1: \documentclass[12pt,preprint]{aastex}
2:
3: %\usepackage{graphicx}
4: %\usepackage{epsfig,amsmath}
5:
6: %\usepackage{hyperref}
7: %\bibliographystyle{amsplain}
8:
9: \newcommand{\be}{\begin{eqnarray}}
10: \newcommand{\ee}{\end{eqnarray}}
11: \newcommand{\gm}[3]{\Gamma^{#1}_{\; #2 #3}}
12: \newcommand{\DCE}{ {\bf DCE } }
13: \newcommand{\etamfp}{\eta_\mathrm{mfp}}
14: \newcommand{\sigCrit}{\sigma_\mathrm{crit}}
15: \newcommand{\etaCrit}{\eta_\mathrm{crit}}
16: \newcommand{\rel}{relativistic}
17: \newcommand{\nonrel}{non\-rel\-a\-tiv\-is\-tic}
18: \newcommand{\transrel}{trans-rel\-a\-tiv\-is\-tic}
19: \newcommand{\ultrarel}{ul\-tra-rel\-a\-tiv\-is\-tic}
20: \newcommand{\degg}{^\circ}
21: \newcommand{\Tbn}{\theta_\mathrm{B0}}
22: \newcommand{\TP}{test-particle}
23: \newcommand{\deltime}{\delta t}
24: \newcommand{\utZ}{u_{0}}
25: \newcommand{\gamZ}{\Gamma_0}
26: \newcommand{\mc}{Monte Carlo}
27: \newcommand{\MC}{Monte Carlo}
28: \newcommand{\syn}{synchrotron}
29: \newcommand{\synch}{synchrotron}
30:
31: \begin{document}
32:
33: %\title{Baryon Anisotropy within Gamma-Ray Burst Jets}
34: \title{A Model for Fast Rising, Slowly Decaying Subpulses
35: in $\gamma$-ray Bursts}
36: \author{David Eichler and Hadar Manis}
37: \affil{Department of Physics,
38: Ben-Gurion University, Beer-Sheva 84105, Israel;
39: eichler@bgu.ac.il}
40: %\author{Donald C. Ellison}\affil{Physics
41: %Department, North Carolina State University, Box 8202, Raleigh, NC
42: %27695, U.S.A.; don\_ellison@ncsu.edu}
43:
44: \begin{abstract}
45: Gamma ray bursts (GRB's) often feature subpulses that have a
46: distinctively asymmetric profile -- they rise quickly and decay much
47: more slowly, while their spectrum softens slightly with observer
48: time. It is suggested that these subpulses are caused by slow
49: baryonic clouds embedded within a primary $\gamma$-ray beam, which
50: scatter the $\gamma$-radiation into our line of sight as they
51: accelerate. Good quantitative agreement is obtained with observed
52: light curves and spectral evolution. The kinetic energy that the
53: baryonic component of GRB jets receives from the primary
54: $\gamma$-radiation is predicted to be about equal to the amount of
55: $\gamma$-radiation that is scattered, consistent with observations
56: of afterglow. Several other observational consequences are briefly
57: discussed. The possibility is raised that the time scale of short
58: GRB is established by radiative acceleration and/or baryon
59: injection rather than the time scale of the central engine.
60: \end{abstract}
61:
62: \keywords{$\gamma$-rays:bursts}
63:
64: \section{Introduction}
65:
66: The nature of the central engine of GRB's remains unresolved by
67: observations. It is not known what the primary form of energy
68: outflow is, or how the $\gamma$-radiation near the spectral peak is
69: powered. Popular models invoke a baryonic outflow as the primary
70: form of energy output, and synchrotron emission from electrons that
71: are shock-accelerated in optically thin regions of the outflow
72: (Meszaros \& Rees 1994), presumably by internal shocks caused by the
73: unsteadiness of the flow. Other models suggest that the primary
74: output of the central engine is photons (e.g. Eichler 1994, Eichler
75: \& Levinson 2000, Rees \& Mezsaros 2005) or Poynting flux (e.g.
76: Thompson 1994, Lyutikov \& Blandford 2003). If the observed
77: $\gamma$-radiation (and presumably a residue of pairs) is the
78: primary form of released energy at the photosphere of the outflow,
79: then the question is if and how the $\gamma$-radiation accelerates
80: the baryonic outflow responsible for GRB afterglow.
81:
82: In this {\it Letter} we propose that many $\gamma$-ray bursts are
83: viewed at an angle that is slightly offset from the direction of
84: the primary $\gamma$-ray beam, and that much of the primary
85: radiation is, consequently, observed only after being scattered by
86: slower baryons that have not yet reached their terminal
87: Lorentz factor $\Gamma_{\infty}$. The existence
88: of slow baryons in the flow is not a strong assumption. They could
89: enter the primary beam from the periphery due to the diffusion of
90: slow neutrons from a surrounding wind (Eichler and Levinson 1999;
91: Levinson and Eichler 2003) or possibly by turbulent excitation of
92: transverse motion by Kelvin--Helmholtz instabilities at the side.
93: Moreover, radiatively driven instabilities could bunch baryons
94: within the outflow. Even in the internal shock model, slow material
95: would be necessary to dissipate the energy of fast material, and the
96: shocked material following their collision would still be slower
97: than at the eventual terminal Lorenz factor. The main assumption
98: made here is that the photons, though they may scatter off clouds of
99: high optical depth, are not trapped {\it within} a large optical
100: depth, and that they accelerate the clouds/bunches as they overtake
101: them from the rear.
102: We show that the observed temperature, as
103: defined by the location of the emission peak, should then decrease
104: approximately as $t^{-2/3}$, as reported by Ryde (2004). We also
105: show that the observed time profile has a characteristic fast rise,
106: slow decay typical of sub-pulses\footnote{often referred to as
107: 'FRED's -- for fast rise, exponential decay} often seen in the
108: prompt emission of GRB's. The decay is due largely to the
109: acceleration of the scattering baryons by the radiation pressure of
110: the primary photons, which causes the beam of scattered photons to
111: narrow to below the offset angle.
112:
113: It would not be surprising if matter were accelerated beyond the
114: photosphere because the isotropic equivalent luminosity can be as
115: high as $10^{15}$ Eddington luminosities or more. This would
116: surely be a powerful accelerator of any material that did not
117: already have a Lorentz factor of at least several hundred, even
118: at distances of $10^{13}$ to $10^{14}$ cm, where the photosphere
119: is placed in many models. The acceleration of isolated baryons
120: would be much faster than the hydrodynamical timescale. In the
121: process of being accelerated, the baryonic scattering material
122: would scatter as much radiative energy as the energy it received
123: from the radiation pressure (see below). Thus, radiation scattered
124: by accelerating baryons would represent as significant a part of
125: the GRB energy budget as the baryons.\footnote{Note that even in
126: the internal shock model, where the exiting photons are secondary
127: products of fast baryons overtaking and shocking slightly slower
128: ones, the resultant super-Eddington photon output would
129: pre-accelerate most of the slower baryons before they were
130: overtaken. A self-consistent treatment of this model should
131: therefore include such scattering and pre-acceleration as well.}
132:
133: %The possibility is also briefly considered that
134: %the central engines of short GRB have
135: %the same lifetime as those of long GRB, and that the difference
136: %in the observed duration is due to the difference in the proximity
137: %of the scattering baryons to the source.
138:
139: The idea that most GRB's are observed from an offset viewing angle
140: is well motivated and has been amply discussed in the literature.
141: One motivation is that very bright GRB's such as 990123 are much
142: brighter than the detection threshold and could be viewed by
143: observers that are offset by an angle of several times $1/\Gamma$,
144: just from their kinematic broadening. Another is that the Amati [and
145: Ghirlanda] correlations (Amati et al 2002, Ghirlanda et al 2004) can
146: be explained as kinematically correlated softening of both the peak
147: frequency and isotropic equivalent [jet] energy (Eichler and
148: Levinson 2004, Levinson and Eichler 2005, Eichler and Levinson
149: 2006). Scattered radiation from accelerating baryons can, at some
150: cost to peak brightness, significantly enhance the solid angle over
151: which a GRB could be detected, and, for sources with low
152: $V/V_{max}$, thus enhance the maximum volume $V_{max} \equiv \int
153: r_{max}^3(\theta,\phi)d\Omega/3$ over which a burst can be detected.
154: A considerable fraction of all detected GRB's, or subclasses
155: thereof, may then involve a component of scattered radiation. Many
156: observers could detect such radiation while the baryons that
157: scattered it into the line of sight continue in a slightly different
158: direction, and this separation between baryon direction and prompt
159: $\gamma$-ray direction could be a reason for the delay of strong
160: afterglow in many GRB's (Eichler and Granot 2006). We will also note
161: that at viewing angles $\theta_V$ that are large compared to the jet
162: opening angle $\theta_{\circ}$, $\theta_V \gg \theta_{\circ}$, this
163: scattering can give rise to outliers from the Amati relation.
164:
165: Very recently, Pe'er et al (2007) have independently proposed that
166: the softening of the spectral peaks of GRB subpulses results from
167: viewing material that is, with observer time, increasingly further
168: in angular separation from the line of sight. This must always occur
169: if the jet is extended in angle, but gives a steeper decline of
170: spectral peak frequency with time ($t^{-\alpha}$, $\alpha \sim 1$)
171: than the model proposed here, as well as a steeper decline in
172: luminosity.
