We turn now to the most difficult part: the nature of the beast that produces the GRB-modeling of the Inner Engine. We examine a few general considerations in section 10.1 and then we turn to the Binary Neutron star merger model in 10.2.
The fireball model is based on an "inner engine" that supplies the energy and accelerates the baryons. This "engine" is well hidden from direct observations and it is impossible to determine what is it from current observations. Unfortunately, the discovery of afterglow does not shed additional direct light on this issue. However it adds some indirect evidence from the association of the location of the bursts in star forming regions.
Once the cosmological origin of GRBs was established we had two
direct clues on the nature of the "inner engine": the rate
and the energy output. GRBs occur at a rate of about one per
106 years per galaxy
[56]
and the total energy is ~ 1052 ergs. These estimates assume
isotropic emission. Beaming with
an angle 
changes these
estimates by a factor 4
/
2 in the rate
and 2 /
4 in the total
energy involved. These estimates are also
based on the assumption that the burst rate does not vary with cosmic
time. The observations that GRB hosts are star forming galaxies
[16,
129,
130,
131,
128]
indicates that the rate of GRBs may follow the star formation rate
[193,
194,
195].
In this case the bursts are further and they take place at a lower rate
and have significantly higher energy output.
The fireball model poses an additional constraint: the inner engine
should be capable of accelerating ~ 10-5
M
to
relativistic energies. One can imagine various scenarios in which
1052 ergs are generated within a short time. The requirement
that this energy should be converted to a relativistic flow is much more
difficult as it requires a "clean" system with a very low but non
zero baryonic load. This requirement suggests a preference for models
based on electromagnetic energy transfer or electromagnetic energy
generation as these could more naturally satisfy this condition (see
[267,
224,
69,
227]).
Paczynski [46]
has recently suggested a unique hydrodynamical model in which
1054 ergs are dumped into an atmosphere with a decreasing
density profile. This is a cosmological variant of Colgate's
[262]
galactic model.
This would lead to an acceleration of fewer and fewer baryons
and eventually to a relativistic velocities. Overall one could say
that the "baryonic load" problem is presently the most bothersome
open question in the "fireball model".
The recent realization that energy conversion is most likely via internal shocks rather than via external shocks provides additional information about the inner engine: The relativistic flow must be irregular (to produce the internal shocks), it must be variable on a short time scale (as this time scale is seen in the variability of the bursts), and it must be active for up to a few hundred seconds and possibly much longer [36] - as this determines the observed duration of the burst. These requirements rule out all explosive models. The engine must be compact (~ 107cm) to produce the observed variability and it must operate for a million light crossing times to produce a few hundred-second signals.
There are more than a hundred GRB models [268]. At a certain stage, before BATSE, there were probably more models than observed bursts. Most of these models are, however, galactic and those have been ruled out if we accept the cosmological origin of GRBs. This leaves a rather modest list of viable GRB models: binary neutron star mergers - NS2Ms - [35] (see also [269, 270, 53, 271, 272]), failed supernova [273], white dwarf collapse [267] and hypernova [46]. All these are based on the formation of a compact object of one type or another and the release of its binding energy. With a binding energy of ~ 5 × 1053 ergs or higher, all these models have, in principle, enough energy to power a GRB. However they face similar difficulties in channeling enough energy to a relativistic flow. This would be particularly difficult if indeed 1053 ergs are needed, as some recent burst have indicated. Paczynski's hypernova is an exception as in this model all the energy is channeled initially to a non-relativistic flow and only later a small fraction of it is converted to relativistic baryons. All these models are consistent with the possibility that GRBs are associated with star forming regions as the life time of massive stars is quite short and even the typical life time of a neutron star binary (~ 108 yr) is sufficiently short to allow for this coincidence. -
Other models are based on an association of GRBs with massive black holes associated with Quasars or AGNs in galactic centers (e.g. [274]). These are ruled out as all GRBs with optical afterglow are not associated with such objects. Furthermore, such objects do not appear in other small GRB error boxes searched by Schaefer et al. [126]. From a theoretical point of view it is difficult to explain the observed energy and time scales with such objects.