0709.4307/ms.tex
1: \documentclass [12pt,preprint]{aastex}
2: % for a referee version
3: %\documentclass [manuscript]{aastex}
4: %documentclass [article]{aa}
5: \usepackage{epsfig}
6: %
7: \begin{document}
8: \voffset-1cm
9: \newcommand{\gsim}{\hbox{\rlap{$^>$}$_\sim$}}
10: \newcommand{\lsim}{\hbox{\rlap{$^<$}$_\sim$}}
11: 
12: 
13: \title{The rapid decline of the prompt emission in
14:        Gamma-Ray Bursts}
15: 
16: 
17: \author{Shlomo Dado\altaffilmark{1}, Arnon Dar\altaffilmark{1} and A. De
18: R\'ujula\altaffilmark{2}}
19: 
20: \altaffiltext{1}{dado@phep3.technion.ac.il, arnon@physics.technion.ac.il,
21: dar@cern.ch.\\
22: Physics Department and Space Research Institute, Technion, Haifa 32000,
23: Israel.}
24: \altaffiltext{2}{alvaro.derujula@cern.ch; Theory Unit, CERN,
25: 1211 Geneva 23, Switzerland. \\
26: Physics Department, Boston University, USA.}
27: 
28: 
29: \begin{abstract}
30: 
31: Many gamma ray bursts (GRBs) have been observed with the Burst-Alert and 
32: X-Ray telescopes of the Swift satellite. The successive `pulses' of these 
33: GRBs end with a fast decline and a fast spectral softening, until they are 
34: overtaken by another pulse, or the last pulse's decline is overtaken by a 
35: less rapidly-varying `afterglow'. The fast decline-phase has been 
36: attributed, in the currently-explored standard fireball model of GRBs, to 
37: `high-latitude' synchrotron emission from a collision of two conical 
38: shells. This high latitude emission does not explain the observed 
39: spectral softening. In contrast, the temporal behaviour and the spectral 
40: evolution during the fast-decline phase agree with the predictions of the 
41: cannonball model of GRBs.
42: 
43: \end{abstract}
44: 
45: \keywords{$\gamma$ rays: burst-radiation mechanisms: non-thermal X-rays: flare} 
46: 
47: \section{Introduction}
48: 
49: 
50: Since the launch of the Swift spacecraft, precise data from its Burst 
51: Alert Telescope (BAT) and X-Ray Telescope (XRT) have been obtained on the 
52: spectral and temporal behaviour of the X-ray emission in $\gamma$-ray 
53: bursts (GRBs) and X-ray flashes (XRFs). These data have already been used 
54: to test the most-studied theories of long duration GRBs and their 
55: afterglows (AGs), the {\it Fireball} (FB) models (see, e.g., Piran 1999, 
56: 2000, 2005; Zhang \& M\'esz\'aros~2004; M\'esz\'aros~2002, 2006,
57: Zhang~2007; and 
58: references therein) and the {\it Cannonball} (CB) model [see, e.g.,~Dar \& 
59: De R\'ujula~2004 (hereafter DD2004); Dado, Dar \& De R\'ujula~2002, 2003, 
60: 2007a, 20007b (hereafter DDD2002, DDD2003, DDD2007a, DDD2007b), and 
61: references therein].
62: 
63: The general behaviour of the Swift X-ray light curves has been described 
64: as `canonical' (Nousek et al.~2006; O'Brien et al.~2006; Zhang et 
65: al.~2007), and is illustrated in Figs.~\ref{f1}a,b, \ref{f2}a for XRF 060218, GRB 060904a 
66: and GRB 061121.  When measured early enough, the X-ray emission has peaks 
67: that coincide with the $\gamma$-ray peaks of the GRB. The prompt emission 
68: has a fast decline after the last detectable peak of the GRB. In most 
69: cases, the rapid decline ends within a couple of hundreds of seconds. 
70: Thereafter, it turns into a much flatter `plateau', typically lasting 
71: thousands to tens of thousands of seconds. Finally, the X-ray light curve, 
72: within a time order of one day, steepens into a power-law decline which 
73: lasts until the X-ray AG becomes too dim to be detected. Often, there are 
74: also X-ray peaks during the fast-decline phase or even later, not 
75: coinciding with a detectable $\gamma$-ray activity. There is a continuous
76: transition of X-ray light curve shapes from the `canonical' ones to the
77: ones that are well described by a single-power decay, e.g.~GRB 061126,
78: see Fig.~\ref{f2}b.
79: 
80: 
81: Neither the general trend, nor the frequently complex structure of the 
82: Swift X-ray data were correctly predicted by (or can be easily 
83: accommodated within) the standard FB models (see, e.g., 
84: Piran~1999, 2000, 2005; ~Zhang \& M\'esz\'aros 2004, for 
85: reviews of the pre-Swift standard FB model, and Kumar et al.~2007; 
86: Burrows \& Racusin~2007; Kocevski \& Butler~2007: Urata et al.~2007; 
87: Zhang, Liang \& Zhang~2007; Yonetoku et al.~2007; Liang et al.~2007
88: for recent comparisons with Swift data). 
89: 
90: The situation in the CB model (Dado, Dar \& De R\'ujula~2004, hereafter 
91: DDD2004) is different. The model offered a good description, based on a 
92: specific synchrotron-radiation (SR) mechanism, of the AGs of all 
93: `classical' GRBs (DDD2002, DDD2003) of known redshift, and allowed one to 
94: extract the relevant parameters of the CBs of GRBs and XRFs. The 
95: consequent predictions for the `prompt' $\gamma$-rays, based on an 
96: explicit inverse Compton scattering (ICS) mechanism, were simple and 
97: successful (DD2004). As shown in DDD2007a,b (and references therein) for 
98: `Swift-era' data, the CB model, with no modification, correctly predicts 
99: the temporal and spectral behaviour of the prompt and AG phases.
100: 
101: 
102: In this paper we confront the Swift's observations with the predictions of 
103: the FB and CB models for the spectral evolution during the fast-decline 
104: phase of the prompt emission. In the FB model this phase was interpreted 
105: (e.g., M\'esz\'aros~2006; Liang et al.~2006; O'Brien et al.~2006; Yamazaki 
106: et al.~2006) as the `curvature effect' or `high-latitude' emission of 
107: colliding shells (Fenimore et al.~1996; Kumar \& Panaitescu~2000; 
108: Dermer~2004). Relative to photons centrally emitted on the line of sight 
109: to an on-axis observer, photons from off-axis latitudes arrive later and 
110: with smaller number density and energy. The consequent spectral behaviour 
111: is entirely different from that observed (see, e.g., Zhang et al.~2007). 
112: In the CB model the properties of the fast-declining phase are also 
113: dominantly `geometrical'. A GRB's $\gamma$-ray pulses and their sister 
114: X-ray flares are made by ICS of light in a `glory reservoir' bathing the 
115: circumburst material (DD2004). This light becomes, in a very specific 
116: manner, less abundant and more radially-directed with distance from the 
117: parent star. These simple facts result in the correct description of the 
118: temporal behaviour and spectral evolution of GRBs, before, during, and 
119: after the fast-decline phase.
120: 
121: In the CB model it is possible in principle to fit the spectral energy 
122: flux of a GRB in a given energy band, as a function of time, and determine 
123: the parameters partaking in a complete prediction of the spectrum at any 
124: time in the fit interval. But the public Swift spectral data is limited to 
125: a `hardness ratio' between the counting rates in the 1.5-10 keV and 
126: 0.3-1.5 keV bands (Evans et al.~2007). To convert these rates into a more 
127: explicit spectral information one must correct for instrumental 
128: efficiency, subtract the background and correct for X-ray absorption in 
129: the host galaxy, in the IGM, and in our Galaxy. 
130: The unabsorbed spectra as functions of time is not generally available. 
131: However, the unabsorbed spectral energy flux in the X-ray band, 
132: parametrized as  $F_\nu\! \propto\! \nu^{-\beta}\, t^{-\alpha}$, is available
133: in the form of the 
134: fitted time-dependent power-law spectral index, $\beta(t)$, for a set
135:  of X-ray light curves measured with Swift's XRT 
136: (Zhang et al.~2007 and {\it http://swift.physics.unlv.edu/xrt.html}). Such a 
137: parametrization is not a faithful description of an exponentially cutoff 
138: power-law, a Band function, or the spectrum predicted by the CB 
139: model,  similar to the Band function for typical parameters (DD2004). 
140: Moreover, to extract $\beta(t)$, the data from 
141: different time intervals was coadded, smoothing the time-dependence 
142: of the {\it effective} fitted photon index. This can be seen by comparing 
143: the effective indices $\Gamma(t)$ in $dN_\gamma/dE\!\propto\!E^{-\Gamma}$,
144: reported in the cited web-page, with
145: the hardness ratios reported  
146: in the Swift light-curve repository (Evans et al.~2007).
147: The spectral indices $\beta$
148: and $\Gamma$ are related by $\beta\! =\! \Gamma\! -\! 1$.
149: 
150: We do not have all of the information needed for a decisive comparison 
151: between the spectral behaviour during the fast decline phase of the prompt 
152: emission. But the variation with time of the hardness ratio and of the 
153: effective spectral index during the fast decline phase are so spectacular 
154: and well correlated to the light curves, that an approximate analysis 
155: suffices to prove our points. We demonstrate this for the hardness ratios of five
156: Swift GRBs with well sampled X-ray light curves during the fast decline 
157: phase and for fourteen other GRBs with an extracted effective time-dependent
158: spectral index. 
159: 
160: 
161: \section{High-latitude emission in the fireball model} 
162: 
163: In the FB model, GRB pulses are produced by synchrotron radiation emitted 
164: by a shock-accelerated electrons, following collisions between conical 
165: shells ejected by a central engine (Rees \& M\'esz\'aros~1994, see Zhang et 
166: al.~2007 for detailed discussion). Consider a spherical shell, arbitrarily 
167: thin, that expands with a Lorentz factor 
168: $\gamma\!\equiv\!1/\sqrt{1\!-\!\,(v/c)^2}$. 
169: Assume that when two shells collide at a radius 
170: $R$, all points emit isotropically in their rest frame an 
171: arbitrarily short pulse of radiation. Let $t\!=\!0$ be the time of arrival of 
172: the first photons on the line of sight to the center of the conical 
173: shell. Photons emitted from a shell's polar angle $\theta$ 
174: arrive at $t\!=\!R\, (1\!-\!\cos\theta)/c$.  
175: If the radiation has a power-law spectrum in the shell's rest 
176: frame, ${\tilde\nu}^{-\beta}$, the spectral 
177: energy flux seen by the observer has the form (Kumar \& 
178: Panaitescu~2000)
179: $F_\nu\! \propto\! \nu^{-\beta}\, \delta^{2+\beta}$,  
180: where $\delta\!\equiv\!1/\{\gamma\, [1\!-\! (v/c)\, \cos\theta]\}$. Thus, 
181: for $\gamma\!\gg\! 1$, 
182: the high latitude emission from a shell collision obeys
183: $F_\nu\! \propto\! \nu^{-\beta}\, (t+t_0)^{-(2+\beta)}$,
184: with $t_0= R/(2\, \gamma^2 \, c)$.
185: Note that the spectral behaviour does not change during 
186: the temporal power-law decline. 
187: This is in contradiction with the observed rapid spectral softening.
188: 
189: To confront this problem Liang et al.~(2006) {\it assumed} that the 
190: high-latitude spectral index $\beta$ is time-dependent but the temporal 
191: index still satisfies $\alpha(t) \!=\! 2\!+\! \beta(t)$. Although 
192: structured 
193: jet models (M\'esz\'aros, Rees \& Wijers~1998; Zhang \& M\'esz\'aros~2002; 
194: Ross, Lazzati \& Rees.~2002) may yield a time-varying $\beta$, there is 
195: no reason why it should depend on an angle defined by the position of the 
196: observer. Indeed, the relation $\alpha(t)\!=\!2\!+\!\beta(t)$ is badly 
197: violated in canonical light curves (Zhang, Liang \& Zhang 2007).
198: We conclude that the curvature effect in the currently explored
199: fireball models does not agree with the data.
200: 
201: \section{The CB model and its predictions}  
202: 
203: In the CB model (e.g., DD2004 and references therein), GRBs and their AGs 
204: are produced by jets of highly relativistic CBs of ordinary matter (Shaviv 
205: \& Dar~1995; Dar~1998; Dar \& Plaga~1999). Long-duration GRBs originate 
206: from CBs ejected in core collapse supernova explosions. The `engine' of 
207: short GRBs is much less well established, it could be the merger of 
208: compact objects, e.g.~neutron stars, and/or mass-accretion episodes 
209: on compact objects in close binaries (e.g., 
210: microquasars), or even phase transitions of increasingly compactifying 
211: stars (neutron stars, hyper stars, or quark stars).
212: 
213: The pre-GRB ejecta of the parent stars create `windy' environments of 
214: `circumburst' material. The early luminosity of the event (a core-collapse 
215: supernova for long GRBs) permeates this semitransparent material with a 
216: temporary constituency of scattered, non-radially-directed photons: a {\it 
217: glory} of visible or UV light, with an approximately `thin-bremsstrahlung' 
218: spectrum (DD2004). The $\gamma$-rays of a single pulse of a GRB are 
219: produced as a CB coasts through the glory. The electrons enclosed in the 
220: CB boost the energy of the glory's photons, via inverse Compton 
221: scattering, to $\gamma$-ray energies. The initial fast expansion of the 
222: CBs and the radially-increasing transparency of the windy environment 
223: result in the exponential rise of a GRB pulse.  As a CB proceeds, the distribution of 
224: the glory's light becomes more radially directed, its density decreases. 
225: Consequently, the energy of the observed photons is continuously shifted 
226: to lower energies as their number plummets. These trends were observed in 
227: CGRO/BATSE data (Giblin et al.~2002; Connaughton~2002; Ryde \& 
228: Svensson~2002, DD2004).  During a GRB pulse the spectrum softens and the 
229: peak energy decays with time as a power law. This is also the behaviour of 
230: the X-ray flares of a GRB, which are either the low-energy tails of 
231: $\gamma$-ray pulses, or fainter and softer signals with the same origin 
232: (DDD2007a). Typically, the fast decline of the prompt emission in the 
233: $\gamma$-ray and X-ray bands is taken over, within few minutes of 
234: observer's time, by the `afterglow' --synchrotron emission from swept-in 
235: ISM electrons spiraling in the CB's enclosed magnetic field.
236: 
237: 
238: The above effects can be explicitly analized (DD2004), and 
239: summarized to a good approximation
240: in a {\it master formula} (DDD2007a)
241:  for the temporal shape and spectral evolution of the energy fluence
242: of an ICS-generated $\gamma$-ray pulse (or X-ray flare):
243: \begin{equation}
244: F_E^i\propto E\, {d^2N_\gamma^i\over dt\,dE} \propto
245: \Theta[t-t_i]\;
246: e^{-[\Delta t_i/(t-t_i)]^m}\,
247:  \left\{1-e^{-[\Delta t_i/(t-t_i)]^n}\right\}\, E\,{dN_\gamma^i(E,t) \over dE}\; ,
248: \label{GRBlc}
249: \end{equation}
250: where `{\it i}' denotes the i-th pulse, produced by
251: a CB launched at (an observer's) time $t_i$. In Eq.~(\ref{GRBlc}),
252: the time scale is set by $\Delta t_i$, with $\gamma\, \delta\, c\, \Delta t_i/(1+z)$  
253: the radius of transparency of the glory, within which its photons are
254: approximately isotropic. In $\Delta t_i$ time units, a pulse rises 
255: as $Exp[-1/t^m]$, $m\!\sim\!1$ to 2, and decreases as $1/t^n$, $n\!\sim\!2$. Finally,
256: $E\, dN_\gamma^i/dE$ is the spectral function of the glory's 
257: photons, up-scattered by  the CB's electrons, and discussed anon.   
258: 
259: 
260: The glory has a thin thermal-bremsstrahlung spectrum:
261: $\epsilon\, {dn_\gamma / d\epsilon} \!\sim\!
262: (\epsilon/\epsilon_g)^{1-\alpha_g}\,e^{-\epsilon/\epsilon_g}$,
263: with a typical (pseudo)-temperature
264: $ \epsilon_g \!\sim\!1$ eV, and index $\alpha_g\!\sim\!1$.
265: During the $\gamma$-ray phase of a GRB,
266: the Lorentz factor $\gamma$ of a CB stays put at its initial value,
267: for the deceleration induced by
268: the collisions with the ISM has not yet had a significant effect
269: (DDD2002, DD2007).
270: Let $\theta$ be the observer's angle relative to the direction of motion
271: of a CB and let the corresponding Doppler factor be
272: $\delta\!=\!1/ \{\gamma\,[1\!-\!\,(v/c)\, \cos\theta]\}$. 
273: Let $\theta_i$ be the angle of incidence of the initial
274: photon onto the CB, in the parent star's rest system.
275: The energy of an observed photon, 
276: Compton scattered in the glory by an electron comoving with a
277: CB at redshift $z$, is given by
278: $E\!=\!\gamma\, \delta\, \epsilon \, (1\!+\!\cos\theta_i)/(1+z)$.
279: The predicted GRB prompt spectrum is (DD2004):
280: \begin{equation}
281: E\, {dN\over dE} \sim \left({E\over T}\right)^{1-\alpha_g}\,
282:  e^{-E/T}+ b\,(1-e^{-E/T})\, \left({E \over T}\right)^{-p/2} .
283: \label{GRBspec}
284: \end{equation}
285: The first term, with $\alpha_g\!\sim\! 1$, is the result of Compton
286: scattering by the bulk of the CB's electrons, which are comoving with it.
287: The second term in Eq.~(\ref{GRBspec}) is induced by
288: a very small fraction of
289: `knocked on' and Fermi-accelerated electrons, whose initial spectrum
290: (before Compton and synchrotron cooling) is $dN/dE_e\propto E_e^{-p}$,
291: with $p\approx 2.2$. Finally, $T$ is the effective (pseudo)-temperature
292: of the GRB's photons:
293: \begin{equation}
294: T\equiv{4\, \gamma\, \delta\,\epsilon_g\,
295: \langle 1+\cos\theta_i\rangle / [3\, (1+z)]}\,.
296: \label{ICST}
297: \end{equation}
298: For a semi-transparent glory $\langle\cos\theta_i\rangle$ would be
299: somewhat smaller than zero.
300: 
301: For $b={\cal{O}}(1)$,
302: the energy spectrum predicted by the CB model, Eq.~(\ref{GRBspec}),
303: bears a striking resemblance
304: to the Band function (Band et al.~1993) traditionally used to model the
305: energy spectra of GRBs. For many Swift GRBs the spectral observations
306: do not extend to energies much bigger than $T$, or the value of $b$
307: in Eq.~(\ref{GRBspec}) is relatively small, so that the first term
308: of the equation provides a very good approximation.
309: This term coincides with the `cut-off
310: power-law' spectrum recently used to model
311: GRB spectra. It yields a `peak-energy' (the maximum of $E^2\, dN/dE$
312: at the beginning of a pulse)
313: $E_p\!=\!(2\!-\!\alpha_g)\, T\!\approx\! T$ for
314: $\alpha_g\!\sim \!1$. At later times, the CB is sampling the glory at distances
315: for which its light is becoming increasingly radial, 
316: $\langle1\!+\!\cos\theta_i\rangle\!\to\!1/r^2\!\propto\!1/t^2$ in Eq.~(\ref{ICST}).
317: The value of $E_p(t)$ consequently decreases as: 
318: \begin{equation}
319: E_p(t)\approx E_p(t_i)\,
320: \left[1-{t-t_i\over \sqrt{\Delta t_i^2+(t-t_i)^2}}\right]\, .
321: \label{Epi}
322: \end{equation}
323: The light-curve generated by a  sum of pulses is well 
324: approximated (DDD2007a) by:
325: \begin{equation}
326: F_E
327: \approx \sum_i\,A_i\, \Theta[t-t_i]\;
328: e^{-[\Delta t_i/(t-t_i)]^2}\,
329:  \left\{1-e^{-[\Delta t_i/(t-t_i)]^2}\right\}\,\left[{E/ 
330: E_p(t)}\right]^{1-\alpha_g} \
331: e^{-E/E_p(t)}\,
332: \label{GRBXlc}
333: \end{equation}
334: until ICS is overtaken by synchrotron radiation.
335: 
336: In   X-rays the distinction between a prompt and an afterglow period
337: can be made precise, they correspond to the successive
338: dominance of the two radiation mechanisms: ICS and SR.
339: The  actual form of the SR-dominated
340: AG spectral energy flux, $F_\nu$,
341: we have discussed very often (DDD2007a,b and references therein).  
342: Suffice it to recall that (for cases as the ones we discuss here,
343: whose AG can be well fit with a single dominant or average CB)
344: the shape of the observed $F_\nu$, corrected for absorption, is 
345: determined by 
346: $\gamma_0\,\theta$, and its time scale is
347: determined by a deceleration time, $t_0$, at which $F_\nu$
348: achromatically `bends down' towards its asymptotic behaviour,
349: $F_\nu \propto \nu^{-\beta}\, t^{-\beta-1/2}$.
350: Typically $\beta\sim 1.1$ (DDD2007b).
351:  
352: \subsection{The hardness ratio in the CB model}
353: 
354: For a case in which the X-ray-absorption  factor $A(E)$ is known,
355: we have given enough information to predict the hardness ratio (HR)
356: from the X-ray energy flux  in a given energy band. For the late
357: SR-dominated phase, this is trivial. A look at an X-ray light curve, such that of
358: Fig.~\ref{f1}a, tells one the time at which the fast-decline ends, meaning
359: that SR starts to dominate. From that time onwards, the HR
360: is that corresponding to  $F_E \!\propto\! E^{-\beta}$, which would be 
361: HR $\approx\! 0.18$ for an
362: unabsorbed flux with $\beta\!=\!1.1$.
363: 
364: In the ICS-dominated phase, the shape of the flux determines 
365: the number of flares one ought to fit, one in Fig.~\ref{f1}a, for instance. 
366: If one uses Eq.~(\ref{GRBXlc}) with $\alpha_g\!=\!1$, each pulse 
367: is fit with 4 parameters:  $t_i$,  $\Delta t_i$, 
368: $E_p(t_i)$, and $A_i$. For the rapid-decline phase, it
369: suffices to consider the main or the latest few flares, since 
370: the last factor in Eq.~(\ref{GRBXlc}) suppresses 
371:  the relative contribution of earlier flares by the time 
372: the data sample the later ones. Once the parameters are
373: fixed, the HR is determined by the quotient of the integrals
374: $\int dE \,A(E)\,d^2N_\gamma/dEdt$ in the two Swift X-ray energy bands.
375: This rosy picture is clouded by two facts: the data for the {\it integrated}
376: flux in the 0.3-10 keV interval is very insensitive to the values of 
377: $E_p(t_i)$, which the fit consequently returns with very large errors;
378: we do not know $A(E)$.
379: 
380: We studied numerically the HR of a pulse given
381: by Eq.~(\ref{GRBXlc}), in the large interval $0\!<\!t\!-\!t_i\!<\!10\,\Delta t_i$,
382: for an exaggerated  range of $E_a$ in $A(E)\!\approx\!Exp[-(E_a/E)^3]$.
383: We found that
384: \begin{equation}
385: {\rm HR}_i(t)=B\,e^{-\;\left[{\Delta E/ E_p(t_i)}\right]\,
386: \left[1-(t-t_i)/\sqrt{\Delta t_i^2+(t-t_i)^2}\,\right]^{-1}}\; ,
387: \label{HR}
388: \end{equation}
389: is a fair approximation, with $\Delta E$ an effective interval between the bands in the HR. 
390: More explicitly, if $B$ and $\Delta E/E_p(t_i)$ are {\it fit}, the approximation
391: is good to a few \% for a typical $E_p(t_i)\!>\! 200$ keV, deteriorating
392: to $\sim\!40$\% for an extreme and atypical $E_p(t_i)\!=\!30$ keV.
393: We shall consequently fit $B$ and $\Delta E/E_p(t_i)$ in comparing theory and data for the HR.
394: 
395: For times at which the late-time tail of a single pulse dominates, the HR satisfies
396: \begin{equation}
397: {\rm HR}_i(t) \to B\,e^{-\left[\Delta E / E_p(t_i)\right]\,
398: \left[2\,(t-t_i)^2/ \Delta t_i^2\right] }
399: \label{HRapprox}
400: \end{equation}
401: with  precision increasing with $t$.
402: 
403: 
404: \section{Hardness ratios: case studies}
405: 
406: The Burst Alert Telescope (BAT) on Swift has detected nearly 250 GRBs or 
407: XRFs whose X-ray emission was followed with its X-Ray Telescope (XRT) from 
408: $\sim\!70$ s after trigger until it faded away. Incapable of discussing 
409: all these observations, we first study five cases, which we view as representative, 
410: and which have well-sampled X-ray fluxes and hardness ratios during the 
411: fast-decline and the ensuing AG phase. They are: the `clean' single-peak 
412: XRF 060218, GRB 060904a with its 4 X-ray flares during the fast decline 
413: phase, the simpler two-flare GRB 061121, the duller GRB 061126, for 
414: which the XRT observations began late and the bright GRB 061007
415: with an approximate single power-law afterglow.
416: 
417: \noindent 
418: {\bf XRF 060218:} 
419: This single-peak XRF provides one of the best testing grounds of theories, 
420: given its proximity, which resulted in very good sampling and statistics. 
421: The BAT data lasted 300s, beginning 159s after trigger, with most of the 
422: emission below 50 keV (Campana et al.~2006; Liang et al.~2006). The 
423: prompt X-ray emission lasted more than 2000s, during which the peak energy 
424: evolved from 54 keV to $\leq 5$ keV at the start of the fast-decline 
425: phase. The flux and HR data are shown in Figs.~\ref{f1}a,c. During the afterglow 
426: phase, the HR seems to decrease gradualy from $\sim\! 0.8$ at 6.2 ks 
427: to $\sim\!0.25$ at 72 ks. In the CB model such a trend could be 
428: produced by diminishing absorption along the line of sight to the CBs. 
429: 
430: The HR 
431: from unabsorbed synchrotron radiation with a typical $\alpha\!=\!1.1$ is
432: HR $\approx 0.18$, well below the reported HR for the absorbed flux 
433: of XRF 060218 (Evans et al.~2007).  
434: Pian et al.~(2008) reported that the extinction derived 
435: from the equivalent width of the Na I D absorption line in the spectrum of 
436: the associated SN2006aj is $E(B-V)\!=\!0.13\pm 0.02$, consistent with 
437: Galactic extinction and no 
438: extinction in the host. The H column density needed to fit 
439: the Swift X-ray prompt spectrum was NH = 
440: $6\times 10^{21}\, {\rm cm^{-2}}$ (Campana et al.~2006), implying 
441: $E(B-V)\!\approx\! 1$, considerably greater than the total 
442: extinction (in the Galaxy, the intergalactic medium and the host galaxy)
443: derived from optical emission and absorption lines, as 
444: well as from the optical colours of the afterglow, measured by Mirabal et 
445: al.~(2006). These authors stress that NH is not really NH, but a 
446: proxy for the heavier elements that dominate the X-ray photoelectric 
447: absorption, and that the relatively small extinction implies a 
448: dust-deficient medium such as the stellar wind of a Wolf-Rayet progenitor, 
449: which has enough column density to be the location of this excess 
450: photoelectric X-ray absorption and relatively dust-deficient medium. The 
451: use of this large NH deduced from the prompt emission to infer the 
452: late X-ray afterglow spectrum may have resulted in the very large 
453: $\Gamma\!\approx\! 4.4\pm 1.0$ reported in 
454: {\it http://swift.physics.unlv.edu/xrt.html}.
455:  
456:  
457: \noindent
458: {\bf GRB 060904a:} 
459: The BAT detected a weak emission of $\gamma$ rays for about a 
460: minute, with several small peaks before the main burst,
461: also seen by the Konus-Wind and 
462: Suzaku satellites (Yonetoku et al.~2007). The XRT  
463: followed the fast decline of the main burst and saw three additional 
464: flares, as shown in Fig.~\ref{f1}b. 
465: A rapid spectral softening was observed during both the prompt tail
466: phase and the decline phase of the X-ray flares, see Fig.~\ref{f1}d.
467: Due to a second GRB (0060904b) being 
468: detected just 1.5 hours later, Swift slewed away from GRB 060904a, so that
469: there were no data during a couple of hours until the 
470: XRT returned  to follow its fading afterglow.
471: After correcting for absorption (Yonetoku et al.~2007), 
472: the photon spectral index during the AG phase was found to be 
473: $\Gamma\! =\!2.1\pm 0.1$.
474: 
475: 
476: \noindent
477: {\bf GRB 061121:}
478: The $\gamma$-ray burst started with a bright precursor which 
479: lasted 10s. Then, 50s later, there was a much brighter 
480: burst of $\gamma$ rays. Swift had already turned its XRT when the 
481: second $\gamma$-ray flare occurred and  the X-ray emission was measured
482: during the actual event and its subsequent rapid decline, as shown in
483: Fig.~\ref{f2}a. After the rapid decline, the photon spectral index, corrected for
484: absorption, was $\Gamma\!=\!2.05\pm 0.15$ (Page et al.~2007).
485: 
486: 
487: 
488: \noindent {\bf GRB 061126:} This very long burst had four main overlapping 
489: peaks, the last peak ending $\sim\!25$s after trigger, but low-level 
490: emission was detected until $\sim\!200$s later. The RHESSI satellite also 
491: detected this burst, and also saw $\gamma$-ray emission for $\sim\!25$s. 
492: The XRT detected the X-ray emission only long after the prompt emission 
493: had faded. These late data are shown in Fig.~\ref{f2}b. The photon spectral index 
494: after correcting for absorption (Perley et al.~2007) is $\Gamma=2.00\pm 
495: 0.07$, and is time-independent, suggesting that the entire XRT light curve 
496: is that of the synchrotron afterglow of GRB 061126.
497: 
498: \noindent {\bf GRB 061007:} 
499: This long bright burst lasted $75\pm 5$ s. Its lightcurve showed three 
500: large peaks, and a smaller peak starting at 75 s, rising to a maximum 
501: at 79 s and declining with a very long and fast 
502: decay. The XRT began follow-up observations 80 s after trigger. The 0.3--10 
503: keV light curve (Fig.~\ref{f3}a) shows a single power-law decline with a slope of 
504: $1.6\pm 0.1$. In the CB model this is the tail of a 
505: cannonical AG whose `plateau' ended before the XRT began its 
506: observations. The predicted photon spectral index (DDD2007b), 
507: $\Gamma=\alpha+0.5=2.10\pm 0.10$, is consistent with the best fit 
508: spectral index, $\Gamma=2.03\pm 0.10$, shown in Fig.~\ref{f3}d.
509: 
510: 
511: 
512: \subsection{Hardness ratios: CB-model results}
513: In the CB-model the SR-dominated X-ray afterglow, if corrected for 
514: absorption, has a time-independent photon spectral index, 
515: $\Gamma\!\sim\!2.1$, and 
516: a constant hardness ratio. This expectation is consistent, within 
517: observational errors, with the Swift data in all the cases we considered,
518: with the possible exception of XRF 060218, whose complex situation
519: regarding absorption corrections we have reviewed.
520: The spectral behaviour is much more complex during the prompt emission.
521: 
522: Since XRF 060218 is a single-flare event, its light curve and the 
523: evolution of its HR, shown in Figs.~\ref{f1}a,c, are simple. The agreement 
524: between the model expectations and the XRT observations is satisfactory. 
525: The CB-model parameters are specified in Table~1. Multi-flare events such 
526: as GRB 060904a and, to a lesser extent, GRB 061121, require 
527: multi-parameter fits; the number of peaks we fit and their relevant 
528: parameters are also specified in Table~1. The way the HR of these 
529: bursts 
530: predictably follows the ups and downs of the flux is quite impressive, 
531: compare Fig.~\ref{f1}b with \ref{f1}d, and 
532: Fig.~\ref{f2}a with \ref{f2}c. For GRBs 061126 (figs.~\ref{f2}b,c and \ref{f3}b)
533: and 061007 (figs.~\ref{f3}a,c,d), 
534: the available data covers only the SR-dominated X-ray AG where, as 
535: expected, the HR ratio is constant. Note in Fig.~\ref{f2}b that, although the 
536: late time behaviour of the flux has the shape predicted by the CB model, 
537: the measured points lie systematically above the prediction. Such a 
538: discrepancy may result from a decreasing X-ray absorption along the line 
539: of sight to the AG source. The fluxes reported in the SWIFT XRT repository 
540: (Evans et al. 2007) assume a constant absorption during the entire 
541: measurements.  In the CB model, the jet of CBs moves hundreds of parsecs 
542: during the observations, and the absorption may decrease with time as the 
543: jet approaches the halo of the host galaxy.
544:   
545: 
546: \section{CB-model results for the effective spectral index}
547: 
548: The spectral index, $\Gamma(t)\!=\!\beta(t)+1$ of many GRBs,
549: extracted from an empirical power-law parametrization,
550: $ F_\nu\!\propto\! \nu^{-\beta}\,t^{-\alpha}$, is reported in
551: {\it http://swift.physics.unlv.edu/xrt.html},
552: and discussed in more detail for a selected set of bright GRBs by
553: Zhang et al.~(2007). As reported in the
554: introduction, these results on $\Gamma(t)$ may themselves be a rough description 
555: of rapidly-varying spectra potentially having an exponential energy-dependence,
556: as in Eq.~(\ref{GRBXlc}). Yet, we may define an {\it effective} index via the logarithmic  
557: derivative of the prompt ICS spectrum. For a single pulse in Eq.~(\ref{GRBXlc}):
558: \begin{equation} 
559: \Gamma_{\rm eff}(E,t-t_i) =-E\; {d\,{\rm log} F_E\over dE}\Bigg|_{E=\widetilde E}=
560:                   \alpha_g+{\widetilde E\over E_p(t-t_i)}\,,
561: \label{Geff}
562: \end{equation}
563: where ${\widetilde E}$ is an effective constant energy, $t$ is the time
564: after trigger, and $\alpha_g\!\approx\! 1$ is defined in Eq.~(\ref{GRBspec}).
565: For the synchrotron afterglow, the CB model predicts a power-law 
566: spectrum with roughly a constant photon index $\Gamma_{\rm SR}$, 
567: and a late-time temporal power-law decline with a power-law index 
568: %$\alpha\!=\! \Gamma_{\rm SR}-1/2$ 
569: (DDD2007b):
570: \begin{equation}
571: \alpha= \Gamma_{\rm SR}-1/2.
572: \label{alpha}
573: \end{equation}
574: 
575: In the data analysis in
576: {\it http://swift.physics.unlv.edu/xrt.html}, for lack of sufficiently
577: large statistics, different time intervals were coadded, smoothing the time-dependence 
578: of the fitted spectral index.
579: For an `effective-index' study of the results of this data analysis, a
580: single-pulse approximation is adequate to the description of a GRB's 
581: $\Gamma(t)$ at the end of the prompt phase and during the fast decline.
582: In this approximation, for a pulse starting at $t\!=\!t_i$, followed by a 
583: SR-dominated afterglow, the rough CB-model prediction is:
584: \begin{equation}
585: \Gamma_{\rm eff}\sim \left[1+ 
586: {\widetilde E \over E_p(t)}\right]\, \Theta[t_{\rm AG}-t]\,\Theta[t-t_i]+ 
587: \Gamma_{\rm SR}\, \Theta[t-t_{\rm AG}]\, ,
588: \label{Gammat}
589: \end{equation}
590: where  $t_{\rm AG}$ is the time at 
591: which the SR `afterglow' takes over the ICS `prompt' emission.
592: The assumed rather abrupt transition from the ICS-dominated
593: first term in Eq.~(\ref{Gammat}), to the second SR-dominated term,
594: is justified by Eqs.~(\ref{Epi}, \ref{GRBXlc}). Indeed, the late decline
595: of the ICS-dominated term is exponential in the square of the time.
596: 
597: In Figs.~\ref{f4} to  \ref{f6} we compare Eq.~(\ref{Gammat}) with the
598: results for $\Gamma(t)$ for twelve GRBs from the cited web-site
599: for which the measurements are good. The figures show how the
600: extracted $\Gamma(t)$ reflects the expected very abrupt transition.
601: Our simple description of the
602: observations in terms of three parameters [$t_i$, $\widetilde E/E_p(t_i)$
603: and $\Delta t$, listed in Table 2], is satisfactory. Also listed in Table 2
604: are the values of $\Gamma_{\rm SR}$, and the
605: values of $\alpha\!+\! 1/2$ from our CB-model
606: fits to the synchrotron-radiation afterglow. 
607: They are in fair agreement with Eq.~(\ref{alpha}).
608: 
609: \section{Approximate results on more GRBs}
610: 
611: Other authors have analized many more GRBs than we have in this paper.
612: Zhang et al.~(2007), for instance, confronting the failure of
613: the high-latitude emission of the FB model to 
614: explain the rapid softening of the tail of the prompt emission
615: in sixteen `clean-tail' bright GRBs, proposed an empirical parametrization 
616: of the X-ray light curve during this phase. Its spectral evolution 
617: can be rewritten as a time-dependent exponentially cutoff power-law: 
618: \begin{equation}
619: F_E\propto \left[{E\over E_c(t)}\right]^{1-\alpha_g}\, 
620: e^{-E/E_c(t)},\;\;\;\; E_c(t)=E_c(t_i)\,\left({t-t_i\over t_i}\right)^{-k}\, .
621: \label{Zhang}
622: \end{equation} 
623: For  $t\gg t_i$, this is the evolution predicted by the CB model (DD2004), 
624: provided one identifies $E_c(t)\!=\!E_p(t)$.
625: Indeed, $E_p(t)\!\approx\!E_p(t_i)$ for $t-t_i\!\ll\!\Delta t_i$,
626: while for $t-t_i\!\gg\!\Delta t_i$,  
627: $E_p(t)\!\approx\! E_p(t_i)\, [(t-t_i)/\Delta t_i]^{-k}/2$,
628: with $k\!=\!2$, see Eqs.~(\ref{Epi},\ref{HRapprox}).
629: These limiting behaviours may be interpolated by the empirical 
630: parametrization of Zhang et al.~(2007), in their chosen narrow 
631: range of $t$, with a constant $k\!\leq \!2$ (they find $1\!\leq\!k\!\leq\!1.6$).  
632: These authors also discern GRBs without a 
633: rapid spectral softening during the fast decline. These seem to us
634: to be cases whose spectral evolution is poorly measured,
635: or cases, like GRBs 061126 and 061007, whose `fast decline phase' is not 
636: the end of the prompt emission but the late decline 
637: of a canonical AG  whose plateau phase ended
638:  before the beginning of the XRT observations
639: (DD20007b).
640: 
641: 
642: \section{Conclusions}
643: 
644: The spectrum of the $\gamma$-ray peaks and X-ray flares of a GRB or an XRF 
645: is predicted in the CB model: it is the spectrum of the `glory's light', 
646: Compton-boosted by the electrons in a CB (DD2004). The time evolution of 
647: the spectrum traces the voyage of the CB through this `target' light. 
648: Though the model predicts the spectrum and its evolution at all 
649: frequencies and times, we have focused on the very rapid decline of the 
650: flux at the end of a pulse, and the equally swift spectral softening. 
651: Their understanding is simple: the glory's `target' light is light 
652: scattered by the circum-burst matter, and its spectrum is exponentially 
653: cutoff. Its number density, and the flux of a pulse, decrease roughly as 
654: $1/r^2\!\propto\!1/t^2$. Simultaneously, the target light is becoming more 
655: radial, so that the characteristic energy of the up-scattered radiation 
656: also decreases as $1/t^2$. These simple facts, explicitly reflected in 
657: the predicted `master formula', Eqs.~(\ref{Epi},\ref{GRBXlc}), result in an 
658: excellent description of the observations.
659: 
660: Lacking access to detailed spectral analyses, we have used Swift
661: data on hardness ratios, uncorrected for X-ray absorption  (Evans et al.~2007), 
662: as well as the effective spectral index of the unabsorbed 
663: spectrum reported in {\it http://swift.physics.unlv.edu/xrt.html}.
664:  We have demonstrated that the spectral time dependences snugly
665: trace their expected correlation to the corresponding flux variations.
666: This test of the CB model validates it once again. Yet,
667: carefully time-resolved absorption corrections would allow even more conclusive 
668: tests. Time-resolved corrections are 
669: important because, in the CB model, the line of sight to the hyperluminal CBs 
670: changes significantly during the long afterglow phase (e.g., DD2004) sweeping
671: different regions of the host galaxy and  the 
672: IGM. The changing absorption may  induce flickering of the observed 
673: X-ray light curve and X-ray spectrum. In fact, the scintillation-like 
674: behaviour in many X-ray light curves and spectra 
675: (see Figs.~\ref{f1},\ref{f2},\ref{f3}), if not 
676: instrumental, may be due to the motion of the CBs in the host 
677: galaxy. This motion may also 
678: explain (Dado, Dar \& De R\'ujula, in preparation) the reported time-dependence 
679: of the equivalent widths of intergalactic absorption systems detected in the 
680: afterglow of GRB 060206 (Hao et al.~2007, but see also Thone et al.~2007).
681: 
682: At least for GRBs or XRFs with a `canonical' light curve, the 
683: transition in time from a rapidly falling X-ray decline to a much less steep plateau
684: --accompanied by the simultaneous and even more pronounced change in the 
685: spectrum that we have studied-- reflect
686: one of the most discontinuous transitions seen in astrophysical data.
687: In the CB model this transition is not attributed to the continued activity of a 
688: steadily energizing engine, but to the passage from one to another dominant 
689: radiation mechanism:
690: inverse Compton scattering versus synchrotron radiation. The transition is so fast
691: because the late decline of the ICS contribution of Eqs.~(\ref{Epi}, \ref{GRBXlc})
692: is exponential in time, a consequence of the exponential
693: cutoff (in energy) of the thin-bremsstrahlung
694: spectrum of the up-scattered light (DD2004).
695: 
696: \noindent
697: {\bf Acknowledgment:} The authors would like to thank an
698: anonymous referee for useful information, comments and suggestions.
699: 
700: \begin{thebibliography}
701: 
702: \bibitem[1993]{Band1993}
703: Band, et al.~1993, ApJ, 413, 281
704: 
705: \bibitem[2007]{Band2007}
706: Burrows, D. N. \& Racusin, J.~2007, arXiv:astro-ph/0702633
707: 
708: \bibitem[2006]{Campana2006}
709: Campana, S., et al.~2006, Nature, 442, 1008
710: 
711: \bibitem[2002]{DDD2002}
712: Connaughton, V.~2002, ApJ, 567, 1028
713: 
714: \bibitem[2002]{DDD2002OXAG}
715: Dado, S., Dar, A. \& De R\'ujula, A.~2002, A\&A, 388, 1079
716: 
717: \bibitem[2003]{DDD2003RADIOAG}
718: Dado, S., Dar, A. \& De R\'ujula, A.~2003, A\&A, 401, 243
719: 
720: \bibitem[2004]{DDD2004XRF}
721: Dado, S., Dar, A. \& De R\'ujula, A.~2004, A\&A, 422, 381
722: 
723: \bibitem[2007a]{DDD2007X}
724: Dado, S., Dar, A. \& De R\'ujula, A.~2007a, arXiv:0706.0880
725: 
726: \bibitem[2007b]{DDD2007XBREAKS}
727: Dado, S., Dar, A. \& De R\'ujula, A.~2007b, arXiv:0712.1527 
728: 
729: \bibitem[2007]{Dai2007}
730: Dai, X., et al., 2007, ApJ, 658, 509
731: 
732: \bibitem[1998]{Dar1998}
733: Dar, A.~1998, ApJ, 500, 93
734: 
735: \bibitem[1999]{DP1999}
736: Dar, A. \& Plaga, R.~1999, A\&A, 349, 259
737: 
738: 
739: \bibitem[2004]{DD2004}
740: Dar, A. \& De R\'ujula, A.~2004, Physics Reports, 405, 203
741: 
742: \bibitem[2004]{Dermer2004}
743: Dermer, C. D., 2004, ApJ, 614, 284
744: 
745: \bibitem[2007]{Evans2007}
746: Evans, P. A.  et al.~2007, A\&A, 469, 379 
747: 
748: \bibitem[2004]{FMN1996}
749: Fenimore, E. E., Madras, C. D., \& Nayakshin, S., 1996, ApJ, 473, 998
750: 
751: \bibitem[2002]{Giblin2002}
752: Giblin, T. W. et al.~2002, ApJ, 570, 573 
753: 
754: \bibitem[2007]{Hao2007}
755: Hao, H., et al.~2007, ApJ, 659, 99
756: 
757: \bibitem[2007]{Kocevski2007}
758: Kocevski, D. \& Butler, N. 2007, arXiv:0707.4478 
759: 
760: \bibitem[2007]{Kumar}
761: Kumar, P.,  et al.~2007, MNRAS, 376, L57
762: 
763: \bibitem[2000]{KP}
764: Kumar, P. \& Panaitescu, A. 2000, ApJ, 541, L51
765: 
766: \bibitem[2006]{Liang2006}
767: Liang, E. W., et al.~2006, ApJ, 646, 351
768: 
769: \bibitem[2007]{Liang2007}
770: Liang, E. W., et al.~2007, arXiv:0708.2942
771: 
772: \bibitem[2002]{Mesz2002}
773: M\'{e}sz\'{a}ros, P.~2002, ARA\&A, 40, 137
774: 
775: \bibitem[2006]{M'esz'aros2006}
776: M\'{e}sz\'aros, P.~2006, Rept. Prog. Phys.   69,  2259
777: 
778: \bibitem[1998]{M'esz'aros1998}
779: M\'{e}sz\'{a}ros, P., Rees, M. J. \& Wijers, R. A. M. J.~1998, ApJ, 499, 301
780: 
781: 
782: \bibitem[2006]{Mirabal2006}
783: Mirabal, N., et al. 2006, ApJ, 643, L99
784: 
785: \bibitem[2006]{Nous2006}
786: Nousek, J., et al.~2006, ApJ, 642, 389
787: 
788: \bibitem[2006]{O'Brien2006}
789: O'Brien, P. T., et al. 2006, ApJ, 647, 1213 
790: 
791: \bibitem[2007]{Page2007}
792: Page, K. L., et al. 2007, arXiv:0704.1609
793:   
794: \bibitem[2006]{Panaitescu}
795: Panaitescu, A., et al.~2006, MNRAS, 369, 2059
796: 
797: \bibitem[2007]{Perley2007}
798: Perley, D. A., et al.~2007, arXiv:astro-ph/0703538
799: 
800: \bibitem[2005]{Pian2005} 
801: Pian, E., 2006, Nature, 442, 1011
802: 
803: \bibitem[1999]{Piran1999}
804: Piran, T.~1999, Physics Report, 314, 575
805: 
806: \bibitem[2000]{Piran2000}
807: Piran, T.~2000, Physics Report, 333, 529
808: 
809: \bibitem[2005]{Piran2005} 
810: Piran, T.~2005, RvMP, 76, 1143
811: 
812: \bibitem[2002]{Ryde2002}
813: Ryde, F. \& Svensson, R.~2002, ApJ, 566, 210 
814: 
815: \bibitem[2002]{Rossi2002}
816: Rossi, E., Lazzati, D. \& Rees, M. J.~2002, MNRAS, 332, 945
817: 
818: \bibitem[1995]{SD1995}
819: Shaviv, N. J. \& Dar, A.~1995, ApJ, 447, 863
820: 
821: \bibitem[2007]{Thone2007}
822: Thone, C. C. et al.~2007, arXiv:0708.3448  
823: 
824: \bibitem[2007]{Urata2007}
825: Urata, Y., et al.~2007, 2007arXiv0707.2826
826: 
827: \bibitem[2007]{Yamazaki2006}
828: Yamazaki, R., et al.~2006, MNRAS, 369, 311
829: 
830: \bibitem[2007]{Yonetoku2007}
831: Yonetoku, D., et al.~ 2007, arXiv:0708.3968 
832: 
833: \bibitem[2007]{ZM2007}
834: Zhang, B.~2007, ChJAA, 7, 1
835: 
836: \bibitem[2002]{ZM2002}
837: Zhang, B. \& M\'esz\'aros, P.~2002, ApJ, 581, 1236
838: 
839: \bibitem[2004]{ZM2004}
840: Zhang, B. \& M\'esz\'aros, P.~2004, IJMPA, 19, 2385
841: 
842: \bibitem[2006]{Zhang2006}
843: Zhang, B., et al.~2006, ApJ, 642, 354
844: 
845: \bibitem[2007]{Z2007}
846: Zhang, B. B., Liang, E. W. \&  Zhang, B.~2007, ApJ, 666.1002 
847: 
848: 
849: \end{thebibliography}
850: 
851: \newpage
852: \begin{deluxetable}{lllllc}
853: \tablewidth{0pt}
854: \tablecaption{CB-model afterglow parameters}
855: \tablehead{
856: \colhead{Parameter} & \colhead{060218} & \colhead{060904a}& 
857: \colhead{061121}& \colhead{061126}& \colhead{061007}}
858: \startdata
859: $t_1\,[{\rm s}]$         & $-1080$ & 41.08  & 52.48 &  --- & ---\\
860: $\Delta t_1\,[{\rm s}]$  &  1977 & 16.02  & 12.44 &  --- &--- \\
861: $\Delta E/E_p(t_1) $     &  0.19 & 0.0452 & 0.061 & --- & --- \\
862: $t_2\,[{\rm s}]$         & --- & 252.8  & 96.88 &  --- &---\\
863: $\Delta t_2\,[{\rm s}]$  & --- & 27.75  & 18.80 & --- & ---\\
864: $\Delta E/E_p(t_2)    $  & ---  & 0.0177 & 0.0014&   --- & --- \\
865: $t_3\,[{\rm s}]$         & --- & 629.7  &  --- & --- & ---\\
866: $\Delta t_3\,[{\rm s}]$  & ---  &  44.0  &  ---     &  --- & ---\\
867: $t_4\,[{\rm s}]$         & ---  & 703.4  &  --- &  --- & ---  \\
868: $\Delta t_4\,[{\rm s}]$  &  ---     & 747.3  & --- &  --- &--- \\
869: \hline
870: $t_0\,[{\rm s}]$         &   183  &  821  & 248&    263 &  40  \\
871: $\gamma\,\theta$         & 4.28  & 1.25  & 1.42  &  1.12& $\ll 1$ \\
872: $p$                     & 2.20   & 2.20  & 2.20  &  1.90 & 2.26  \\
873: \enddata
874: \label{t1}
875: \end{deluxetable}
876: 
877: \newpage
878: \begin{deluxetable}{llllllc}
879: \tablewidth{0pt}
880: \tablecaption{Parameters in the description of the photon spectral index
881: $\Gamma(t)$. The values of $\Gamma_{\rm SR}$ are from the Swift
882: public data in {\it http://swift.physics.unlv.edu/xrt.html}. 
883: The values of $\alpha\!+\! 1/2$ are from our CB-model
884: fits to the synchrotron-radiation afterglow. In the model the two last columns ought
885: to be equal, see Eq.~(\ref{alpha}). }
886: \tablehead{
887: \colhead{GRB} & \colhead{$t_i$[s]} & \colhead{$\Delta$ [s]}&
888: \colhead{$\widetilde E/E_p(t_i)$}&\colhead{$t_{\rm AG}$ [s]}&
889: \colhead{$\Gamma_{\rm SR}$} & \colhead{$\alpha+1/2$}}
890: \startdata
891: 061126  &  ---  & --- &  --- & --- & $1.93\pm 0.12$  & 1.95 \\
892: 061007  &  --- & --- &  ---   & --- & $2.10\pm 0.20$ & 2.13 \\
893: 070129 & 243  & 487 &  0.57 &  1050  & $2.28\pm 0.22 $ & 2.14\\
894: 061222A &  108 & 113  &   1.09 & 195 & $2.15\pm 0.08$ & 2.15\\
895: 061121  &  64 & 5.15 &  0.0035 &161 & $1.99\pm 0.13 $& 2.10 \\
896: 061110A &   3.7  & 219  & 1.056 & 261 & --- & 1.80 \\
897: 060814 & 109 & 295 &  0.68 & 360 & $2.20\pm 0.10$ & 2.16\\
898: 060729 & 131 & 146 & 1.48 & 300 & $2.10\pm 0.10$  &  2.10\\
899: 060510B & 190  &  57  & 0.036 & 460 & $2.14\pm 0.15$ & --- \\
900: 060211A &  0  & 325  &  0.44 & 371& $2.03\pm 0.12$ & 2.04 \\
901: 050814  &  12    & 365  & 0.54 & 361& $1.91\pm 0.09$ & 1.93 \\
902: 050724  &  0     & 154 &  0.23 & 320  & $1.88\pm 0.16$ & 1.86\\
903: 050717  &  0     & 194  & 0.19 & 195 & $1.85\pm 0.12 $ & 1.84\\
904: 050716  &  31    & 96  &   0.037 &496& $1.97\pm 0.11$ & 1.88\\
905: 
906: \enddata
907: \label{t2}
908: \end{deluxetable}
909: 
910: 
911: \newpage
912: 
913: \begin{figure}[]
914: \centering
915: %\vspace{-1cm}
916: \vbox{
917: \hbox{
918: \epsfig{file=fig1a.eps,width=8.0cm}
919: \epsfig{file=fig1b.eps,width=8.0cm}
920: }}
921: 
922: \vbox{
923: \hbox{
924: %\hskip 3.cm
925:  \epsfig{file=fig1c.eps,width=8cm}
926:  \epsfig{file=fig1d.eps,width=8cm}
927: }}
928: %\vspace{2cm}
929: \caption{
930: Comparisons between Swift XRT observations 
931: (Evans et al.~2007) and the CB model predictions.
932: {\bf Top left (a):} The light curve of XRF 060218.
933: {\bf Top right (b):} The light curve of GRB 060904a. 
934: {\bf Bottom left (c):} The hardness ratio of XRF 060218.
935: {\bf Bottom right (d):} The hardness ratio of GRB 060904a.
936: }
937: \label{f1}
938: \end{figure}
939: 
940: \newpage
941: \begin{figure}[]
942: \centering
943: \vspace{-1cm}
944: \vbox{
945: \hbox{
946: %\hskip 2.cm
947: \epsfig{file=fig2a.eps,width=8.0cm}
948: \epsfig{file=fig2b.eps,width=8.0cm}
949: % \hskip -.3cm
950: }}
951: %\vspace{2cm}
952: \vbox{
953: \hbox{
954: %\hskip -.3cm
955:  \epsfig{file=fig2c.eps,width=8cm}
956:  \epsfig{file=fig2d.eps,width=8cm}
957: }}
958: %\vspace{1cm}
959: \caption{
960: Comparisons between Swift XRT observations
961: (Evans et al.~2007) and the CB model predictions. 
962: {\bf Top left (a):} The light curve of GRB 061121.
963: {\bf Top right (b):} The light curve of GRB 061126. 
964: {\bf Bottom left (c):} The hardness ratio of GRB 061121.
965: {\bf Bottom right (d):} The hardness ratio of GRB 061126.
966: }
967: \label{f2}
968: \end{figure}
969: 
970: 
971: \begin{figure}[]
972: \centering
973: %\vspace{-1cm}
974: \vbox{
975: \hbox{
976: \epsfig{file=fig3a.eps,width=8.0cm}
977: \epsfig{file=fig3b.eps,width=8.0cm}
978: }}
979: 
980: \vbox{
981: \hbox{
982: %\hskip 3.cm
983:  \epsfig{file=fig3c.eps,width=8cm}
984:  \epsfig{file=fig3d.eps,width=8cm}
985: }}
986: %\vspace{2cm}
987: \caption{
988: Comparisons between Swift XRT observations (Evans et al.~2007)
989: and the CB-model predictions.
990: {\bf Top left (a):} The light curve of GRB 061007.
991: {\bf Top right lef (b):} The photon spectral index of GRB 061126.
992: {\bf Bottom left (c):} The hardness ratio of GRB 061007.
993: {\bf Bottom right (d):} The photon spectral index of GRB 061007.
994: $\Gamma$ values are from {\it http://swift.physics.unlv.edu/xrt.html}.
995: }
996: \label{f3}
997: \end{figure}
998: 
999: 
1000: \begin{figure}[]
1001: \centering
1002: %\vspace{-1cm}
1003: \vbox{
1004: \hbox{
1005: \epsfig{file=fig4a.eps,width=8.0cm}
1006: \epsfig{file=fig4b.eps,width=8.0cm}
1007: }}
1008: 
1009: \vbox{
1010: \hbox{
1011: %\hskip 3.cm
1012:  \epsfig{file=fig4c.eps,width=8cm}
1013:  \epsfig{file=fig4d.eps,width=8cm}
1014: }}
1015: %\vspace{2cm}
1016: \caption{
1017: Comparisons between the effective photon spectral 
1018: index in the 0.3-10 keV X-ray band as inferred from 
1019: observations of GRBs with the Swift XRT, 
1020: and the CB model approximate prediction, Eq.~(\ref{Gammat}).
1021: {\bf Top left (a):} GRB 070129.
1022: {\bf Top right (b):} GRB 06122A.
1023: {\bf Bottom left (c):} GRB 061121.
1024: {\bf Bottom right (d):} GRB 061110A.
1025: $\Gamma$ values are from {\it http://swift.physics.unlv.edu/xrt.html}.
1026: }
1027: \label{f4}
1028: \end{figure}
1029: 
1030: 
1031: \begin{figure}[]
1032: \centering
1033: %\vspace{-1cm}
1034: \vbox{
1035: \hbox{
1036: \epsfig{file=fig5a.eps,width=8.0cm}
1037: \epsfig{file=fig5b.eps,width=8.0cm}
1038: }}
1039: 
1040: \vbox{
1041: \hbox{
1042: %\hskip 3.cm
1043:  \epsfig{file=fig5c.eps,width=8cm}
1044:  \epsfig{file=fig5d.eps,width=8cm}
1045: }}
1046: %\vspace{2cm}
1047: \caption{
1048: Comparisons between the effective photon spectral
1049: index in the 0.3-10 keV X-ray band as inferred from
1050: observations of GRBs with the Swift XRT,
1051: and the CB model approximate prediction, Eq.~(\ref{Gammat}).
1052: {\bf Top left (a):} GRB 060814.
1053: {\bf Top right (b):} GRB 060729.
1054: {\bf Bottom left (c):} GRB 060501B.
1055: {\bf Bottom right (d):} GRB 060211A.
1056: $\Gamma$ values are from {\it http://swift.physics.unlv.edu/xrt.html}.
1057: }
1058: \label{f5}
1059: \end{figure}
1060: 
1061: \begin{figure}[]
1062: \centering
1063: %\vspace{-1cm}
1064: \vbox{
1065: \hbox{
1066: \epsfig{file=fig6a.eps,width=8.0cm}
1067: \epsfig{file=fig6b.eps,width=8.0cm}
1068: }}
1069: 
1070: \vbox{
1071: \hbox{
1072: %\hskip 3.cm
1073:  \epsfig{file=fig6c.eps,width=8cm}
1074:  \epsfig{file=fig6d.eps,width=8cm}
1075: }}
1076: %\vspace{2cm}
1077: \caption{
1078: Comparisons between the effective photon spectral
1079: index in the 0.3-10 keV X-ray band as inferred from
1080: observations of GRBs with the Swift XRT,
1081: and the CB model approximate prediction, Eq.~(\ref{Gammat}).
1082: {\bf Top left (a):}  GRB 050814.
1083: {\bf Top right (b):} GRB 050724.
1084: {\bf Bottom left (c):} GRB 050717.
1085: {\bf Bottom right (d):} GRB 050716.
1086: $\Gamma$ values are from {\it http://swift.physics.unlv.edu/xrt.html}.
1087: }
1088: \label{f6}
1089: \end{figure}
1090: 
1091: \end{document}
1092: 
1093: 
1094: