1: \documentclass[a4paper,preprint]{aastex}

2: %\documentstyle[emulateapj,times,psfig]{article}

3: \def\ni{\noindent}

4: \def\spose#1{\hbox to 0pt{#1\hss}}

5: \def\lta{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

6:      \raise 2.0pt\hbox{$\mathchar"13C$}}}

7: \def\gta{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

8:      \raise 2.0pt\hbox{$\mathchar"13E$}}}

9: \def\bp{\bar{\rm p}}

10: 

11: \renewcommand{\thefootnote}{\fnsymbol{footnote}}

12: 

13: \begin{document}

14: 

15: 

16: 

17: \title{THE ENERGY OF LONG DURATION GRBS}

18: 

19: \author{T. Piran\footnote{tsvi@phys.huji.ac.il}}

20: \affil{Racah Institute, The Hebrew University, Jerusalem 91904,

21: Israel}

22: %\authoremail{tsvi@phys.huji.ac.il}

23: \author{P. Kumar\footnote{pk@ias.edu}}

24: \affil{IAS, Princeton, NJ 08540, USA}

25: %\authoremail{pk@ias.edu}

26: \author{A. Panaitescu\footnote{alin@astro.princeton.edu}}

27: \affil{Dept. of Astrophysical Sciences, Princeton University, NJ

28: 08544, USA}

29: %\authoremail{alin@astro.princeton.edu}

30: \author{L. Piro\footnote{piro@ias.rm.cnr.it}}

31: \affil{Istituto Atrofisica Spaziale, CNR, Via Fosso del Cavaliere,

32: Rome 00133 Italy}

33: %\authoremail{piro@ias.rm.cnr.it}

34: 

35: 

36: 

37: 

38: \begin{abstract}

39: The energy release in gamma-ray bursts is one of the most useful

40: clues on

41:  the nature of their ``inner engines". We show that, within the framework

42:  of the relativistic external shocks afterglow model, the narrowness of the observed

43:  X-ray luminosity of GRB afterglows, implies that the energy of the GRB jets

44:  after the early afterglow phase spans less than one order of magnitude.

45:  This result is not affected by uncertainties in the electrons energy,

46:  magnetic field strength, and external medium density. We argue that the

47:  afterglow kinetic energy is within a factor of two of the initial energy

48:  in the relativistic ejecta, therefore the energy output of the central

49:  engine of long duration GRB has an universal value.

50: \end{abstract}

51: 

52: Subject headings: gamma-rays: bursts - ISM: jets and outflows -

53: radiation mechanisms: non-thermal - shock waves

54: 

55: %\maketitle

56: %\section{introduction}

57: 

58: %\begin{multicol}

59: In the last three years we have learned a great deal about long

60: duration gamma-ray bursts (GRBs). The Italian-Dutch satellite

61: BeppoSAX provided angular position of several dozen long bursts

62: to within about 3 arc-minutes which enabled follow up

63: observations  in the x-ray (see Piro, 2000), optical, milli-meter

64: and radio frequencies which has provided a wealth of information

65: on these explosions. These observations are described well by the

66: relativistic fireball model (see e.g. Piran, 1999). According to

67: this model the energy from the central source is deposited in

68: material that moves with speed very close to the speed of light.

69: The kinetic energy of this material is converted to the observed

70: electromagnetic radiation as a result of collisions between fast

71: moving material that catches up with slower moving ejecta, and

72: the shock heated circum-burst medium. The nature of the ``inner

73: engines" that expels relativistic material, which is responsible

74: for the the GRBs, is not determined yet. The energy of the

75: relativistic matter ejected by the ``inner engine", $E_{rel}$, is

76: one of the most important clues on its nature. Our goal is to

77: find a reliable estimate of $E_{rel}$.

78: 

79: Given an observed $\gamma$-ray fluence and the redshift to a

80: burst one can easily estimate the energy emitted in $\gamma$-rays,

81: $E_{\gamma,iso}$  assuming that the emission is isotropic.

