Sursauts gamma: les gros titres ne disent pas tout

La découverte des sursauts gamma date de la fin des années 60, mais leur origine est longtemps restée mystérieuse. Les récentes observations du satellite Swift de la NASA ont donné lieu à de gros titres dans de prestigieuses publications, affirmant que le mystère vieux de 35 ans des sursauts gamma avait finalement été résolu dans le cas des sursauts brefs; les sursauts longs, par contre, présenteraient un comportement inattendu. Mais ce comportement est-il vraiment inattendu et le mystère est-il réellement dissipé?

Gamma-ray bursts (GRBs) - intense but brief flashes of gamma rays - were first discovered accidentally by US military satellites in 1967, and have since become a major puzzle for astrophysics. By 1992, however, observations mainly with the Burst and Transient Source Experiment (BATSE) on-board NASA's Compton Gamma Ray Observatory, had provided compelling evidence that GRBs originate mostly at large cosmological distances and, moreover, divide into two distinct classes: short hard-spectrum bursts (SHBs) with a typical duration of less than one second, and long soft bursts that typically last longer than two seconds. However, the nature of the GRBs remained a mystery.

A significant breakthrough came when the Italian and Dutch space agencies put BeppoSAX into orbit in 1996. This X-ray satellite localized GRBs in its field of view with arcminute precision and led to the discovery of X-ray, optical and radio afterglows for long-duration GRBs. These afterglows faded relatively slowly and enabled subarcsecond localization of the long GRBs, as well as measurement of their cosmological redshifts and absolute brightness, identification of their star-forming galaxies and finally their progenitors - ultrarelativistic jets ejected from supernova explosions due to the core collapse of massive stars (see CERN Courier June 2003 p5 and p12). Yet despite this impressive progress, many important questions regarding long GRBs remained unanswered. What type of core-collapse supernova produces them? What sort of remnant is left over? What is the true production mechanism? Moreover, despite extensive searches no afterglow was detected for the SHBs, and their redshifts, intrinsic brightness, host galaxies and progenitors remained unknown.

This situation has changed dramatically in the past few months after the successful launch in November 2004 of Swift, NASA's multi-wavelength observatory dedicated to the study of GRBs. Its main missions are to detect GRBs, measure their properties, localize their sky positions with sufficient precision shortly after detection, and communicate these positions automatically to other space- and ground-based telescopes in order to discover and follow up the afterglows in a broad range of wavelengths, soon after the beginning of the bursts. By the end of September 2005, Swift had detected and localized 70 GRBs. Three of these - 050509B, 050724 and 050813 - were SHBs and follow-up observations have discovered elliptical host galaxies at redshifts 0.225, 0.258 and 0.722, respectively. Shortly after Swift's detection and localization of SHB 050509B, NASA's High Energy Transient Explorer satellite, HETE-2, which had been launched in 2000, detected and localized another SHB, 050709, on 9 July 2005 (Gehrels et al. 2005 and Villasenor et al. 2005). Follow-up measurements have found and measured its X-ray and optical afterglows, which led in turn to the discovery of its host - a star-forming young galaxy at redshift 0.16 (Hjorth et al. 2005 and Fox et al. 2005).

The observed brightness and energy fluence, and the measured redshifts of the SHBs imply that their intrinsic brightness is smaller than that of typical long GRBs by two to three orders of magnitude. Moreover, their inferred total emitted radiation, assuming isotropic emission, is smaller by four to five orders of magnitude. So it is quite possible that SHBs are seen at relatively small redshifts because they are intrinsically faint and cannot be seen from large cosmological distances. However, it is not clear why around 20% of the bursts observed by BATSE are SHBs, but only 5% of those seen by Swift.

Has the mystery been solved?

These observations have led to recent press releases by NASA and some prestigious universities, and the publication of articles in astrophysical journals and in Nature, Science and Scientific American, which claim that "the 35 year old mystery of GRBs" has finally been solved and that SHBs have been proven to be produced by the merger of neutron stars or of a neutron star and a stellar black hole in close binary systems. But is this so?

The relatively small redshifts of SHBs and their association with both star-forming spiral galaxies and elliptical galaxies containing mainly old stars appears consistent with their origin in the merger of neutron stars in binary systems, as the merger usually takes place a long time after the formation of the neutron stars. The idea is that a large number of the neutrinos and antineutrinos that are emitted in the merger collide with each other outside the merging stars and annihilate into electron-positron pairs, which form a fast expanding fireball that produces the GRB (Goodman et al. 1987). Later, it was suggested that instead of spherical fireballs, mergers produce highly relativistic jets along the rotation axis, which can produce shorter and brighter GRBs through, for example, inverse Compton scattering of ambient light around the merging stars.

At first sight the merger scenario seems consistent with the observations, but a more careful examination raises serious doubts. The cosmic rate of such mergers as a function of redshift can be calculated from general relativity using the observed properties of galactic neutron-star binaries and their production rate, which must be proportional to the measured star formation rate. Despite the small statistics the redshift distribution of the SHBs detected by Swift and HETE-2 appears inconsistent with the theoretical expectations from the merger model.

A second problem concerns an X-ray flare observed by the Chandra X-ray Observatory in the afterglow of SHB 050709 on day 16 after the burst. In the fireball models of GRBs, X-ray flares in the afterglow are interpreted as due to "re-energization" of the afterglow by the central engine (Zhang and Meszaros 2004). The final merger in a neutron-star binary due to gravitational wave emission, however, takes place in less than a millisecond and produces a black hole. It is hard to imagine that the remnant can "re-energize" the X-ray afterglow after 16 days, a time scale one billion times larger than a millisecond. On the other hand, in the alternative "cannonball" model of GRBs, X-ray flares are produced when the highly relativistic jets from the central engine (in this case mass accretion on a compact object) encounter density changes in the interstellar medium (Dado et al. 2002). Indeed, SHB 050709 took place not far from the centre of a galaxy where star formation produces strong winds and density irregularities.

Other scenarios for SHB production have been dismissed as unfavoured by the observations, but this may have been premature. Accretion-induced collapse of neutron stars in compact neutron-star/white-dwarf binaries is consistent with all the observations. Origin in a supernova collapse was ruled out for 050509B and 050709 by follow-up measurements with powerful optical telescopes, but only for these SHBs; much larger statistics are needed to conclude that SHB production in a type Ia supernova is unlikely. Origin in soft gamma-ray repeaters (SGRs), which are anomalous pulsars that occasionally produce GRBs, was ruled out by the claim that they are too faint to be observed at the measured redshifts of SHBs. Consider, however, the burst emitted on 27 December 2004 by the galactic SGR 1806-20. It was the brightest GRB ever recorded from any astronomical object, beginning with a short 0.2 second spike that was followed by a much longer and dimmer tail modulated by the 7.55 second period of the pulsar. Had it taken place in a distant galaxy, the spike, if detected, would have been classified as an SHB.

The maximum distance from which such a spike could be detected depends on the uncertain distance of SGR 1806-20 and, if it was relativistically beamed into a small solid angle like ordinary long GRBs, on its viewing angle. A viewing angle three to four times smaller than that for the spike presumably beamed from the superburst of SGR 1806-20 would be sufficient to make it look like a normal SHB at a redshift of z ∼ 0.25 (Dar 2005). Moreover, SGRs may be born not only in core collapse supernova explosions but, for example, in the accretion-induced collapse of white dwarfs, as suggested by the fact that, despite their young age, only one of the four known SGRs is located inside a supernova remnant. This may explain why SHBs are produced both in elliptical galaxies with old stellar populations and in star-forming spiral galaxies with young stellar populations.

Unsurprising behaviour

Because of its higher sensitivity, Swift can see deeper into space than any previous gamma-ray satellite. Indeed, the 14 long GRBs localized by Swift for which a redshift has been reported have a mean redshift of  = 2.8. This is twice the mean of  = 1.4 for the 43 GRBs with a known redshift that have been localized by BeppoSAX, HETE-2 and the interplanetary network over the past seven years.

