7 Summary and discussion

We have identified eight different isotopes in the COMPTEL instrumental line background, namely ²D, ²²Na, ²⁴Na, ²⁸Al, ⁴⁰K, ⁵²Mn, ⁵⁷Ni, and ²⁰⁸Tl. These isotopes can account for the major instrumental background lines. Some minor instrumental lines, however, remain unidentified at this time. In addition, we have studied the variation of the event rate of these isotopes with time and incident cosmic-ray intensity, and determined the average activity of spacecraft materials.

These results provide valuable insight into the physical processes that give rise to instrumental background lines in $\gamma$-ray detectors in low-Earth orbits. Because of their importance for MeV astronomy, these issues have repeatedly been studied (see, e.g., Dean et al. 1991 and references therein). Obviously, each detector has its own, unique instrumental background. Nevertheless, background investigations such as those reported in this work add to a growing pool of background experience to be found in the literature, which will prove to be important for the design of future instruments and the understanding of their data.

The successful modelling of the time variation of the background contributions from the long-lived isotopes ²²Na, ²⁴Na, ⁵²Mn, and ⁵⁷Ni in COMPTEL supports the conclusion that activation occurs predominantly during SAA passages. On average, CGRO passes through the SAA 6-8 times each day. In 1991 (at solar maximum and hence at minimum SAA-proton flux) the daily average of the incident flux of SAA protons with ${E}_{\rm p} > 100$ MeV was about 38 times higher than the corresponding incident galactic cosmic-ray proton flux (Dyer et al. 1994), illustrating the importance of SAA passages with respect to the total deposited radiation dose. The SAA-proton flux varies strongly with altitude and solar cycle. For example, during its mission the altitude of CGRO ranged from 350-500 km, which corresponds to a variation of the SAA-proton flux of a factor $\sim$10, while the solar activity results in a variation by a factor $\sim$2 at these altitudes (Stassinopoulos 1989). According to our activity model (Varendorff et al. 1997) the anisotropy of the SAA-proton flux at low satellite altitudes (Watts et al. 1989) has an important effect on the radiation dose and hence on activation. These findings are consistent with earlier studies of GRS-SMM instrumental lines, in which activation due to (anisotropic) radiation-belt protons during SAA transits has been identified as major source of (long-lived) line background and its variation with time (see Kurfess et al. 1989; Share et al. 1989, and references therein).

In contrast, the activities of short-lived isotopes are mostly determined by the incident primary cosmic-ray and albedo-neutron fluxes. The enhanced production of these isotopes during SAA transits is of little importance for their background contribution outside the SAA, where astrophysical data are recorded, because of their rapid decay, which precludes build-up.

Two strategies for reducing instrumental line background in future low-energy $\gamma$-ray detectors due to activation of long-lived isotopes present themselves. First, care should be taken to avoid detector materials which are easily activated by SAA protons (such as aluminium) or which carry primordial radioactivity. Second, instruments should be placed in orbits that minimize radiation-belt proton dosage, such as a low-altitude equatorial orbit below the radiation belts and outside the SAA, or a high-altitude orbit above the radiation belts (as scheduled for the INTEGRAL observatory). The flux of (low-energy) cosmic rays is significantly lower in a low-altitude orbit as compared to a high-altitude orbit - at the cost of a much higher albedo-neutron flux. Another strategy for limiting the impact of remaining instrumental lines in future instruments is, e.g., to exploit the expected improvement in energy resolution to identify (and model) a large number of instrumental background components. A comprehensive treatment of the implications of the COMPTEL instrumental line (and continuum) background for future Compton telescopes is beyond the scope of this paper and will be given in a separate publication.

Finally, although investigations of background lines are of interest in their own right, they are ultimately motivated by the struggle to eliminate background events in astrophysical analyses. The COMPTEL instrumental background lines, particularly those from long-lived isotopes, are a concern below about 4 MeV. At these energies detailed understanding of the dataspace structure and variability of the instrumental lines is indispensable for optimizing many astrophysical investigations, such as that of the cosmic diffuse gamma-ray background (see e.g. Kappadath et al. 2000; Weidenspointner et al. 2001) or of the galactic 1.8 MeV line emission from ²⁶Al (see e.g. Oberlack 1997; Plüschke et al. 2000).

Comparing event rates due to $\gamma$-ray line sources and due to instrumental background it is evident that source fluxes cannot be derived from global event rates with COMPTEL. This is only possible in imaging analysis, which exploits the characteristic cone-like distribution of source events in a three-dimensional data space spanned by the scatter angle $\bar{\varphi}$ and the direction (zenith and azimuth) of the scattered photon (see Schönfelder et al. 1993). Instrumental background components exhibit comparatively smooth distributions in this data space, allowing for their separation from the signal. Further exploitation of existing knowledge on the dataspace structure of instrumental lines will remedy some of the difficulties in the search for astrophysical $\gamma$-ray lines (see e.g. Morris et al. 1997b). Nevertheless, data selections that increase the signal-to-background ratio are clearly of great value.

The importance of instrumental background suppression/rejection can be illustrated with the original report of nuclear de-excitation lines from the Orion region (Bloemen et al. 1994), which would have corresponded to an event rate of about 6 10^-4 s^-1 spread over a rather broad energy range of 3-7 MeV. As was recognized in our recent analyses, the variations in the ²⁴Na event rate over the different observation periods of this galactic region are capable of producing spatial inhomogeneities in the background of that order, which then can be falsely attributed to a celestial signal. Only correlations of the supposed Orion signal with data space domains where ²⁴Na contamination is strongest provided the hints to suspect an instrumental background artifact as the origin of the claimed detection (Bloemen et al. 1999).

Acknowledgements

It is a pleasure to acknowledge laboratory efforts of J. Macri, and of R. Georgii, C. Wunderer, and W. Plass in measuring primordial radioactivity in D1 PMTs. We are also pleased to acknowledge work on hadron simulations for estimating activation of spacecraft materials by P. Jean. The COMPTEL project is supported by the German government through DARA grant 50 QV 90968, by NASA under contract NAS5-26645, and by The Netherlands Organization for Scientific Research NWO.