173:
174: \section{A Simple Model}
175:
176: To calculate the luminosity of scattered radiation as a function of
177: time, we assume that the scattering material is introduced into the
178: primary photon beam at rest or with purely radial motion, and that
179: the size of the photon source is negligible. The photons are thus
180: completely combed radially and their flux in the scatterer frame is
181: $F^\prime = \frac{1-\beta}{1+\beta}F$ (Landau and Lifschitz 1962),
182: where $F$ is the flux in the source frame at radius $r$. Assue
183: first, for convenience, that the cloud is optically thin. The
184: acceleration in the scatterer frame is given by
185: \begin{equation}
186: d\beta_s/d\tau = F^\prime \sigma_T/m_ic^2 =
187: \frac{1-\beta}{1+\beta}F\sigma_T/m_ic^2
188: \end{equation}
189: and thus
190: \begin{equation}
191: d\beta/dt_{source} = \Gamma^{-3}d\beta_s/d\tau = \Gamma^{-3}F^\prime
192: \sigma_T/m_ic^2 = \Gamma^{-3}\frac{1-\beta}{1+\beta}F\sigma_T/m_i
193: c^2. \label{dbdt}
194: \end{equation}
195: Integrating this equation in the late time limit
196: %behavior of $(1-\beta)$,
197: when $ (1+\beta)\sim 2$, $Fdt_{source}\propto d\beta/(1-\beta)^{5/2}$,
198: we obtain $(1-\beta) \propto (\int
199: Fdt_{source})^{-2/3}$. The received flux by a distant observer at
200: viewing angle $\theta_V$ as a function of observer time $t$ is given
201: by (Rybicki and Lightman 1979)
202: \begin{equation}
203: dP_e/d\Omega = [\gamma(1-\beta \cos\theta_V)]^{-4}dP^\prime
204: /d\Omega^\prime ,
205: \end{equation}
206: and the late time behavior when $1/\Gamma \ll \theta_V$ is, for
207: constant F, $dP_e/d\Omega \propto (1-\beta)^3 \propto t^{-2}$.
208:
209: Note that for a powerful GRB with isotropic equivalent luminosity
210: $L_{iso} = 10^{53}L_{53}$ erg/s, and baryons beginning from rest
211: at radius $r = 10^{12}r_{12}$ cm, the acceleration time up to
212: $\Gamma \le (10^{8}L_{53}/r_{12})^{1/3}$ is less than the
213: hydrodynamical expansion time (Eichler 2004) for $r_{12}, L_{53}
214: \sim 1$. Thus, the model predicts {\it a priori}, given the
215: relevant range of parameters for the central engine and host star
216: envelope ($L_{53} \sim 1$, $r_{12} \sim 1$), that the baryons
217: naturally obtain a Lorentz factor of several hundred.
218:
219: It is possible that the baryonic cloud is optically thick when
220: injected into the primary beam. In this case, the back end is
221: compressed by the radiation pressure and a reverse shock is sent
222: through the cloud (which could result in particle acceleration and a
223: nonthermal component in the scattered radiation), and
224: the average acceleration is reduced by the optical depth $\tau$.
225: The scattered radiation emerges from the back end of the cloud,
226: after only one or very few scatterings, so, in the frame of the
227: cloud, the forward hemisphere is shaded. {\it When the cloud
228: accelerates beyond $\Gamma = 1/\theta_V$, the observer's line of
229: sight emerges from the shadow, and a sudden turn on of the scattered
230: radiation is seen.} The turn on is just at the value of $\beta$
231: where the flux of scattered radiation detected by the observer is
232: near maximum. Alternatively, the optical depth $\sigma_T
233: \int^{\infty}_r n(r^{\prime})[1-\beta(r^{\prime})]dr^{\prime}$ of
234: any given parcel of baryons also drops, due to the acceleration,
235: much faster than the expansion. The photosphere can {\it
236: self-organize} in the sense that a sudden drop in optical depth is
237: then both the cause and effect of a sudden drop in $[1-\beta]$.
238: %As
239: %long as the cloud is optically thick, the scattered radiation must
240: %then escape backwards in the frame of the cloud, which is
241: %illuminated from behind, and, in the source frame, all of the
242: %scattered radiation lies {\it outside} the $1/\Gamma$ beaming cone.
243: Finally, the cloud may be optically thin. This would imply that
244: much of the primary radiation escapes unscattered. This possibility
245: seems to be allowed, at present, by observations: A GRB as powerful
246: as 990123, for example, though only occurring once per $\sim 10^3$
247: bursts, is $\sim 10^3$ times as powerful as the typical GRB, and ,
248: having a high peak frequency, is a logical candidate for primary
249: emission. The proposed model can thus accommodate a rather large
250: range of initial optical depths, as long as the key assumption is
251: maintained that the photons are not trapped within the baryons as
252: they would be if everything
253: were distributed smoothly. %i.e. before
254: %they lose most of their energy to adiabatic expansion. This
255: %renders unnecessary (though does not prohibit) recovery of the
256: %energy lost to baryons with internal shocks.
257:
258: In Fig.~\ref{fig_1}, we have plotted $\hat t \equiv (\sigma_T/m_ic^2)\int (1-\beta
259: \cos \theta_V)F dt_{source}$ as a function of $\beta$, the velocity
260: of the scatterer in units of $c$. Here $F$ is to be taken at the
261: instantaneous position of the scatterer. We have also assumed a
262: plane parallel geometry, which is valid when the acceleration is
263: rapid compared to the expansion time. Assuming constant $F$, we have
264: then plotted in Fig.~\ref{fig_2} the observed light curve
265: $dP_e/d\Omega$ as a function of $\hat{t}$. Here, we have assumed
266: that the baryonic cloud is optically thin, so that its emission can
267: be seen by the observer even when $\Gamma \ll 1/\theta$. It is seen
268: even in this case that the characteristic time asymmetry that is the
269: signature of GRB subpulses is nicely reproduced by the model as seen
270: in Fig.~\ref{fig_2}. In the case where the cloud is injected - or
271: the radiation first transmitted to the observer - at finite $\beta$,
272: the rise is even sharper relative to the decay, and the peak can be
273: a true cusp ( Fig.~\ref{fig_3}).
274:
275: %Because we have selected
276: %particularly sharp,
277: %asymmetric subpulses out of a rather varied selection, we have
278: %tailored our assumptions to give the sharpest, most asymmetric
279: %peaks.
280: %We have assumed that the injection of baryons into the
281: %primary radiation beam is instantaneous - or that the optical
282: %thinning of the flow is sudden because of very rapid acceleration.
283: %Neither assumption is perfect, and indeed, the observed rises tend
284: %to be a bit more gradual than our calculated rises. However, we
285: %defer detailed quantitative analysis of the sharp rises to late%r
286: %work. We have also assumed that the polarization of the primary
287: %radiation is perpendicular to the observer's line of sight;
288: %unpolarized primary radiation would give a somewhat rounder peak.
289:
290: The late time scaling behavior of the spectral location of the peak
291: is easily calculated by noting that the initial peak photon
292: frequency in the source frame $\nu_{pi}$ is seen in the scatterer
293: frame as $\nu_{pi}^\prime = \nu_{pi}/\Gamma (1+\beta)$ while the
294: final peak frequency, as seen by the observer, is $\nu_{pf} =
295: \nu_{pi}/\Gamma^2(1+\beta)(1-\beta\cos\theta_V)$. This implies that
296: $\nu_{pf}=\nu_{pi}m_ic^2
297: \frac{d\beta}{dt}\Gamma/[(1-\beta)F\sigma_T]$ where equation
298: (\ref{dbdt}) has been used with
299: $dt/dt_{source} = (1-\beta\cos\theta_V)$.
300: %$^2(1+\beta)$and $1-\beta \ll 1-\cos\theta_V =$,
301: For F constant,
302: and $\beta$ close to unity, the previous result that $(1-\beta)\sim
303: 1/2\Gamma^{2}\propto t^{-2/3} $ implies that
304: %$\sim2\Gamma^2(1-\beta cos\theta)$, and
305: \begin{equation}
306: \nu_{pf} = -\nu_{pi}\frac{dln(1-\beta)}{dlnt}\Gamma/\hat{t}
307: %\nu_p/2\Gamma^2(1-\cos\theta_v)\propto\Gamma^{-2}\propto (1-\beta)
308: \propto t^{-2/3}
309: \end{equation}
310: in good agreement with observations (Ryde 2004).
311:
312: There are several free parameters in the model, including the
313: viewing angle, the initial velocity and radius at which the
314: baryonic cloud is injected into the jet (or at which it becomes
315: optically thin), the rate at which baryons are injected into the jet
316: (e.g. suddenly or gradually) and the inevitable decline of the
317: primary luminosity with time. Nevertheless, we believe that the
318: fast rise, slow decay is a generic feature of both the observations
319: and theoretical predictions. We find that the peculiar shape is
320: insensitive to the viewing angle, the luminosity and radius; these
321: parameters basically rescale the x and/or y axis. The decline of the
322: luminosity with time is expected and could be the reason the
323: theoretical tails are slightly more prolonged than the ones actually
324: observed (e.g. Fig. 3), but, because the cloud moves almost as fast
325: as the photons, the Langrangian derivative of the luminosity is
326: considerably smaller than the Eulerian derivative, so this effect is
327: likely to be small. By the same token, there could be a small
328: component of the primary beam near the viewing angle, which could
329: make the luminosity decline more gradually with time than if the
330: primary beam is entirely within a narrow pencil shape.
331: % Note that it is
332: %conceivable that the cloud is injected, or becomes optically thin,
333: %at a value of $\beta^*$ that exceeds the one that optimizes the
334: %flux scattered into the observer's direction. This could give a
335: %nearly instantaneous rise, if the cloud is injected into the beam
336: %fast enough, followed by a light curve that is already in decline
337: %-- i.e. ($dL/dt$) finitely negative -- just to the right of the
338: %maximum.
339: %The maximum is then nearly vertical on the left
340: %and already in decline - ie. ($dL/dt $) finitely negative -
341: %arbitrarily close on the right.
342: %This results in a cusp-like appearance of the subpulse, which is
343: %observed once or twice per year. An example of such a pulse is GRB
344: %910602, as shown in Fig.~\ref{fig_3}. That such profiles are seen
345: %only about one percent of the time suggests that this is rare.
346:
347: %Other factors could also help diversify the shapes of subpulses,
348: %and the variation from burst to burst. For example, the $t^{-2/3}$
349: %dependence of the softening spectral peak assumes that the radiation
350: %pressure is constant in time. There could be an intrinsic decline in
351: %the primary luminosity as the cloud accelerates, in which case the
352: %spectral softening, if the spectrum remains constant in the source
353: %frame, is slightly shallower relative to the decay of the
354: %luminosity. Correlation between the two time dependences could thus
355: %test the degree of acceleration.
356: % Despite the various factors that influence the shape of the subpulse,
357: % the characteristic rise-decay asymmetry of a FRED-type subpulse
358: %and the reported time dependence of the location of the spectral
359: %peak $\nu_{peak} \propto t^{-2/3}$ are nicely accounted for by the
360: %picture of acceleration of a scattering cloud by radiation
361: %pressure.
362:
363: In the case that the scatterer is optically thin, the model predicts
364: correlation of polarization with intensity (Fig. 2). If the primary
365: radiation is unpolarized, and the cloud is being accelerated
366: through the Lorentz factor $\gamma= 1/\theta_V$, then the peak of
367: the subpulse should correspond to $\Gamma \sim 1/\theta_V$. This
368: corresponds to a $\pi/2$ scattering in the frame of the cloud and
369: should therefore correspond to maximum polarization.
370:
371: It is easy to see that any parcel of energy $dE_{\gamma}$ that
372: emerges as scattered radiation by relativistic baryons is about
373: equal to the kinetic energy $dE_b$ that is imparted to baryons as a
374: result of its scattering: In the instantaneous rest frame of the
375: baryon, in the limit of elastic Thomson scattering, the scattered
376: radiation has on the average zero momentum, $dp_{\gamma}^\prime =
377: 0$, and all its original energy $dE_\gamma^\prime$, while the
378: scatterer has gotten all the momentum, $ dp_b^{\prime} =
379: dE_\gamma^\prime/c$, and essentially none of the energy,
380: $dE_{b}^\prime \sim 0$. In the source frame, the ratio of energy in
381: scattered photons $dE_\gamma = \Gamma (dE_{\gamma}^\prime + \beta c
382: dp_{\gamma}^\prime)$ to the energy they impart to the scatterer
383: $dE_b=\Gamma\beta cdp_b^\prime$ is $dE_{\gamma}/dE_b = \beta^{-1}
384: $, so that in the limit $\beta \sim 1$, the energy that ultimately
385: remains in the scattered radiation is nearly equal to the kinetic
386: energy that it imparted to the scatterer. {\it This provides a basis
387: for why the kinetic energy in the baryons, as inferred from
388: afterglow data, is comparable to that in the prompt emission.}
389: Allowing for the possibility of only partial coverage, then in fact
390: a significant fraction of the primary radiation remains unscattered,
391: in which case the energy in prompt emission should consistently
392: exceed the energy inferred from afterglow data. This is consistent
393: with the results of Eichler \& Jontof-Hutter (2005), which indicate
394: that the prompt emission, corrected for viewing angle, is typically
395: 3 to 15 times larger than the estimated kinetic energy
396: (Lloyd-Romming \& Zhang 2004) inferred from the X-ray afterglow
397: after 10 hours.
398: % For the case described
399: %above of an optically thick cloud thinning into a self organized
400: %photosphere, a more careful treatment of the radiative transfer near
401: %and beyond the photosphere is clearly desirable, but beyond the
402: %scope of this letter. More generally, we should assume that baryons
403: %can be injected at any optical depth (perhaps right at the base of
404: %the flow) and the solid angle coverage, optical depth of the
405: %baryons, and their initial velocity at the injection radius may be
406: %significantly constrained by careful observations and statistics of
407: %the time behavior of subpulses within GRB. These observational
408: %constraints should improve with time, and may serve as useful
409: %diagnostics of the transverse structure of GRB jets and the baryons
410: %within them.
411:
412: \section{Implications for GRB Statistical Correlations}
413:
414: We can
415: imagine three classes of observers: a) those in the direct line of
416: the baryon beam, b) those that are slightly offset from the baryon
417: beam and c) those that are at large offset angles from the baryon
418: beam. For simplicity of discussion, we assume here that the baryon
419: beam is a filled in cone of opening angle $\theta_{\circ}$. The
420: direct line observers see more or less what is scattered/emitted by
421: the baryons, appropriately blue shifted by the factor
422: $\Gamma_{\infty}$. Those slightly offset by angle $\Delta \theta
423: \equiv \theta_V - \theta_{\circ} \ge 0$, nevertheless have a good
424: chance to observe the baryons even after they have reached their
425: terminal Lorentz factor if $\Delta \theta \Gamma_{\infty}$ is not
426: too large. In contrast to observers in the beam, they see the
427: emitted radiation blue shifted by only the lower Doppler factor
428: $D\equiv 1/\Gamma_{\infty}[1-\beta cos(\Delta \theta)]$, and the
429: total fluence they measure that is contributed by any given pencil
430: beam scales as $D^3$. In the case of $\Delta \theta \Gamma_{\infty}
431: \gg 1$, which implies $D\sim 1/\Gamma_{\infty}[1-cos\Delta \theta]$,
432: and $\Delta \theta \ll\theta_{\circ}$, the fraction of the total
433: beam that contributes to the radiation in the observer's direction
434: is proportional to $1-cos\Delta \theta = 1/D$, so that the total
435: fluence is proportional to $1/D^2$. This is essentially the Amati
436: relation (Eichler and Levinson 2004). That the dynamic range of the
437: Amati relation covers two orders of magnitude in frequency suggests
438: that this interpretation invokes a range of $ \Delta \theta
439: \Gamma_{\infty} \le 10$.
440:
441: Finally, at large $\theta_V$, $\Delta \theta \gg 1/\Gamma_{\infty}$, the
442: offset may be too large for the observer to detect any significant
443: contribution from the baryons moving at terminal Lorentz factor.
444: However, if the baryons at some point accelerated through the
445: Lorentz factors $\Gamma \sim 1/\theta_V \ll \Gamma_{\infty}$, then such an observer could
446: nevertheless see a $\gamma$-ray pulse from the baryons as they were accelerating
447: through the Lorentz factor $\Gamma_V \equiv 1/\theta_V$ lasting the acceleration time
448: at $\Gamma = \Gamma_V$. The peak frequency at the peak of the
449: light curve is half the intrinsic peak frequency of the source,
450: independent of $\theta_V$.
451: Scattering by slow ($\Gamma \sim 1/\theta_V \ll 1/\theta_{\circ}$)
452: baryons, in contrast to the viewing angle effect when $\Gamma \Delta
453: \theta \ge 1 $, widens the observable photon beam relative to the
454: baryon beam without significantly altering the observed spectrum. It
455: thus introduces one-sided scatter into the Frail, Amati and
456: Ghirlanda etc. correlations in that it lowers the observed fluence
457: (and ultimately the inferred $E_{iso}$ ) %- relative to what an
458: %observer in the direct path of the primary photons would have
459: %observed -
460: without altering the inferred (via afterglow breaks) opening angle
461: of the jet or observed spectral peak. At a given spectral peak,
462: therefore, there should always be outliers that appear underluminous
463: in the context of these correlations, or overly hard spectra for a
464: given $E_{iso}$. This is consistent with observations (e.g. Butler
465: et al 2007and references therein ).
466:
467:
468: The angular profile of the time integrated scattered radiation has,
469: during the acceleration phase of the scatterer, a "universal"
470: structure and a systematic correlation between fluence and observed
471: duration: For any given source, the amount of energy $E(\theta_V)$
472: that fills a cone of opening angle $\theta_V$ is proportional to
473: $1/sin\theta_V$ when the observer is well outside the cone (or
474: annulus) of primary emission $\theta_{\circ}$. This follows from the
475: fact that the amount of energy scattered by a scatterer at energy
476: $\Gamma mc^2$ is proportional to $\Gamma$, while the observer at
477: viewing angle $\theta_V$ does not see any scattered radiation after
478: the scatterer has accelerated much beyond $\Gamma \sim
479: 1/sin\theta_V$. Because the emission cone's solid angle $ \Omega_V$
480: at the observed peak of the subpulse goes as $1-cos \theta_V$, the
481: observer sees a fluence ${\cal F} = E(\theta_V)/ \Omega_V d_l^2$
482: (where $d_l$ is the luminosity distance) that scales as ${\cal F}
483: \propto 1/sin\theta_V (1-cos\theta_V)\sim 2/\theta_V^3$, and this
484: would induce some scatter in the prompt radiative output as
485: inferred from the Frail anti-correlation between isotropic
486: equivalent energy and apparent opening angle (the latter being
487: inferred from afterglow breaks). Specifically, GRB subpulses
488: observed at very large $\theta_V$ ($\gg1/\Gamma_{\infty})$ would
489: appear less energetic, though their spectra might not be especially
490: atypical.
491: %between fluence and peak frequency. %Moreover, the decrease in
492: %apparent radiative output with $\theta_V$ is only for observers with
493: %large enough $\theta_V$, $\theta_V \gg \theta_{\circ}$, that
494: %the primary jet can be taken as a pencil beam. %If, on the other
495: %hand, the jet and embedded baryons are extended in solid angle,
496: %and the observer is just at the edge, then the scatterer need not
497: %be too much slower than the primary fireball in order to scatter
498: %some of it towards the observer, and the scattering does not
499: %much alter the fluence measured by the observer or the observer's
500: %estimate of the jet energy.
501: % Moreover, the Amati and Ghirlanda relations have not, to our knowledge, been
502: %tested on individual subpulses. The above discussion, therefore,
503: %does not obviously impact the significance of the Frail, Amati, and
504: %Ghirlanda relations other than to provide a reason for residual
505: %scatter and outliers at very large offset viewing angles.}
506: Low energy GRB such as GRB 980425 might thus lie well off the Frail
507: relation because they are observed at large viewing angle. In the
508: case of large viewing angle, the observed duration $\Delta t_{peak}$
509: of the subpulse peak scales as $\Gamma_{peak} \sim 1/sin\theta_V$
510: because, although the observed time lapse is compressed as $\Delta
511: t_{peak}/\Delta t_{peak,source} \sim 1/\Gamma_{peak}^2$, the
512: acceleration time in the source frame goes as
513: $[-dln(1-\beta)/dt_{source}]^{-1} \propto \Gamma_{peak}^3$. The
514: number of photons scattered during this interval, in the limit of
515: constant source luminosity, scales as $\Delta t_{peak}$.
516: %Finite photon energy and
517: %flux thresholds, and declining source luminosity all weaken this
518: %otherwise linear correlation. %This is in good agreement with figure
519: %2 (upper right) in Butler et al (2007), which shows that
520: %$n_{\gamma}$ scales as $T_{90}^{\alpha}$, for $\alpha$ slightly less
521: %than unity.
522: The peak flux $E_{\gamma}(\theta_V)/\Delta t_{peak}\Omega_V d_l^2$
523: is proportional to $1/\Omega_V d_l^2 = 1/2\pi(1-cos\theta_V)
524: d_l^{2}$. Assuming a flux detection threshold independent of $\Delta
525: t$ and that $d_{l,max}^2$ scales roughly as $1/ \Omega_V$ (which
526: neglects non-Euclidian cosmological effects) the maximum volume of
527: detectability $V_{max}\sim d_{l,max}^3 \Omega_V$ scales as $
528: 1/(1-cos\theta_V)^{1/2}$, %Thus, bursts observed at large angle due
529: %to scattering by baryons at an early stage of acceleration have,
530: %through that stage, an output of
531: %scattered radiation%\footnote{Such an
532: %observer would be unlikely to detect the jet break and by "inferred"
533: %we mean that such an observer correctly guesses the solid angle
534: %subtended by the pulse he observes. The guess for a very nearby weak
535: %burst would likely be that $ sin\theta_V \sim 1$ and that the jet
536: %energy is therefore close to the isotropic equivalent energy.}
537: that is only $\sim[(1-cos\theta_{\circ})/(1-cos\theta_V)]^{1/2} $
538: times that of a typical GRB. This suggests that GRB seen at large
539: $\theta_V$ due to scattering by slow baryons should be relatively
540: infrequent, and not terribly contaminate or obscure general trends
541: in GRB statistics. However, a
542: % If there were no other
543: %issues that bear on the relative frequency offrom this viewing
544: %angle, would be detected only $
545: %[(1-cos\theta_{\circ})/(1-cos\theta_V)]^{1/2} $ times as frequently.
546: %They would tend to be of short duration, underluminous, and hard. A
547: more detailed analysis of the effects of scattering by slow baryons
548: on GRB statistics, well beyond the scope of this paper, should
549: include empirical estimates of the relative occurrence of short
550: duration, hard, underluminous GRB, and the fraction of GRB output
551: that can or needs to be attributed to slow baryons as opposed to
552: those already at terminal Lorentz factor. The likelihood that GRBs
553: can have different classes of host stars, and thus different modes
554: of baryon injection into the fireball can also be a complex matter
555: that affects the statistics of GRB parameters.
556: %
557: %For example, a
558: %GRB such as GRB 980425, which has an estimated output of several
559: %times $10^{48}$ ergs (i.e. only $10^{-3}$ of the usual estimate for
560: %a typical GRB) would be observed only once per $10^3$ bursts, which
561: %is consistent with recorded observations.
562: %\propto E_{jet}$, where $E_{jet}$ is the inferred jet energy
563:
564: \section{Summary and Further Discussion}
565:
566: We have proposed that at least part of the baryons in GRB are
567: frequently bunched, either because they are injected in bunches, or
568: because radiatively driven instabilities bunch them. The primary
569: radiation that scatters off them is seen by an offset observer as
570: fast rise, slow decay subpulses, of the sort typically seen in
571: GRB's. The peak frequency as seen by the observer decays roughly as
572: $t^{-2/3}$, in
573: agreement with the data analysis of Ryde (2004). Because this
574: softening is kinematic, the power spectrum of rapid variations
575: originating in the source, if they survive time of flight dispersion
576: due to the finite size of the scattering region, should, even in the
577: limit of zero scatterer size, be softened in the same way as the
578: spectral peak. This could provide a future confirmation of the
579: model. Another possible observational consequence of the model
580: (cleanest if the scatterers are optically thin) is that the
581: polarization should correlate positively with the received flux in
582: the subpulse.
583:
584: In constructing the simplest mathematical model for scattering by
585: primary $\gamma$-rays by slow material, we have assumed that the
586: scatterer is point-like in solid angle, and that the radiation is
587: radially combed. In reality, any given observer may see a
588: superposition of radiation from a finite range of directions,
589: including primary emission beamed directly at him. Moreover, the
590: finite angular spread $\Delta \theta$ of the primary photons, which
591: may not be negligible if and when $\Gamma$ approaches $1/\Delta
592: \theta$, is determined by the collimation profile imposed by the
593: host star. These considerations undoubtedly vary from burst to
594: burst, and can, along with other variable factors, provide the rich
595: diversity of individual light curves and spectra found among GRB's.
596:
597: We might even conjecture that the difference between short and long
598: GRB's lies not so much in the lifetime or energy of the central
599: engine, but rather in the acceleration time of baryons in the path
600: of the fireball and/or the duration of baryon injection into the
601: primary beam. Differences in the apparent acceleration time would
602: likely accompany the differences in the corresponding host stars
603: (e.g. a white dwarf or neutron star merger that collapsed into a
604: black hole would accelerate baryons much closer to the central
605: engine), and could also be due at least in part to differences in
606: the observer's viewing angle. This interpretation of short GRB's
607: would be consistent with the view that they are typically observed
608: from a much larger viewing angle off the jet axis than long GRB's
609: and {\it appear} to have lower energy than long GRB's. It must be
610: kept in mind, however, that this model for short GRB's would be
611: constrained by any millisecond variability observed. It would also
612: be worth looking for "breakout flashes" at large viewing angle,
613: which could occur when a GRB fireball is just breaking through the
614: uppermost layer of its post main sequence host star. The baryons in
615: the way of the GRB fireball just as it is breaking out would be
616: accelerated on a timescale much shorter than the hydrodynamical
617: timescale, so the observed duration would be short. The short term
618: (millisecond) variability, however, would be washed out by the
619: scattering, and this would distinguish it from other short GRB's.
620: Standard short GRB's, by the same token, may simply be breakout
621: flashes from merged white dwarfs or neutron stars.
622:
623: We are grateful to Dr. A. Celloti for calling our attention to the
624: data analysis of Ryde (2004) and for several discussions in that
625: early stage. DE acknowledges the hospitality of the Kavli Institute
626: at the University of California, Santa Barbara, and organizers of
627: its program on Astrophysical Outflows at which these discussions
628: took place. We also thank Dr. Y. Lyubarsky for helpful discussions.
629: We are grateful to Drs. S. Patel and U. Griv for technical
630: assistance in downloading and plotting the data. We acknowledge
631: support from the US-Israel Binational Science Foundation and the
632: Israeli Science Foundation's Center of Excellence Program.
633:
634:
635: \begin{thebibliography}{}
636: \bibitem {}Amati, L., et al., 2002, A \& A, 390, 81
637: \bibitem {}Butler, N.R., Kocevski, D. Bloom, J.S. and Curtis J.L
638: astro-ph 07061275
639: \bibitem {}Eichler, D. 1994, ApJ Supp., 90, 877
640: \bibitem {}Eichler, D. 2004, ApJ, 608, L81
641: \bibitem {}Eichler, D. \& Granot, J., 2006, ApJ, 641, L5
642: \bibitem {}Eichler, D. \& Jontof-Hutter, D. 2005, ApJ, 635, 1182
643: \bibitem {}Eichler, D., \& Levinson, A., 1999, ApJ, 521, L117
644: \bibitem {}Eichler, D. \& Levinson A., 2000, ApJ, 529, 146
645: \bibitem {}Eichler, D. \& Levinson, A., 2004, ApJ. 614, L13
646: \bibitem {}Eichler, D. \& Levinson, A., 2006, ApJ, 649, L5
647: \bibitem {}Ghirlanda, G., Ghisellini, G. \& Lazzati, D. 2004, ApJ, 616, 331
648: \bibitem {} Landau,L.D. and Lifshitz, E.M. {\it The Classical Theory of Fields,} (1962) Pergamon Press, p. 127
649: \bibitem {}Levinson, A. \& Eichler, D., 2003, ApJ, 594, L19
650: \bibitem {}Levinson, A. \& Eichler, D., 2005, ApJ, 629, L13
651: \bibitem {}Lloyd-Ronning, N., and Zhang., B, 2004 ApJ, 613, 477
652: \bibitem {}Lyutikov, M. \& Blandford, R., 2003, HEAD, 35, 622
653: \bibitem {}Meszaros \& Rees 1994 \emph{Mon Not Astron Soc}, 269, L41
654: \bibitem {} Pe'er, A., Ryde, F., Wijers, R. A. M. J., Meszaros, P., \& Rees, M. J., 2007, astro-ph 0703734
655: \bibitem {}Rees, M.J. \& Meszaros, P. 2005, ApJ, 628, 847
656: \bibitem {} Rybicki, G.B. and Lightman, A.P. {\it Radiative Processes in
657: Astrophysics} (1979) John Wiley and Sons
658: \bibitem {}Ryde, F. 2004, ApJ, 611, L41
659: \bibitem {}Thompson, C., 1994, MNRAS, {\bf 270}, 480
660: \end{thebibliography}{}
661:
662:
663: \begin{figure}
664: \epsscale{.90} \plotone{f1.eps} \caption{The luminosity and time,
665: both as measured by an observer viewing the GRB at an offset angle
666: $\theta_V = 10$ degrees, are displayed as functions of $\beta$ of
667: the scattering plasma. It is assumed that the plasma is optically
668: thin. The normalization of L and r, the distance of the plasma from
669: the central engine, are arbitrary. }\label{fig_1}
670: \end{figure}
671:
672:
673: \begin{figure}
674: \epsscale{.90} \plotone{f2.eps} \caption{GRB 911104 fitted by L(t)
675: as prescribed in Figure 1. The curves labeled $L_{sc}$ and
676: $L_{\perp}$ show the respective contributions of the polarizations
677: in and perpendicular to the scattering plane under the assumption of
678: an unpolarized primary pencil beam. The count rate assumes a
679: spectrum of dN/dE$\propto E^{-1}$, from an assumed detector
680: threshold energy $E_{th} = 30 KeV$ to a maximum energy in the
681: source frame $E_{max}$ of 2 MeV. \label{fig_2}}
682: \end{figure}
683:
684:
685: \begin{figure}
686: \epsscale{.90}
687: \plotone{f3.eps}
688: \caption{GRB 910602 fitted by L(t) as prescribed in Figure 1, but
689: with the assumption that the scattering material is injected with
690: $\beta = \beta^*$. The part of the curve to the left of the observed
691: rise is therefore not physical.} \label{fig_3}
692: \end{figure}
693:
694: \end{document}
695: