Gamma-ray burst afterglows

0.25 MB·

by  Bing Zhang

Gamma-RayBurst Afterglows
7
Bing Zhang
0
0 Department of Physics, Universityof Nevada, Las Vegas,USA
2 Email: [email protected]

n
a
J

0 Abstract
1
Extended,fadingemissionsinmulti-wavelengthareobservedfollowing Gamma-ray bursts(GRBs).Recentbroad-

2 band observational campaigns led by the Swift Observatory reveal rich features of these GRB afterglows. Here we
v reviewthelatestobservationalprogressanddiscussthetheoreticalimplicationsforunderstandingthecentralengine,
4 composition, andgeometric configuration of GRBjets, as well astheirinteractions with theambient medium.
7
7
1 Keywords:
1 Gamma-RayBursts, SwiftObservatory, X-rays,optical, radio
6
0
/
h
1. Introduction long-term broad band afterglow (M´esza´ros & Rees
p
1997; Sari et al. 1998); and a short-lived reverse
-
o Gamma-raybursts(GRBs)arethemostviolent shock propagating into the ejecta itself gives rise
r
t explosionsintheuniverse.Theyarerelativisticout- to a possible short-term optical/IR flash and a ra-
s
a flows launched during collapses of massive stars or dio flare (M´esza´ros & Rees 1997, 1999; Sari & Pi-
: mergersofcompactobjects.Regardlessofthenature ran1999a,b).The relativistic ejecta arelikely colli-
v
i oftheexplosion,agenericfireballshockmodel(Rees mated (Rhoads 1999;Sarietal.1999),andthe jets
X &M´esza´ros1992,1994;M´esza´ros&Rees1993,1997, may have substantial angular structures (Zhang &
r for reviews see Piran 1999, 2005; M´esza´ros 2002, M´esza´ros2002;Rossietal.2002).Thisgeneralthe-
a
2006;Zhang&M´esza´ros2004)isfoundsuccessfulto oretical framework has been successful to interpret
interpret the broad GRB phenomenology. Accord- mostofthe observationaldatainthe pre-Swiftera.
ingtothismodel,theejectaisintrinsicallyintermit- ThesuccessfullaunchandoperationofNASA’s
tent and unsteady, and is composed of many mini- broadband(gamma-ray,X-ray,UV&optical)GRB
shellswithawiderangeofbulkLorentzfactors.In- mission Swift (Gehrels et al. 2004) opens a brand
ternalshocks(Rees&M´esza´ros1994)arelikelyde- neweraintheGRBstudy.Thepromptslewingcapa-
veloped before the global fireball is decelerated by bility ofthe X-RayTelescope (XRT, Burrowsetal.
the ambient medium, which are generally believed 2005a)andUV-OpticalTelescope(UVOT,Roming
to be the emission sites of the observed prompt et al. 2005) allows the satellite to swiftly catch the
GRBemission.Thefireballisdeceleratedatalarger very early X-ray and UV/optical signals following
distanceaftersweepingenoughinterstellarmedium the GRB prompt emission detected by the Burst
whoseinertiabecomesnoticeable,andtheblastwave AlertTelescope(BAT,Barthelmyetal.2005a).The
gradually enters a self-similar deceleration regime preciselocalizationsmadebyXRTforthemajority
(Blandford-McKee1976).Upondeceleration,apair of the bursts make it possible for ground-basedfol-
of shocks forms. A long-lived forward shock prop- low up observations of most bursts. We now have
agating into the ambient medium gives rise to the unprecedented information about GRB afterglows,
Accepted forpublication inElsevier
1997; Sari et al. 1998; Chevalier & Li 2000). A
systematic test of the afterglow closure-relations
(e.g.Table1ofZhang&M´esza´ros2004)suggests
however that a fraction of bursts do not satisfy
0
~ -3 anyafterglowmodel(Willingale etal.2006).
V – PostJetbreakphase(IV):Occasionallyobserved
I
following the normal decay phase, typically with
a decay slope ∼ −2, satisfying the predictions of
II the standard jet model (Rhoads 1999; Sari et al.
~ -0.5 t :104-105 s
III b3 1999) or the structured jet model (Rossi et al.
~ -1.2 2002;Zhang& M´esza´ros2002).
– X-ray flares (V): Appear in nearly half of GRB
t :102-103 s t :103-104 s ~ -2
b1 b2 IV afterglows. Sometimes multiple flares harbor in
one GRB. Typically have very steep rising and
decaying slopes (Burrows et al. 2005b; Falcone
et al. 2006; Romano et al. 2006) with δt/t ≪
1. Appear in both long-duration (Falcone et al.
Fig. 1. A canonical X-ray afterglow lightcurve revealed by
2006) and short-duration GRBs (Barthelmy et
SwiftXRTobservations (fromZhang et al.2006)
al.2005b;Campanaet al.2006a),andboth hard
which sheds light onto many outstanding problems GRBsandsoftX-rayflashes(Romanoetal.2006).
in the pre-Swift era (Zhang & M´esza´ros 2004 for a Except for the normal decay and the jet-break
summary):e.g.centralengine,compositionandge- phases, all the other three components were not
ometric configuration of the GRB fireball, and its straightforwardlyexpectedinthepre-Swiftera1.As
interactionwiththe ambientmedium. ofthetimeofwriting,thesteepdecayphaseandX-
ray flares are better understood, while the shallow
decayphase isstilla mystery.
2. AcanonicallightcurveofX-rayafterglows
2.1. Steepdecay phase: tailof theprompt emission
OneofthemajordiscoveriesofSwiftistheiden-
tification of a canonical X-ray afterglow behavior
(Nousek et al. 2006; Zhang et al. 2006; O’Brien et The generally accepted interpretation of the
al.2006,Chincarinietal.2005;seeFig.1).Although steep decay phase is the tail emission due to the
different afterglow lightcurves may vary from one so-called “curvature effect” (Fenimore et al. 1996;
another,theyareallcomposedofseveralofthefive Kumar & Panaitescu 2000; Zhang et al. 2006;
componentsillustratedinFig.1. Panaitescuetal.2006a;Dyksetal.2006).Thebasic
– Steep decay phase (I): Typically smoothly con- assumption of this interpretation is that the GRB
nected to the prompt emission (Tagliaferri et al. emission region is disconnected from the afterglow
2005; Barthelmy et al. 2005b), with a temporal region (the external shock), and that the emission
decay slope ∼ −3 or steeper (sometimes up to ∼ from the GRB emission region ceases abruptly.
−10, e.g. Vaughan et al. 2006; Cusumano et al. This is consistent with the conjectures of internal
2006;O’Brienet al.2006)extending to ∼(102− shocks or other internal dissipation mechanisms
103)s.Usuallyhaveadifferentspectralslopefrom (e.g. magnetic fields reconnection, etc). Since it is
the laterafterglowphases. generally assumed that the ejecta has a conical ge-
– Shallow decay phase (II): Typically with a tem- ometry, the curvature of the radiation front causes
poral decay slope ∼ −0.5 or flatter extending to apropagationdelayforhigh-latitudeemissionfrom
∼ (103−104)s, at which a temporal break is ob- the line of sight. Combining with the variation of
servedbefore the normaldecay phase (e.g. Cam- the Doppler factor at different latitudes, one gets a
pana etal.2005;De Pasqualeet al.2006).There simple prediction
is nospectralevolutionacrossthe break.
– Normal decay phase (III): Usually with a decay 1 The flare-like signature was seen by Beppo-SAX, but it
slope∼−1.2,andusuallyfollowsthe predictions
was interpreted as the onset of the afterglow (Piro et al.
ofthestandardafterglowmodel(M´esza´ros&Rees 2005).
2
α=2+β (1) – Pe’eretal.(2006)suggestthattheemissionfrom
the relativistically expanding hot plasma “co-
for the emission outside the Γ−1 emission cone,
coon”associatedwiththeGRBjetcouldalsogive
wheretheconventionF ∝t−αν−β isadopted.The
ν riseto the steepdecayphaseobservedbySwift.
salientfeature ofthis interpretationis thatitcould
Motivated by the discovery of the spectrally
bedirectlytestedsincebothαandβ couldbemea-
evolving tails in GRB 050724 (Campana et al.
sured directly from the observational data, given
2006b)andGRB060614(Gehrelsetal.2006;Zhang
thattwocomplicationsaretreatedproperly(Zhang
et al. 2007a), recently Zhang et al. (2007c) per-
etal.2006):First,forinternalemissions,everytime
formed a systematic time-dependent spectral anal-
when the central engine restarts, the clock should
ysis of 17bright steep decay tails.They found that
be re-setto zero2. Inthe log−log lightcurves,this
while 7 tails show no apparent spectral evolution,
usually introduces an “artificial” very steep decay
the other 10 do. A simple curvature effect model
if the GRB trigger time (which is usually taken as
invoking an angle-dependent spectral index cannot
t = 0) significantly leads the time zero point (t0) interpretthedata.Thissuggeststhatthecurvature
of the corresponding emission episode. Second, the
effect is not the sole factor to control the steep de-
observeddecayisthesuperpositionofthecurvature
cay tail phase at least in some bursts. Zhang et al.
decay and underlying afterglow decay from the ex-
(2007c) show that some of the spectrally evolving
ternal shock.One needs to subtract the underlying
tailsmaybeinterpretedassuperpositionofthecur-
afterglow contribution before performing the test.
vature effect tail and an underlying central engine
Thecredibilityofthecurvatureeffectinterpretation
afterglow,whichissoftbutdecays“normally”.Such
to the steep decay phase is that by properly tak-
a component has been seen in GRB 060218 (Cam-
ing into account the two effects mentioned above,
pana et al. 2006b),which cannot be interpreted by
the steep decay is consistent with Eq.(1) with t0 the standard external shock afterglow (Willingale
shifted to the beginning ofthe lastpulse ofprompt
etal.2006).ThestrongspectralevolutionsinGRB
emission(Liang etal.2006)atleastinsomecases.
050724, GRB 060218, and GRB 060614, however,
Besides the standard curvature effect model,
cannot be interpreted with such a model. They are
otherinterpretationsforthesteepdecayphasehave
interpreted as internal shock afterglows by Zhang
beendiscussedinthe literature.
etal.(2007c).
– Insomecases,thesteep-decayslopemaybeshal-
lower than the expectation of the curvature ef-
fect3. This would suggest that the central en- 2.2. X-rayflares:restartingthecentralengine
gine may not die abruptly or the shocked region
maynotcoolabruptly,butratherdecaywithtime
The X-ray flares have the following observa-
gradually,leadingtoadecayingafterglowrelated
tional properties (Burrows et al. 2005b; Chincar-
to the central engine (Fan & Wei 2005;Zhang et
ini et al. 2007): Rapid rise and fall times with
al. 2006). This was taken by Fan et al. (2006) to
δt/t ≪ 1; many light curves have evidence for
peak
interpret the abnormal power law decay of GRB
a same decaying afterglow component before and
060218(Campanaetal.2006b).
after the flare; multiple flares are observed in some
– Yamazakietal.(2006)studythe curvatureeffect
bursts with similar properties; large flux increases
of an inhomogeneous fireball (mini-jets). They
at the flares; typically degrading fluence of flares
found that the decay tail is generally smooth,
withtime,butinrarecases(e.g.GRB050502B)the
but sometimes couldhave structures, which may
flare fluence could be comparable with that of the
interpretthe small-scalestructure in some of the
promptemission;flaressoftenastheyprogress;and
decaytails.
later flares are less energetic and more broadened
than early flares. These properties generally favor
2 Wenoticethatforexternalshockrelatedemissions,taking
the interpretation that most of them are not asso-
the GRB trigger time as the time zero point is generally
required (Lazzati & Begelman 2006; Kobayashi & Zhang ciated with external-shock related events. Rather
2006). they are the manifestations of internal dissipations
3 It is worth noticing that generally a decay slope steeper at later times, which requires restarting the GRB
than the curvature effect prediction is not allowed, unless
central engine (Burrows et al. 2005b; Zhang et al.
the jet is very narrow. Usually, even if the intrinsictempo-
2006; Fan & Wei 2005; Ioka et al. 2005; Wu et al.
ral decay slope is steeper than Eq.(1), the curvature effect
nonetheless takes over to define the decay slope. 2006;Falcone et al.2006;Romano et al.2006;Laz-
3
zati & Perna 2006). Compared with the external promptemissionis over.
shock related models, the late internal dissipation The late internal dissipation model of X-ray
models have the following two major advantages flaresisalsotestedbyLiangetal.(2006).Thesame
(Zhang et al. 2006): First, since the clock needs to logic of testing the steep decay component is used.
bere-seteachtimewhenthecentralenginerestarts, The starting assumption is that the decay of X-ray
it is very natural to explain the very sharp rising flaresarecontrolledbythecurvatureeffectafterthe
and falling lightcurves of the flares. Second, ener- abruptcessationofthe internaldissipation,sothat
getically the late internal dissipation model is very Eq.(1)isassumedtobevalid.Aftersubtractingthe
economical. While in the refreshed external shock underlying forward shock afterglow contribution,
models a large energy budget is needed (the injec- Liang et al. (2006) search for the valid zero time
tion energy has to be at least comparable to that points (t0) for each flare to allow the decay slope
alreadyintheblastwaveinordertohaveanysignif- satisfying the requirement of the curvature effect
icantinjection signature,Zhang & M´esza´ros2002), model. If the hypothesis is correct, t0 should be
the internal model only demands a small fraction generally before the rising segment of each flare.
of the prompt emission energy to account for the Thetestingresultsareimpressive:Mostoftheflares
distinctflares. indeed have their t0 at the beginning of the flares.
The leading candidate of the late internal dis- This suggeststhatthe internaldissipationmodelis
sipation model is the late internal shock model. In robust for most of the flares. It is worth emphasiz-
such a model, the collisions could be between the ing that even the late slow bump at around 1 day
fastshellsinjectedlaterandtheslowshellsinjected following the short GRB 050724 (Barthelmy et al.
earlierduringthepromptphase(e.g.Zouetal.2006; 2005c;Campanaetal.2006a)satisfiesthecurvature
Staff et al. 2006) or between two shells injected at effect model, suggesting that the central engine is
later times (see Wu et al. 2006 for a categoriza- still active even at 1 day after the trigger. This is
tion of different types of collisions). One concern is alsoconsistentwiththelateChandraobservationof
whether later collisions between two slow shells in- this burst (Grupe et al. 2006a) that indicates that
jected during the prompt phase could give rise to the afterglow resumes to the pre-flare decay slope
theobservedX-rayflares.Thisisgenerallynotpos- after the flare.
sible.Theinternalshockradiuscanbeexpressedas Having identified the correct model for the
Ris ∼ d0/(βf −βs) ∼ Γ2sd0, where d0 is the initial flare phenomenology, one is asked about a funda-
separationbetween the two colliding shells, β and mental question: how to restart the central engine.
f
β are the dimensionless velocities of the fast and No central engine models in the pre-Swift era have
s
slowshells,respectively,andΓ istheLorentzfactor specifically predicted extended activities far after
s
oftheslowshell.Thesecondapproximationisvalid thepromptemissionphase.PromptedbytheX-ray
ifΓ ≫Γ .Inordertoproducelateinternalshocks, flare observations, the following suggestions have
f s
the two slow shells must both have a low enough been made recently, and none is proved by robust
Lorentzfactorsothatatthetimeofcollisiontheydo numericalsimulationsatthe moment.
not collide with the decelerating blastwave.Also in – Fragmentationorgravitationalinstabilitiesinthe
ordernottocollidewitheachotherearlier,theirrel- massivestarenvelopes.Kingetal.(2005)argued
ative Lorentz factor ∆Γ must be very small. When thatthecollapseofarapidlyrotatingstellarcore
they collide, the internal energy [∝ (Γ −1) ≪ 1] leads to fragmentation.The delay of accretionof
sf
is usually too small to give rise to significant emis- some fragmentsafter the majoraccretionleadto
sion (where Γ ∼ (Γ /Γ +Γ /Γ )/2 is the rela- X-rayflaresfollowingcollapsar-relatedGRBs.
sf f s s f
tive Lorentz factor between the two shells). Should – Fragmentation or gravitational instabilities in
suchacollisionoccur,mostlikelyithasnointerest- the accretion disk. Observations of GRB 050724
ing observational effect (see Lazzati & Perna 2006 (Barthelmy et al. 2005c; Campana et al. 2006a;
for more detailed discussion on this issue). Gener- Grupeetal.2006a),ashortGRBassociatedwith
ally, in the internal shock model the observed time an elliptical host galaxy that is consistent with
sequence reflects the time sequence in the central the compact star merger progenitor model, re-
engine (Kobayashiet al.1997).As a result, the ob- vealthatitisalsofollowedbyseveralX-rayflares
served X-ray flares (102 − 105)s after the prompt startingfrom10sofsecondsallthewayto∼105s.
emissionmustimplythatthecentralenginerestarts The properties of these X-ray flares are similar
during this time span, say,as late as days after the to those in long GRBs. The requirement that
4
bothlongandshortGRBs shouldproduce X-ray 2.3. Shallow decay phase: stillamystery
flares with similar properties prompted Perna et
al. (2006) to suggest that fragmentation in the The shallowdecayphasecouldfollowthesteep
accretion disk, the common ingredient in both decayphaseorimmediatelyfollowthepromptemis-
long and short GRB models, may be the agent sion(O’Brienet al.2006;Willingale etal.2006).It
forepisodic accretionthatpowersthe flares. isverylikelyrelatedtotheexternal-shock-originaf-
– Magnetic barrier. Based on the MHD numerical terglow. However, the origin of this shallow decay
simulations in other contexts and theoretical ar- phase is more difficult to identify, since there exist
guments, Proga & Zhang (2006) argue that the severaldifferentpossibilitiesthatarenoteasytodif-
magneticbarrierneartheblackholemayactasan ferentiate among each other from the X-ray obser-
effectivemodulatoroftheaccretionflow.Theac- vations. The fact that the spectral index does not
cretion flow can be intermittent in nature due to change across the temporal break from the shallow
theroleofmagneticfields.Thismodeldoesnotre- decay phase to the normal decay phase rules out
quiretheflowbeingchopped(e.g.duetofragmen- the models that invoke crossing of a spectral break
tationorgravitationalinstabilities)atlargerradii acrossthe band.The nature ofthe breakshouldbe
before accretion, although in reality both pro- then eitherhydrodynamicalorgeometrical.
cessesmayoccuraltogether.Themagneticbarrier Followingmodelshavebeendiscussedinthelit-
model is in accordance with the magnetic origin erature.
of X-ray flares based on the energetics argument – Energyinjectioninvokingalong-termcentralen-
(Fanetal.2005c). gine. The most straightforward interpretation of
– NS-BHmerger.FlaresinGRB050724(Barthelmy the“shallower-than-normal”phaseisthattheto-
etal.2005c)pose greatchallengeto the previous tal energy in the external shock continuously in-
compact star merger models. Numerical simula- creases with time. This requires substantial en-
tionofNS-NSmergerstypicallygivesashortcen- ergy injection into the fireball during the phase
tralenginetimescale(0.01-0.1)s,ifthefinalprod- (Zhangetal.2006;Nouseketal.2006;Panaitescu
uctisaBH-torussystem(Aloyetal.2005).Inor- et al. 2006a). There are two possible energy in-
der to account for the late time flares in 050724, jection schemes (Zhang etal. 2006;Nousek et al.
Barthelmyetal.(2005c)suggestapossibleNS-BH 2006). The first one is to simply invoke a long-
mergerprogenitorsystem.Numericalsimulations lasting central engine, with a smoothly varying
of BH-NS merger systems have been performed. luminosity, e.g. L ∝ t−q (e.g. Zhang & M´esza´ros
Although X-ray flares at 100s of seconds or later 2001).Inordertogiveinterestinginjectionsigna-
stillchallengethemodel,extendedaccretionover ture q <1 is required;otherwisethe totalenergy
severalsecondscouldbereproduced(Faberetal. intheblastwaveessentiallydoesnotincreasewith
2006;cf.Rosswog2005). time.Suchapossibilityisvalidforthecentralen-
– NS-NSmergerwithapostmergermillisecondpul- ginesinvokingaspinningdownpulsar(Dai&Lu
sar. Dai et al. (2006a) argue a possible solution 1998; Zhang & M´esza´ros 2001; Fan & Xu 2006)
for the extended X-ray flares following merger- oralong-lastingBH-torussystem(MacFadyenet
type GRBs. Numerical simulations have shown al. 2001). One possible signature of this scenario
that the product of a NS-NS merger may not be thatdifferentiatesitfromthevarying-Γmodeldis-
aBH(Shibataetal.2005),iftheNSequation-of- cussed below is that a strong relativistic reverse
state is stiff. Instead, the final product may be a shockisusuallyexpected,ifattheshockinteract-
differentially-rotatingmassiveneutronstar.Ifthe ingregiontheσ-parameter(theratiobetweenthe
initialmagneticfieldsoftheNSisnotstrong,the Poyntingfluxandthekineticflux)isdegradedto
α−Ω dynamoactionwouldinduce magnetic ex- belowunity (Dai2004;Yu&Dai2006).Alterna-
plosions that give rise to late internal shocks to tively,ifσ isstillhighattheshockregion,there-
produceX-rayflares(Daietal.2006a). verseshockmaybeinitiallyweak,butwouldstill
– Multi-stagecentralengine.Gao&Fan(2006)and becomerelativisticiftheenginelastslongenough
Staff et al. (2006) proposed multi-stage central (i.e. this is effectively a rather thick shell, Zhang
engine modelsto interpretX-rayflares. & Kobayashi2005). The observational data sug-
gestarangeofqvalueswithtypicalvalueq ∼0.5.
Thisisdifferentfromtherequirementoftheana-
5
lytical pulsarmodel (q =0). However,numerical the shallow decay phase. For those cases with a
calculations suggest that a pulsar model can fit straight shallow decay lightcurve, one needs to
someoftheXRTlightcurves(Fan&Xu2006;De incorporate the steep decay tail to mimic the
Pasqualeetal.2006;Yu &Dai2006). observations.
– Energy injection from the ejecta with a wide Γ- – Off-beamjetmodel.Geometricallyonecaninvoke
distribution.Thismodelinvokesadistributionof an off-beam jet configuration to account for the
the Lorentzfactorofthe ejectawithlow-Γejecta shallowdecay.Eichler&Granot(2006)showthat
lagging behind the high-Γ ones, and only piling ifthelineofsightisslightlyoutsidetheedgeofthe
uptotheblastwavewhenthehigh-Γpartisdecel- jet that generates prominent afterglow emission,
erated (Rees & M´esza´ros 1998). In order to pro- a shallow decay phase can be mimicked with the
duceasmoothpowerlawdecay,theΓ-distribution combination of the steep decay GRB tail. Toma
needs to be close to a powerlaw with M(>Γ)∝ etal.(2006)discussedasimilarmodelwithinthe
Γ−s.Asignificantenergyinjectionrequiress>1. frameworkofthe patchyjet models.
Thetemporalbreakaround(103−104)ssuggests – Two-component jet model. A geometric model
acutoffofLorentzfactoraroundseveral10’s,be- invoking two jet components could also fit the
lowwhichsbecomesshallowerthanunity(Zhang shallow-decay data, since additional free param-
et al. 2006). Granot & Kumar (2006) have used eters are invoked (Granot et al. 2006; Jin et al.
this propertyto constrainthe ejecta Lorentzfac- 2006).
tordistributionofGRBswithintheframeworkof – Precursormodel.Iokaetal.(2006)suggestthatif
this model. The reverse shock of this scenario is thereisaweakprecursorleadingthemainburst,a
typicallynon-relativistic(Sari&M´esza´ros2000), shallowdecayphasecanbeproducedasthemain
since the relative Lorentz factor between the in- fireballsweepsthe remnantsofthe precursor.
jectionshellandtheblastwaveisalwayslowwhen – Varying microphysics parameter model. One
the formerpiles upontothe latter. could also invoke evolution of the microphysics
– Delayed energy transfer to the forward shock. shockparameterstoreproducetheshallowdecay
Analytically, the onset of afterglow is estimated phase(Iokaetal.2006;Fan&Piran2006;Granot
to be around t = max(t ,T), where t ∼ etal.2006;Panaitescuetal.2006b).
dec γ γ
5 s(EK,52/n)1/3(Γ0/300)−8/3(1 + z) is the time – Dust scattering model. Shao & Dai (2006) sug-
scale at which the fireball collects Γ−1 of the gestthatsmallanglescatteringofX-raysbydust
rest mass of the initial fireball from the ISM, couldalsogiverisetoashallowdecayphaseunder
and T is the duration of the explosion. The so- certainconditions.
called “thin” and “thick” shell cases correspond Can different possibilities be differentiated by
to t >T andt <T,respectively(Sari&Piran themoreabundantdata?Itseemstobeachalleng-
γ γ
1995; Kobayashi et al. 1999). Numerical calcula- ing task. The author is inclined to the first three
tions suggest that the time scale before entering interpretationsontheabovelist.Forthetwoenergy
the Blandford-McKee self-similar deceleration injectionmodels,oneexpectsdifferentreverseshock
phaseislong,oforderseveral103s(Kobayashi& signatures (i.e. relativistic reverse shock for the
Zhang2006).This suggeststhat ittakestime for long-termcentralenginemodelandnon-relativistic
thekineticenergyofthefireballtobetransferred reverseshock for the varying-Γmodel). This would
to the medium. In a high-σ fireball, there is no give different radio emission properties at early
energy transfer during the propagation of a re- times. On the other hand, the uncertainty of the
verseshock(Zhang&Kobayashi2005).Although compositionofthecentralengineoutflow(e.g.theσ
energy transfer could happen after the reverse parameter)wouldmakethereverseshocksignature
shock disappears, this potentially further delays oftheformermodelmoreobscured.Thedelayeden-
the energy transfer process (although detailed ergytransfermodel(thethirdoneontheabovelist)
numerical simulations are needed to verify this). is the simplest.Ifitis correct,the so-calledshallow
The shallow decay phase may simply reflect the decay phase is nothing but a manifestation of the
slow energy transfer process from the ejecta to onset of afterglow (Kobayashi& Zhang 2006).The
theambientmedium.Thismodel(e.g.Kobayashi peak time can be then used to estimate the bulk
&Zhang2006)predictsasignificantcurvatureof Lorentzfactorofthefireball(whichis∼100orless
the lightcurves. This is consistent with some of for standard parameters). This might be the case
the lightcurves that show an early “dip” before for atleastsomeofthe bursts.
6
3. Optical observations

Inthepre-Swiftera,the afterglowobservations t−1/2
ℜ < 1
were mainly carried out in the optical and radio ν
bands. The late time optical/radio observations
(t ,F )
havebeenfocusedonidentifyingtemporalbreaksin t−2 p,r ν,p,r
tthhee “lijgehttbcureravkess,”w(sheiechFraariel egtenael.ra2l0ly01i;nBtelropormeteedt aals. flux t5 or t1/2 t1/2
2003; Ghirlanda et al. 2004; Dai et al. 2004; Fried-
man & Bloom 2005; Liang & Zhang 2005 for com- ℜ > 1 (t ,F ) t−1
pilationsofthejetbreakdatainthepre-Swiftera). ν p,f ν,p,f
Broad-bandmodelingwascarriedoutforahandful
ofwellobservedbursts(Panaitescu&Kumar2001,
2002; Yost et al. 2003), and the data are generally
time
consistent with the standard external shock after-
glowmodel.Insomecases,veryearlyopticalflashes Fig. 2. Theoretically expected early optical afterglow
have been discovered (e.g. GRB 990123, Akerlof lightcurvesincludingemissionfrombothreverseandforward
et al. 1999; GRB 021004, Fox et al. 2003a; GRB shocks (from Zhang etal. 2003).
021211,Foxetal.2003b;Lietal.2003a),whichare
generally interpreted as emission from the reverse 2003)intheISMmodel4.Thethicksolidlineshows
shock (Sari & Piran 1999a;M´esza´ros& Rees 1999; twopeaks:thefirstpeakfollowedby∼t−2 decayis
Kobayashi & Sari 2000; Kobayashi 2000; Wang et the reverseshock emissionpeak time,which is typ-
al.2000;Fanetal.2002;Kobayashi&Zhang2003a; ically at the shock crossing time (tdec). The second
Zhang et al. 2003; Wei 2003; Kumar & Panaitescu peak followed by ∼ t−1 is the forward shock peak,
2003; Panaitescu & Kumar 2004; Nakar & Piran which corresponds to the time when the typical
2004). Early radio flares have been detected in a synchrotronfrequencyνm crossesthe opticalband.
sample of GRBs (Frail et al. 2003), which are also Depending on parameters, the forward shock peak
attributed to the reverse shock emission (Sari & couldbeburiedbelowthereverseshockcomponent
Piran 1999a; Kobayashi & Sari 2000; Soderberg & (the thin solidline). One thereforehas two cases of
Ramirez-Ruiz 2003). The expectation for Swift be- optical flashes: Type I (rebrightening) and Type II
forethelaunchhasbeenthatUVOTwouldcollecta (flattening).
good sample of early afterglow lightcurves to allow A unified study of both reverse shock and
adetailedstudy ofGRB reverseshocks. forward shock emission suggests that Type I
lightcurves should be generally expected, if the
shock microphysics parameters (ǫ , ǫ , p, etc) are
e B
3.1. Early optical afterglows: where is the reverse
the same in both shocks. On the other hand, these
shock emission?
microphysics parameters may not be the same in
both shocks. In particular, if the central engine
In the Swift era, UVOT has been regularly is strongly magnetized, as is expected in several
collecting optical photons ∼ 100s after the burst progenitor models, the outflow likely carries a pri-
triggers for most GRBs. Ground-based robotic mordial magnetic field, which is likely amplified
telescopes (e.g. ROTSE-III, PAIRITEL, RAP- at the shocks. It is then possible to have R =
B
TOR, P60, TAROT, Liverpool, Faulkes, KAIT, (ǫ /ǫ )1/2 ≫ 1 in some cases. This is actually
B,r B,f
PROMPT, etc) have promptly observed most tar- the condition to realize the Type II lightcurves
gets whenever possible. A good list of early optical (Zhangetal.2003).Inordertointerpretthebright
detections have been made. However, the majority opticalflashandthesubsequentTypeIIlightcurves
of bursts have very dim or non-detection of opti- in GRB 990123and GRB 021211,one typically re-
cal afterglows(Roming et al. 2006a).This suggests quires R ∼ 10 or more (Fan et al. 2002;Zhang et
B
that in most cases the reverse shock, if any, is not
significant.
Figure 2 displays the theoretically predicted 4 Forwindmodels,seeWuetal.(2003);Kobayashi&Zhang
early optical afterglow lightcurves (Zhang et al. (2003b); Kobayashi et al.(2004).
7
al. 2003; Kumar & Panaitescu 2003; Panaitescu & (Fan et al. 2005b). Another Type II (flattening)
Kumar2004). lightcurve was detected from GRB 060111B(Klotz
The ǫ treatment is based on a purely hydro- etal.2006).Marginalreverseshocksignaturesmay
B
dynamicaltreatmentofshockswithmagneticfields be present in GRB 050525A (Blustin et al. 2006;
putinbyhand.Invokingastrongmagneticcompo- Shao&Dai2005),GRB050904(Gendreetal.2006;
nentinthereverseshockregionraisesthenecessity Wei et al. 2006), GRB 060117 (Jelinek et al. 2006)
totreatthedynamicsmorecarefullywithadynam- and GRB 060108 (Oates et al. 2006). Data sug-
icallyimportantmagneticfield.Zhang&Kobayashi gest a second type of optical flashes, which tracks
(2005)studiedthereverseshockdynamicsandemis- thegamma-raylightcurves(forGRB041219A,Ves-
sion for an outflow with an arbitrary σ parameter. trand et al. 2005). These optical flashes are likely
Theyfoundthatthemostfavorablecaseforabright related to internal shocks (M´esza´ros & Rees 1999),
optical flash (e.g. GRB 990123 and GRB 021211) probablyneutronrich(Fanetal.2005b).
is σ ∼ 1, i.e. the outflow contains roughly equal There are however cases that clearly show no
amount of energy in magnetic fields and baryons. reverseshockcomponentinthebrightopticalafter-
This is understandable: For a smaller σ, the mag- glows.GRB061007(Mundelletal.2006;Schadyet
neticfieldinthereverseshockregionissmaller,and al.2006b)issuchacase.Reachingapeakmagnitude
the synchrotronemission is weaker(see also Fan et < 11 (similar to 9th magnitude of GRB 990123),
al. 2004). For a larger σ, the magnetic field is dy- both the X-ray and optical lightcurves show single
namicallyimportant,whosepressuredominatesthe power law decaying behavior from the very begin-
outflowregion.Theshockbecomesweakordoesnot ning(∼80safterthetrigger).Thissuggestsastrong
existatall(whenσ islargeenough). external forwardshock emission with enormous ki-
The lack of bright optical flashes such as those netic energy (Mundell et al. 2006) or a structured
observedinGRB990123andGRB021211isthere- jet with very early jet break (Schady et al. 2006b).
fore not surprising. In order to have a bright Type The reverse shock emission in this case is believed
II flash, one needs happen to have an outflow with to peak atthe radioband(Mundell etal.2006).
σ ∼1,whilebothlargerandsmallerσ’swouldlead
to not very significant optical flashes. Even with-
outadditionalsuppressioneffects,anon-relativistic 3.2. Bumps and flares
shock with σ = 0 would generally give a reverse
shockpeak flux belowthe forwardshock peak level Wiggles andbumps havebeen observedinsev-
(Kobayashi 2000; Nakar & Piran 2004; Zhang & eral pre-Swift GRB optical afterglows (e.g. GRB
Kobayashi 2005). On the other extreme, a high-σ 021004,Hollandetal.2003;GRB030329,Lipkinet
flowwouldleadtoveryweakreverseshockemission al.2004).Modelstointerpretthesevariabilitiesusu-
ornoreverseshockatall(Zhang&Kobayashi2005). allyinvokeexternalshockrelatedprocesses,suchas
Thus the tight early UVOT upper limits (Roming density fluctuation, inhomogeneous jets, refreshed
etal.2006a)arenotcompletelyoutofexpectation. shocks, or multiple component jets (Lazzati et al.
Additional mechanisms to suppress optical 2002;Heyl&Perna2003;Nakaretal.2003;Berger
flashes have been discussed in the literature. Be- et al. 2003a; Granot et al. 2003; Ioka et al. 2005).
loborodov (2005) argues that Compton cooling of Early optical lightcurves may contain neutron de-
electrons by the prompt MeV photons may be a caysignatures(Beloborodov2003;Fanetal.2005a).
way to suppress the optical flashes. Kobayashi et Iokaetal.(2005)pointedoutthatsomeopticalfluc-
al. (2006) suggest that a dominant synchrotron- tuations are difficult to interpret within any exter-
self-Compton process in the reverse shock region nal shock related schemes, and they require reacti-
wouldsuppressthesynchrotronopticalemission.Li vationofthe centralengine.
etal.(2003b)andMcMahonetal.(2006)suggesta That erratic X-ray flares generally require late
pair-richreverseshockwithweakopticalemission. centralengineactivitiesraisesthequestionwhether
Despiteofthegeneraldisappointments,several some opticalflashes/flaresarealsodue to the same
brightopticalflasheshavebeendetectedintheSwift origin(butsofterandevenlessenergetic,e.g.Zhang
era,whichcouldbegenerallyinterpretedwithinthe 2005). Recent optical afterglow observations reveal
reverse/forwardshock model discussed above. The that anomalous optical afterglows seem to be the
IRafterglowofGRB041219A(Blakeetal.2005)is norm (Stanek et al. 2006; Roming et al. 2006c).
wellmodeledbyaTypeI(rebrightening)lightcurve Although some of them could be accommodated
8
within the external shock relatedmodels, some op- 4. Global properties
ticalflaresdoshowsimilarpropertiesasX-rayflares
(e.g.δt/t<1,Romingetal.2006c),whichdemands Combiningthebroad-bandafterglowproperties
latecentralengineactivities.Forexample,theopti- fordifferenttypes ofGRBs,onecanpeerintosome
cal fluctuations detected in the short GRB 060313 globalpropertiesofGRB afterglows.
opticalafterglows(Romingetal.2006b)maybebet-
ter interpreted as due to late central engine activi-
tiesthanduetodensityfluctuations(Nakar&Gra- 4.1. GRBradiative efficiency
not2006).
Effortstomodelopticalflaresusingthelatein- One interesting question is the GRB radiative
ternal shock model have been carried out recently efficiency, which is defined as η = E /(E +E ),
γ γ K
(Wei et al. 2006; Wei 2006). The results suggest where E and E are isotropic gamma-ray energy
γ K
that for plausible parameters, even the traditional andkineticenergyoftheafterglow,respectively.The
reverse shock optical flashes such as those in GRB reasonwhy η is important to understand explosion
990123,GRB041219AandGRB060111Bcouldbe mechanism is that it is related to the energy dissi-
interpretedwithin the late internalshockmodel. pationmechanismofthepromptemission,whichis
notidentified.Thestandardpictureisinternalshock
dissipation,whichtypicallypredictsseveralpercent
3.3. Optically bright vs. optically dark; optically
radiative efficiency (Kumar 1999; Panaitescu et al.
luminous vs.optically dim
1999,cf.Beloborodov2000,Kobayashi&Sari2001).
Other mechanisms (e.g. magnetic dissipation) may
In the previous optical follow up observations, havehigherefficienciesalthoughdetailedprediction
GRBsaregenerallydividedintotwocategories,op- is not available.It is of great interest to estimate η
tically bright and optically dark ones (e.g. Jakobs- fromthedata,whichcanpotentiallyshedlightonto
sonetal.2004;Roletal.2005).Thelattertypically the unknownenergydissipationprocess.
account for ∼ 50% of the total population5. The Inordertoestimateη,reliablemeasurementsof
discovery of the early optical flash of GRB 021211 bothE andE areneeded.WhileE couldbedi-
γ K γ
(Foxetal.2003b;Lietal.2003)intheHETE-2era rectly measured from the gamma-ray fluence if the
had led to the ansatz that as long as observations GRB redshift is known, the measurement of E is
K
are performed early enough, most dark bursts are nottrivial,whichrequiresdetailedafterglowmodel-
not dark.This is now provennot the case (Roming ing. In the pre-Swift era, attempts to estimate E
K
et al. 2006a). Among the possible reasons of opti- andηusinglatetimeafterglowdatahavebeenmade
caldarkness,foregroundextinction,circumburstab- (e.g.Panaitescu& Kumar2001,2002;Freedman&
sorption,andhighredshiftarethe best candidates. Waxman 2001; Berger et al. 2003b; Lloyd-Ronning
AmongtheopticallybrightGRBs,itisintrigu- & Zhang 2004). The Swift XRT observations sug-
ing to discover that there are two sub-categories, gest a substantial shallow decay phase in a good
namelyopticallyluminousandopticallydim(Liang fractionofGRBs(Fig.1).Ifthisisduetoenergyin-
&Zhang2006;Nardinietal.2006;Kannetal.2006). jection,thenE isafunctionoftime.Theη values
K
Therest-framelightcurvesofGRBswithknownred- measuredusingthe late time dataarenolongerre-
shiftsarefoundtofollowtwo“universal”tracks.The liable. For a constant energy fireball, ideally early
rest-frame 10-hour luminosities of the bursts with afterglows may be used to study radiative loss of
known redshifts show a clear bimodal distribution. thefireball.Howevertheshallowdecayphasedueto
Theopticallydimburstsallappeartolocateatred- energy injection smears the possible signature and
shifts lower than ∼ 1 (Liang & Zhang 2006). The makessuchadiagnosisdifficult.
originof sucha clear dichotomyis unknown, but is AsystematicanalysisofGRBradiativeefficien-
likely related to different total explosion energy in- ciesusingthefirsthandSwiftdataiscarriedoutby
volvedinthe twogroupsofbursts. Zhangetal.(2007b).Similaranalysesusingsecond-
hand data for smaller samples of bursts were car-
ried out by Fan & Piran (2006) and Granot et al.
5 SwiftUVOTdoesnotdetectopticalafterglowsfor∼67%
(2006).Theconclusionsemergingfromthesestudies
oftheSwiftbursts.Combiningwithground-basedfollowups,
the non-detection rate is ∼45% (P. Roming, 2006, private suggestthatinmostcasestheefficiencyisveryhigh
communication). (e.g.>90%)ifEK rightafterthe burstis adopted.
9
However, using E at a later time when the injec- the achromatic behavior, was not robustly estab-
K
tion process is over one typically gets η ∼ several lished in any of these bursts. The best case was
percent. The nature of the shallow decay phase is GRB 990510 (Harrison et al. 1999), in which clear
therefore essentialto understand the efficiency. For multi-color optical breaks were discovered, which
example,iftheshallowdecayphaseisduetocontin- are consistent with being achromatic. The radio
uous energy injection, the GRB radiativeefficiency dataarealsoconsistentwithhavingabreakaround
mustbeveryhigh-causingproblemstotheinternal thesametime.However,basedonradiodataalone,
shockmodel.If,however,theshallowdecayissimply one cannot robustly fit a break time that is con-
duetothe delayofenergytransferintothe forward sistent with the optical break time (D. Frail, 2006,
shock,theGRBradiativeefficiencyisjusttheright private communication). Most of other previous
oneexpectedfromtheinternalshockmodel.Onein- jet breaks were claimed using one-band data only,
terestingfindingofZhangetal.(2007b)isthatX-ray mostly inoptical,andsometimesinX-rayorradio.
flashes may notbe intrinsically less efficientGRBs, With these “jet break times” t , several empirical
j
as was expected in the pre-Swift era (Soderberg et relationshavebeendiscussedinthe literature.
al. 2004; Lloyd-Ronning & Zhang 2004). Analyses – Frail relation: Frail et al. (2001) found that the
show that atthe earlydecelerationtime, XRFs are beaming-corrected gamma-ray energy is essen-
as efficient as harder GRBs (see also Schady et al. tially constant, i.e. E θ2 = E ∼ const. Since
γ,iso j j
2006a). the standard jet model predicts t ∝ E1/3 θ8/3
j γ,iso j
One of major breakthroughs made by Swift is
(Sari et al. 1999), this relation is generally con-
the discoveries of the afterglows of short-duration −1
sistentwithE ∝t .
γ,iso j
GRBs and identifications of their host galaxies
– Ghirlandarelation:Ghirlandaetal.(2004)found
(Gehrels et al. 2005; Fox et al. 2005; Villasenor et
that the beaming-correctedgamma-rayenergyis
al.2005;Hjorthetal.2005;Barthelmyetal.2005c;
notconstant,butisrelatedtotherest-framespec-
Bergeretal.2005).Theseobservationssuggestthat tralpeak energy(E ) throughE ∝E2/3.Again
shortGRBslikelyhavedistinctprogenitorsystems, p p γ,j
expressingE in terms ofE andt , this re-
γ,j γ,iso j
which are consistent with compact star (NS-NS, lation is effectively E ∝ E1/2 t1/2. Notice that
NS-BH, etc) mergers. As far as the radiative effi- p γ,iso j
the Ghirlanda relation and the Frail relation are
ciencyisconcerned,ontheotherhand,shortGRBs
incompatible witheachother.
arerathersimilartolongGRBs(Zhangetal.2007b;
– Liang-Zhangrelation:Liang&Zhang(2005)took
seealsoBloometal.2006andLeeetal.2005).This
one step back. They discard the jet model, and
suggests that both types of GRBs share the same
onlypursueanempiricalrelationamongthreeob-
radiationphysics.
servables,namelyE ,E andthe optical band
p γ,iso
breaktimet .TherelationgivesE ∝E0.52 t0.64.
b p γ,iso b
4.2. Wherearethejet breaks? Itisevidentthatiftbisinterpretedasthejetbreak
time, the Liang-Zhang relation is rather similar
to the Ghirlanda relation. However, the former
If GRB outflows are collimated into the typi-
has the flexibility of invoking chromatic tempo-
caljetangleθ ,anachromaticafterglowsteepening
j
ralbreaksacrossdifferentbands.Soviolatingthe
breakshouldbeobservedinallenergybandsatthe
Ghirlanda relation in other wavelengths (e.g. in
timewhenthebulkLorentzfactorofthejetsatisfies
Γ−1 =θ (Rhoads1999;Sarietal.1999).Thistime the X-rayband,Sato etal.2006)doesnotneces-
j
sarilydisfavorthe Liang-Zhangrelation.
is calledjet breaktime t .
j
It has been highly expected that the multi-
Identifying GRB jet breaks in the afterglow
wavelength observatory Swift would clearly detect
lightcurvesisessentialtounderstandthe geometric
achromaticbreaksinsomeGRBstoverifythelong-
configurationandthetotalenergeticsofthejets.In
invoked GRB jet scenario. The results are however
the pre-Swift era, a list of “jet breaks” have been
discouraging.After detecting nearly 200 bursts, no
“identified” in the optical (sometimes X-ray and
“textbook” version jet break is yet detected in any
radio) afterglows (see Frail et al. 2001; Bloom et
GRB. The lack of detections may be attributed
al. 2003;Ghirlanda et al. 2004;Friedman & Bloom
partiallytotheintrinsicfaintnessoftheSwiftafter-
2005; Liang & Zhang 2005 for compilations of the
glows,andpartiallytotheverylowrateoflatetime
jet breakdata). We use quotationmarkshere since
optical follow-up observations. Achromatic breaks
the “smoking-gun” feature of the jet breaks, i.e.
10