3 Characteristics of instrumental linebackground

The particle and photon environment of COMPTEL produces coincident interactions in the D1 and D2 detectors that pass all the logic and electronic criteria for a valid event. In the following, an overview is given of different types of instrumental background events in terms of their interaction process and location, as illustrated in Fig. 2. This classification of background events (van Dijk 1996) provides a simple and versatile framework for discussing the COMPTEL instrumental background. ToF is of prime importance in identifying and rejecting instrumental background events. A schematic representation of the ToF distribution of valid events is depicted in Fig. 3.

Figure 2: An illustration of the main channels for the triggering of valid events due to instrumental background (adopted from van Dijk 1996). For simplicity, BATSE was omitted in this schematic view of CGRO. The various event types are explained in the text

$\bullet$ Event types A and B: events caused by the double scattering of a single photon. Any photon created in CGRO may produce this type of background event, which, if the scattering is from D1 to D2, is identical to a proper celestial event. It follows that the ToF distribution of forward scattered single photon events is identical to that of celestial photons and peaks around 5 ns in the ToF forward peak. Depending on where they originate, forward scattered single photon events may be rejected. In particular, many of the photons that originate from below the D1 detector (such as type B events) can be eliminated by a selection on the scatter angle $\bar{\varphi}$ (see e.g. the event selections in Appendix A). The remaining instrumental background from forward scattered single photons is therefore mostly due to photons originating in, around and above the D1 detector (such as type A events); type B events are negligible. Line emissions appear in energy space, particularly in $E_{\rm tot}$ and E₁-E₂ space (see below).

Single photons may also scatter in D2 before interacting in D1. These backward scattered single photon events, however, are identifiable by their ToF distribution, which is confined in the backward peak at ToF values of about - 5 ns. Finally, high-energy neutrons can undergo a double scattering process analogously to photons, however, many of these neutron events can be rejected by their PSD value.

Figure 3: A schematic representation of the ToF distribution of valid events. Three major components can be discerned: the ToF backward peak and forward peak, centered at ToF values of about -5 ns and +5 ns, respectively, and an underlying continuum distribution. The backward peak is composed of all types of background events originating in and around the D2 detector (these are not specified individually in this illustration). The ToF forward peak contains the celestial signal as well as background events originating in and around the D1 detector, mostly of types A and C. The ToF continuum is dominated by background events of types D, E, and F originating in the instrument structure between the two detectors and the spacecraft structure in general. The relative magnitudes of the different components, which depend on $E_{\rm tot}$, are only represented approximately

$\bullet$ Event types C and D: events caused by two or more photons both spatially and temporally correlated (so-called multiple photon or cascade events). In general, multiple photon events are more efficient in generating a background event than single photon events, since the probability for coincident interactions in both detectors increases with the number of emitted photons. Emission of two or more photons can occur from a small region on a time-scale shorter than the coincidence window of 40 ns. Nucleons that have been excited above the first nuclear level, e.g. by proton or neutron interactions, may promptly emit a cascade of photons. The multitude of nuclear excitation levels often results in a rather featureless continuum distribution in energy space. If, however, only a few transition levels are involved, characteristic features appear in E₂ and E₁-E₂ space (see below). Multiple photons may also arise from $\beta$-particle bremsstrahlung, or the annihilation of a positron. Also, high-energy neutrons may induce the emission of one or more $\gamma$-ray photons outside of the D1 scintillator (thereby eluding rejection by PSD) and some may interact in the D2 detector. More complicated nuclear reactions such as the spallation (break-up) of a nucleus or the initiation of a shower of secondaries by an incident cosmic-ray particle or neutron may also produce multiple photons.

Frequently, the emission of photons is nearly simultaneous, i.e. on time-scales much shorter than the coincidence window. The ToF value of these event types is determined by the location of the emitting nucleus relative to the D1 and D2 detectors. Multiple photons originating in the vicinity of the D1 detector (such as type C events) will peak slightly below a ToF value of 5 ns (see Sect. 4.2) and therefore contribute to the ToF forward peak. Photons emitted by nuclei in the spacecraft material between the D1 and D2 detectors (such as type D events) will interact in the two detectors near-simultaneously and have ToF values that are broadly distributed around zero, while photons originating in and around the D2 detector will contribute to the ToF backward peak. The full, double-peaked ToF distribution of all multiple photon events not only reflects the location of the emitting nucleus, but also the mass distribution of the entire spacecraft relative to the D1 and D2 detectors.

