First Identification of Host Galaxies for Short Gamma-Ray Bursts

It’s hard to learn much about a celestial explosion if you know neither its host environment nor how far away it was. For a given apparent brightness, distance determines energy output and the environment suggests some mechanisms and precludes others. For the majority class of gamma-ray bursts (the so-called long GRBs, which last longer than a few seconds), the breakthrough came in 1997, when the first detection of optical and x-ray afterglows made it possible to pinpoint them to young star-forming galaxies at cosmological distances measurable by redshift. Nowadays it is widely accepted that long GRBs are caused by unusually energetic supernova explosions of massive young stars (see Physics Today, August 2005, page 21 ).

There is, however, a distinct minority class of GRBs—those with burst durations shorter than two seconds—for which the determinations of distances and hosts are only now becoming available. The 6 October issue of Nature reports the first two localizations of short GRBs—one that erupted on 9 May of this year ¹ and the second ^2–4 on 9 July. A subsequent issue will report on the 24 July short GRB, the third to be successfully localized. ^5,6

What do these first few localizations tell us about the intrinsic differences between the short and long GRBs? The short GRBs appear to be less luminous—in prompt gammas and in x-ray, optical, and radio afterglows—than their longer-duration cousins by two or three orders of magnitude. Also, whereas the long GRBs are always found in populations of young stars, typically in spiral galaxies glowing with active star formation, the three short GRBs were found in quite different surroundings: Two appear to come from elliptical galaxies dominated by old stars, ^1,5,6 and the other was pinpointed to a non-star-forming neighborhood within an elderly spiral galaxy. ⁴ Another important distinction is that the three short GRBs, despite being an order of magnitude closer than most long GRBs, show no sign of the supernovae that are almost always found in conjunction with the closest of the long GRBs.

Such different observational character strongly suggests a totally different astrophysical origin for the short GRBs. All the evidence, thus far, favors the presumption that a short GRB heralds the final spectacular merger of a matched or mixed pair of compact stellar objects—neutron stars or black holes—that have been orbiting each other for 10⁸ or 10⁹ years.

Among the few otherwise plausible alternatives, magnetar disruptions—catastrophic magnetic rearrangements of very young neutron stars—are excluded because the short GRBs, with energies exceeding 10⁴⁸ ergs, are too energetic. Furthermore, their stellar neighborhoods are too geriatric to harbor magnetars (see Physics Today, May 2005, page 19 ). The old-star populations also argue against the supernova origin of short GRBs, as does the brevity of the bursts and the absence of optical evidence of associated supernovae.

Swift and HETE

Why have astrophysicists had to wait until this year for the localization of short bursts? Whereas a short GRB typically lasts only a fraction of a second, long bursts usually continue for tens of seconds. The first, and crudest, directional information from a GRB comes from a gamma-burst detector aboard an orbiting satellite. The longer the burst’s duration, the better the initial localization, which serves to point x-ray and optical telescopes in roughly the right direction for successive refinement. The transient afterglows they see can, in the best cases, locate the source to within 0.1 arcsecond.

Since 1997, the Italian—Dutch BeppoSAX or-biter and NASA’s High Energy Transient Explorer (HETE), in conjunction with followup observations of afterglow by larger telescopes, had localized more than a hundred long GRBs—but not a single short one—to their host galaxies.

The launch of NASA’s Swift satellite last November began a new observational era. Among Swift’s principal purposes was the prompt localization of short GRBs. To that end, it carries an x-ray telescope (XRT) that can be slewed in less than a minute to point toward a GRB recorded by BAT, the orbiter’s wide-field burst-alert gamma telescope. Such prompt response, it was hoped, would allow the XRT to detect and locate a rapidly fading x-ray afterglow with sufficient accuracy to direct much bigger orbiting and ground-based telescopes to the appropriate patch of sky.

The 9 May short GRB, the first and weakest of the three, was recorded by Swift and analyzed by a team led by Neil Gehrels (NASA Goddard Space Flight Center). ¹ Over the following days, bigger telescopes scrutinized the local region of sky defined by the handful of x-ray photons recorded by the XRT in two hours after the 40-ms gamma burst. They found no longer-lasting afterglow at x-ray, optical, or radio wavelengths. But the XRT localization, by itself, identified the probable host as a large elliptical galaxy at redshift z = 0.22. That implies a distance of about 3 billion light years. The average redshift of the long GRBs localized since 1997 is ten times larger.

The two short GRBs discovered in July were brighter and therefore more informative than their May predecessor, even though their redshifts imply that all three were about equally distant. The 9 July short GRB was discovered and localized by the HETE team, led by George Ricker of MIT. ² The five-year-old HETE satellite carries both gamma and x-ray detectors, but it does not have Swift’s rapid-slewing capability. By a stroke of fortune, however, HETE’s x-ray imagers were pointing in just the right direction to localize the burst.

In addition to the 100-ms hard-gamma peak, HETE recorded a surprising broad second peak of x rays that lasted for more than 100 seconds. Because that extended peak was too faint to have been detected by the earlier GRB satellites, it doesn’t compromise the designation of the burst as a classic short-duration GRB.

The ultimate localization that pinpointed the 9 July burst unambiguously to a quiet corner of an old spiral galaxy at z = 0.16 resulted from a symbiosis between the initial HETE localization and afterglow measurements over succeeding days by the orbiting Chandra X-Ray Observatory , ground-based telescopes, and the Hubble Space Telescope (see figure 1). ^3,4

Collimation

The burst’s light curve—the record of how afterglow fades with time—observed by HST yields two important conclusions. First, it convincingly precludes the existence of a supernova associated with the GRB. And second, a sudden steepening of the light curve after a week suggests that the 9 July GRB’s emission of radiation and relativistic ejecta is collimated into a narrow jet with opening angle about 15°. Such an abrupt change in light-curve slope is taken to be a relativistic effect of light generated by a narrow cone of high-speed ejecta slowing down in the ambient medium.

There is abundant evidence of the collimation of long GRBs. If most short bursts are similarly collimated, the relative observed brightness for a given distance straightforwardly implies that the short bursts release at least a hundred times less energy in gammas than do the long bursts. And it implies that there are many more short GRBs out there than the small fraction that happen to be beamed toward us.

The 24 July event, detected by Swift, ⁵ provides a striking look at details of the compact-object merger that presumably triggered it. Figure 2 shows BAT’s record of gammas detected in the first few minutes. A hard-gamma spike at 0.1 s is followed by softer-gamma emission with a second prompt peak at 1.1 s and a faint late enhancement—similar to the one HETE found—centered at about 80 s.

The burst’s x-ray afterglow, recorded by Swift’s XRT and eventually by Chandra, allowed ground-based telescopes to find optical and radio afterglows and confirm that the host was an old elliptical galaxy at z = 0.26 with very little star formation. ⁶

Surpisingly, the steady fading of the x-ray afterglow after the first two seconds was interrupted by three flare-ups. The first, similar to HETE’s unexpected finding, peaked at about one minute, and the last came six hours later. “Such an extended scenario of activity is hard to reconcile with model simulations of a fast, clean merger of two neutron stars,” says Gehrels. Instead, the Swift team argues, the episodic flaring may well indicate the stretching, breaking, and piecemeal consumption of a neutron star by a black-hole partner several times its mass.

The supernova scenario explains why long GBRs should be strongly collimated. But compact-object merger models are less clear about collimation of short GRBs. ⁷ Evidence from the 24 July burst is contradictory: A break in the slope of the infrared light curve ⁶ a day after the burst suggests collimation with an opening angle of about 10°. But the apparent absence of a corresponding break in Chandra x-ray data casts doubt on that collimation.

If the short GRBs are generally not collimated, then their energy output in gammas is not very much less than the 10⁵¹ ergs typical of the long GRBs. Such high energies are not obviously inconsistent with merger models. If, as now seems likely, the short GRBs are indeed caused by mergers of compact objects, their collimation is an important issue for estimating the rate at which an upgraded LIGO gravitational-wave detector should expect to detect signals from such mergers.

Happily the localization of three short GRBs in just three months holds out the hope that astrophysicists will be able to confront their merger scenarios with many more bursts in the near future. In fact, a very faint short GRB, detected by Swift on 13 August, has been tentatively localized to a cluster of galaxies at a redshift of 0.7, much farther away than its three predecessors.