Next Contents Previous

6.1.5. The Gamma Ray Background

Diffuse gamma-ray radiation was first detected in 1972 by the OSO-3 satellite (see Kraushaar et al. 1972) and has been confirmed by a number of other satellite missions. In 1991, the Compton Gamma Ray Observatory (CGRO) was launched. On board was the Energetic Gamma Ray Experiment Telescope (EGRET) which had an order of magnitude more sensitivity and lower instrumental background than any previous gamma-ray detectors launched into orbit. The primary scientific objective of EGRET was to perform an all-sky survey for gamma-ray radiation with photon energies greater than 30 MeV. Prior to CGRO, there was a well-established correlation between the observed diffuse emission and the column density of interstellar gas in our Galaxy. However, their remained hints of residual emission along lines of sight that didn't intersect any Galactic gas thus hinting at an extragalactic origin. These hints, however, were unable to be verified by other missions due either to premature instrumental failure or just lack of instrumental sensitivity.

To date EGRET has detected very strong diffuse emission from the Galactic plane. At higher galactic latitudes, this diffuse emission is considerably weaker thus facilitating the detection of point sources. By late 1995, 36 points sources have been identified with a class of active galactic nuclei known as Blazars (see Impey 1996). Blazars are strong radio sources and often show very strong polarized optical emission. Some Blazars exhibit superluminal motion (see below) and most have high time variability of their emission (at all wavelengths). The identification of Blazars as gamma-ray sources then opens up the possibility that their redshift-smeared emission can produce an extragalactic background. However, this background has not yet been detected by CGRO as its a rather difficult measurement. The intensity of the background is low and there is no predicted spectral signature to look for. Furthermore, the instrumental background is still rather high and little can be done to correct this as the background is produced by interactions between cosmic-rays and the material the detector is made out of. In addition, a secure detection of a diffuse gamma-ray background of cosmological origin requires a very good model of the Galactic emission which needs to be subtracted out.

The current model of gamma-ray emission from the Galaxy (see Bertsch et al. 1993) is in very good agreement with the EGRET data and so it becomes a matter of understanding the instrumental response of EGRET to high precision. A preliminary analysis of many high galactic latitude fields indicates that some residual extragalactic background gamma-rays are present. The overall spectrum is a power law with a slope of -2.2 ± 0.2 but the normalization/calibration of this spectrum is not yet available from the data. In addition, EGRET confirms the presence of one spectral feature noted earlier, namely a significant enhancement of flux, above the power law fit, at energies approx 3 Mev (known as the 3 Mev bump). If we therefore assume that the gamma-ray background represents the sum of discrete sources, such as AGNs or QSOs, then both the slope of the power law and the presence of 3 Mev bump needs to be accounted for.

As a starting point, one can use the observation that discrete gamma ray sources tend to be associated with radio loud quasars. Dermer and Schlickeiser (1992) have used a linear scaling between radio flux and gamma ray flux and have integrated over the radio source counts to reach the conclusion that radio sources can not be the sole contributors to the gamma-ray background (keeping in mind that the overall background level is still uncertain). Furthermore, detailed studies of a few gamma-ray AGN have shown them to have the wrong spectral slope. Thus it seems that additional sources are required. Perhaps these are the more numerous radio-quiet QSOs and/or lower luminosity AGNs. Still, the possible existence of a source-less gamma-ray background can't yet be ruled out. One model which would produce this appeals to late matter-anti matter annihilation occurring sometime between z = 10-100. In this model (see Stecker and Salamon 1996), the 3 Mev bump is produced by the redshifted 70 Mev annihilation line associated with the decay of relativistic pions. Hence, it is clearly important to see if the data can yield a firm distinction between a discrete source background or a truly diffuse background that comes from matter-anti-matter annihilation (the latter of which would be difficult/impossible to understand from our current cosmological models).

The Mystery of Gamma Ray Bursters

Another observation which points to the importance of discrete sources is the profound mystery of Gamma Ray Bursters (GRBs) Although these have been known for 20 years, CGRO has discovered several hundred new ones using the BATSE detector. In general, the GRBs have burst durations that range from a few seconds to a few minutes. In a couple of cases, burst durations of less than a second have been detected. On March 1, 1994 a very bright burst was detected. A search for a radio counterpart of that burst was made for approx 100 post-burst days with no detection recorded (see below). To date, no GRB has yet had an observed counterpart at other wavelengths. Part of this is due to the low positional accuracy (only about 10 degrees) that BATSE is capable of providing.

The real mystery of the GRBs is revealed in Figure 6-2 which shows the distribution of approx 1000 GRBs plotted in galactic coordinates. Here it can be plainly seen that the distribution of sources is isotropic in nature. Even though the Galaxy is known to be a strong source of gamma-ray emission, the burst phenomenon does not appear to be associated with these processes. To explain the isotropic nature of these sources, three classes of models have been suggested: 1) The Oort-cloud model in which GRBs are actually objects spherically distributed around our own solar system; 2) they are distributed in a very extended spherical halo around the Galaxy; 3) they are of cosmological origin. This failure to even know the distance scale that corresponds to the GRBs by 12 orders of magnitude qualifies them as being the one astrophysical phenomenon that we know least about!

Figure
 6-2

Figure 6-2: Spatial Distribution of detected gamma ray bursters. The isotropic distribution strongly implies a cosmological origin. This figure was current as of June 1997 and is courtesy of NASA and the Compton Observatory Science Support Center.

The Oort-cloud model suffers from the lack of any known reasonable physical object which might produce gamma-rays (evaporating mini-black holes have been suggested but too high of space density seems to be required). The extended halo model appeals to collisions of objects with neutron-stars to generate the gamma-rays. Such a situation 1) requires a velocity dispersion of the halo which is considerably larger than observed and 2) suggests an anisotropy in the source counts should be observed towards nearby galaxies like M31 as a similar process would be occurring in its halo. Thus we are left a cosmological origin as the most reasonable(!) of the hypothesis.

Consistent with a cosmological origin is the observation that, in addition to their isotropic distribution, there are more bright bursts than faint bursts. This suggests that the distribution of these objects is bounded by some limit and that we are near its center. The lack of fainter bursts argues that the more numerous and more distant objects yield fluxes that are below the detection threshold of the instrument. A cosmological origin would suggest that galaxies are the likely hosts of GRBs. In this case, many models predict that post-burst flux-densities at cm wavelengths of a few 10's of milli- Janskys should appear in a few weeks following the initial burst. So far, no positive detections have occurred including the best case of the March 1,1994 outburst whose field was observed at radio wavelengths for 100 post-burst days (see Dessenne et al. 1996; Koryani et al. 1995; Frail et al. 1994).

Because of the poor angular resolution of the BATSE detector and the subsequent positional uncertainty of a few degrees, deep imaging studies that look for the optical counterpart are not practical as there will be literally millions of candidates to choose from over the field of view corresponding to the GRB position. To optimize the detection of GRB host galaxies requires accurate burst localizations. These can be obtained by using a widely separated network of satellites that are sensitive to gamma-ray bursts (in addition to CGRO there are a number of military satellites available that use this signature to search for nuclear test-ban violations), using precise time-of-arrival measurements. Through triangulation, positional error boxes of a few arcminutes can be realized. Larson et al. (1996) have made deep near-infrared measurements of the six GRB error boxes which have the lowest positional uncertainties as determined in this manner. Each of the six fields does contain an obvious galaxy whose angular size and flux level is consistent with it being located at high redshift and hence is the candidate host galaxy of the GRB. If galaxies are the hosts of GRBs, what then is the physical source of the radiation?

There seem to be two plausible physical mechanisms that could generate GRBs. The first appeals to collisions between neutron stars (ns). The burst is a manifestation of a thermonuclear runaway on the surface of the neuron star. Alternatively, a burst could be generated as the signature of the release of large amounts of gravitational binding energy. Either kind of event should produce optical and soft X-ray photons as the higher-energy radiation produced by the burst interacts with the surrounding medium and is absorbed and re-radiated as lower energy photons. This provides the motivation for searching for optical/soft X-ray counterparts. The main question about this method of producing GRBs revolves around the environmental requirements that a galaxy must have to facilitate ns-ns collision (or a collision of another star with a ns).

The second source of GRB appeals to a mechanism which is already known to be operative. There exists a class of about 20 objects discovered to date which are known as superluminal radio sources. These objects are galaxies with active nuclei that emit bipolar jets of material. From our observational perspective, the time evolution of the angular separation of the jets on the plane of the sky apparently requires faster than light motion. This paradox can be understood if are looking down essentially a beam pipe of relativistic out-flowing material as explained in Chapter 2. To date, practically all of the known superluminal sources have been detected by EGRET (the non-detections are thought to be the result of "bent" jets - von Montigny et al. 1995). so we know that relativistic beaming and the associated Lorentz boosting of photon energies can produce gamma rays. In addition, this relativistic jet of material should contain electrons and positrons which will annihilate and produce the 0.511 MeV annihilation line. In the observer's frame, this radiation can be greatly blueshifted through the motion of the jet and then later redshifted via universal expansion. The superposition of many of these beamed jets could produce the observed 3 Mev bump in the gamma-ray background. Since the acceleration mechanism of these jets is still unknown (one plausible model appeals to radiation pressure that accelerates a plasma to a bulk relativistic velocity) it is difficult to understand how this process might produce an actual burst of Gamma rays which is significantly above their continuous production. More to the point, if relativistic beaming by distant extragalactic sources is the main producer of GRBs then the intrinsic number of GRBs in the Universe must be quite large as only a fraction of these relativistic beam pipes will point at the earth. One then has to wonder what the effect of a large space density of these "plasma guns" is on galaxies that are trying to form at high redshift and might be located near one.

At present, there are over 100 theories (see Appendix) that purport to explain GRBs (obviously theorists love unexplained phenomena) and to this we add one more here. The principle requirement of these theories is to produce an isotropic background. Within the gravitational instability paradigm framework, there may exist a natural system for the production of GRBs that is associated with structure formation. Recall that the Jeans Mass at the time of recombination is similar to that of a globular cluster. Let's assume that globular clusters with an order of magnitude more density than those presently observed, formed at z approx 10-50. What would be the dynamical fate of these clusters? Such clusters would have very short dynamical timescales and collapse and form stars quickly. The key is to have a system with a relaxation timescale which is approximately the same as the stellar evolutionary timescale. This would be approx 107 years. Spitzer and Hart (1971) parameterize the relaxation time for globular clusters as

Equation 2   (2)

An initial cluster of N approx 105 and M approx 105 Msun which is confined to a radius of rh approx 0.5 pc (corresponding to rho approx 10-20 g cm-3) gives a relaxation time of approx 107 years. This is the timescale over which equipartition of energy occurs and the massive stars sink to the center of the cluster and the lower mass stars are heated,possibly to escape velocity. Thus, as the core collapses due to relaxation (and dynamical friction) the ns remnants naturally sink to the center thus greatly increasing the probability of a ns-ns binary (one of these is known to exist in our galaxy). Energy loss from gravitational radiation of the ns-ns binary system will eventually cause them to merge and possibly produce a GRB. The rest of the cluster stars would have evaporated away by this time. Hence, the equations of stellar dynamics tell us that if a sufficiently dense perturbation is present to form a dense stellar cluster, the resulting match between the dynamical relaxation timescale and the stellar evolutionary timescale produces a natural breeding grounds and confinement mechanism for ns-ns binaries. As this is expected to occur at very high redshift, before any clustering sets in, the distribution of these objects would be isotropic and there would be no surviving remnant today as the lighter stars would have evaporated. Pretty cool, huh?

Next Contents Previous