Swift record redshift so far is z = 6.29, for GRB 050904, which looks like an ordinary burst with an ordinary afterglow (Haislip et al. 2005). This redshift is comparable to that of the most distant quasar measured to date, and in the standard cosmological model it corresponds to a look-back time of nearly 14 billion years, to when the universe was only one billion years old. Thus this single GRB already indicates that star formation and core-collapse supernova explosions took place at this early cosmic time, and together with previous measurements shows that the rate of star formation has not declined between z = 1.4 and z = 6.29. It also demonstrates that long GRBs and their optical afterglows, which are more luminous than any known astronomical object by many orders of magnitude, can be used as excellent tools for studying the history of star formation, galaxies, and intergalactic space since the time of the early universe.

The X-ray telescope on-board Swift also recorded, in fine detail, the evolution of X-ray afterglows of a dozen or so GRBs, from right after the burst until they became undetectable. Most of these X-ray afterglows demonstrate a universal behaviour: an initial fast fall-off within the first few minutes followed by a shallow decline over the next few hours, which afterwards steepens gradually into a much faster power-law decline (see figure 1). In some cases Swift has also detected X-ray flares superimposed on this universal behaviour. These observations were presented in Nature and Science as complete surprises that cannot easily be explained by current theoretical models of GRBs (see CERN Courier October 2005 p11). This may be true for the popular fireball models of GRBs, but it is not true for the cannonball model, which correctly predicted these effects long ago.

The fast initial fall-off and the gradual roll over of the shallow decline to a later power-law decline were in fact already indicated by observations in 1998. Figure 2 shows the comparison with these observations of the universal behaviour predicted from the cannonball model in 2001. Moreover, an X-ray flare had also already been seen in 1997 by BeppoSAX in the afterglow of GRB 970508 (Pian et al. 2001) and can also be explained in the cannonball model.

In the cannonball model, the early X-ray afterglow originates in thin bremsstrahlung from a rapidly expanding plasmoid - the cannonball - which stops expanding within a few observer minutes after ejection. Synchrotron emission from the ionized interstellar electrons, which are swept into the decelerating cannonball, then takes over. The shallow decline followed by the roll-over into a power-law decline is a simple effect of off-axis viewing of decelerating jets in the interstellar medium, which has been observed in many optical afterglows but misinterpreted by fireball models. The flares are caused by collisions of the jet with density jumps in the interstellar medium produced by stellar winds and supernova explosions.

In conclusion, it seems that the localization of SHBs by Swift and HETE-2, which led to the discovery of their afterglows, the identification of their host galaxies and the measurements of their redshifts, have been over-interpreted. While these are undoubtedly observational breakthroughs, the origin of SHBs is still an unsolved mystery. Nevertheless, the small redshifts of SHBs are good news for gravitational wave detectors such as LIGO and LISA, in particular, if SHBs are produced mainly by mergers of neutron stars or a neutron star and a black hole in binaries, as first suggested in 1987. Moreover, the observed behaviour of the early X-ray afterglows of long GRBs and the X-ray flares - both claimed to be a complete surprise and unexpected in the fireball models - were predicted correctly long ago, like many other features of long gamma-ray bursts, by the cannonball model.

Further reading

S Dado et al. 2002 A&A 388 1079.

A Dar 2005 http://arxiv.org/abs/astro-ph/0509257.

D B Fox et al. Nature 2005 437 845.

N Gehrels et al. 2005 Nature 437 851.

J Goodman et al. 1987 ApJ 314 L7.

J Haislipet al. 2005 http://arxiv.org/abs/astro-ph/0509660 .

J Hjorth et al.. 2005 Nature 437 859.

E Pian et al.. 2001 A&A 372 456.

J S Villasenor et al.. 2005 Nature 437 855.

B Zhang and P Meszaros 2004 J. Mod. Phys. A 19 2385.