4. BURSTS IN THE SWIFT ERA

The launch of the Swift satellite in 2004 ushered in a new era of extensive data collection and analysis on GRBs, at wavelengths ranging from optical to MeV energies. This resulted in a number of interesting new discoveries, which have motivated various refinements and reappraisals as well as new work on theoretical models, as discussed at greater length in the next sections.

Swift is equipped with three instruments: the Burst Alert Telescope (BAT), the X-Ray Telescope (XRT) and the UV Optical Telescope (UVOT). The BAT detects bursts and locates them to about 2 arcminutes accuracy. This position is then used to automatically slew the spacecraft, typically within less than a minute, re-pointing the high angular resolution XRT and UVOT instruments towards the event. The positions are also rapidly sent to Earth so that ground telescopes can follow the afterglows.

A surprising new result achieved by Swift was that in a large fraction of the bursts the X-ray afterglow shows an initial very steep time decay, starting after the end of the prompt -ray emission. This then is generally followed by a much shallower time decay, often punctuated by abrupt, large amplitude X-ray flares, lasting sometimes for up to ~ 1000 s, which then steepens into a power law time decay with the more usual (pre-Swift) slope of index of roughly -1.2 to -1.7 [56, 57]. A final further steepening is sometimes detected, ascribed to beaming due to a finite jet opening angle. The initial steep decay may be ascribed to the evanescent radiation from high latitudes > ^-1 relative to the line of sight [58, 59], while the ensuing shallow decay phase may be due to continued outflow of material after the prompt emission has ended [60], which may undergo occasional internal shocks resulting in X-ray flares, e.g. [56, 61, 12]. The subsequent steepening can be ascribed to the previously known forward shock gradual deceleration and the beaming induced jet break. These structures in the X-ray afterglow light curves are present both in long and short bursts.

Figure 3. Light curves of GRB060428A from the Swift XRT [62].

The number of GRB redshifts obtained underwent a rapid expansion after the launch of Swift (currently in excess of 200), thanks to the rapid localization allowing large grounds-based telescopes to acquire high quality spectra while the afterglow was still bright. The most distant ones are intrinsically the brightest, typically E_iso 10⁵⁵ erg, the current record holder being GRB090423 at a spectroscopically confirmed redshift z = 8.2 [63], and GRB090429B, at a photometric redshift z ~ 9.4 [64].

With the increasing statistics, LGRBs are contributing to a better understanding of the high- redshift universe. They provide spectroscopic information about the chemical composition of the intervening intergalactic medium at epochs when the Universe was as low as 1/20th of its present age. Also, since LGRBs are the endpoints of the lives of massive stars, their rate is approximately proportional to the star formation rate. This gives information at high redshift where the rate is highly uncertain. There can be evolutionary biases, such as a dependence of LGRBs on the metalicity of host galaxies, which must be taken into account [65, 66].

Swift succeeded in finally localizing the host galaxies of a number of short GRBs (SGRBs). e.g. [67, 68]. Unlike long GRBs, the SGRBs typically originate in host galaxies with a wide range of star formation properties, including low formation rates. The host properties are substantially different than those of LGRBs [69, 70, 71], indicating a different origin. Furthermore, nearby SGRBs show no evidence for simultaneous supernovae [72], as do many long bursts. These results reinforce the interpretation that SGRBs arise from an old population of stars, probably due to mergers of compact binaries such as double neutron star or neutron star-black holes [72, 8, 9].

Short GRBs are found to have generally a lower isotropic-equivalent luminosity and total energy output E_iso than LGRBs, typically E_iso ~ 10⁵⁰ ergs, with a weak afterglow, and in the few cases where a jet break has been measured, the jet opening angle appears to be wider than in LGRBs, _j ~ 5° - 25° [73, 74]. Another new result was the discovery, in about 25% of SGRBs, of a longer (~ 100 s) light curve tail with a spectrum softer than the initial episode [75, 76]. This is puzzling in the context of double neutron star or neutron star-black hole mergers, since numerical simulations suggest that the disk of disrupted matter is accreted in at most a few seconds, e.g. [72]. A longer accretion timescale, however, may occur if the disk is highly magnetized [77], or if the compact merger results in a temporary magnetar whose magnetic field holds back the accretions disk until the central object collapses to a black hole [78].

4.1. Other types of bursts

The demarcation into two classes of bursts is however too simplistic to be the whole story. Some bursts fit neither category. For example, some bursts detected by Swift are extreme magnetar flares caused by the sudden readjustment (and release of stored energy) in the magnetosphere of a highly magnetized ( 10¹⁴ G) neutron star. These are of interest for phenomenologists but are a confusing complication for those seeking correlations between the observable parameters of bursts.

But the Swift spacecraft has revealed another type of object that is of great interest, and which was a surprise: bursts characterized by unusually persistent and prolonged emission, and located at the centre of the host galaxy. These are interesting both to astrophysicists and to relativists, as they may be triggered by a long-predicted effect that has not before been conclusively detected: the tidal disruption of a star by a massive hole.

Tidal capture and disruption of stars attracted interest back in the 1970s, when theorists started to address the dynamics of stars concentrated in a high-density `cusp' surrounding the kind of black hole expected to exist in the centres of galaxies (and perhaps in some globular star clusters as well). It was recognized that stars could be captured and swallowed by the central hole if they were in a `loss cone' of near-radial orbits.

If the central hole is sufficiently massive, tidal forces at the horizon may be too gentle to disrupt to star while it is still in view, in which case it is captured without any conspicuous display. For a solar-type star, this requires ~ 10⁸ M; for white dwarfs the corresponding mass is ~ 10⁴ M. (And neutron stars are swallowed whole by black holes with masses above about 10 M - this is important for the gravitational wave signal in coalescing binary stars, as discussed elsewhere in this volume). For a spinning hole, the cross-section for capture, and the tidal radius for disruption, depend on the relative orientation of the orbital and spin angular momenta. (Stars on orbits counter-rotating with respect to the hole are preferentially captured: this is a process that would reduce the spin of a hole in a galactic nucleus.)

When stars are swallowed before disruption, they can be treated as point mass particles moving in the gravitational field of the hole; their interactions among themselves can be treated the same way, except insofar as star-star collisions are important. But the physics is much messier in the cases when the tidal radius is outside the hole and the star is disrupted rather than swallowed whole. This phenomenon has been studied since the 1970s, first via analytic models (e.g. [79, 80]) and subsequently by progressively more powerful numerical simulations (e.g. [81, 82], etc.). In the Newtonian approximation the tidal radius is R_t ~ R_∗(M_BH / M_∗)^1/3. There are several key parameters: the type of star; the pericentre of the star's orbit relative to the tidal radius, and the orientation of the orbit relative to the hole's spin axis. In most astrophysical contexts, the captured stars would be on highly eccentric orbits (i.e the orbital binding energy would be small compared to that of a circular orbit at the tidal radius). If the pericentre is of order R_t, the star will be disrupted, and the debris will be continue on eccentric orbits, but with a spread of energies of order the binding energy of the original star. Indeed nearly half the debris will escape from the hole's gravitational field completely; the rest will be on more tightly bound (but still eccentric ) orbits, and would be fated to dissipate further, forming a disc much of which would then be accreted into the hole. A pericentre passage at (say) 2 or 3 times R_t would not disrupt a star completely, but would remove its envelope, and induce internal oscillations, thereby extracting orbital energy and leaving the star vulnerable on further passages. On the other hand, as first discussed by [79], a star that penetrates far inside the tidal radius (but not so close to the hole that it spirals in) will be drastically distorted and compressed by the tidal forces, perhaps to the extent that a nuclear explosion occurs, leading to a greater spread in the energy of the debris than would result from straight gas dynamics.

There have in recent years been detailed computations of these processes, and also of the complicated and dissipative gas dynamics that leads to the accretion of the debris, and the decline of the associated luminosity as the dregs eventually drain away. There are two generic predictions: the debris enveloping the hole should initially have a thermal emission with a power comparable to the Eddington luminosity of the hole; and at late times, when the emission comes from the infall of debris from orbits with large apocentre, the luminosity falls as L t^-5/3.

There has been much debate about the role of tidal capture in the growth of supermassive holes, and the fueling of AGN emission, and many calculations of the expected rate, taking account of what has been learnt about the masses of holes, and the properties of the stellar populations surrounding them. Some flares in otherwise quiescent galactic nuclei, where the X-ray luminosity surges by a factor 100, have been attributed to tidal disruptions.

But tidal disruption is included in this chapter mainly because of a remarkable burst detected by Swift, Sw J164449.3 [83], located at the centre of its host galaxy, and which was exceptionally prolonged in its emission. This is perhaps the best candidate so far for an event triggered by tidal capture of a star. The high energy radiation, were this model correct, would come from a jet generated near the hole. Modeling is still tentative, and is difficult because there is no reason to expect alignment between the angular momentum vectors of the hole and of the infalling material. But the inner disc (and therefore the inner jet) would be expected to align with the hole, though it is possible that the jet is deflected further out by material with different alignment (c.f. [84]).

Be that as it may, this exceptional burst offers model-builders an instructive `missing link' between the typical long (`Type 1') burst, involving a massive star, and the jets in AGNs which are generated by processes around supermassive holes.