1: %\documentclaLin_Yi-Fungss[]{article}

2: %\documentstyle[12pt,aasms4]{article}

3: \documentstyle[11pt,aaspp4]{article}

4: %\documentstyle[emulateapj]{article}

5: %\usepackage{emulateapj}

6: \newcommand\beq{\begin{equation}}

7: \newcommand\eeq{\end{equation}}

8: \def\mic{{\,\mu{\rm m}}}

9: 

10: \begin{document}

11: \title{$\gamma$-ray burst afterglows as probes of galactic and

12: intergalactic dust}

13: \author{Rosalba Perna\altaffilmark{1} and Anthony Aguirre}

14: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden 

15: Street, Cambridge, MA 02138; rperna, aaguirre@cfa.harvard.edu}

16: \altaffiltext{1}{Harvard Society of Fellows}

17: %\submitted{Accepted by The Astrophysical Journal Letters} 

18: 

19: \begin{abstract}

20: 

21: The amount and properties of high-redshift galactic and intergalactic

22: (IG) dust are largely unknown, but could be investigated using

23: multi-wavelength photometry of high-$z$ objects that have a known

24: intrinsic spectrum.  Observations of $\gamma$-ray burst (GRB)

25: afterglows appear to support the theoretical model of an adiabatic

26: blast wave expanding into an external medium.  In this model, the

27: synchrotron peak flux is independent of frequency, providing a flat

28: spectrum when observed over time, and therefore allowing

29: straightforward measurement of the relative attenuation of afterglow

30: flux in widely separated bands.  Applying this method to dust

31: extinction, we show that for a sample of afterglows which have been

32: corrected by galactic extinction, comparison between the number counts

33: of peak fluxes in $X$-ray versus optical can provide constraints on an

34: intergalactic component of dust.  A similar technique can probe the

35: redshift-dependence of extinction in GRB-forming regions without

36: requiring an assumed relation between extinction and reddening by the

37: dust.  Probing systematic changes in extinction with redshift --

38: particularly in IG and/or non-reddening dust -- is crucial to a proper

39: interpretation of the Type Ia Supernova Hubble diagram and similar

40: observations, and useful in understanding GRB progenitor environments.

41:        

42: \end{abstract}

43: 

44: \noindent {\em Subject headings:} cosmology: observations --

45: dust, extinction -- $\gamma$-rays: bursts

46: 

47: \section{Introduction}

48: 

49: There is now substantial evidence that $\gamma$-ray bursts (GRBs)

50: originate at cosmological distances (e.g. Metzger et al. 1997) from

51: very powerful explosions (e.g. Kulkarni et al. 1998, 1999).  They have

52: been detected at very high redshifts, making them useful probes of the

53: universe out to early cosmological epochs.  In addition, the

54: $\gamma$-ray emission is followed by delayed emission at longer

55: wavelengths, from the $X$-ray to the radio band (Costa et al. 1997;

56: van Paradijs et al. 1997; Frail et al. 1997).  This afterglow is

57: described reasonably well as synchrotron radiation, emitted when a

58: relativistic shell collides with an external medium (Paczy\'nski \&

59: Rhoads 1993; Katz 1994; Waxman 1997a,b; Wijers, Rees \& M\'esz\'aros

60: 1997; Sari, Piran \& Narayan 1998). Afterglows, like their high energy

61: counterpart, are also expected to be detected out to very high

62: redshift. The absorption-line systems and the Ly$\alpha$ forest

63: visible in their spectra can therefore be used to trace the evolution

64: of metallicity in the universe, and to constrain, or possibly measure,

65: the epoch at which re-ionization of the universe occurred (e.g.  Lamb

66: \& Reichart 2000).

67: 

68: In this {\em Letter} we point out other uses that multi-wavelength

69: observations of afterglows might have in cosmology. Whereas it is by

70: now clear that GRBs (and consequently their afterglows) are far from

71: being standard candles, the afterglows do have a very interesting

72: property: the flux at the peak of the synchrotron spectrum is

73: independent of time for adiabatic hydrodynamic evolution (Katz 1994;

74: M\'esz\'aros \& Rees 1997; Sari, Piran \& Narayan 1998), and the

75: adiabatic shock model has received robust support from observations

76: (Waxman 1997a,b).  Because the frequency of the peak flux smoothly

77: decreases with time, this provides a strong theoretical connection

78: between observations at various widely-separated frequencies which is

79: lacking in other high-$z$ objects such as quasars or galaxies.

80: 

81: This property makes GRB afterglows well suited to absorption studies

82: of both the GRB immediate environment and host galaxy, and of the

83: intervening intergalactic medium (IGM).  One such possibility is the

84: study of intergalactic dust, which may play a role in observations of

85: high-$z$ Supernovae (SNe).  In the past few years, observations of

86: Type Ia SNe by two separate groups have revealed a progressive dimming

87: of SNe at high redshift with respect to the predictions of a

88: matter-dominated universe or even an open universe with a zero

89: cosmological constant (Riess et al. 1998; Perlmutter et al. 1998).

90: This dimming has been interpreted as evidence for acceleration in the

91: cosmic expansion, probably caused by a positive cosmological constant.

92: Two important systematic effects which must be accounted for in

93: drawing this conclusion are evolution in the supernova intrinsic

94: brightness and dust obscuration; both groups have discussed these at

95: length. However, their dust corrections, based on reddening, rely on

96: the assumption that dust has everywhere the same characteristics as in

97: the Milky Way, which might not be the case.  As Aguirre (1999) has

98: shown, a scenario in which galaxies expel a significant fraction of

99: their metals in winds or by radiation pressure ejection of dust, and

100: in which very small grains are selectively destroyed or retained by

101: galaxies, can self-consistently provide a viable candidate for an

102: intergalactic (IG) dust component which is more 'grey' (i.e. less

103: reddening) than dust in the Milky Way. Dust reddening properties also

104: vary significantly within and between galaxies.

105:  

106: GRB afterglows are observed in widely-separated bands at which even

107: grey dust will attenuate radiation very differently, and, as explained

108: above, a known relation exist between the expected fluxes in those

109: bands in the absence of extinction. This {\em Letter} shows that,

110: given a sample of GRB afterglows for which a reliable correction for

111: the host galaxy absorption has been made, a comparison between the

112: number count distribution of peak fluxes in various bands can

113: sensitively test the presence of a cosmological distribution of grey

114: dust. Whereas this test is insensitive to 

115: the choice of model parameters made, such as type of cosmology,

116: shape of the GRB luminosity function, redshift evolution of the GRB rate,

117: etc. (because these parameters influence in the {\em same} way

118: the number counts in all bands, and we are {\em

119: comparing} data in different bands), it does

120:  rely on the assumption of the constancy of the

121: peak flux in the various bands. In reality, several effects 

122: can contribute to a departure from the simple, ideal behavior

123: (see e.g. Meszaros, Rees \& Wijers 1998), and we discuss how

124: and to what extent these can be corrected for, so that our test

125: would still remain possible.

126: 

127: Besides probing the existence of IG dust, we show that the same type

128: of multi-wavelength study of GRB afterglows can provide a useful probe

129: of the evolution of metallicity (or dust-to-gas ratio) in the

130: GRB-forming regions of galaxies.  Independent probes of metallicity

131: evolution (i.e. not related to GRB sites) can then yield information

132: on the specific environments of GRBs in relation to galaxies, helping

133: to elucidate the nature GRB progenitors. Conversely, assuming a GRB

134: progenitor type, one could obtain information about dust in the

135: environment of that type of object, at very high $z$.

136: 

137: \section{Number counts of GRB afterglow peak fluxes}

138: 

139: To begin with, we adopt the simplest unbeamed synchrotron model

140: (Waxman 1997a,b).  Under the assumption that the magnetic field energy

141: density in the shell rest frame is a fraction $\xi_B$ of the

142: equipartition value, and that the power-law electrons carry a fraction

143: $\xi_e$ of the dissipated energy, the observed frequency at which the

144: synchrotron spectral intensity of the electrons peaks is

145: \begin{equation}

146: \nu_m(t)=2.4\times 10^{16}\left(\frac{1+z}{2}\right)^{1/2}

147: \left({\xi_e\over 0.2}\right)^2 

148: \left({\xi_B\over 0.1}\right)^{1/2} 

149: E_{52}^{1/2}t_{\rm hr}^{-3/2}\;{\rm Hz}\;,

150: \label{eq:num}

151: \end{equation}

152: where $z$ is the cosmological redshift of the source. For $\Omega=1$ and

153: $H_0=70~{\rm km~s^{-1}~Mpc^{-1}}$, the observed intensity at $\nu_m$ is

154: $

155: F_{\nu_m}\sim 1\;{\rm mJy}\;n_1^{1/2}\left({\xi_B\over 0.1}\right)^{1/2}

156: E_{52}

157: $

158: for a burst at a redshift $z\sim 1$.

159: Due to the combination of several effects, the afterglow flux falls off with

160: redshift very slowly (Lamb \& Reichart 2000; Ciardi \& Loeb 2000). For

161: an Einstein-De Sitter cosmology, Ciardi \& Loeb find 

162: $

163: F_{\nu_m}\propto (1+z)^{(p-3)/4}[1-(1+z)^{-1/2}]^{-2}\;,

164: $

165: where $p$ is the power-law index of the electron energy

166: distribution. Both the GRB and the afterglow observations are well

167: fitted by the value $p\approx 2.5$ (Sari, Narayan \& Piran 1998; Kumar

168: \& Piran 1999).  We fix the constant of proportionality so that

169: $F_{\nu_m}= 1$ mJy at $z= 1$.  In absence of absorption, the peak

170: luminosity of a given burst is then given in terms of the luminosity

171: distance $D_L(z)$ by $L_{\nu_m} = 4\pi D_{\rm L}^2(z)F_{\nu_m}/(1+z)$

172: (see e.g. Hogg 1999 and references therein).  At each redshift, we

173: allow for a scatter in the burst luminosity by assuming a probability

174: distribution that is log-normal in $L_{\nu_m}$.

175: %\begin{equation}

176: %P(L_{\nu_m},z)dL_{\nu_m} = \frac{1}{\sqrt{2\pi\sigma^2}}

177: %\exp\left\{-\frac{[\ln(L_{\nu_m})-\ln(L_\star)]^2}{2\sigma^2}\right\}

178: %{dL_{\nu_m}\over L_{\nu_m}}\;.

179: %\label{eq:pl}

180: %$\end{equation}

181: The mean of this distribution

182: %, $\langle

183: %L_{\nu_m}(z)\rangle=L_\star\exp\left(\sigma^2/2\right)$, 

184: is chosen so that

185: $\langle F_{\nu_m}(z)\rangle$ evolves according to Ciardi \& Loeb's relation.

186: We choose $\sigma=0.5$ in our computations. However, as explained in the

187: following, our results are not very sensitive to the particular choice of parameters.

188: 

189: The number (per unit time)

190: of GRB afterglows with peak flux above a given threshold $S$ is given by

191: \beq

192: N(>S)=\int_0^{z_{\rm max}} dz \,R(z)\,{dV_c\over dz}

193: \int_{L_{\rm min}(z,S)}^\infty dL_{\nu_m}

194: P(L_{\nu_m},z)\;.

195: \label{eq:nc}

196: \eeq

197: Here $R(z)$ is the GRB formation rate, which we assume to be proportional to the

198: star formation rate (SFR), as given by the most recent observations of Steidel 

199: et al. (1999), and $z_{\rm max}$ is the redshift limit of the sample.

200: The comoving volume is given by $dV_c=4(c/H_0)^3(1+z-\sqrt{1+z})^2(1+z)^{-7/2} 

201: d\Omega dz$ (e.g. Hogg 1999) in our adopted cosmology. 

202: The minimum luminosity that a burst at redshift $z$ must have in order

203: to be detected with flux $>S$ is given by 

204: \beq

205: L_{\rm min}(z,S) = \frac{4\pi D_{\rm L}^2(z) S}{(1+z)\exp\{-[\tau_{\rm dust}(z)

206: +\tau_{\rm abs}(z)]\}}\;.

207: \label{eq:lmin}

208: \eeq

209: Here $\tau_{\rm abs}$ is the optical depth to photoelectric absorption, for which

210: there is a contribution from the host galaxy itself and a contribution from the 

211: intergalactic medium (IGM). $X$-rays are not affected by photoabsorption, 

212: if observed at high enough frequency (see below). Optical light is

213: unabsorbed if observed below the Ly$\alpha$ resonance

214: frequency $\nu_\alpha(z)=2.47\times 10^{15}/(1+z)$~Hz, so the $V$-band 

215: ($\nu=5.4\times 10^{14}$ Hz) and longer wavelengths will not be affected by

216: photoabsorption if one only considers bursts with redshifts up to $z\sim 3.5$.

217: We will show our results by assuming a sample with $z_{\rm max}$=3. 

218: 

219: The optical depth of dust at redshift $z$ is indicated by $\tau_{\rm dust}$

220: in Equation~(\ref{eq:lmin}). We assume

221: $\tau_{\rm dust}(z)=\tau_{\rm gal}(z)+\tau_{\rm grey}(z)\;,$

222: and consider several possibilities. 

223: 

224: For the grey dust component, we take the model as in Aguirre (1999):

225: dust is composed of equal masses of graphite and silicate spherical

226: grains of radii $a$, with a grain-size distribution $dN(a)/da \propto

227: a^{-3.5}$, $0.1\mic \le a \le 0.25\mic$.  This dust has a very flat

228: extinction curve blueward of the $V$ band.  We take a dust

229: density $\Omega_{\rm dust}(z=0) = 4.5\times 10^{-5}$, and assume that

230: $\Omega_{\rm dust}(z) \propto \int dt SFR$, where the $SFR$ is the

231: same as that assumed for the GRB rate.  Using publicly available

232: extinction data (see Laor \& Draine 1993) these assumptions give a

233: cosmologically `interesting' extinction of $\sim 0.2$ mag to $z=0.5$

234: in the observed $B$-band.

235: 

236: For the host galaxy absorption we

237: assume the extinction law to be the same as in our Galaxy, and model, for

238: an observed wavelength $\lambda$, 

239: \beq

240: A(\lambda,z) = k(z)\xi\left(\frac{\lambda}{1+z}\right)

241: \left(\frac{N_{\rm H}(z)}{10^{21}{\rm cm}^{-2}}

242: \right)\;,

243: \label{eq:ext}

244: \eeq with $\xi(\lambda)$ fitted as in Pei (1992). We then consider two

245: types of evolutionary scenarios for the dust-to-gas ratio $k(z)$;

246: model (1): k(z) = const = k(0), with k(0) = 0.78 as in the Galaxy;

247: model (2): k(z) traces the integrated star-formation rate. These

248: should represent the most extreme possibly evolutionary paths for the

249: dust-to-gas ratio: the second effectively assumes instantaneous mixing

250: of metals throughout all cosmic gas, whereas the first effectively

251: assumes either that cosmic metallicity does not evolve, or

252: alternatively that metals are so poorly mixed that while cosmic

253: metallicity evolves, the metallicity near the GRBs does not.  Model

254: (1) could better represent a scenario in which GRBs are associated

255: with coalescent compact objects (e.g. Eichler et al. 1989, Narayan,

256: Paczy\'nski \& Piran 1992), and therefore they would not be expected

257: to occur in particularly dense or dusty regions. On the other hand, if

258: GRBs are associated with the death of massive stars (e.g. Woosley

259: 1993, Paczy\'nski 1998), then they are likely to occur in star-forming

260: regions.  The mean metallicity expected for such regions is

261: substantially higher than the cosmic mean.

262: This is a situation which could be more

263: closely described by our dust model (2)

264: \footnote{Note that, if GRBs are associated with star-formation

265: regions, then also the density in their surrounding medium is expected

266: to be on average higher. However, extinction is only sensitive to the

267: product of column density and metallicity, and therefore we 

268: parameterize our models with only one of them, that is the dust-to-gas ratio.}.

269: 

270: In Equation~(\ref{eq:ext}), $N_{\rm H}(z)$ is the average column

271: density intercepted by the GRB photons as they escape a host galaxy at

272: a redshift $z$.  To estimate this function, we have performed a Monte

273: Carlo calculation using the model of Ciardi \& Loeb (2000) to obtain

274: the probability distribution $N_H(M,z)$ for GRBs occurring (with

275: spatial distribution proportional to the gas density squared) in the

276: disks of spirals embedded in halos of mass $M$, combined with the

277: Press-Schechter formalism for the distribution of $dN(M,z)/dM$.  The

278: resulting distributions are normal in $log(N_H)$. As $z$ runs from 0

279: to 4, the distribution's center increases roughly linearly from 21.27

280: to 21.48, and its width decreases linearly from 0.915 to 0.73.\footnote{For

281: (GRB density) $\propto$ (gas density), the mean values are similar but

282: about 0.25 dex smaller.}  This should be regarded not as an accurate

283: and detailed model of the expected column densities (for example it

284: does not include the evolution of the galaxies' gas fraction), but

285: rather as a rough check on the possible systematic increase of GRB

286: host-galaxy extinction. Our two models, based on this estimate,

287: bracket the more detailed models of extinction evolution of Calzetti

288: \& Heckman (1999) and Totani \& Kobayashi (1999) which apply to the

289: normal stellar population in galaxy disks.

290: 

291: \section{The effects of dust on GRB number counts}

292: 

293: We present results for number count distributions with an observed

294: peak flux in $X$-ray, V, and J bands. Observations in the {\em hard}

295: $X$-ray band are necessary for a proper comparison which is not biased

296: by photoelectric absorption within the host galaxy. This should not be

297: a problem for a 5 KeV band at which {\em Swift} can observe:

298: Ghisellini et al. (1999) find that, at this energy, a burst at $z=0$

299: in a medium with solar metallicity will be unaffected by absorption

300: for column densities $N_{\rm H}\la 10^{23}$ cm$^{-2}$. For a burst at

301: $z\sim 2$ in a solar metallicity environment, the constraint becomes

302: $N_{\rm H}\la 10^{24}$ cm$^{-2}$. Evidence for an unusually high

303: column density to a burst can be found by the time-dependence of the

304: UV flux (Perna \& Loeb 1998), $K_\alpha$ fluorescent line emission

305: (Ghisellini et al. 1999), and, clearly, heavy reddening of the optical

306: emission. For a proper analysis such bursts should not be included in

307: the sample.

308: 

309: To probe the presence and type of dust extinction, the observed

310: optical-near IR bands seem the most appropriate. As explained in \S 2,

311: the $V$ band and longer wavelengths are unaffected by photoelectric

312: absorption in the IGM up to $z \sim 3.5$.  Therefore, if the 

313: sample is limited to redshifts lower than this, the on-board {\em

314: Swift} $V$ photometry could be used to determine the peak $V$ flux

315: (although ground-based NIR data may be necessary for obtaining the

316: {\em rest-frame} UBV photometry necessary for accurate de-reddening).

317: 

318: Figure 1 shows a comparison between number counts in $X$-ray, V, and J

319: bands for a sample of GRBs for which perfect corrections for Galactic

320: and GRB host-galaxy extinction have been made, while the grey

321: component (which would not be accounted for by UV/optical

322: de-reddening) has not been subtracted out. Figure 2 shows the same

323: curves, where host galaxy extinction -- quantified by models (1) and

324: (2) of \S 2 -- has {\em not} been corrected. In these figures, the

325: horizontal offset of two curves indicates the amount of extinction,

326: whereas the vertical {\em change} in that offset shows the evolution

327: of extinction with increasing redshift.

328: 

329: In a realistic case, one could hope to correct for host galaxy

330: extinction only imperfectly: there will be errors in the photometry

331: and $K$-correction of the rest-frame UVB fluxes, errors in determining

332: the appropriate extinction law ($R_V$, see Cardelli, Clayton \& Mathis

333: 1989), and variations in the extinction law even for a given

334: $R_V$. The extinction may also be time-dependent (Waxman \& Draine,

335: 1999).  However, we find that even if the correction leaves a

336: `residual', the statistical difference is still readily

337: measurable. We performed a Monte Carlo simulation of data

338: drawn from the $X$-ray distribution and of data drawn from the

339: $V$-band distribution corrected for galactic extinction but not for a

340: grey intergalactic component (as in Figure 1).  To each flux was assigned a

341: fractional error that was randomly generated from a Gaussian

342: distribution of width $\sigma_{\rm err}$. We took $\sigma_{\rm

343: err}=0.5$, which corresponds to about half to 1 magnitude error in

344: extinction correction, larger than the quoted errors on $A_v$ for GRBs

345: observed so far (see e.g. Bloom et al. 1998, Vreeswik et al. 1999).

346: By performing a Kolmogorov-Smirnov test (see e.g. Press et al. 1992)

347: with the two sets of simulated data, we found that with $N\ga 70$

348: bursts the two distributions can be distinguished with a confidence

349: level $\sigma\ga 3$. 

350: 

351: Note that, if the host galaxy extinction is decreasing with $z$ (as in

352: our model (1), or as in Calzetti \& Heckman 1999 and Totani \&

353: Kobayashi 1999), then galactic extinction would never mimic the

354: effects of IG grey dust. On the other hand, if the likely scenario is

355: as in model (2), then our ability to distinguish IG dust from

356: increasing host galaxy extinction would depend on the accuracy with

357: which host galaxy dust correction can be made.

358: 

359: A very clear distinction can also be made between the distributions

360: described by model (1) and (2) in Figure 2. A random simulation as

361: described above showed that with $N\ga 20$ bursts the two $V$-band

362: distributions in the case where there is no grey dust component can be

363: distinguished with a confidence level $\sigma\ga 3$. Similar results

364: hold for the corresponding distributions which also include grey dust.

365: In these simulations we have assumed the same $\sigma_{\rm err}$ as

366: above. However, when no correction for galactic extinction is

367: required, this error is likely to be much smaller.  This study is

368: therefore very sensitive to the conditions (i.e. metallicity and

369: density) in the environment of GRBs. Combining such analysis with

370: independent probes of the metallicity and density evolution of the

371: medium in galaxies and in the IGM\footnote{Spectra of the GRBs

372: themselves could also be used to such purposes, as shown by Lamb \&

373: Reichart (2000).}, significant information can therefore be obtained

374: about the location of GRBs within galaxies.  This is particularly

375: relevant in order to distinguish scenarios in which GRBs are due to

376: coalescence of two compact objects with respect to scenarios in which

377: GRBs are associated with the death of massive stars.  Since reddening

378: and extinction are measured rather independently, the technique can

379: also give information about the reddening properties of dust in GRB

380: progenitor environments.

381: 

382: We need to emphasize once more that our method is rather insensitive

383: to the choice of model parameters made, such as type of cosmology,

384: shape of the GRB luminosity function, GRB evolution with redshift,

385: etc.  This is because these parameters influence in the {\em same} way

386: the number counts in all bands, and our method is based on a {\em

387: comparison} of data in different bands.  What can change (due for

388: example to a different distribution of the bursts with the redshift)

389: are statistical quantities such as the number of bursts required to

390: distinguish between two models with a given confidence level, but we

391: found the difference in this number required for various models to be

392: rather marginal.  The main uncertainty of this method lies in the

393: fact, already discussed in \S 1, that the perfect, adiabatic model

394: with a constant peak flux is a clear approximation that does not take

395: into account several effects which might cause a departure from this

396: simple behavior. Meszaros et al.  found that a number of factors

397: (such as angular anisotropy of the fireball, properties of the

398: environment, later re-energization of the afterglow, etc.)  can lead to

399: either an increase or to a decrease of the peak flux with time,

400: depending on what the particular conditions (such as for example

401: density gradients in the external medium) are for a given burst. This

402: means that, most likely, such disturbances should not give rise to

403: systematic effects, but rather would lead to variations equivalent to

404: having a larger scatter in the errors on the fluxes.  

405: If this is indeed the case, the effects produced by grey dust should

406: not be easily washed out: a random simulation of data with errors in

407: peak fluxes drawn from a distribution with $\sigma_{\rm err}$=2

408: (i.e. a fractional error of 200\%) showed that the distributions in

409: Figure 1 can still be distinguished with a confidence level

410: $\sigma\ga3$ if the number of afterglows is $\ga 200$. Clearly, the

411: situation would become more difficult if these disturbing effects,

412: instead of being stochastic, had a trend which would systematically

413: reduce the emission in one (or more) bands with respect to the

414: others. However, unless these hypothetical systematics have also an

415: evolution with redshift, it should be possible to correct for them by

416: measuring ratios between peak fluxes in various bands for a sample of

417: bursts at low $z$ (which would only be marginally affected by IG

418: dust). A systematic decrease in the flux at peak frequency could also

419: mimic the difference between a dusty and a relatively dust-free 

420: GRB environment, but again only at one redshift; {\em evolution} in the

421: GRB environment should still be measurable. 

422: Such a study, to be statistically significant, would need a

423: large number of observed afterglows. This should be possible with {\em

424: Swift}, which will observe $\sim 300$ bursts per year, and will

425: monitor their emission from the $X$-ray to the V band (and other

426: instruments will very likely obtain NIR photometry of the

427: afterglows). It will be an ideal tool for such a study.

428: 

429: \section{Conclusions}

430: 

431: One of the outstanding problems in observational cosmology is to

432: understand the type and amount of dust existing at high-$z$, whether

433: outside galaxies or within them. Traditional methods of estimating

434: extinction require a known dust reddening curve and an accurate model

435: for the intrinsic UV/optical spectrum of the observed object, both of which may

436: be rather uncertain; very few existing techniques allow the measurement of

437: absolute extinction.

438: 

439: In this {\em Letter}, we have proposed GRB afterglows as probes of

440: dust in cosmology. Like QSOs and galaxies they can be observed to very

441: high $z$.  In addition, there is a well-based theoretical spectrum

442: which is flat between $X$-ray and near infra-red frequencies. This

443: allows very straightforward estimates of dust extinction {\em

444: independent of its reddening properties} (though the reddening would

445: be observable and useful) in GRB environments or in the intergalactic

446: medium; the former can help help determine the properties of GRB

447: progenitor environments (including the dust properties), while the

448: latter is crucial for careful interpretation of the Type Ia

449: Supernova Hubble diagram.

450: 

451: While relying on the assumption that afterglow evolution is well

452: described by the simplest adiabatic blast model, with a large number

453: of afterglows (such as {\em Swift} will provide), the technique should

454: be applicable even if deviations from the theory occur.  

455: 

456: 

457: \begin{references}

458: 

459: \reference{} Aguirre, A. 1999, ApJ, 525, 583

460: 

461: \reference{} Bloom, J. S. et al. 1998, ApJ, 508, L21

462: 

463: \reference{} Calzetti, D. \& Heckman, T. M. 1999, ApJ, 519, 27

464: 

465: \reference{} Cardelli, J. A., Clayton, G. C., \& Mathis, J. S. 1989, ApJ, 345, 245

466: 

467: \reference{} Ciardi, B. \& Loeb, A. astro-ph/0002412

468: 

469: \reference{} Costa, E. et al. 1997, Nature, 387, 783

470: 

471: \reference{} Eichler, D., Livio, M., Piran, T. \& Schramm, D. N. 

472: 1989, Nature, 340, 126

473: 

474: \reference{} Frail, D. A. et al. 1997, ApJ, 483L, 91

475: 

476: \reference{} Freedman, D. L. \& Waxman, E. 1999, astro-ph/9912214

477: 

478: \reference{} Ghisellini, G., Haardt, F., Campana, S., Lazzati, D.,

479: \& Covino, S. 1999, ApJ, 517, 168

480: 

481: \reference{} Hogg, D. 1999, astro-ph/9905116

482: 

483: \reference{} Katz, J. I. 1994, ApJ, 422, 248

484: 

485: \reference{} Kulkarni, S. R. et al. 1998, Nature, 393, 35 

486: 

487: \reference{} Kulkarni, S. R. et al. 1999, Nature, 398, 389 

488: 

489: \reference{} Kumar, P. \& Piran, T. 1999, astro-ph/9909014

490: 

491: \reference{} Lamb, D. Q. \& Reichart, D. E. 2000, ApJ in press, astro-ph/9909002

492: 

493: \reference{} Laor, A.  \& Draine, B. 1993, ApJ, 402, 441

494: 

495: \reference{} M\'esz\'aros, P., \& Rees, M. J. 1997, ApJ, 476, 232 

496: 

497: \reference{} Metzger, R. et al. 1997, Nature, 387, 878

498: 

499: \reference{} Narayan, R., Paczy\'nski, B. \& Piran, T. 1992, ApJ, 395, L83  

500: 

501: \reference{} Paczy\'nski, B. 1998, ApJ, 494, L45

502: 

503: \reference{} Paczy\'nski, B., \& Rhoads, J. E. 1993, ApJ, 418, L5 

504: 

505: \reference{} Pei, Y. 1992, ApJ, 395, 130

506: 

507: \reference{} Perlmutter, S., et al. 1998, Nature, 391, 51

508: 

509: \reference{} Perna, R. \& Loeb, A. 1998, ApJ, 501, 467 

510: 

511: \reference{} Press, W. H., Teukolski, S. A., Vetterling, W. T., \&

512: Flannery, B. P. 1992, ``Numerical Recipes'' , Cambridge Univ. Press

513: 

514: \reference{} Riess, A. G., et al. 1998, AJ, 116, 1009

515: 

516: \reference{} Sari, R., Piran, T., \& Narayan, R. 1998, ApJ, 497, L17

517: 

518: \reference{} Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickinson, M., 

519: Pettini, M. 1999, ApJ, 519, 1

520: 

521: \reference{} Totani, T. \& Kobayashi, C. 1999, ApJ, 526, L65

522: 

523: %\reference{} Valageas, P. \& Silk, J. 1999, A\&A, 347, 1

524: 

525: \reference{} van Paradijs, J., et al. 1997, Nature, 386, 686

526: 

527: \reference{} Vreeswijk, P. M. et al. 1999, astro-ph/9904286

528: 

529: \reference{} Waxman, E. 1997a, ApJ, 485, L5

530: 

531: \reference{} Waxman, E. 1997b, ApJ, 489, L33

532:  

533: \reference{} Waxman, E. \& Draine, B. astro-ph/9909020

534: 

535: \reference{} Wijers, R. A. M. J., Rees, M. J., \& M\'esz\'aros, P. 1997, MNRAS, 288, L51

536: 

537: \reference{} Woosley, S. E. 1993, ApJ, 405, 273

538: 

539: \end{references}

540: 

541: \newpage

542: 

543: \begin{figure}[t]

544: \centerline{\epsfysize=5.7in\epsffile{fig1.ps}}

545: \caption{Cumulative number count distribution of GRB afterglow peak fluxes

546: in the $X$, J, and V bands 

547: for a sample corrected by galactic extinction but not by the extinction 

548: due to an intergalactic grey dust component.}

549: \label{fig:1}

550: \end{figure}

551: 

552: \begin{figure}[t]

553: \centerline{\epsfysize=5.7in\epsffile{fig2.ps}}

554: \caption{Cumulative number count distribution of GRB afterglow peak fluxes 

555: in the $X$, and V bands

556: (upper panel) and the $X$ and J band (lower panel). In both panels, the dust models (1)

557: and (2) refer to the host galaxy dust models

558:  described in the text. For each model, the upper curve

559: has only the component from galactic extinction, while the lower curve includes 

560: also an intergalactic grey dust component.}

561: \label{fig:2}

562: \end{figure}

563: 

564: 

565: \end{document}

566: 

567: 

568: