GRB Fireball Physics: Prompt and Early Emission

We review the fireball shock model of gamma-ray burst prompt and early afterglow emission in light of rapid follow-up measurements made and enabled by the multi-wavelength Swift satellite. These observations are leading to a reappraisal and expansion of the previous standard view of the GRB and its fireball. New information on the behavior of the burst and afterglow on minutes to hour timescales has led, among other results, to the discovery and follow-up of short GRB afterglows, the opening up of the z>6 redshift range, and the first prompt multi-wavelength observations of a long GRB-supernova. We discuss the salient observational results and some associated theoretical issues.

from the detailed X-ray light curves, starting on average 100 seconds after the γ-ray trigger, that result from the prompt XRT observations of BAT-detected bursts. These observations suggest a canonical X-ray afterglow ( Fig. 1; [14,15]) with one or more of the following stages (note that the numerical subscript enumerates each stage): (1) An initial steep decay F X ∝ t −α 1 with temporal index 3 < ∼ α 1 < ∼ 5, and an energy spectrum F ν ∝ ν −β 1 with energy spectral index 1 < ∼ β 1 < ∼ 2 (photon index 0 < ∼ Γ < ∼ 1), extending up to a time 300 s < ∼ t 1 < ∼ 500 s; (2) A subsequent flatter decay F X ∝ t −α 2 with 0.2 < ∼ α 2 < ∼ 0.8 and energy index 0.7 < ∼ β 2 < ∼ 1.2, at times 10 3 s < ∼ t 2 < ∼ 10 4 s (in some cases interspersed with flares, see [15]; (3) A "normal" decay F X ∝ t −α 3 with 1.1 < ∼ α 3 < ∼ 1.7 and 0.7 < ∼ β 3 < ∼ 1.2 (generally unchanged from the previous stage, i.e. β 3 ≈ β 2 ), up to a time t 3 ∼ 10 5 s, or in some cases longer; (4) In some cases, a steeper X-ray decay at the end, F X ∝ t −α 4 , with 2 < ∼ α 4 < ∼ 3, after t 4 ∼ 10 5 s, resembling what is expected from jet breaks; (5) In about half the afterglows, one or more X-ray flares or bumps are observed in the light curve, sometimes starting as early as 100 s after trigger, and sometimes as late as 10 5 s. The energy in these flares ranges from a percent up to a value comparable to the prompt emission (in GRB 050502b). The rise and decay times of these flares is unusually steep, depending on the reference time t 0 , behaving as (t − t 0 ) ±α fl with 3 < ∼ α fl < ∼ 6, and energy indices which can be also steeper than during the smooth decay portions. The flux level after the flare usually decays to the value extrapolated from the value before the flare rise. Figure 1. Schematic features seen in early X-ray afterglows detected with the Swift XRT instrument (e.g. [15,14]); see text.
Very high-redshift bursts: Another major advance achieved by Swift was the detection of the long burst GRB 050904, which broke the z > 6 redshift barrier. This burst was very bright, both in its prompt γ-ray emission (E γ,iso ∼ 10 54 erg) and in its X-ray afterglow. Prompt ground-based optical/IR upper limits and a J-band detection suggested a photometric redshift z > 6 [16]. Spectroscopic confirmation with the 8.2 m Subaru telescope gave z = 6.29 [11]. There are several striking features to this burst.
One is its enormous X-ray brightness, exceeding for a full day the X-ray brightness of the most distant quasar known to-date, SDSS J0130+0524 -and exceeding it by a factor of 10 5 in the first minutes [17]. The implications for the use of GRBs as a tool for probing the IGM are thought-provoking. Another notable feature was its extremely variable X-ray light curve, showing many large amplitude flares throughout the first day (Fig. 2). A third exciting feature is its brief, very bright IR flash [18], comparable in brightness to the famous m V ∼ 9 optical flash in GRB 990123. Figure 2. Prompt and afterglow lightcurves for the redshift z = 6.29 GRB 050904 at hard X-ray (BAT, 50 keV), X-ray (XRT, 5 keV), and near-infrared (J-band, 1.1 µm; and I-band, 9500Å) frequencies. A bright early flare is seen in the IR and X-ray; in the X-ray this is followed by extensive flaring activity for more than a day after the burst. Power-law decay indices measured in the IR (α o ) are indicated, where available; a likely jet break is seen at t ≈ 2.3 days [20]. Data from [16,18,19,20].

Short bursts:
The third major advance from Swift was the discovery and localization of short GRB afterglows. As of July 2006, eleven definitive short bursts have been localized by Swift, and HETE-2 has discovered two short bursts with subsequent afterglow emission; in addition, the short burst GRB 051103 was localized by the IPN to a region intersecting the nearby galaxies M81 and M82. In seven of the Swift short bursts and two HETE-2 bursts an X-ray afterglow was detected and observed, with GRBs 050709, 050724, 051221A, 060313, and 060121 (as well as the candidate short burst GRB 051227) also showing optical afterglows, and GRBs 050724 and 051221A being detected in the radio [21]. These are the first afterglows detected for short bursts, and yielded the first host galaxy identifications. The hosts are of early type (ellipticals) in roughly half the cases, and otherwise appear to be dwarf irregular galaxies -albeit with evidence for old as well as young stellar populations [22]. The redshifts of five of them are in the range 0.15 to 0.5, while GRB 050813 was initially thought to have z = 0.7 [23] but may be associated with a galaxy cluster at z ≈ 1.8 [21], and the redshift of the HETE-2 GRB 060121 is estimated to be z ≈ 4.6 [24]. The median z (as yet unaffected by these two relatively high redshifts) is z med = 0.26, which gives a (1 + z) that is 1/3 that of the long bursts. While there is evidence for star formation in roughly half the host galaxies, overall the host properties correspond to expectations for a progenitor population of neutron star or neutron star-black hole binary systems, the most often-discussed progenitor candidates. Such hosts would also be compatible with other progenitors involving old compact stars.
The first short burst afterglow seen by Swift was that of GRB 050509B, a low luminosity (E γ,iso ∼ 2 × 10 48 erg) burst with a simple power-law X-ray afterglow which could only be followed for ∼ 10 4 s [25]. The subsequent afterglows of GRB 050724 (E γ,iso ∼ 3 × 10 50 erg; [26]) and GRB 051221A (E γ,iso ∼ 1.5 × 10 51 erg; [22]) were brighter, and could be followed in X-rays for at least 10 5 s [27,28]. In particular, the X-ray afterglow of GRB 050724 is remarkable in that it exhibits a typical X-ray light curve as observed from long GRBs by Swift (apart from the absence of a slow-decay phase). It shows a fast early decay, with significant X-ray flares, one at 100 s and another at 3 × 10 4 s. The first flare has the same fluence as the prompt emission, while the late flare has roughly 10% of that. The interpretation of this activity poses interesting challenges, as discussed below.
The GRB-SN connection: The fourth major advance from Swift was its observation, with BAT, XRT and UVOT, of an unusually long (∼ 2000 s), soft burst, GRB 060218 [29], soon found to be associated with SN2006aj, a nearby (z = 0.033) type Ic supernova [30,31,32,33]. The optical light curve of this supernova peaked earlier than for most other known supernovae, and the time of core collapse is constrained to within a day of the GRB trigger. This was the first time that a connected GRB and supernova event was observed starting in the first ∼100 s in X-ray and UV/optical light, and the results are of great interest. The early X-ray light curve shows a slow rise and plateau followed by a drop after ∼ 10 3 s; the spectrum is initially dominated by a power-law component, with an increasing thermal component that dominates after ≈3000 s. Perhaps the most interesting interpretation involves shock break-out of a semi-relativistic component in a WR progenitor wind [29], although see also [34]. After this a more conventional X-ray power-law decay follows, and a UV component peaking at a later time can be interpreted as due to the slower supernova envelope shock. Finally, a radio-bright component is seen that presents an interesting comparison to other local and GRB-related type Ic supernovae [33]. A distinct GRB/SN detection based on Swift afterglow observations is that associated with GRB 050525A [35].
Prompt optical observations: A final major advance enabled by Swift has been a wealth of information about the prompt and early-time optical emission of GRBs. Thanks to Swift alerts and an array of ground-based robotic telescopes (as well as the UVOT), the detection of optical counterparts during and shortly after (<100 s) the gamma-ray trigger is becoming almost routine. The Swift era began with the detection in optical [36] and near-infrared [37] wavelengths of a bright flash associated with GRB 041219A (localized in real time by INTEGRAL). Apart from exhibiting the first optical flash with brightness (after correcting for Galactic extinction) similar to GRB 990123, this burst showed the first evidence for correlated low-and high-energy emission, intepreted as optical emission from the same internal shocks that produced the GRB [36,37]. A similar level of correlated optical emission was later observed from GRB 050820A [38]; however in this case it was accompanied by an uncorrelated component that connected smoothly to the later afterglow. Prompt optical emission that extrapolates smoothly to the later afterglow seems to be the more usual case; however, bright flares consistent with a reverse-shock interpretation are seen on occasion [39,40]. The most exciting single observation is probably the I = 14.1 mag optical flash from GRB 050904 at z = 6.3 observed with TAROT [18], which rises to its peak after the end of gamma-ray emission and is brighter, and decays more quickly than, the later afterglow -all properties consistent with an origin in this burst's reverse shock.

