Swift and Fermi observations of the early afterglow of the short Gamma-Ray Burst 090510
M. De Pasquale, P. Schady, N.P.M. Kuin, M.J. Page, P.A. Curran, S. Zane, S.R. Oates, S.T. Holland, A. A. Breeveld, E.A. Hoversten, G. Chincarini, D. Grupe, Fermi/LAT, Fermi/GBM collaborations. Minor changes in the authorlist
aa r X i v : . [ a s t r o - ph . H E ] O c t Draft version November 1, 2018
Preprint typeset using L A TEX style emulateapj v. 04/20/08
SWIFT
AND
FERMI
OBSERVATIONS OF THE EARLY AFTERGLOW OF THE SHORT GAMMA-RAYBURST 090510
M. De Pasquale ∗ , P. Schady , N.P.M. Kuin , M.J. Page , P.A. Curran , S. Zane , S.R. Oates , S.T. Holland ,A.A. Breeveld , E. A. Hoversten , G. Chincarini , D. Grupe , Fermi /LAT and
Fermi /GBM Collaborations Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, RH5 6NT, UK NASA/Goddard Space Flight Center, Greenbelt, MD 20771, USA Department of Astronomy & Astrophysics, Pennsylvania State University, 525 Davey Lab, University Park, PA 16802, USA Universita degli Studi di Milano Bicocca. Piazza della Scienza, 3, 20126 Milano, Italy Osservatorio Astronomico di Brera (INAF). Via E. Bianchi 46, 23807 Merate (LC), Italy
Draft version November 1, 2018
ABSTRACTWe present the observations of GRB090510 performed by the
Fermi
Gamma-Ray Space Telescopeand the
Swift observatory. This is a bright, short burst that shows an extended emission detectedin the GeV range. Furthermore, its optical emission initially rises, a feature so far observed only inlong bursts, while the X-ray flux shows an initial shallow decrease, followed by a steeper decay. Thisexceptional behavior enables us to investigate the physical properties of the GRB outflow, poorlyknown in short bursts. We discuss internal shock and external shock models for the broadband energyemission of this object.
Subject headings: gamma rays: bursts INTRODUCTION
With the availability of a relatively large sam-ple of Gamma-Ray Bursts (GRBs), we came torecognize that they comprise of two large classes(Kouveliotou et al. 1993): the so-called short-hardGRBs (duration . & Swift (Gehrels et al. 2004) in the optical and the X-ray rangeas soon as ∼
100 s after the burst. In this paper, wepresent the study of the short GRB090510 with
Swift and
Fermi in a broad energy range, which extends fromthe optical up to a few GeV.We report our observations and analysis in §
2; in § § X =10 n X n for cgs units and F ∝ t − α ν − β , where F is energyflux, t is the time from the trigger of Swift
Burst AlertTelescope (BAT; Barthelmy et al. 2005a) and ν the fre-quency. Errors are reported at 1 σ , unless otherwise spec-ified. We assume a cosmology in which H = 71 km s − Mpc − , Ω m = 0 .
27, Ω λ = 0 .
73. All the fluxes, times andfrequencies are measured in the observer’s frame. OBSERVATIONS AND DATA ANALYSIS.
BAT data ∗ Corresponding authors: M. De Pasquale: [email protected];M. J. Page: [email protected]; K. Toma: [email protected];V. Pelassa: [email protected].
At 00:23:00 UT, May 10th, 2009, BAT, which oper-ates in the 15-350 keV range, triggered on GRB090510(Hoversten et al. 2009).
Swift slewed immediately to theburst. The duration was T = 0 . ± .
07 s. Detailedanalysis of the BAT data is shown in Ukwatta et al.(2009).
XRT data
The
Swift
X-ray Telescope(XRT; 0.3-10 keV; Burrows et al. 2005) began toobserve the X-ray afterglow of GRB090510 at T+98 s.The lightcurve (Fig. 1) shows an initial slow flux decline.Observations were interrupted when the source enteredthe Earth constraint at T+1.9 ks. When they resumed,at T+5.1 ks, the flux was much lower. A broken pow-erlaw fit of the lightcurve gives as best fit parametersan early decay slope α X, = 0 . ± .
