5.4. Gamma-ray bursts

Gamma-ray bursts were first detected by the military Vela satellites in 1967. These are short flashes of gamma-rays (typically with a peak energy around 100 keV), lasting between a few milliseconds and tens of minutes. Already in the early 1990s, the BATSE experiment on board the Compton Gamma-Ray Observatory demonstrated that the rate of gamma-ray bursts is about 1-2 per day, and that they are distributed isotropically in the sky. The data from BATSE also showed that gamma-ray bursts (GRBs) come in two distinct classes: (i) short-duration ( 2 sec) with hard spectra, and (ii) long duration with softer spectra (Kouveliotou et al. 1993, Fishman 2001).

Due to the relatively poor localization capabilities of gamma-ray detectors (which did not allow identification of the sources in other wavebands), two thousand bursts have been detected before it became possible to determine the distance to the bursts. The isotropic sky distribution argued for either a cosmological origin or an extremely local one (e.g., the Galactic halo). This situation changed dramatically since the operation of the Italian-Dutch BeppoSAX satellite in 1997. BeppoSAX was able to determine the positions of bursts to within arcminutes. The discovery of rapidly declining "afterglows" in the X-ray, optical and radio bands (Costa et al. 1997, van Paradijs et al. 1997, Metzger et al. 1997, Frail et al. 1999) allowed for redshift determinations that immediately placed GRBs at cosmoligical distances. I should note that to date, only afterglows of the long-duration bursts have been observed. Thus, it is not even clear if the short bursts produce afterglows (the best limit to date, for the burst GRB 020531, failed to detect an afterglow candidate [down to V ~ 25] about 20 hours after the burst; Salamanca et al. 2002). Typically, the afterglows decay with time as t^-, with in the ranges 1-2. This behavior was predicted by a model in which a fireball is expanding into a homogeneous external medium (Mészáros and Rees 1997).

At the time of this writing redshifts have been determined to more than two dozen GRBs, and they usually lie in the z = 0.5-1.5 range (although redshifts as high as 4.5 have been recorded). The observed fluxes imply energies of up to 10⁵⁴ ergs for isotropic emission. However, there is increasing evidence, in the form of kinks in the afterglow light curve, and in polarization detected in a few bursts, that gamma-ray bursts are in fact collimated into narrow jets. If the observer's line of sight is within the jet solid angle, _j, then as long as the Lorentz factor satisfies _j^-1/2, the light-cone is within the jet boundary. However, as the jet decelerates, eventually drops below _j^-1/2. Consequently, the (transverse) emitting area starts to grow more slowly (as r_{|| j}^-1/2 instead of (r_|| / )²), resulting in a break (faster decay) in the light curve (Rhoads 1997, Mészáros and Rees 1999, Livio and Waxman 2000), consistent with observations in GRB 990123 (Kulkarni et al. 1999, Fruchter et al. 1999). When collimation is taken into account, the average total energy of GRBs is estimated to be of the order of 2 × 10⁵¹ ergs (Frail et al. 2001, Panaitescu and Kumar 2001).

Based on the energetics, and the fact that the production of astrophysical jets typically relies on the collimation and acceleration provided by an accretion disk around a compact object (Livio 2000), the most popular models for GRBs involve the formation of stellar-mass black holes, surrounded by a debris torus from which mass accretion onto the central object occurs (Mészáros 2002). The two most likely progenitors to produce such a configuration are the core collapse of massive stars (Woosley 1993, Paczynski 1998), or the ongoing merger of neutron star-neutron star or black hole-neutron star binaries (e.g., Paczynski 1986, Goodman 1986, Eichler et al. 1989). In both cases the main source of energy is gravitational, even though tapping into the (large) spin energy of the black hole is also possible in principle (Blandford and Znajek 1977, Krolik 1999b). A schematic describing how a GRB and its afterglow may be generated by internal and external shocks (respectively) in a collimated jet arising from a stellar collapse, is shown in Figure 22.

Figure 22. Schematic gamma-ray burst from internal shocks and afterglow from external shock, arising from a relativistic jet produced by the collapse of a massive star. Internal shocks produce -rays and neutrinos, external shocks produce -rays, X-rays, optical and radio emission. Adapted from Mészáros (2002).

In particular, the supernova SN 1998bw (of the relatively rare type Ic, caused by the explosion of a massive star that had rid itself of its hydrogen envelope prior to exploding; Filippenko and Sargent 1986) was found to be approximately coincident in both position and time with the relatively weak GRB 980425, at redshift z = 0.0085 (Galama et al. 1998). Furthermore, the supernova light curve was found to be consistent with the formation of a black hole [Iwamoto et al. 1998). In a few other bursts (e.g., GRB 011121) the optical and near infrared afterglow light curve exhibits a "bump" with a time delay and amplitude that are consistent with resulting from a supernova explosion occuring simultaneously with the GRB (e.g., Greiner et al. 2001).

The contributions of HST to this exciting field have been in a few areas.

First, by resolving the host galaxies and being able to pinpoint the GRB on the host, HST has shown unambiguously that at least some GRBs are not associated with galactic nuclei. In fact, they sometimes occur far from the nucleus or in spiral arms (Sahu et al. 1997, Andersen et al. 2002). Second, the Hubble Space Telescope has shown (Fruchter et al. 1999, Fruchter 2002) that the colors of the hosts of GRBs corresond to the bluest colors observed for galaxies in the Hubble Deep Fields (see Section VI). This means that GRBs with afterglows occur preferentially in galaxies with high star formation rates - a finding that is consistent with GRBs being associated with collapses of massive stars. Furthermore, the fact that in almost all cases the GRB's position was found to be within the extent of the rest-frame ultraviolet image (where star formation is intense) of the hosts, also argues in favor of the "collapsar" model for the long-duration bursts. Neutron star binaries, on the other hand, are born with "kicks" of as much as a few hundred km/sec, which could drive them out of the star-forming regions.

The Hubble Space Telescope also helped to firm the association of at least some GRBs with supernova explosions. In GRB 011121, for example, HST detected an intermediate-time flux excess ("red bump") that was redder in color relative to the GRB afterglow. This "bump" could be well described by a redshifted Type Ic supernova that exploded approximately at the same time as the GRB (Bloom et al. 2002, Garnavich et al. 2003). Near-infrared and radio observations of the afterglow further provided evidence for extensive mass loss ( ~ 2 × 10^-7 M yr^-1) from the massive stellar progenitor of GRB 011121 (Price et al. 2002).

To conclude this topic, it is very likely that every time a GRB goes off, a black hole is born. Gamma Ray Bursts therefore offer, in principle at least, a direct measure of the massive-star formation rate in the early universe. Future observations with the HETE-2 spacecraft and with the Swift multi-wavelength GRB afterglow mission (equipped with -ray, X-ray and optical detectors), as well as follow-ups with HST and other observatories from the ground and space, will hopefully reveal the true nature of these most dramatic explosions (including the short-duration bursts).