D. A. Frail (National Radio Astronomy Observatory, USA/Caltech University, USA); P. A. Price (Research Schoool of Astronomy and Astrophysics/Caltech University, USA); E. Berger (Caltech, USA); M. Wieringa, R. Subrahmanyan, R. Wark (ATNF)

Gamma-ray bursts may be the most luminous events in the Universe but they have not given up their secrets easily. For the past several years we have been part of an international effort to unveil these mysterious sources responsible for such explosive events. Now, thanks to observations made at the Australia Telescope Compact Array and elsewhere, we have found a direct link between gamma-ray bursts and the death throes of massive stars.

Any successful gamma-ray burst model must be capable of releasing an enormous amount of energy on a time scale of tens of seconds. One of the more promising candidates has been the collapsar model, in which a massive star, more than 20 times the mass of our own Sun, ends a lifetime of nuclear burning and undergoes core collapse, forming a black hole at its centre. The outer layers of the star are driven outward at speeds of 20,000 to 60,000 kilometres per second in a brilliant supernova explosion. Not to be outdone, the newly formed black hole, powered by infalling material, forms collimated beams of material along its rotation axis at speeds approaching that of light (Figure 1). The interaction of different shocks within the jet is responsible for the gamma-ray emission, while particle acceleration from the ultra-relativistic shock which is driven into the circumburst medium gives rise to long-lived X-ray, optical and radio afterglow emission, dwarfing the light from the supernova.

For some time now evidence has been slowly accumulating in favour of the collapsar model. All welllocalised gamma-ray bursts occur in host galaxies which are actively undergoing star formation, sometimes at a rate of several hundred solar masses per year. Likewise, the radial distribution of gammaray bursts closely follows the (stellar) UV light from their host galaxies. Substantial gas and dust along the line-of-sight to gamma-ray bursts has also been inferred from the absorption of low-energy X-rays and extinction of optical afterglows. In many cases the dust extinction is large enough to produce optically dark gamma-ray bursts which are visible only in the radio and X-ray bands. While these indirect indicators have been telling us that gamma-ray bursts originate in the same dusty, gas-rich environments as massive stars, the decisive evidence in favour of the collapsar model has remained elusive.

Fortunately, the collapsar model provides two powerful observational tests. The massive star progenitor undergoes prodigious mass loss prior to collapse, shedding nearly all of its hydrogen envelope. The relativistic shock driven outward must propagate through this gas and hence we expect to see the mass-loss density profile imprinted on the spectral and temporal evolution of the afterglow emission. Observations at radio wavelengths are especially important in detecting the wind signature since they probe the rise and fall of the afterglow emission as the shock propagates through this density gradient.

Another inevitable consequence of the collapsar model is that a supernova explosion will occur simultaneously with the gamma-ray burst. The light from the afterglow and the supernova can be distinguished from each other since the former undergoes a pure power-law decay, while the latter exhibits a rise to maximum several weeks after the burst, followed by an exponential decay. Similarly, the optical spectrum of the afterglow is a featureless (synchrotron) continuum, whereas the supernova spectrum has characteristic red colours. Detecting the weak supernova signal in the presence of the afterglow and the host galaxy is not an easy task. Prior claims of late-time red bumps and stellar wind signatures have been made but not together and not with a high degree of confidence. With the gamma-ray burst of 21 November 2001 (also known as GRB 011121) we had our long-awaited opportunity to search for both.

GRB 011121 began innocuously enough. It was detected as a 30-second long burst at 18:47 UT by the Dutch/Italian satellite BeppoSAX, and the subsequent ground-based response by our group and others quickly identified the optical and radio afterglows. A pleasant surprise awaited us when an optical spectrum, taken at the Baade 6.5-m telescope, showed that the redshift of this burst was z = 0.36, making GRB 011121 the nearest cosmological gamma-ray burst known to date. By contrast, most of the two dozen bursts with distance determinations lie at redshifts between one and two. Since nearby gamma-ray bursts are ideal for carrying out tests of the collapsar model we immediately began a large observing campaign on GRB 011121.

Broadband optical observations were undertaken at the Anglo-Australian Telescope (AAT) with the newly commissioned IRIS2 instrument, and the du Pont 2.5-m and Baade 6.5-m telescopes at Las Campanas Observatory. These observations enabled us to determine the spectrum and the power-law decline of the optical afterglow at early times. When the light curves were extrapolated to a time between 15 and 75 days after the burst we found that our Hubble Space Telescope measurements taken at this time were an order of magnitude brighter than expected. Both the light curve of this flux excess and its observed spectral shape were consistent with the expectations for a supernova of type Ib or Ic.

Meanwhile at the Compact Array we had made seven epochs of observations of the afterglow at 4.8 and 8.7 GHz, spanning from one day after the burst to 70 days later. By comparing the early evolution of the optical afterglow with that of the radio afterglow, we were able to eliminate all other potential afterglow models, including an isotropic or highly collimated outflow expanding into a constant density medium. Taken together the optical and radio data are consistent with a gamma-ray burst exploding into a windblown circumburst medium. Moreover, careful modelling of these data tell us that the mass loss of this pre-supernova star was 10^-7 solar masses per year, typical of evolved stars.

If all long-duration gamma-ray bursts are due to the core collapse of massive stars, as these observations certainly suggest, then we are likely witnessing the birth of a stellar black hole. Looking ahead, this raises the exciting possibility that in the future gamma-ray bursts could be used to trace star formation when the Universe was very young. The extreme luminosity of gamma-ray bursts and their accompanying afterglows means that in principle they could be detected out to redshifts of 10 or higher, at a time when the first stars were still being formed.

back to highlights page