limits on the complete sample are 1.05 10^-9 photons cm^-2 s^-1 keV^-1 for the 3-10 MeV band and 1.32 10^-10 photons cm^-2 s^-1 keV^-1 in the 10-30 MeV band, and about the same in as the limit in the 1-3 MeV band (9.7 10^-9 photons cm^-2 s^-1 keV^-1), see Paper I). The 2 upper limits are shown in Fig.1, together with the extension of the average spectrum of Seyfert galaxies at several 100 keV derived from OSSE observations (Johnson et al. 1994, Z95).

Table 1. Individual 2 upper limits for Seyfert galaxies observed by COMPTEL in Phase I

Fig. 1. COMPTEL 2 Upper Limits Compared to X-Ray Samples of Johnson et al. (1994) and Z95

This result is not surprising when compared to the analysis of Z95 who analysed (non-simultaneous) Ginga and OSSE data of 9 Seyfert galaxies and found that the average spectra can be described by an exponentially truncated power law with photon indices around the canonical X-ray values of 1.8-2.0 and an e-folding energies of several 100 keV. The OSSE data alone indicate even steeper spectra with e-folding energies of  40-50 keV (Johnson et al. 1994). Simultaneous broad-band observations with XTE and OSSE could better constrain this parameter in the future. The best fit spectrum for the complete sample of Seyferts from Z95 is shown in Fig.1 for comparison. It is obvious that the upper limits we derive are still substantially above the level of emission expected from the work of Z95.

It has been suggested that the thermal appearance of the Seyfert X-ray spectra could be due to a highly anisotropic nonthermal source emitting most photons toward the disk, so that the observed spectrum is mostly due to reflection and Comptonization by the disk (Mannheim 1995a). A natural anisotropy of this kind develops for pair cascades produced by ultrarelativistic protons (Lorentz factor ) accelerated in a magnetized disk wind ('hadronic jet') as they cool by photo-production of secondary particles in the radiation field of the disk. Pions are produced by head-on collisions with UV photons (energy ) from the inner disk. In the rest frame of the proton, the UV photons appear with energies which leads to catastrophic energy losses when , thereby stopping further proton acceleration. The pions subsequently decay giving rise to an anisotropic cascade irradiating the disk hemisphere. Infrared photons originating in the heated dust torus surrounding the central object appear with energies in the proton rest frame giving rise to Bethe-Heitler pairs. Owing to the solid angle subtended by the infrared photons, the Bethe-Heitler pair distribution is nearly isotropic. The pairs produce synchrotron -rays in the 3-10 MeV energy range. The expected energy flux in the 3-10 MeV range is maximally of the same order as the Compton reflected component, but does not contradict the already derived flux limits at MeV due to the rather flat spectrum (Fig.2). From our observation alone, no constraints on this extra component can be derived, as the upper limits are of about the same magnitude in as the fluxes of the OSSE observations at several tens of keV (Z95).

Fig. 2. Comparison of COMPTEL 2 Upper Limits to those derived from the CXB (solid lines) which is scaled to the keV emission (dashed lines) derived by Z95. The dashed-dotted line represents the maximal flux expected from a Bethe-Heitler pair component in the hadronic jet model of Mannheim (1995a)

Tighter constraints on the MeV emission of Seyferts than those obtained from the individual and cumulative observations can be derived from the recent results on the extragalactic background derived by Kappadath et al. (1996), who found that the MeV bump in the background (e.g. Gruber 1992) was an artifact owing to the detector background caused by charged particles being dependent on the geomagnetic rigidity at which individual measurements were conducted. Kappadath et al. (1996) conclude that there is no MeV bump, and that the spectrum of the CXB can be described by a power law from 100 keV to hundreds of MeV, as originally found by Mazets et al. (1975). The photon index of the power law connecting hard X-ray and MeV -rays lies in the range 2.5-3, significantly flattening toward EGRET energies where (Kniffen et al. 1996).

Recent modeling of the CXB (Comastri et al. 1995, Z95) has shown that the CXB at several tens of keV can be described by the superposition of Seyfert galaxies (more generally radio quiet AGN) at various levels of obscuration. Assuming similar spectra for all Seyferts from hard X-rays to MeV energies, the steep spectrum of the CXB places stronger constraints on the -ray flux from Seyferts than the actual observations reported in this paper.

This can be seen from Fig.2, which shows the 2 upper limits derived from the COMPTEL data together with the average X-ray spectrum of Seyferts from Z95, plus the spectrum of the CXB (following Mazets 1975, Kappadath 1996 and Kniffen 1996) scaled to the Z95 spectrum. Even neglecting the possible contribution of other AGN and Supernovae type Ia to the CXB, the persistent emission from Seyferts must lie a factor of 10 below the COMPTEL upper limits to be consistent with the CXB. Accordingly, these constraints imply that the anisotropic nonthermal cascades either have extremely large disk/observer flux ratios obtaining values or, more likely, that they do not contribute substantially to the reflected component. This is in agreement with the non-detection of  TeV neutrinos with the Fréjus proton-decay experiment (Mannheim 1995b). A diffuse neutrino background with an energy flux comparable to the CXB as proposed by Stecker et al. (1991) therefore cannot be expected from radio-quiet AGN.

The main contribution to the primary X-ray emission from radio-quiet AGN responsible for the reflection hump seems to come from coronal plasma near the inner accretion disk (e.g., Haardt and Maraschi 1993) or from a nonthermal (non-cascade) source with an intrinsic turnover at  keV.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de