03, break time t X = 1 . +0 . − . ks, late decay slope α X, = 2 . ± . χ = 112 with 77 degrees of fredom (d.o.f.), which is stillmarginally acceptable (chance probability P = 0 . UVOT and other optical data
The
Swift
Ultra Violet and Optical Tele-scope (UVOT; 160-800 nm; Roming et al. 2005;Poole et al. 2008) began settled exposures at T+97 s.The lightcurve of the optical afterglow, produced byrenormalizing all individual filters to white, as describedin Oates et al. (2009), is shown in Fig. 1. The op-tical emission rises until ∼ . χ = 23.9 with 19d.o.f.) by a broken powerlaw (Beuermann et al. 1999)with a smooth break. The best fit parametersare: α Opt, = − . +0 . − . ; t peak = 1 . +0 . − . ks; α Opt, = 1 . +0 . − . . Adding a constant does notimprove the fit significantly, suggesting a small hostgalaxy contribution. Very Large Telescope observations(Rau et al. 2009) provide a spectroscopic redshift of De Pasquale et al. z = 0 . Fermi spectral pa-rameters, the isotropic equivalent energy of GRB090510is E iso = 1 . × ergs in the 10 keV - 30 GeV restframe (Abdo et al. 2009). Fermi data
GRB090510 triggered both instruments on-board the
Fermi observatory (Guiriec et al. 2009;Ohno, & Pelassa 2009). The Gamma-ray Burst Moni-tor (GBM; 8 keV - 40 MeV) observed the burst duringthe prompt emision phase, and after an autonomous re-pointing, the Large Area Telescope (LAT; 20 MeV - morethan 300 GeV) began observations and detected a long-lasting (up to 200s) high-energy (up to 4 GeV) emission.The analysis and interpretation of the prompt emissionis presented in Abdo et al. (2009). Follow up observa-tions lasted until 1500 s when the source was occulted bythe Earth, and resumed ∼ β γ = 1 . ± .
1. The lightcurve shows nosignificant feature and is best fitted by a powerlaw. Theonset of the GBM emission (T+0.013 s) is a sensiblereference for this temporal fit and yields a decay index α γ = 1 . ± .
07 ( χ /dof = 9.4/7) (Fig. 1). For thebinned spectral analysis described in § DISCUSSION
GRB090510 was a short burst with a relatively brightafterglow, and E iso is among the highest for this class(Graham et al. 2009). The early rise of the optical fluxis so far unique in short GRBs. More importantly,GRB090510 shows high energy emission up to the GeVrange, until T+200 s.Energetic short GRBs with optical transients, such asGRB050724 (Barthelmy et al. 2005b), typically have ex-tended emission (EE) detected by BAT and XRT, fol-lowing the hard emission spike (Troja et al. 2008). IfGRB090510 had occurred at z = 0 .
26, as GRB050724, itwould have produced a flux in the BAT range of a few10 − ergs cm − s − until ∼
100 s, and we would haveclassified it as an EE-GRB.The nature of this high energy emission is neverthelessnot easy to understand. EE often fades slowly for a fewhundreds of seconds, then vanishes with a slope whichcan be as fast as α ∼
7; after this sudden drop a late af-terglow with a typical decay slope α ∼ . Swift observations seems due to the usual FSmechanism, such as in GRB051221 (Burrows et al. 2006)and GRB 061201 (Stratta et al. 2007).We propose and discuss separately 2 scenarios to ex-plain the emission after the initial spike: in the first one,the emission is due to both external FS and internalshock, (IS, Rees & M´esz´aros 1994) while in the secondthe emission is due to FS alone.
X-ray internal shock, optical external shock
In the first scenario we assume that the initial X-ray (until the break at ∼ . γ -ray flux areIS emission, while the FS is responsible for the opti-cal lightcurve and the late X-ray flux. In particular,the optical rise may be due to the onset of FS emis-sion, detected 10 − s after the trigger in long GRBs(Oates et al. 2009). This model can explain the differentbehavior of the early X-ray/LAT and optical lightcurves.We constrain some physical properties of the FSblastwave. The initial Lorentz factor of the ejecta isΓ = 1 . × E / n − / t − / peak, (Sari 1997), where E is the isotropic kinetic energy, n the environment den-sity in cm − , and t peak is the peak time. The max-imum FS flux is at the synchrotron characteristic fre-quency ν m , and it is F ν m = 1 . × E ǫ / B, − n / µ Jy(Granot & Sari 2002), where ǫ B is the fraction of inter-nal energy in the magnetic field.These parameter values must be consistent with Γ & σ lower limit on t peak >
730 s and a peak flux F ≃ µ Jy. The first constraint can be written E n − > . × . If ν m is just below the optical band, the con-straint on the flux becomes E ǫ / B, − n / ≃ . × − .Assuming ǫ B, − ≃
1, the model is consistent with obser-vations for E & .
4, while n ≃ . × − E − .The XRT and LAT fluxes can be explained by ISsynchrotron emission with ν c < ν Opt < ν sa < ν X <ν m , where ν c is the synchrotron cooling frequencyand ν sa is the synchrotron self-absorption frequency(Guetta & Granot 2003). The synchrotron luminosity,estimated from the 100s spectral energy distribution(SED, Fig. 2), is L ≃ erg s − . We find that forthis value of L and for p = 2 . ǫ e, − = 5 . ǫ B, − = 33,Γ = 410, t v = 3 × − s, where p , ǫ e , Γ and t v are the index of the powerlaw electron energy distri-bution, the fraction of energy given to electrons, thebulk Lorentz factor and the variability timescale respec-tively, we have ν m ≃
210 keV, F(1.7 keV) ≃ µ Jy,which is within a factor ∼ ≃ . × − µ Jy, which is consistent within2 σ of the data. The cut-off energy for pair productionis hν γγ ≃ . ∼ ν sa ≃ .
32 keV.Compared with the scenario presented in § to explain an early (few seconds) FS emissiononset in a low density environment expected for a shortburst. However, it needs some fine tuning of parame-ters. The optical rise slope α Opt, = − . α = − ν m werecrossing the optical band and, in general, broad FS on-set rises are also expected for outflows observed off-axis(Panaitescu & Verstrand 2008). However, a bright andhard event such as GRB090510 is difficult to reconcilewith the latter scenario, which predicts soft and dimprompt emission (Yamazaki et al. 2002). Optical, X-ray and GeV emission from externalshock.
A second possibility is that the afterglow ofGRB090510, including the emission detected by LAT,is entirely produced by the FS propagating in a constantdensity medium (Sari et al. 1998). According to themodel, the broad afterglow spectrum consists of threesegments: a low-energy tail, of spectral slope β = − / ν m < ν < ν c , where β = ( p − / ν c , the third segment has β = p/
2. Forcomparison with this spectral template, we built up 5SEDs, at 100 s, 150 s, 1 ks, 7 ks, and 12 ks (Fig. 3),all including UVOT and XRT data, and LAT data werealso included in the first SED. LAT data were accumu-lated between 10 s and 200 s (i.e. well after the endof the prompt emission seen in the GBM), and renor-malized to 100 s using the decay index of α γ = 1 . β = − / β = β + 1 / χ /d.o.f.=110.3/83) and is shown in Fig. 3 and Table 2.The FS alone could successfully describe the spectrumover 9 decades of frequency. A break between X-ray and γ -ray ranges is fitted, at E b ≃
300 MeV, but not con-strained. It is studied more precisely by fitting the 100 sSED alone, freezing N H and E ( B − V ) at the 5 SEDs fitresults and leaving β and β free to vary (see Table 2).This fit fulfills the relation β = β + 1 / σ ) anda significant break is found (3.6 σ ), although it yields aslightly harder β (1.8 σ ) than the 5 SED fit shown inFig. 3. The LAT emission shows no spectral evolution,even at early times. Therefore, to better characterize thehigh-energy spectrum, the SED at 100 s was rebuilt in-cluding LAT data between 0.38 s and 200 s (see Table 2and Fig. 2). A significant break ( > . σ including sys-tematics) between 10 and 133 MeV was found. However,including this selection of LAT data in the 5 SED fityields a worse fit ( χ /dof = 125.3/83) than that shownin Fig. 3.In this FS interpretation, the initial increase of the op-tical emission is due to ν m approaching the optical band.The X-ray is already decaying because it lies above ν m .In order to verify whether the required physical parame-ters are plausible, we impose the following constraints: i) F (1 ks) ≃ . µ Jy at 10 Hz and ii) ν m (1 ks) ≃ Hz.Adopting the expressions for F ν m , ν m , and ν c fromGranot & Sari (2002), the constraints above lead to The self-absorption frequency is not relevant in this study. the following equations: ǫ B, − ≃ E − / ǫ − e, − ξ − p ν os , n ≃ . × − (1330) p − . E − ǫ e, − ξ p ν − p − os , where ξ p = 3( p − / ( p −
1) and ν os = ν m (1 ks) / Hz. Weverified that the synchrotron self-Compton cooling is notsignificant for t ≤ . ν c (1 . > Hz fora reasonable range of parameters (Nakar et al. 2009). Avery low, but not implausible, density is suggested.The flux at 100MeV is F ν>ν c ≃ . × − (2 . × − ) ( p − . E ǫ e, − ( t/ s ) (2 − p ) / ( hν/ − p/ ξ p ν ( p − / os µ Jy(1)This is consistent with the LAT data at 100s, provided E ǫ e, − ≃ p ≈ . ξ p ≈
1, and ν os ≈
1. For theseparameters, ν c ≪ t & F ∝ t (2 − p ) / ∼ t − . at t & > E / n − / − is required for the FSonset time to be . α = 3 β / . ± .
06, clearly inconsistent with theobserved α X, = 0 . ± .
03. Secondly, if the X-ray breakat t = 1 . α X, = 2 . ± .
10. How-ever, a fit of the whole lightcurve with a smooth brokenpower-law ( § α Opt, =1 . +0 . − . , incompatible with the X-ray decay. Finally,although the error bars are quite large, we notice that,taken at face value, the slope of the UVOT spectrum inthe 1 ks SED is negative (Fig. 3), suggesting that ν m may be already below the optical at that epoch.The above mentioned flaws imply that the simpleFS model is not viable to explain the properties ofGRB090510. However, this model relies on highly ide-alized assumptions, and it is known that several GRBafterglows do not strictly follow its simple predictions.Plausible effects that may affect the predictions and easethe comparison with GBR090510 are:- a phase of energy injection (Sari & M´esz´aros 2000),or an evolution of the microphysical parameters of theblast wave (Panaitescu et al. 2006); both may cause anearly shallow decay of the X-ray flux;- the transit of ν m slightly after the jet break, whichcould explain a shallow late optical decay.With the present data, we are unable to distinguishbetween energy injection and microphysical parameterevolution. As for the X-ray decay post jet break, hydro-dynamical simulations show that initially the jet decayslope can reach α ≃ α X, = 2 . ± .
10 could be achieved. Therefore,with some extensions, the FS model could arrange thetemporal properties, although some fine tunings wouldbe needed. De Pasquale et al. CONCLUSIONS
We have reported the
Swift and
Fermi observationsof the short GRB090510, an event endowed with brightprompt and afterglow emission, and detected in the GeVrange up to 200 s after the trigger. The initial X-rayemission shows a slow decay up to ≃ . ≃ . ν m . This interpretationdoes not require extremely high values of Lorentz factor,should the density of the environment be very low. Wealso find that reasonable values for the physical param-eters can lead to the observed properties, which mightfavour the model, although some fine tuning is neces-sary. The second scenario assumes that the FS producesthe full spectrum of the emission, observed from the op-tical to the GeV band. The γ -ray, X-ray and opticalspectrum can be reproduced by the template FS spectralmodel and the required physical parameters are plausi-ble. Although the simple FS model fails to reproduce the observed temporal behavior, extensions of this modelcould accommodate the temporal mismatch. In order toidentify the origin of the GeV component of GRBs like090510, more case studies will be necessary. Fortunately,we have very promising prospects for other simultaneous Fermi and
Swift observations of short GRBs, which willprovide us with more measurements to shed light on theproperties of this class of events.The
Fermi
LAT Collaboration acknowledges supportfrom a number of agencies and institutes for both de-velopment and the operation of the LAT as well as sci-entific data analysis. These include NASA and DOEin the United States, CEA/Irfu and IN2P3/CNRS inFrance, ASI and INFN in Italy, MEXT, KEK, and JAXAin Japan, and the K. A. Wallenberg Foundation, theSwedish Research Council and the National Space Boardin Sweden. Additional support from INAF in Italy andCNES in France for science analysis during the opera-tions phase is also gratefully acknowledged. SZ aknowl-edges STFC support. This work used data supplied bythe UK Swift SDC at the University of Leicester.
REFERENCESAbdo A.A., et al. 2009, in preparationAtwood, W., Abdo, A. A., Ackermann, M., et al. 2009 ApJ, 697,1071Barthelmy, S.D., Barbier, L.M., Cummings, J.R., et al. 2005,SSRv, 120, 143Barthelmy, S.D., Chincarini, G., Burrows, D.N., et al. 2005,Nature, 438, 944Beuermann, K., Hessmann, F.V., Reinsch K. et al. 1999, A&A,352, 26Burrows, D.N., Hill, J.E., Nousek, J.A., et al. 2005, SSRv, 120,165Burrows, D.N., Grupe, D., Capalbi, M., et al. 2006, ApJ, 653, 468Evans, P.A., Beardmore, A.P., Page, K.L. et al. 2007, A&A 469,379-385Evans P.A., Beardmore, A.P., Page K.L., et al. 2009,MNRAS submitted, arXiv:08123662Ferrero P., Kann, D.A., Zeh, A., et al. 2006, A&A 457, 857Fong, W., Berger, E., & Fox, D. B. 2009, ApJ, submitted,arXiv:0909.1804Gehrels, N., Chincarini, G., Giommi P., et al. 2004, ApJ, 611,1055Gehrels, N., Ramirez-Ruiz, E., & Fox, D. S. 2009, ARAA, 47, 567Graham, J. F., Fruchter, A. S., Levan, A. J., et al. 2009, ApJ,698, 1620Goad, M.R., Page K.L., Godet, O., et al 2007, A&A, 468, 103Granot J. 2007, RMAC, 27, 140Granot, J., & Sari., R. 2002, ApJ, 568, 820Guetta, D., & Granot, J. 2003, ApJ, 585, 885Guiriec, S., Connaughton, V. & Briggs M. 2009, GCN 9336Hoversten, E.A., Krimm, H.A., Grupe, D. et al. 2009, GCN 9331Kouveliotou, C., Meegan, C.A., Fishman, G.J. et al. 1993, ApJ,413, 101Mangano V., Sbarufatti, B., Ukwatta, T.N., et al. 2008,GCN 7208 M´esz´aros P & Rees M. 1993, ApJ405, 278Metzger, B.D., Quataert, E. Thompson, T. A. 2008, MNRAS,385, 1455Nakar E. 2007, PhR, 442, 166Nakar E., Ando, S., Sari, R. 2009, arXiv:0903.2557Norris, J.P., & Bonnel, J.T. 2006, ApJ, 643, 266Oates, S.R., Page, M. J., Schady, P. et al. 2009 MNRAS395, 490Ohno, M. & Pelassa, V. 2009, GCN 9334Panaitescu A., M´esz´aros, P., Burrows, D., et al. 2006, MNRAS,369, 2059Panaitescu A. & Vestrand T. 2008, MNRAS, 387, 479.Perna, R. Armitage, P. J., Zhang, B. 2006, ApJ, 636, 29Poole, T.S., Breeveld, A.A., Page, M.J., et al. 2008, MNRAS,383, 627Rau A., McBreen S., Kruehler T. et al. 2009, GCN 9353Rees, M. J. & Meszaros, P. 1994, ApJ, 430, 93Roming, P.W.A., Kennedy, T.E., Mason, K.O., et al. 2005, SSRv,120, 95Rosswog, S. 2007, RMxAC, 27, 57RSari, R. 1997, ApJ, 489, 37Sari, R., & M´esz´aros P. 2000, ApJ, 535, 33Sari, R., Piran, T., & Narayan R. 1998, ApJ, 497, 17Sari, R., Piran., T., and Helpern, J. 1999, ApJ, 519, 17Stratta, G., D’Avanzo, P., Piranomonte, S. et al. 2007, A&A, 474,827Troja, E., King, A.R., O’Brien, P.T. et al. 2008, MNRAS, 385, 10.Ukwatta., T.N., Barthelmy, S.D., Baumgartner W.H. et al. 2009,GCN 9337Yamazaki, R., Ioka, K., & Nakamura, T. 2002, ApJ, 571, 31Zhang, B., Zhang, B.-B., Virgili, F. J. et al. 2009, ApJ in press,arXiv:0902.2419. arly afterglow of the short GRB090510 5
Time Test Energy Photon flux above 100 MeVbins (s) Statistic index ( ph.cm − .s − )1: 0.38 – 0.48 208.2 0.85 − . . +1 . − . −
2: 0.48 – 0.92 541.8 1.20 − . . +0 . − . −
3: 0.92 – 1.5 192.1 0.93 − . . +2 . − . −
4: 1.5 – 2.5 301.7 1.41 − . . +2 . − . −
5: 2.5 – 5.5 163. 0.76 − . . +3 . − . −
6: 5.5 – 11.5 58. 0.86 − . . +1 . − . −
7: 11.5 – 37. 71. 2.27 − . . +0 . − . −
8: 37. – 69.5 59.9 0.85 − . . +2 . − . −
9: 69.5 – 200. 43. 1.74 − . . +1 . − . −
10: 200. – 400. ∼ . < −
11: 400. – 800. ∼ . < −
12: 800. – 1500. ∼ . < −
13: 4200. – 7200. ∼ . < −
14: 10150. – 13000. ∼ . < −
15: 15800. – 18500. ∼ . < − TABLE 1LAT time-resolved spectroscopy. The source test statistic was defined as twice the difference of the unbinnedlog-likelihood between the null hypothesis (background only) and the alternative hypothesis (presence of a source).Bottom table shows 95% confidence level upper limits on flux for different energy indeces.
SED 5 SED 100 s 100 sLAT dataset (s) 10. – 200. 10. – 200. 0.38 – 200. N H ( × cm ) 1.52 +0 . − . E ( B − V ) (mag) 0.000 +0 . − .
0. (fixed) 0. (fixed) E b (keV) . +0 . − . (100 s)0 . +0 . − . (150 s)0 . +0 . − . (1000 s) < .
001 (7000 s) < .
01 (12000 s) 0.31 +0 . − . +0 . − . β ± +0 . − . +0 . − . E b (MeV) ≃
300 (100 s) [20 − − β β + 1 / +0 . − . +0 . − . TABLE 2Best fit parameters obtained by fitting the 5 SEDs simultaneously or the SED at 100 s only with a double brokenpowerlaw model (see text). N H and E ( B − V ) are host absorption and extinction respectively; E b and E b are the twobreak energies calculated at the epoch in parenthesis. De Pasquale et al. F l u x [ > M e V ] ( ph / c m / s ) −7 −6 −5 −4 −3 −2 prompt emissionlong−lived det.UL index 1.1UL index 0.5UL index 2.5 Time since BAT trigger (s) −1
10 1 10 F l u x d e n s i t y ( Jy ) −11 −10 −9 −8 −7 −6 −5 −4 −3 −2 LAT
GBM
XRTBAT
UVOT
Fig. 1.—
Top:
LAT flux above 100 MeV and best fit to the flux decay (line).
Bottom: energy flux densities averaged in the observedenergy bands: BAT (15 keV – 350 keV, stars); XRT (0.2 keV – 10 keV, crosses); UVOT renormalised to white (diamonds); LAT (100MeV – 4 GeV, filled squares; the average spectral index was used to convert from photon to energy flux) with upper limits for β = 1 . arly afterglow of the short GRB090510 7 −10 −9 −8 −7 −6 −5 −4 −3 C oun t R a t e ( s − k e V − ) r a ti o Energy (keV)
Fig. 2.—
UVOT-XRT-LAT count spectrum, with best fit and residuals shown (see text)
De Pasquale et al.