Prompt Emission Models
The prompt gamma-ray emission of a classical GRB has most of its energy in the energy range 0.1 to 2.0 MeV. The generic photon spectrum is a broken power law [41] with a break energy in the above range, and, typically, power law extensions down into the X-ray, and up into the 100 MeV to GeV ranges (although a number of bursts show a thermal-like spectrum, which sometimes can be fitted with a blackbody [42]).

Gamma-Ray Emission
The simplest model for the gamma-ray emission assumes that a proton crossing a strong shock front with a relative bulk Lorentz factor Γ 21 acquires (in the comoving frame) an internal energy characterized by a random (comoving) Lorentz factor γ p,m ∼ Γ [43]. The comoving magnetic field behind the shock can build up due to turbulent dynamo effects behind the shocks [43,44] (as also inferred in supernova remnant shocks). More recently, the Weibel instability has been studied in this context [45,46,47]. The efficiency of this process remains under debate, but one can parametrize the magnetic field as having a post-shock energy density which is a fraction ǫ B of the equipartition value relative to the proton random energy, [43,48]. Scattering of electrons (and protons) by magnetic irregularities upstream and downstream can lead to a Fermi acceleration process resulting in a relativistic power law distribution of energies N(γ) ∝ γ −p with p ≥ 2. The starting minimum (comoving) Lorentz factor of the thermal electrons injected into the acceleration process, γ e,m would in principle be the same as for the protons, Γ (they experience the same velocity difference), hence both before and after acceleration they would have ∼ (m e /m p ) less energy than the protons. However, the shocks are collisionless, i.e. mediated by chaotic electric and magnetic fields, and can redistribute the proton energy between the electrons and protons, up to some fraction ǫ e of the thermal energy equipartition value with the protons, so γ e,m ∼ ǫ e (m p /m e )Γ [44,49]. The synchrotron spectrum, in its simplest form, peaks at ν m ∼ Γ(3/8π)(eB ′ /m e c)γ 2 m ∼ 2×10 6 B ′ γ 2 m Γ Hz, and has a shape , in the adiabatic limit where the cooling time is longer than the dynamic time. In the opposite (radiative) limit, the spectrum peaks at the cooling frequency ν c (at which cooling time for electrons emitting at ν c equals dynamic time, and the spectrum above is F ν ∝ ν −p/2 . Depending on the parameters, more complicated spectra are possible, discussed e.g. in [50]. The synchrotron spectrum is modified at low energies by self-absorption [44,51], making the spectrum steeper; and it is also modified at high energies, due to inverse Compton effects [44,49,52], extending into the GeV range. However, a number of effects can modify the simple synchrotron spectrum. For instance, the distribution of observed low energy spectral indices β 1 (where F ν ∝ ν β 1 below the spectral peak) has a mean value β 1 ∼ 0, but for a fraction of bursts this slope reaches positive values β 1 > 1/3 which are incompatible with a simple synchrotron interpretation [53]. Possible explanations include synchrotron self-absorption in the X-ray [54] or in the optical range up-scattered to X-rays [55], low-pitch angle scattering or jitter radiation [56,57], observational selection biases [58] and/or time-dependent acceleration and radiation [59], where low-pitch angle diffusion can also explain high energy indices steeper than predicted by isotropic scattering. Other models invoke a photospheric component and pair formation [60]. Pair formation can become important in internal shocks or dissipation regions occurring at small radii, since a high comoving luminosity implies a large comoving compactness parameter. A moderate to high scattering depth can lead to a Compton equilibrium which gives spectral peaks in the right energy range [61,62]. An important aspect is that Compton equilibrium of internal shock electrons or pairs with photospheric photons leads to a high radiative efficiency, as well as to spectra with a break at the right preferred energy and steep low energy slopes [63,64,65]. It also leads to possible physical explanations for the Amati [66] or Ghirlanda [67] relations between spectral peak energy and burst fluence [63,68].

Optical Emission
Prompt optical flashes, defined as optical emission detected while gamma-ray emission is still in progress and exemplified by the original detection from GRB 990123 [69], have been generally interpreted [70,71,72] as radiation from the reverse component of the external shock. However, a prompt optical flash can be produced from either an internal shock or the reverse external shock, or both [73,71]. The decay rate of the optical flux from reverse shocks is very fast (and that of internal shocks, faster yet) compared to the decay of forward shock emission, so that the forward shock component typically dominates within minutes to tens of minutes.
Observations prior to the Swift era demonstrated already that bright flashes such as that seen from GRB 990123 were relatively rare. At the same time, the slow-fading (α ∼ 0.4) optical emission from GRB 021004 and the relatively faint optical flash from GRB 021211 were both argued to have properties consistent with a reverse shock origin [74,75,76,77,78,79].
Optical/UV afterglows are now detected at early times, <100 s, in prompt observations with the Swift UVOT telescope in roughly half the bursts for which an X-ray afterglow was seen; for a detailed discussion see [7]. Of particular interest is the ongoing discussion of whether the "dark GRBs" which remain undetected by the UVOT are really optically deficient, or remain unobserved due to observational biases (e.g., the relatively blue bandpass of UVOT and the high redshifts of the Swift sample; [10]).
The fastest routine responses to Swift alerts are being realized by robotic groundbased telescopes, a large number of which have been brought on-line in recent years. The first substantial discovery yielded by these observatories has been of the gamma-ray correlated component of the prompt optical emission [36,37,38]. This component is not observed in every burst, but the mere fact of this correlation is sufficient to establish its likely origin in the burst's internal shocks [73,71]. When observed, the ratio of the correlated gamma-ray to optical flux densities has been found to be roughly 10 5 to one.
In contrast to bursts with reverse-shock flare or gamma-ray correlated emission, the typical burst is now revealed to either exhibit a single power-law decay from early times [80,81,82], or to exhibit a flat or rising light curve [83,84] before it enters the standard power-law afterglow decay. The initial brightness of the typical counterpart is V ∼ 14 to 17 mag, which has made observations challenging for the usual <1 m telescopes employed.
There are a number of possible reasons for the observed faintness of early optical emission from typical GRBs. Suppression of internal shock emission can be provided by self-absorption in the optical, coupled to the lower flux implied by the ν 1/3 low-energy asymptote of the synchrotron spectrum peaking at ∼ MeV [73]. Suppression of reverse shock emission, on the other hand, may indicate the absence or weakness of the reverse shock, e.g. if the ejecta are highly magnetized [73]. Alternatively, the deceleration might occur in the thick-shell regime (T ≫ t dec ), resulting in the reverse shock being relativistic and boosting the optical spectrum into the UV [85] (but a detection by UVOT might then be expected, unless the decay is faster than the typical 100 to 200 s UVOT response time). Another possibility, for a high comoving luminosity, is copious pair formation in the ejecta, causing the reverse shock spectrum to peak in the IR [86]. Both GRBs 990123 and 050904 (see below) have a relatively large E γ,iso ∼ 10 54 erg, so the latter seems a promising option. Even without pairs, more accurate calculations of the reverse shock [87,88] find the emission to be significantly weaker than what was estimated earlier.
Yet another possibility is that the cooling frequency in reverse shock is not much larger than the optical band frequency. In this case the optical emission from the reverse shock drops to zero very rapidly soon after the reverse shock has crossed the ejecta and the cooling frequency drops below the optical and there are no electrons left to radiate in the optical band [88].
That said, a few persuasive observations of reverse-shock optical emission have been made in the Swift era, so it can finally be said that GRB 990123 does not stand alone. The early optical/IR emission from GRB 041219 would have rivalled that seen from GRB 990123 if not for the large Galactic extinction along the line of sight [36], and of the three distinct peaks observed by PAIRITEL, the second may represent a reverse shock contribution [37] -if so, a relatively small Lorentz factor, Γ < ∼ 70, is implied. Observations of GRB 050525A with UVOT [89] and GRB 060111B with TAROT (in a unique time-resolved tracking mode; [40]) show the "flattening" light-curve familiar from GRB 021211 (as well as, probably, GRB 990123) that is termed the "type II" light curve by [75]. Evolution of the optical flux in this manner is supposed to indicate the presence of magnetized ejecta or a Poynting-flux dominated outflow.
An alternate case may be provided by GRB 060117, observed by the FRAM sky monitor telescope of the Pierre Auger Observatory [39]. This burst is the largest-fluence GRB observed by Swift to-date, and its optical flash was also very bright, peaking at R ≈ 10 mag and thus rivalling GRB 990123. Although their quality is not high, the data suggest that the forward shock peak is distinguished above the decay of the reverse shock flux, providing a potential first example of the "type I" two-peaked lightcurve that is predicted for a hydrodynamic (nonmagnetized) outflow [75].
Finally, the most exciting prompt robotic IR detection (and optical non-detection) is that of GRB 050904 [18,16], which is also though to be due to a reverse shock [90,91] (see, however, [92]). This object, at the unprecedented "very high" redshift of z = 6.29 [11], had an X-ray brightness exceeding for a day that of the brightest X-ray quasars [17], and its optical/IR brightness in the first 500 s (observer time) was comparable to that of GRB 990123, with a similarly steep time-decay slope α ∼ 3.

Early Afterglow Models
The afterglow generally becomes important after a time where the deceleration time is t dec ∼ (r dec /2cΓ 2 ) and T is the duration of the prompt outflow; t ag then marks the beginning of the self-similar blast-wave regime where Γ ∝ r −3/2 ∝ t −3/8 (in the adiabatic regime; Γ ∝ r −3 ∝ t −3/7 in the radiative regime).
Denoting the frequency and time dependence of the afterglow spectral energy flux as F ν (t) ∝ ν −β t −α , the late X-ray afterglow phases (3) and (4) described above are similar to those known previously from Beppo-SAX (for a review of this earlier behavior and its modeling see e.g. [93]). The "normal" decay phase (3), with temporal decay indices α ∼ 1.1 − 1.5 and spectral energy indices β ∼ 0.7 − 1.0, is what is expected from the evolution of the forward shock in the Blandford-McKee self-similar late time regime, under the assumption of synchrotron emission.
The late steep decay phase (4) of §1, occasionally seen in Swift bursts, is naturally explained as a jet break, when the decrease of the ejecta Lorentz factor leads to the light-cone angle becoming larger than the jet angular extent, Γ j (t) > ∼ 1/θ j (e.g. [93]). It is noteworthy, however, that this final steepening has been seen in less than ∼ 10% of the Swift afterglows, and even then for the most part only in X-rays. The corresponding optical light curve breaks have been few, and not well-constrained. This is unlike the case with the ∼ 20 Beppo-SAX bursts, for which many achromatic breaks were reported in the optical [94], while in some of the rare cases where an X-ray or radio break was reported it occurred at a different time [95]. The relative paucity of optical breaks in Swift afterglows may be an observational selection effect due to the larger median redshift, and hence fainter and redder optical afterglows at the same observer epoch, as well as perhaps reluctance to commit large telescope time on the more frequentlyreported Swift bursts (an average, roughly, of two per month with Beppo-SAX versus two per week with Swift).

Steep Decay
Among the new early afterglow features detected by Swift, the steep initial decay phase F ν ∝ t −3 − t −5 in X-rays of the long GRB afterglows is one of the most puzzling. There could be several possible reasons for this. The most immediate of these would be the cooling following cessation of the prompt emission (internal shocks or dissipation). If the comoving magnetic field in the emission region is random [or transverse], the flux per unit frequency along the line of sight in a given energy band, as a function of the electron energy index p, decays as  [71]. In some bursts this may be the explanation, but in others the time and spectral indices do not correspond well.
Currently the most widely considered explanation for the fast decay, either in the initial phase (1) or in the steep flares, attributes it to off-axis emission from regions at θ > Γ −1 (also termed curvature effects or high-latitude emission [96]). In this case, after line-of-sight gamma-ray emission has ceased, the off-axis emission observed from θ > Γ −1 is (Γθ) −6 smaller than that from the line of sight. Integrating over the equal arrival time region, this flux ratio becomes ∝ (Γθ) −4 . Since the emission from θ arrives (Γθ) 2 later than from θ = 0, the observer sees the flux falling as F ν ∝ t −2 , if the flux were frequency independent. For a source-frame flux ∝ ν ′−β , the observed flux per unit frequency varies then as i.e. α = 2 + β. This high-latitude radiation, which for observers outside the line cone at θ > Γ −1 would appear as prompt γ-ray emission from dissipation at radius r, appears to observers along the line of sight (inside the light cone) to arrive delayed by t ∼ rθ 2 /2c) relative to the trigger time, and its spectrum is softened by a Doppler factor D ∝ t −1 into the X-ray observer band. For the initial prompt decay, the onset of the afterglow (e.g. phases 2 or 3), which also comes from the line of sight, may overlap in time with the delayed high-latitude emission. In equation (2) t 0 can be taken as the trigger time, or some value comparable or less than by equation (1). This can be used to constrain the prompt emission radius [97]. When t dec < T , the emission can have an admixture of high-latitude and afterglow contributions, and since the afterglow has a steeper spectrum than the high-latitude emission (which has a prompt spectrum), one can have steeper decays [98]. Values of t 0 closer to the onset of the decay also lead to steeper slopes. Structured jets, when viewed on-beam, produce essentially the same slopes as homogeneous jets, while off-beam observing can lead to shallower slopes [99]. For the flares, if their origin is assumed to be internal (e.g. some form of late internal shock or dissipation) the value of t 0 is just before the flare, near the observed time of flare onset [100]. This interpretation appears, so far, compatible with most of the Swift afterglows [15,14,101]. Alternatively, the initial fast decay could be due to the emission of a cocoon of exhaust gas [102], where the temporal and spectral index are explained through an approximately power-law behavior of escape times and spectral modification of multiplyscattered photons. The fast decay may also be due to the reverse shock emission, if inverse Compton interactions up-scatter the (primarily synchrotron) optical photons into the X-ray range. The decay starts after the reverse shock has crossed the ejecta and electrons are no longer accelerated, and may have both line-of-sight and off-axis components [103]. This poses strong constraints on the Compton y parameter, and cannot explain steeper decays with α > 2, or α > 2 + β if the off-axis contribution dominates. Models involving bullets -whose origin, acceleration and survivability is unexplained -could give a prompt decay index α = 3 to 5 [104], but imply a bremsstrahlung energy index β ∼ 0 which is not observed in the fast decay, and require fine-tuning. Finally, a "patchy shell" model, where the Lorentz factor is highly variable in angle, would produce emission with α ∼ 2.5. Thus, such mechanisms may explain the more gradual decays, but not the more extreme α = 5 to 7 values encountered in some cases.

Shallow Decay
The slow decay portion of the X-ray light curves (α ∼ 0.3 to 0.7), quite ubiquitously detected by Swift, is not entirely new, having been detected in a few cases by Beppo-SAX. This, as well as the appearance of wiggles and flares in the X-ray light curves several hours after the burst, were the motivation for the "refreshed shocks" scenario [105,106]. Refreshed shocks can flatten the afterglow light curve for hours or days, even if the ejecta is all emitted promptly at t = T < ∼ t γ , but with a range of Lorentz factors, say M(Γ) ∝ Γ −s , where the lower Γ shells arrive much later to the foremost fast shells which have already been decelerated. Thus, for an external medium of density ρ ∝ r −g and a prompt injection where the Lorentz factor spread relative to ejecta mass and energy is M(Γ) ∝ Γ −s , E(Γ) ∝ Γ −s+1 , the forward shock flux temporal decay is given by [106] It needs to be emphasized that in this model all the ejection can be prompt (e.g. over the duration ∼ T of the gamma ray emission) but the low Γ portions arrive at (and refresh) the forward shock at late times, which can range from hours to days. That is, it is not the central engine which is active late; rather, its effects are seen late. Fitting such refreshed-shock models to the shallow decay phases in Swift bursts [107] leads to a Γ distribution which is a broken power law, extending above and below a peak around ∼ 45.
An alternate version of refreshed shocks does envisage central engine activity extending for long periods of time, e.g. < ∼ day (in contrast to the < ∼ minutes engine activity in the model above). Such long-lived activity may be due to continued fall-back into the central black hole [108] or to a magnetar wind [109]. One characteristic of both types of refreshed models is that after the refreshed shocks stop and the usual decay resumes, the flux level shows a step-up relative to the previous level, since new energy has been injected.
From current analyses, the refreshed shock model is generally able to explain the flatter temporal X-ray slopes seen by Swift, both when it is seen to join smoothly on the prompt emission (i.e. without an initial steep decay phase) or when seen after an initial steep decay. Questions remain concerning the interpretation of the fluence ratio in the shallow X-ray afterglow and the prompt gamma-ray emission, which can reach E X /E γ < ∼ 1 [98]. This requires a higher radiative efficiency in the prompt gamma-ray emission than in the X-ray afterglow. One could speculate that this might be achieved if the prompt outflow were Poynting-dominated. Alternatively, a highly-efficient afterglow might emit a large fraction of its energy in other (unobserved) bands, e.g. in the GeV or IR. Or [110] a previous mass ejection might have emptied a cavity into which the ejecta moves, leading to greater efficiency at later times, or otherwise causing the energy fraction going into electrons to increase ∝ t 1/2 .

X-ray Flares
Refreshed shocks can also explain some of the X-ray flares whose rise and decay slopes are not too steep. However, this model encounters difficulties with the very steep flares with rise or decay indices α ∼ ±(5 − −7), such as inferred from the giant flare of GRB 050502b [111] around 300 s after the trigger. Also, the flux level increase in this flare is a factor ∼ 500 above the smooth afterglow before and after it, implying a comparable energy excess in the low-versus high-Γ material. An explanation based on inverse Compton scattering in the reverse shock [103] can explain a single flare at the beginning of the afterglow, as long as the subsequent flux decay is not too steep. For multiple flares, models invoking blastwave interaction with a lumpy external medium have generic difficulties explaining steep rises and decays [15], although extremely dense, sharp-edged lumps, if they exist, might satisfy the steepness criterion [112].
Currently the more widely-considered model for the flares ascribes them to late central engine activity [15,14,101]. The strongest arguments in favor of this are that the energy budget is more easily satisfied, and the fast rises and decays more straightforward to explain. In such a model the flare energy can be comparable to the prompt emission, and the fast rise comes naturally from the short time variability leading to internal shocks (or to rapid reconnection), while the rapid decay may be explained as the high-latitude emission following the flare, with t 0 reset to the beginning of each flare (see discussion in [100]). Considering the full population of X-ray flares, some are well-modeled by refreshed forward shocks, while in others this is clearly ruled out and a central engine origin is better suited [113]. Aside from the phenomenological desirability based on energetics and timescales, a central engine origin is conceivable, within certain time ranges, based on numerical models of the core collapse scenario for long bursts. These invoke the core collapse of a massive stellar progenitor, where continued infall into the fast-rotating core can continue for a long time [108]. However, large flares with a fluence which is a sizable fraction of the prompt emission occurring hours later remain difficult to understand. It has been argued that gravitational instabilities in the infalling debris torus can lead to lumpy accretion [114]. Alternatively, if the accreting debris torus is dominated by magneto-hydrodynamic (MHD) effects, magnetic instabilities can lead to extended, highly time-variable accretion [115].

Short Burst Afterglows
Swift and to a lesser extent HETE-2 have provided the first bonafide short burst afterglows, with the first observations beginning (in some cases) within <100 s of the trigger, leading in turn to arcsec and sub-arcsec localizations, host galaxy identifications, and redshifts (although, as yet, none via direct absorption spectroscopy).
In the first short burst afterglow, GRB 050509b [25], the extrapolation of the prompt BAT emission into the X-ray range, along with the XRT light curve from 100 to 1000 s, can be fitted with a single power law of α ∼ 1.2. The X-ray coverage was sparse due to orbital constraints and the faintness of the afterglow, and the number of detected X-ray photons was small. No optical transient was detected, but an elliptical host galaxy was identified at z = 0.225 (e.g. [21]). In GRB 050709 an optical transient was identified, as well as a host galaxy [116], an irregular galaxy at z = 0.16 (observations also ruled out any supernova association).
GRB 050724 was relatively bright, and in addition to an X-ray afterglow observed with the XRT and Chandra [26], yielded decaying optical and radio afterglows [117]. This burst, as with roughly half the short burst afterglows seen to-date, is associated with an elliptical host galaxy. It also had a low-luminosity, soft gamma-ray extension of the short-hard gamma-ray component (which would have been missed by BATSE), and it had an interesting X-ray afterglow extending beyond 10 5 s, with no jet break seen to the limits of the final Chandra observation [27]. The soft gamma-ray extension, lasting up to 200 s, connects smoothly with the beginning of the XRT afterglow, which has α ∼ 2 between 100 and 300 s, and then enters a much steeper decay with α ∼ 5 to 7 out to ∼ 600s, followed by a more moderate decay α ∼ 1. An unexpected feature is a strong flare peaking at 5 × 10 4 s, with an energy 10% that of the prompt emission, and an amplitude which represents a 10-fold increase over the preceding slow decay.
GRB 051221A, the next panchromatic short burst afterglow, has an extended light curve in the X-ray [28] and optical [22]. This was the brightest short burst observed by Swift to date, and occurred in a star-forming galaxy at redshift z = 0.5464. The X-ray light curve also provides the clearest evidence of "standard" energy injection, with a flattening from 1.4 to 3.4 hours after the burst which increases the afterglow energy by a factor of ∼3, and is moreover reflected in contemporaneous radio detections from the VLA.
With seven afterglows and five relatively secure redshifts in hand via host galaxy identifications, the distribution of short bursts in redshift and among host galaxy types -including an equal number of spiral/irregular and elliptical hosts -is typical of an old ( > ∼ Gyr) population of progenitors, such as neutron star (NS) binaries or black holeneutron star binaries [118].
The main challenges posed by the short burst afterglows are the relatively long, soft tail of the prompt emission, and the strength and late occurrence of the flares. A possible explanation for the extended long soft tails (∼100 s) may be that the compact binary progenitor is a black hole -neutron star system [26], since analytical and numerical arguments ( [119], and references therein) suggest that disruption of the NS and its disappearance into the black hole may lead to a complex and extended accretion phase significantly longer than for double neutron stars. The flares, for which the simplest interpretation might be as refreshed shocks (compatible with a short engine duration T < ∼ t γ ∼ 2 s, for ejecta with an extended Lorentz factor distribution), require the energy in the slow material to be at least ten times more energetic than the fast material responsible for the prompt emission in the specific case of the GRB 050724 flare at 10 4 s. The rise and decay times are moderate enough for this interpretation. Another interpretation invokes the accretion-induced collapse of a neutron star in a binary, leading to a flare when the fireball created by the collapse hits the companion [120], which might explain moderate energy one-time flares. However, for repeated, energetic flares, as also seen from long bursts, the total energetics are easier to satisfy if one postulates late central engine activity (lasting at least half a day), containing ∼ 10% of the prompt fluence [26]. A possible way to produce this might be via temporary "choking" of an MHD outflow [115] (c.f. [121]), which might imply linear polarization of the X-ray flare [122]. Such MHD effects could plausibly also explain the initial ∼100 s soft tail. However, a justification for substantial > ∼ 10 5 s features remains so far on tentative grounds.
The similarity of the X-ray afterglow light curves to those of long bursts is, in itself, an argument in favor of the prevalent view that the afterglows of both long and short bursts can be described by the same paradigm, independently of any differences in the progenitors. This impression is reinforced by the fact that the X-ray light curve temporal slope is, on average, that expected from the usual forward shock afterglow model, and that in two short bursts (so far) there is evidence for what appears to be a jet break [116,28,22]. However, while very similar, the first-order differences are revealing: the average isotropic energy of the short bursts is a factor of ∼100 smaller, while the average jet opening angle (based on two breaks and one lower limit) is a factor of ∼ 2 larger [116,27,28,22]. Using the standard afterglow theory, the bulk Lorentz factor decay can be expressed through Γ(t d ) = 6.5(n o /E 50 ) 1/8 t For the first three well-studied afterglows, GRBs 050709, 050724, and 051221A, these equations, together with the standard afterglow expressions for the flux level as a function of time before and after the break, lead to fits [123,28,22] which are not completely determined, allowing for GRBs 050709 and 051221A either a very low or a moderately low external density, and for GRB 050724 a moderately low to large external density. The main uncertainties are in the jet break times, which are poorly sampled, and in the absence of high-quality radio data for any burst (GRB 050724 had a reasonable radio flux, but too much variability in all bands). More bright short bursts like GRB 051221A will be needed to improve the jet break statistics substantially.

The Supernova Component
GRB-associated supernovae have been regularly observed as "red bumps" in the light curves of low-redshift GRBs, and more rarely, by direct spectroscopic observation at 7 to 30 days after the GRB. The observation of the GRB-supernova GRB 060218/SN 2006aj with Swift, however, has raised the prospect of an entirely new component of emission at early times, namely, the shock breakout of the simultaneous supernova explosion. The thermal component of the prompt XRT emission, which evolves into the UVOT range over the course of the first few hours after the burst, has been interpreted as the shock break-out of the GRB/SN in an optically-thick wind of a Wolf-Rayet progenitor star [29] (but see also [34]). It is important that the wind be sufficiently dense; given the known redshift of this burst, the blackbody radius of the early XRT/UVOT thermal component is roughly 100 times the size of a Wolf-Rayet star. At the same time, however, the absence of hydrogen features (e.g. Hα) in the SN spectrum requires such a progenitor, and disallows a red giant progenitor, which would produce a type II SN.
The observation of a single such event in one year of Swift observations offers the tantalizing prospect of gathering more examples with each additional year of Swift operations. In particular, intensive ground-based observations tracking the evolution of this thermal component to near-infrared, millimeter, and radio wavelengths will be useful in confirming or refuting the shock-breakout interpretation.