82: $E_{\gamma,iso}$ can also be estimated from the BATSE catalogue

83: by fitting the flux distribution to theoretical models (Cohen \&

84: Piran, 1995; Schmidt, 2001). As afterglow observations proceeded,

85: alarmingly large values (Kulkarni et al. 1999) ($3.4 \times

86: 10^{54}$ergs for GRB990123) were measured for $E_{\gamma,iso}$.

87: However, it turned out (Rhoads, 1999; Sari Piran \& Halpern,

88: 1999) that GRBs are likely beamed and $E_{\gamma,iso}$ would not

89: then be a good estimate for the total energy emitted in

90: $\gamma$-rays. We define instead: $E_\gamma =

91: (\theta^2/2)E_{\gamma,iso}$. Here $\theta$ is the effective angle

92: of $\gamma$-ray emission, which can be estimated from $t_{b}$,

93: the time of the break that appears latter in the afterglow light

94: curve (Rhoads, 1999; Sari Piran \& Halpern, 1999): $\theta =0.12

95: (n/E_{51})^{1/8} t_{b,days}^{3/8}$,  where $E_{51}$ is the

96: isotropic-equivalent energy kinetic energy during the adiabatic

97: fireball phase, discussed below, in units of $10^{51}$ergs, and

98: $t_{b,days}$ is the break time in days. Recently Frail et al.

99: (2001, hereafter F1) estimated $E_\gamma$ for 18 bursts, finding

100: typical values around $10^{51}$ergs. While $E_\gamma$ is closer to

101: $E_{rel}$ it is still not a good estimate.  First, we have to

102: take an unknown conversion efficiency of energy to $\gamma$-rays

103: into consideration: $E_{rel} = \epsilon^{-1} E_\gamma

104: =\epsilon^{-1} (\theta^2/2) E_{\gamma,iso}$. Second, the large

105: Lorentz factor during the $\gamma$-ray emission phase, makes the

106: observed $E_\gamma$ rather sensitive to angular inhomogeneities

107: of the relativistic ejecta (Kumar \& Piran, 2000).

108: 

109: We consider here another quantity: $E_{K,ad}$, the kinetic energy

110: of the ejecta during the adiabatic afterglow

111: phase\footnote{\baselineskip 9pt The external relativistic shock

112: becomes adiabatic about 1/2 an hour after the explosion, and

113: furthermore the loss of energy during the earlier radiative phase

114: is typically not large.}. Clearly: $E_{rel} \ge

115: \overline{E_\gamma} + E_{K,ad}= \epsilon E_{rel} + E_{K,ad}$,

116: where $\overline{E_\gamma}$, is the angular average of

117: $E_\gamma$. The inequality arises from possible energy losses

118: during the early afterglow radiative phase. However observations

119: of long time tails of GRBs suggest that, unless this energy is

120: radiated at extremely high energy channel, this losses are not

121: large (Burenin et al, 1999, Giblin et al, 1999; Tkachenko et al.,

122: 2000) . Therefore, with $\epsilon \approx 10\%$ (Kobayashi, Piran

123: \& Sari, 1997) ($\epsilon$ cannot be too close to unity otherwise

124: there won't be afterglow) we expect that $E_{K,ad} \approx

125: E_{rel}$ to within better than a factor of 2. Hereafter we drop

126: the subscript $kin$ denoting $E=E_{K,ad}$.

127: 

128: The purpose of this paper is to determine the spread of $E$ using

129: the x-ray afterglow flux. The advantage of the method presented

130: here is that it is independent of the uncertain density of the

131: ISM, and in fact all other parameters, except for the observed

132: width of the distribution of the jet opening angle.

133: 

134: One way of determining how $E$ is distributed is to figure out the

135: energy for individual bursts by modeling their afterglow emission

136: over a wide range of frequency and time. This procedure, carried

137: out for 8 well studied bursts (Panaitescu \& Kumar 2001,

138: hereafter PK01), gives the mean energy to be $\sim

139: 3\times10^{50}$ergs and the standard deviation of the log of

140: energy distribution to be about 0.3. However, the detailed

141: modeling of individual GRB afterglow emission is cumbersome and

142: time consuming, and hard to carry out for a large sample of

143: bursts especially considering that we need data in radio, optical

144: and x-ray bands with good time coverage for this kind of an

145: analysis to be useful.

146: 

147: Moreover, this procedure is not necessary if we simply want to

148: know the width of energy distribution. For this we can use the

149: x-ray afterglow flux at a fixed time after the explosion. The

150: width of the distribution of this flux, an easily measurable

151: quantity, yields the width of the distribution for the energy

152: release in the explosion. This method is described below. The

153: observation of the x-ray flux should be carried out at a

154: sufficiently late time such that the angular variations and

155: fluctuations across the surface of the ejecta have been smoothed

156: out. This occurs several hours after the explosion when the bulk

157: Lorentz factor of the ejecta has decreased to about 10 at which

158: time we see a good fraction of the relativistic ejecta.

159: Conveniently, this is also when the observed x-ray in the 2-10

160: keV band is above the cooling frequency in which case the

161: observed flux is independent of the density of the medium in the

162: vicinity of the explosion (Kumar 2000). Furthermore, it is best to

163: carry out observations while $\Gamma > \theta^{-1}$, that is

164: before the jet begins its sideways expansion, which makes the

165: interpretation of the observed flux much simpler. In five cases

166: of GRB afterglow light curves where we see the effect of the

167: finite opening angle of explosion as steeping of the light curve,

168: we find the effect manifests itself at least one day after the

169: explosion. Thus, the above two requirements suggest it is best to

170: consider the x-ray afterglow flux between several hours and a day

171: after the explosion.

172: 

173: The x-ray afterglow fluxes from GRBs have a power law dependence

174: on $\nu$ and on the observed time $t$ (Piro, 2000): $f_\nu(t)

175: \propto \nu^{-\beta} t^{-\alpha}$ with $\alpha \sim 1.4$ and

176: $\beta \sim 0.9$. The observed x-ray flux per unit frequency,

177: $f_x$, is related, therefore, to, $L_x$, the isotropic luminosity

178: of the source at redshift, z by:

179: \begin{equation}

180: L_x(t) = {4 \pi d_L^2 \over (1+z)} f_x(t)(1+z)^{\beta-\alpha}

181: \equiv f_{x}(t) Z(z) \ ,

182: \end{equation}

183: where $Z(z)$ is a weakly varying function of $z$. For bursts with

184: $0.5<z<4$ and with $\beta-\alpha \approx -0.5$ we find $\sigma_Z

185: \approx 0.31$ (for a cosmology with $\Omega_m=0.3$ and

186: $\Omega_\Lambda=0.7$). Here and thereafter we denote by

187: $\sigma_X$ the standard deviation of the $\log(X)$, unless noted

188: otherwise.

189: 

190: Assuming that the x-ray luminosity does not evolve with redshift

191: we can relate the dispersion of $\log(L_x)$ at a fixed observer

192: time after the explosion, $\sigma_{L_x}$, with the observed

193: dispersion $\sigma_{f_{x}}$: $\sigma^2_{L_x}=\sigma^2_{f_{x}} +

194: \sigma^2_Z\approx \sigma^2_{f_{x}} \ .$ Using 21 BeppoSAX bursts

195: (Piro, 2000) we find $\sigma_{f_{x}}\approx 0.43 \pm 0.1$ (see the

196: caption for Figs. 1 and 2 for the details of the observations and

197: the analysis), and therefore, $\sigma_{L_{x}} \approx 0.43$ to

198: within 25\%. This result is supported by 10 x-ray light-curves of

199: GRBs with known red-shifts, and $\alpha$ and $\beta$.

200: 

201: The x-ray flux, in an energy band above the cooling frequency at a

202: fixed time after the burst, depends on the energy per unit solid

203: angle in the explosion (provided that $\Gamma> \theta^{-1}$ at the

204: time of observation), and on the fractional energy taken up by

205: electrons, $\epsilon_e$. The flux does not depend on the density

206: of the surrounding medium, $n$, or its stratification or the

207: fractional energy in the magnetic field, $\epsilon_B$ (Kumar,

208: 2000). The flux has a weak dependence on $n$ when the electron

209: cooling is dominated by the inverse Compton scattering

210: (Panaitescu \& Kumar 2000).

211: 

212: Under these, rather general, conditions the standard synchrotron

213: fireball model implies that the isotropic equivalent flux at

214: frequency $\nu$ above the cooling frequency, at a fixed elapsed

215: time since the explosion, is given by (Kumar 2000; Freedman \&

216: Waxman, 2001):

217: \begin{equation}

218: L_x = \eta_p \left[{d E\over d\Omega}\right]^{(p+2)/4}

219:    \epsilon_e^{p-1}\epsilon_B^{(p-2)/4},

220:    \label{LX}

221: \end{equation}

222: where $dE/d\Omega$ is the energy per unit solid angle, and

223: $\eta_p$ is a constant\footnote{\baselineskip=10pt It is worth

224: emphasizing that equation (2) is independent of the details of

225: the jet structure i.e., the variation of Lorentz factor across

226: the jet, since at about 1/2 a day after the explosion the jet

227: Lorentz factor has dropped to $\sim10$ and we see a good fraction

228: of the entire jet surface.}. Assuming that there is no correlation

229: between the microscopic variables, $\epsilon_e$, $\epsilon_B$,

230: $p$ and $dE /d\Omega$ we obtain from the above equation that

231: $\sigma_{dE/d\Omega}<\sigma_{L_x}$. Using $\sigma_{L_x}\approx

232: 0.43\pm 0.1$ for the 21 BeppoSAX bursts we find that

233: $\sigma_{dE /d\Omega}\leq 0.43\pm 0.1$.

234: 

235:  From $\sigma_{dE /d\Omega}$ we can now obtain $\sigma_E$ that

236: characterizes the distribution of the kinetic energy provided we

237: 

238: know $\sigma_\theta$ using the trivial relation: $\sigma^2_{dE

239: /d\Omega} = \sigma^2_{E}+4 \sigma^2_\theta$. Panaitescu \& Kumar

240: (PK01) and Frail et al. (F01) have estimated the jet opening

241: angles from the observed (or lack of) breaks in the light curves

242: of optical afterglow light curves. For 8 GRBs from the PK01

243: sample we have: $\sigma_\theta \approx 0.31\pm 0.06$ while a

244: sample of 10 bursts from F01 yields $\sigma_\theta \approx 0.28\pm

245: 0.05$. If these values are representative for the whole GRB

246: population we find a marginally viable solution within two

247: $\sigma$ errors of $\sigma_E < 0.2$ (for the PK01  result) and

248: $\sigma_E < 0.27$ (for the F01 data); to get a viable solution we

249: had to take both the values of $\sigma_{L_x}$ one standard

250: deviation above the mean and the value of $\sigma_\theta$ one SD

251: below the mean. This result suggests that there is a narrow

252: energy distribution; the FWHM of $E$ being less than a factor of

253: 5. If $E$ and $\theta$ are correlated the above relation is

254: modified i.e. $\sigma^2_{dE/d\Omega} = \sigma^2_{E}+

255: 4\sigma^2_\theta -4\zeta\sigma_E\sigma_\theta$. Both the PK01 and

256: the F01 data  show that this correlation ($\zeta$) if non-zero is

257: weak, less than 0.35. With such a correlation the allowed  energy

258: distribution could be somewhat (but not significantly) broader.

259: 

260: A stronger constrain on $\sigma_E$ can be obtained using

261: $\sigma_{\epsilon_e}=0.3$ for 8 GRBs analyzed by PK01. It follows

262: from Eq. \ref{LX} that a non-zero value for $\sigma_{\epsilon_e}$

263: makes the distribution for $dE/d\Omega$ and hence $E$ even

264: narrower, however to quantify this effect we need a larger data

265: set.

266: 

267: We have argued before that $E=E_{K,ad}$, discussed here, is a

268: rather good estimate to $E_{rel}$ the total energy emitted by the

269: ``inner engine". The constancy of $E_{K,ad}$ is another indication

270: for it being a good measure of $E_{rel}$. The constancy of

271: $E_{K,ad}$ is also an indication that the assumptions that have

272: lead to Eq. \ref{LX} are justified. Otherwise it would have been

273: remarkable if starting from different levels of initial energy

274: and having different amounts of energy losses the final kinetic

275: energy of the afterglow would converge to a constant value. At

276: present there is no way to tell whether significant amounts of

277: additional energy is released in other forms, e.g. non

278: relativistic particles or neutrinos.  Similar arguments suggest

279: further that this total energy emitted by the ``central engine"

280: does not vary significantly and that it is rather close to the

281: energy estimated here.

282: 

283: The distribution of relativistic kinetic energy during the

284: afterglow phase is narrow, with full width at half maximum less

285: than one decade. These results suggest that the wide distribution

286: of directly and indirectly determined $E_{\gamma,iso}$ results

287: from the distribution of beaming angles, from a variation of

288: $dE/d\Omega$ across the jet, and from a variable efficiency in

289: conversion of kinetic energy to $\gamma$-rays. The fact that GRB

290: engines are "standard" engines in terms of their energy output

291: provide a very severe constraint on the nature of these enigmatic

292: explosions. For instance, in the collapsar model for GRBs the

293: central engine is composed of a black hole (BH) and an accretion

294: disk around it (Woosley 1993; Paczynski, 1998; MacFadyen \&

295: Woosley, 1999). This model has two energy reservoirs which can be

296: tapped to launch a relativistic jet: the BH rotation energy and

297: the gravitaional energy of the disk. Our result of nearly

298: constant energy in GRBs implies that the mass accretion on to the

299: BH plus the possible conversion of rotational energy of the BH to

300: kinetic energy of the jet does not vary much from one burst to

301: another inspite of the fact that both the disk mass and the BH

302: spin are expected to vary widely in the collapse of massive stars.

303: 

304: BeppoSAX is a mission of the Italian Space Agency (ASI) with

305: participation of the Dutch space agency (NIVR). TP acknowledges

306: support by the US-Israel BSF.

307: 

308: 

309: 

310: \begin{references}

311: 

312: 

313: \reference{burenin} Burenin, R.A. et al., 1999, A \& A

314: Supplement, 138, 443

315: \reference{cohen} Cohen, E. and Piran, T., 1995, ApJ 444, L25

316: \reference{frail} Frail, D.A. et al. 2001, astro-ph/0102282, (F01)

317: \reference{freedman} Freedman, D.L., and Waxman, E.,

318: 2001, ApJ 547, 922

319: \reference{giblin} Giblin, T.W., et al. 1999,

320: ApJ 524, L47

321: \reference{kobayashi}Kobayashi, S., Piran, T., and

322: Sari, R., 1997, ApJ 490, 92

323: \reference{kulkarni} Kulkarni et al.

324: 1999, Nature 398, 389

325: \reference{kumarA} Kumar, P., 2000, ApJ

326: 538, L125

327: \reference{kumarB} Kumar, P., and Piran, T., 2000, ApJ

328: 535, 152

329: \reference{macfadyen}MacFadyen, A. I. \& Woosley, S. E.,

330: 1999, ApJ, 524,262

331: \reference{bohdan} Paczynski, B., 1998, ApJ,494L, 45

332: \reference{panaitescuA} Panaitescu, A., and Kumar, P., 2000, ApJ,

333: 543, 66

334: \reference{panaitescuB} Panaitescu, A., and Kumar, P., 2001, ApJ,

335: 560, in press (PK01)

336: \reference{piran} Piran, T. 1999, Physics Reports, 314, 575.

337: \reference{piro}Piro, L., 2000, proc of X-Ray Astronomy 99:

338: Stellar End points, AGN and the Diffuse X-ray background,

339: Sept.6-10 1999, Bologna, N. White ed., in press

340: (astro-ph/0001436)

341: \reference{rhoads} Rhoads, J.E., 1999, ApJ

342: 525, 737

343: \reference{sari} Sari, R., Piran, T., and Halpern, J.P.,

344: 1999, ApJ 519, L17

345: \reference{schmidt} Schmidt, M., 2001, ApJ

346: 552, 36

347: reference{tkachenko} Tkachenko, A., et al. 2000,

348: Astronomy \& Astrophysics, 358, L41

349: \reference{woosley} Woosley,

350: S. E. 1993, ApJ, 405, 273

351: 

352: 

353: 

354: \end{references}

355: 

356: 

357: \clearpage

358: 

359: 

360: \begin{figure}

361: %\centerline{\psfig{figure=f1.eps}}

362: \includegraphics{f1.eps}

363: \caption[]{\footnotesize{The distribution of X-ray fluxes (2-10

364: keV) at t=11 hours after the GRB in  21 afterglows observed by

365: BeppoSAX. The sample includes all the fast observations performed

366: by BeppoSAX on GRB from January 1997 to October 1999. Data are

367: from Piro (2001, proc. of X-ray astronomy 99, and references

368: therein) Stratta et al. (2001, in preparation), Nicastro et al.

369: 1999 (IAUC7213), Feroci et al. (2001 A\&A, in press), Montanari

370: et al. (2001, Proc. of the 2nd wokshop on GRB in the afterglow

371: era, Rome, 17-20 Oct. 200, E. Costa, F. Frontera J. Hjort eds.,

372: in press), Piro et al. (1999, GCN 409) and \'T Zand et al. (2000,

373: ApJ, 545, 266). No X-ray afterglow was detected in GB990217 (Piro

374: et al. 1999, IAUC 7111) to the limiting instrumental sensitivity

375: of $10^{-13}$ erg cm$^{-2}$ s$^{-1}$, 6 hours after the burst. In

376: the case of GB970111 (Feroci et al. 1998 A\&A 332,L29) a

377: candidate was detected, but evidence of fading behaviour is

378: marginal, so we have considered both cases as upper limits

379: (indicated by the arrow in fig).}}

380: \end{figure}

381: \begin{figure}

382: %\centerline{\psfig{figure=f2.eps}}

383: \includegraphics{f2.eps} \caption[]{\footnotesize{Likelihood contour lines

384: (corresponding to 99\%, 90\% and 69\%  confidence levels) in the

385: $\overline{\log(f_x)}$, $\sigma_{f_x}$ plane for the X-ray flux

386: distribution as inferred from 21 GRBs detected by BeppoSAX from

387: January 1997 to October 1999. We determine $\overline{\log(f_x)}$

388: and $\sigma_{f_x}$ by minimizing the likelihood function

389: $S=-\sum_i \ln\left\{\left[2 \pi(\sigma_i^2 +

390: \sigma^2_{f_x})\right]^{-1/2} \exp\left[ - (\log f_{x_i} -

391: \overline{\log f_x})^2/2 (\sigma_i^2 +

392: \sigma^2_{f_x})\right]\right\}$; where $\log f_{x_i}$ and

393: $\sigma_i$ are the observed x-ray flux 11 hr after the onset of

394: the i-th GRB and the associated measurement error respectively.

395: The maximal likelihood is at $\overline{\log(f_x)}=-12.2{\pm0.2}$

396: and $\sigma_{f_x} =0.43^{+.12}_{-.11}$. Two upper limits of

397: $10^{-13}$ ergs cm$^{-2}$ s$^{-1}$ in the 2--10 kev band at 11

398: hours after the bursts are included in this data set. The value

399: of $\sigma_{f_x}$ is 0.42 if we exclude these upper limits.  We

400: have checked a posteriori with a Kolgomorov-Smirnov test that the

401: distribution is consistent with a gaussian (at 90\% confidence

402: level). We also note that the predicted number of X-ray

403: afterglows with a flux lower than about $2 \times 10^{-13}$ is

404: 3.5, consistent with the observed number of objects.}}

405: \end{figure}

406: 

407: 

408: \end{document}

409: