Final Shuttle Mission to Hubble Space Telescope
Gamma-ray bursts (GRBs) are flashes of gamma rays associated with extremely energetic explosions in distant galaxies. They are the most luminous electromagnetic events occurring in the universe. Bursts can last from milliseconds to nearly an hour, although a typical burst lasts a few seconds. The initial burst is usually followed by a longer-lived "afterglow" emitting at longer wavelengths (X-ray, ultraviolet, optical, infrared, and radio).
Most observed GRBs are believed to be a narrow beam of intense radiation released during a supernova event, as a rapidly rotating, high-mass star collapses to form a black hole. A subclass of GRBs (the "short" bursts) appear to originate from a different process, possibly the merger of binary neutron stars.
The sources of most GRBs are billions of light years away from Earth, implying that the explosions are both extremely energetic (a typical burst releases as much energy in a few seconds as the Sun will in its entire 10 billion year lifetime) and extremely rare (a few per galaxy per million years). All observed GRBs have originated from outside the Milky Way galaxy, although a related class of phenomena, soft gamma repeater flares, are associated with magnetars within the Milky Way. It has been hypothesized that a gamma-ray burst in the Milky Way could cause a mass extinction on Earth.
GRBs were first detected in 1967 by the Vela satellites, a series of satellites designed to detect covert nuclear weapons tests. Hundreds of theoretical models were proposed to explain these bursts in the years following their discovery, such as collisions between comets and neutron stars. Little information was available to verify these models until the 1997 detection of the first X-ray and optical afterglows and direct measurement of their redshifts using optical spectroscopy. These discoveries, and subsequent studies of the galaxies and supernovae associated with the bursts, clarified the distance and luminosity of GRBs, definitively placing them in distant galaxies and connecting long GRBs with the deaths of massive stars.
Gamma-ray bursts were first observed in the late 1960s by the U.S. Vela satellites, which were built to detect gamma radiation pulses emitted by nuclear weapons tested in space. The United States suspected that the USSR might attempt to conduct secret nuclear tests after signing the Nuclear Test Ban Treaty in 1963. On July 2, 1967, at 14:19 UTC, the Vela 4 and Vela 3 satellites detected a flash of gamma radiation unlike any known nuclear weapons signature. Uncertain what had happened but not considering the matter particularly urgent, the team at the Los Alamos Scientific Laboratory, led by Ray Klebesadel, filed the data away for investigation. As additional Vela satellites were launched with better instruments, the Los Alamos team continued to find inexplicable gamma-ray bursts in their data. By analyzing the different arrival times of the bursts as detected by different satellites, the team was able to determine rough estimates for the sky positions of sixteen bursts and definitively rule out a terrestrial or solar origin. The discovery was declassified and published in 1973 as an Astrophysical Journal article entitled "Observations of Gamma-Ray Bursts of Cosmic Origin".
Many theories were advanced to explain these bursts, most of which posited nearby sources within the Milky Way Galaxy. Little progress was made until the 1991 launch of the Compton Gamma Ray Observatory and its Burst and Transient Source Explorer (BATSE) instrument, an extremely sensitive gamma-ray detector. This instrument provided crucial data indicating that the distribution of GRBs is isotropic—not biased towards any particular direction in space, such as toward the galactic plane or the galactic center. Because of the flattened shape of the Milky Way Galaxy, sources within our own galaxy would be strongly concentrated in or near the Galactic plane. The absence of any such pattern in the case of GRBs provided strong evidence that gamma-ray bursts must come from beyond the Milky Way. However, some Milky Way models are still consistent with an isotropic distribution.
For decades after the discovery of GRBs, astronomers searched for a counterpart: any astronomical object in positional coincidence with a recently observed burst. Astronomers considered many distinct classes of objects, including white dwarfs, pulsars, supernovae, globular clusters, quasars, Seyfert galaxies, and BL Lac objects. All such searches were unsuccessful,[nb 1] and in a few cases particularly well-localized bursts (those whose positions were determined with what was then a high degree of accuracy) could be clearly shown to have no bright objects of any nature consistent with the position derived from the detecting satellites. This suggested an origin of either very faint stars or extremely distant galaxies. Even the most accurate positions contained numerous faint stars and galaxies, and it was widely agreed that final resolution of the origins of cosmic gamma-ray bursts would require both new satellites and faster communication.
Several models for the origin of gamma-ray bursts postulated that the initial burst of gamma rays should be followed by slowly fading emission at longer wavelengths created by collisions between the burst ejecta and interstellar gas. Early searches for this "afterglow" were unsuccessful, largely due to the difficulties in observing a burst's position at longer wavelengths immediately after the initial burst. The breakthrough came in February 1997 when the satellite BeppoSAX detected a gamma-ray burst (GRB 970228[nb 2]) and when the X-ray camera was pointed towards the direction from which the burst had originated, it detected fading X-ray emission. Ground-based telescopes later identified a fading optical counterpart as well. Once the GRB faded, deep imaging was able to identify a faint, distant host galaxy at the location of the GRB as pinpointed by the optical afterglow.
Because of the very faint luminosity of this galaxy, its exact distance was not measured for several years. Well before then, another major breakthrough occurred with the next event registered by BeppoSAX, GRB 970508. This event was localized within four hours of its discovery, allowing research teams to begin making observations much sooner than any previous burst. The spectrum of the object revealed a redshift of z = 0.835, placing the burst at a distance of roughly 6 billion light years from Earth. This was the first accurate determination of the distance to a GRB, and together with the discovery of the host galaxy of 970228 proved that GRBs occur in extremely distant galaxies. Within a few months, the controversy about the distance scale ended: GRBs were extragalactic events originating within faint galaxies at enormous distances. The following year, GRB 980425 was followed by a bright supernova (SN 1998bw), indicating a clear connection between GRBs and the deaths of very massive stars. This burst provided the first strong clue about the nature of the systems that produce GRBs.
BeppoSAX functioned until 2002 and CGRO (with BATSE) was deorbited in 2000. However, the revolution in the study of gamma-ray bursts motivated the development of a number of additional instruments designed specifically to explore the nature of GRBs, especially in the earliest moments following the explosion. The first such mission, HETE-2, launched in 2000 and functioned until 2006, providing most of the major discoveries during this period. One of the most successful space missions to date, Swift, was launched in 2004 and as of 2009 is still operational. Swift is equipped with a very sensitive gamma ray detector as well as on-board X-ray and optical telescopes, which can be rapidly and automatically slewed to observe afterglow emission following a burst. More recently, the Fermi mission was launched carrying the Gamma-Ray Burst Monitor, which detects bursts at a rate of several hundred per year, some of which are bright enough to be observed at extremely high energies with Fermi's Large Area Telescope. Meanwhile, on the ground, numerous optical telescopes have been built or modified to incorporate robotic control software that responds immediately to signals sent through the Gamma-ray Burst Coordinates Network. This allows the telescopes to rapidly repoint towards a GRB, often within seconds of receiving the signal and while the gamma-ray emission itself is still ongoing.
New developments over the past few years include the recognition of short gamma-ray bursts as a separate class (likely due to merging neutron stars and not associated with supernovae), the discovery of extended, erratic flaring activity at X-ray wavelengths lasting for many minutes after most GRBs, and the discovery of the most luminous (GRB 080319B) and the most distant (GRB 090423) objects in the universe.
While most astronomical transient sources have simple and consistent time structures (typically a rapid brightening followed by gradual fading, as in a nova or supernova), the light curves of gamma-ray bursts are extremely diverse and complex. No two gamma-ray burst light curves are identical, with large variation observed in almost every property: the duration of observable emission can vary from milliseconds to tens of minutes, there can be a single peak or several individual subpulses, and individual peaks can be symmetric or with fast brightening and very slow fading. Some bursts are preceded by a "precursor" event, a weak burst that is then followed (after seconds to minutes of no emission at all) by the much more intense "true" bursting episode. The light curves of some events have extremely chaotic and complicated profiles with almost no discernible patterns.
Although some light curves can be roughly reproduced using certain simplified models, little progress has been made in understanding the full diversity observed. Many classification schemes have been proposed, but these are often based solely on differences in the appearance of light curves and may not always reflect a true physical difference in the progenitors of the explosions. However, plots of the distribution of the observed duration[nb 3] for a large number of gamma-ray bursts show a clear bimodality, suggesting the existence of two separate populations: a "short" population with an average duration of about 0.3 seconds and a "long" population with an average duration of about 30 seconds. Both distributions are very broad with a significant overlap region in which the identity of a given event is not clear from duration alone. Additional classes beyond this two-tiered system have been proposed on both observational and theoretical grounds.
Most observed events have a duration of greater than two seconds and are classified as long gamma-ray bursts. Because these events constitute the majority of the population and because they tend to have the brightest afterglows, they have been studied in much greater detail than their short counterparts. Almost every well-studied long gamma-ray burst has been associated with a rapidly star-forming galaxy and in many cases a core-collapse supernova as well, unambiguously linking long GRBs with the deaths of massive stars.
Events with a duration of less than about two seconds are classified as short gamma-ray bursts. Until 2005, no afterglow had been successfully detected from any short event and little was known about their origins. Since then, several dozen short gamma-ray burst afterglows have been detected and localized, several of which are associated with regions of little or no star formation, including large elliptical galaxies and the intracluster medium. This rules out an association with massive stars, confirming that short events are physically distinct from long events. The true nature of these objects (or even whether the current classification scheme is accurate) remains unknown, although the leading hypothesis is that they originate from the mergers of binary neutron stars. A small fraction of short gamma-ray bursts are probably associated with giant flares from soft gamma repeaters in nearby galaxies.
Gamma-ray bursts are very bright as observed from Earth despite typical immense distances. An average long GRB has comparable bolometric flux to a bright Galactic star despite a distance of billions of light years (compared to a few tens of light years for most stars). Most of this energy is released in gamma rays, although some GRBs have extremely luminous optical counterparts as well. GRB 080319B, for example, was accompanied by an optical counterpart that peaked at a visible magnitude of 5.8, comparable to that of the dimmest naked-eye stars despite the burst's distance of 7.5 billion light years. This combination of brightness and distance requires an extremely energetic source. Assuming the gamma-ray explosion to be spherical, the energy output of GRB 080319B would be within a factor of two of the rest-mass energy of the Sun(the energy which would be released were the Sun to be converted entirely into radiation.) 
No known process in the Universe can produce this much energy in such a short time. However, gamma-ray bursts are thought to be highly focused explosions, with most of the explosion energy collimated into a narrow jet traveling at speeds exceeding 99.995% of the speed of light. The approximate angular width of the jet (that is, the degree of beaming) can be estimated directly by observing "jet breaks" in afterglow light curves: a time after which the slowly-decaying afterglow abruptly begins to fade rapidly as the jet slows down and can no longer beam its radiation as effectively. Observations suggest significant variation in the jet angle from between 2 and 20 degrees.
Because their energy is strongly beamed, the gamma rays emitted by most bursts are expected to miss the Earth and never be detected. When a gamma-ray burst is pointed towards Earth, the focusing of its energy along a relatively narrow beam causes the burst to appear much brighter than it would have been were its energy emitted spherically. When this effect is taken into account, typical gamma-ray bursts are observed to have a true energy release of about 1044 J, or about 1/2000 of a Solar mass energy equivalent. This is comparable to the energy released in a bright type Ib/c supernova (sometimes termed a "hypernova") and within the range of theoretical models. Very bright supernovae have been observed to accompany several of the nearest GRBs. Additional support for strong beaming in GRBs has come from observations of strong asymmetries in the spectra of nearby type Ic supernova  and from radio observations taken long after bursts when their jets are no longer relativistic.
Short GRBs appear to come from a lower-redshift population and are less luminous than long GRBs. The degree of beaming in short bursts has not been accurately measured, but as a population they are likely less beamed than long GRBs or possibly not beamed at all in some cases.
Because of the immense distances of most gamma-ray burst sources from Earth, identification of the progenitors, the systems that produce these explosions, is particularly challenging. The association of some long GRBs with supernovae and the fact that their host galaxies are rapidly star-forming offer very strong evidence that long gamma-ray bursts are associated with massive stars. The most widely-accepted mechanism for the origin of long-duration GRBs is the collapsar model, in which the core of an extremely massive, low-metallicity, rapidly-rotating star collapses into a black hole in the final stages of its evolution. Matter near the star's core rains down towards the center and swirls into a high-density accretion disk. The infall of this material into a black hole drives a pair of relativistic jets out along the rotational axis, which pummel through the stellar envelope and eventually break through the stellar surface and radiate as gamma rays. Some alternative models replace the black hole with a newly-formed magnetar, although most other aspects of the model (the collapse of the core of a massive star and the formation of relativistic jets) are the same.
The closest Galactic analogs of the stars producing long gamma-ray bursts are likely the Wolf-Rayet stars, extremely hot and massive stars which have shed most or all of their hydrogen due to radiation pressure. Eta Carinae and WR 104 have been cited as possible gamma-ray burst progenitors. It is unclear if any star in the Milky Way has the appropriate characteristics to produce a gamma-ray burst.
The massive-star model probably does not explain all types of gamma-ray burst. There is strong evidence that some short-duration gamma-ray bursts occur in systems with no star formation and where no massive stars are present, such as galaxy halos and intergalactic space. The favored theory for the origin of most short gamma-ray bursts is the merger of a binary system consisting of two neutron stars. According to this model, the two stars in a binary slowly spiral towards each other due to the release of energy via gravitational radiation until the neutron stars suddenly rip each other apart due to tidal forces and collapse into a single black hole. The infall of matter into the new black hole in an accretion disk then powers an explosion, similar to the collapsar model. Numerous other models have also been proposed to explain short gamma-ray bursts, including the merger of a neutron star and a black hole, the accretion-induced collapse of a neutron star, or the evaporation of primordial black holes.
The means by which gamma-ray bursts convert energy into radiation remains poorly understood, and as of 2007 there was still no generally accepted model for how this process occurs. Any successful model of GRB emission must explain the physical process for generating gamma-ray emission that matches the observed diversity of light-curves, spectra, and other characteristics. Particularly challenging is the need to explain the very high efficiencies that are inferred from some explosions: some gamma-ray bursts may convert as much as half (or more) of the explosion energy into gamma-rays. Recent observations of the bright optical counterpart of GRB 080319B, whose light curve was correlated with the gamma-ray light curve, has suggested that inverse Compton may be the dominant process in some events. In this model, pre-existing low-energy photons are scattered by relativistic electrons within the explosion, augmenting their energy by a large factor and transforming them into gamma-rays.
The nature of the longer-wavelength afterglow emission (ranging from X-ray through radio) that follows gamma-ray bursts is better understood. Any energy released by the explosion not radiated away in the burst itself takes the form of matter or energy moving outward at nearly the speed of light. As this matter collides with the surrounding interstellar gas, it creates a relativistic shock wave that then propagates forward into interstellar space. A second shock wave, the reverse shock, may propagate back into the ejected matter. Extremely energetic electrons within the shock wave are accelerated by strong local magnetic fields and radiate as synchrotron emission across most of the electromagnetic spectrum. This model has generally been successful in modeling the behavior of many observed afterglows at late times (generally, hours to days after the explosion), although there are difficulties explaining all features of the afterglow very shortly after the gamma-ray burst has occurred.
Currently orbiting satellites detect an average of about one gamma-ray burst per day. Because gamma-ray bursts are visible to distances encompassing most of the observable universe, a volume encompassing many billions of galaxies, this suggests that gamma-ray bursts must be exceedingly rare events per galaxy. Measuring the exact rate is difficult, but for a galaxy of approximately the same size as the Milky Way, the expected rate (for long GRBs) is about one burst every 100,000 to 1,000,000 years. Only a few percent of these would be beamed towards Earth. Estimates of rates of short GRBs are even more uncertain because of the unknown beaming fraction, but are probably comparable.
A gamma-ray burst in the Milky Way, if close enough to Earth and beamed towards it, could have significant effects on the biosphere. The absorption of radiation in the atmosphere would cause photodissociation of nitrogen, generating nitric oxide that would act as a catalyst to destroy ozone. According to a 2004 study, a GRB at a distance of about a kiloparsec could destroy up to half of Earth's ozone layer; the direct UV irradiation from the burst combined with additional solar UV radiation passing through the diminished ozone layer could then have potentially significant impacts on the food chain and potentially trigger a mass extinction. The authors estimate that one such burst is expected per billion years, and hypothesize that the Ordovician-Silurian extinction event could have been the result of such a burst.
There are strong indications that long gamma-ray bursts preferentially or exclusively occur in regions of low metallicity. Because the Milky Way has been metal-rich since before the Earth formed, this effect may diminish or even eliminate the possibility that a long gamma-ray burst has occurred within the Milky Way within the past billion years. No such metallicity biases are known for short gamma-ray bursts. Thus, depending on their local rate and beaming properties, the possibility for a nearby event to have had a large impact on Earth at some point in geological time may still be significant.