$\bullet$ Event type E: events that are both spatially and temporally uncorrelated (the so-called random coincidences). The COMPTEL detectors are continuously exposed to a large flux of $\gamma$-ray photons. This inevitably leads to coincident interactions that qualify as valid events. The photons producing these random coincidences are mostly of local or atmospheric origin. Since the two photons creating the event are unrelated to one another, and, in particular, not correlated in time, these events are uniformly distributed in ToF. Type E events may also involve a neutron instead of a photon triggering the D2 detector, which has no PSD capability.

$\bullet$ Event type F: events caused by two photons that are temporally correlated, but spatially uncorrelated. High-energy cosmic-ray particles or atmospheric neutrons may interact at several different locations along their path through CGRO. Individual interactions include those generating type C and D events (e.g. spallations or showers). The whole interaction chain creating type F events is similar to multiple photon events, the main difference being that their ToF distribution depends on both the location of the interactions as well as the (relativistic) velocity of the primary particle. The ToF distribution of this type of event is broad and covers the entire coincidence window (also reflecting the spacecraft mass distribution).

$\bullet$ Other processes, such as the interaction of a neutron in the D2 detector after producing a photon in the D1 system, or direct ionization losses of $\beta$-particles created in the housings of the D1 scintillators, may also play a role. The event signatures of these and other, more complicated processes, however, will be similar to the event types described above.

Event types A - D may arise from activation by primary cosmic-ray protons or secondary particles as well as atmospheric neutrons. The time between the interaction of the proton or neutron within the telescope material and the actual triggering of a background event varies, since it depends on the decay time of the radio-isotopes produced. These processes and the resulting background events can be crudely separated into "prompt'' and "delayed'' components. For prompt background events the time delay between the primary particle interaction in the instrument and the resulting emission of background photons is shorter than the coincidence window of 40 ns for the triggering of a valid event. Thus the intensity of prompt background components instantaneously follows the (time-variable) incident local cosmic-ray flux (see Sect. 5). For delayed events the time delay is longer than the typical length of the fast-logic veto signal of $\sim$200 ns. In contrast to protons, neutrons can travel to any location in the spacecraft to produce $\gamma$-ray photons without triggering the veto system. Inside the veto domes, particularly in the D1 detector, prompt events can therefore only be produced by neutrons. Both protons and neutrons, however, can produce delayed background events inside as well as outside the veto domes.

As described in Sect. 2, events due to incident celestial photons have ToF values around 5 ns. Only background events with a similar ToF value will therefore interfere with astrophysical analyses. As illustrated in Fig. 3, a major portion of the instrumental background events, including those in the backward peak, can be eliminated by a ToF selection. The most important background event types in the ToF forward peak region around ToF = 5 ns are those originating in the D1 detector (types A and C), and some of the background events produced in the satellite structure (types D, E, and F).

After ToF selection, the majority of the instrumental line background is expected to arise from activation of the D1 detector material because of the relatively high mass density and probability for triggering a background event as compared to the general spacecraft structure. The material composition of the D1 detector system therefore provides important clues as to which radioactive isotopes can effectively be produced and ultimately contribute to the instrumental line background in the low-Earth orbit of CGRO. The D1 support structure and the D1 module and PMT housings are mostly aluminium, the most abundant element in the instrument. The liquid scintillator NE 213A is composed of hydrogen and carbon. The quartz windows in the module housings contain silicon and oxygen. The PMTs and electronics boxes contain, among other elements, copper, nickel, and iron.

Figure 4: The $E_{\rm tot}$, E2, E1, and E1-E2 distribution of instrumental 2.22 MeV photons from 2D production for CDG event selections as determined from Monte Carlo simulations

Figure 5: The $E_{\rm tot}$, E2, E1, and E1-E2 distribution of instrumental 24Na events for CDG data selections as determined from Monte Carlo simulations

Events from the instrumental line background produce conspicuous and characteristic features in energy space that can be exploited to distinguish them from the general (continuum) background. The ToF distributions of the line background and the continuum background are also different: the instrumental line background is concentrated in the forward peak region, while the continuum background in energy is throughout the ToF continuum as well as the ToF forward peak. In the following, the characteristics of single photon (type A) events and multiple photon or cascade (type C) events are illustrated with examples of (background) events from ²D and ²⁴Na, respectively, both major contributors to the instrumental line background (see Sect. 4).

The instrumental 2.22 MeV photons emitted when ²D is produced in the D1 scintillators (see Sect. 4.1) are single photon (type A) background events. The $E_{\rm tot}$, E₂, E₁, and E₁-E₂ distributions of instrumental 2.22 MeV photons for CDG event selections (described in Appendix A.1) as determined from Monte Carlo simulations are depicted in Fig. 4. The $E_{\rm tot}$ distribution exhibits a peak at the energy of the primary photon, while the distributions in E₂ and E₁ are broad and relatively featureless. In E₁-E₂ space the event distribution of type A events follows the diagonal ${E}_1 + {E}_2 = {E}_\gamma$, with ${E}_\gamma = 2.22$ MeV for ²D. The distributions in $E_{\rm tot}$ and E₁-E₂ are the most important characteristics of single photon background events.

The $\beta^-$-decay of ²⁴Na results in the emission of two photons with energies 1.37 MeV and 2.75 MeV causing type C line background events (see Sect. 4.2). The $\beta^-$-particle is of minor importance for the generation of a background event. Since it rarely escapes the support structure, it contributes to the background only through secondary bremsstrahlung photons. The ²⁴Na signature in $E_{\rm tot}$, E₂, E₁, and E₁-E₂ as determined from Monte Carlo simulations is shown, for CDG event selections (see Appendix A.1), in Fig. 5. The most important characteristic of ²⁴Na, and any other isotope emitting multiple photons, is the E₂ distribution, which exhibits peaks at the energies of each of the primary photons. In E₁ the only conspicuous feature is the Compton edge of the 1.37 MeV photon (the Compton edge of the 2.75 MeV photon is suppressed by the CDG data selections). The $E_{\rm tot}$ spectrum is more complex and exhibits less pronounced, line-like features just above the individual photon energies. These come from the absorption of one photon in D2 with the other photon scattering in D1 with an energy deposit near the D1 threshold. In addition, in $E_{\rm tot}$ a shoulder is present at about 3.9 MeV. This is analogous to the sum peak in standard spectroscopy employing a single detector. Under CDG data selections, this shoulder is due to the absorption of the 2.75 MeV photon in D2, with the 1.37 MeV photon scattering in D1 with an energy deposit at the Compton edge. The ²⁴Na event distribution in E₁-E₂ space is also very characteristic: the events cluster along two bands parallel to the E₁ axis located in E₂ at the energies of the two decay photons. The band at 2.75 MeV in E₂ extends in E₁ from the threshold up to the Compton edge of the 1.37 MeV photon. The band at 1.37 MeV in E₂ is suppressed by the CDG data selections.

For other isotopes, if the energy of one of a pair of cascade photons is below the D2 threshold, then the two photons produce a cascade (type C) event only if the lower-energy photon scatters in D1, and the higher-energy photon interacts in D2. The E₂ and E₁-E₂ distribution of such a cascade photon pair is simple: there is only one photopeak in E₂ and only one band in E₁-E₂.

Radioactive decays that result in the (simultaneous) emission of two or more photons can produce line background events other than of type C. For example, an individual photon of the emitted photon multiple can produce a type A event, provided all other photons escape from the instrument without interacting in any of the detectors. As apparent in the E₁-E₂ event distribution of ²⁴Na, however, the $E_{\rm tot}$ signatures corresponding to the photon energies 1.37 MeV and 2.75 MeV are much weaker than the cascade structure. Monte Carlo simulations show that less than 10% of the ²⁴Na background events are due to type A events. This illustrates the fact that multiple photon decays in the D1 detector material are more efficient in generating a background event than decays that result in the emission of a single photon only.

Figure 6: An illustration of the spectral distributions in $E_{\rm tot}$(top) and E2 (bottom) as a function of time. The two spectra plotted with thick lines represent the times of highest (after the second reboost in May 1997) and lowest (from the beginning of the mission until the first reboost, May 1991 - Nov. 1993) contributions from long-lived background isotopes. The three intermediate spectra, plotted with thin lines, cover the time periods Nov. 1993 - Oct. 1994 (lowest), Oct. 1994 - Oct. 1995 (middle), and Oct. 1995 - May 1997 (highest)