Table Of Content

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed1February2008 (MNLATEXstylefilev1.4) Gamma-ray bursts and the history of star formation A. W. Blain1,2 and Priyamvada Natarajan1,3 1 Institute of Astronomy, Madingley Road, Cambridge, CB3 0HA 2 Cavendish Laboratory, Madingley Road, Cambridge, CB3 0HE 3 Department of Astronomy, Yale University,NewHaven, CT06520-8181, USA 0 0 1February2008 0 2 n ABSTRACT a Popular models for the origin of gamma-ray bursts (GRBs) include short-lived mas- J sive stars as the progenitors of the fireballs. Hence the redshift distribution of GRBs 8 should track the cosmic star formation rate of massive stars accurately. A significant 2 proportion of high-mass star formation activity appears to occur in regions that are obscuredfromview inthe opticalwavebandby interstellardust.The amountofdust- 2 enshrouded star formation activity taking place has been estimated by observing the v thermalradiationfromthedustthathasbeenheatedbyyoungstarsinthefar-infrared 8 andsubmillimetre wavebands.Herewediscussanalternativeprobe–the redshiftdis- 6 4 tributionof GRBs.GRBs aredetectable atthe highestredshifts,andbecause gamma 1 rays are not absorbed by dust, the redshift distribution of GRBs should therefore 1 be unaffected by dust extinction. At present the redshifts of GRBs can only be de- 9 termined from the associated optical transient emission; however, useful information 9 about the prevalence of dust-obscured star formation can also be obtained from the h/ ratioofGRBswithandwithoutanassociatedopticaltransient.EightGRBscurrently p have spectroscopic redshifts. Once about a hundred redshifts are known, the popula- - tion of GRBs will provide an important test of different models of the star formation o history of the Universe. r t s Key words: dust,extinction–galaxies:evolution–galaxies:formation–cosmology: a observations – gamma-rays:bursts : v i X r a 1 INTRODUCTION infrared wavebands have been used to estimate the history of star formation activity (Lilly et al. 1996; Madau et al. Gamma-ray bursts (GRBs) are detectable out to the edges 1996;Steideletal.1999);however,absorptionbyinterstellar oftheobservableUniverse,andsoprovideinformationabout dust in star-forming galaxies could significantly modify the the processes occurring within their progenitors at all cos- results (Blain et al. 1999a,c). It is difficult but not impossi- micepochs(Piran1999a,b).IfGRBsariseeitherfrombinary bletocorrecttheseopticallyderivedresultstotakeaccount mergersofmassivestellarremnants(neutronstarsandblack of this effect. By making observations in the near-infrared holes; Paczynski 1986) or from failed supernovae collapsing waveband,wheretheoptical depth of thedust is less, some to form black holes (Woosley 1993), or from the collapse progress has been made (Pettini et al. 1998; Steidel et al. of massive stellar cores (hypernovae; Paczynski 1998), then 1999). However, there are considerable uncertainties in the theywill beassociated with theformation of massive stars. sizeofthecorrectionsthatshouldbeapplied.Itisnowpos- See Hogg & Fruchter (1999) and Holland & Hjorth (1999) sible to detect the energy that has been absorbed and re- for some observational evidence that this is the case. Be- emittedbydustinhigh-redshiftgalaxiesdirectlyintheform cause high-mass stars have very short lifetimes, the rate of ofthermalfar-infraredradiation,whichisredshiftedintothe GRBs should trace the formation rate of massive stars in submillimetre waveband. The sensitive SCUBA camera at the Universe. Hence, if the distribution of the redshifts of the James Clerk Maxwell Telescope has revealed this emis- GRBsisknown,thisshouldallowtheevolutionofthehigh- siondirectly(Smail,Ivison&Blain1997;Bargeretal.1998; mass star-formation rate to be derived (Totani et al. 1997; Hughes et al. 1998; Blain et al. 1999b, 2000; Barger, Cowie Wijers et al. 1998; Hogg& Fruchter1999; Mao & Mo 1999; & Sanders 1999; Eales et al. 1999). It is possible to derive Krumhotz,Thorsett & Harrison 1999). a history of high-mass star formation from the SCUBA ob- Observations of faint galaxies in the optical and near- servations (Blain et al. 1999a,c); at least as much energy (cid:13)c 0000RAS 2 A.W. Blain and Priyamvada Natarajan is inferred tohave been released from dust-enshroudedstar formation activity as in the form of unobscured starlight. Anindependenttestoftherelativeamountofobscuredand unobscuredstarformationactivitywouldbeextremelyvalu- able. The advantage of using GRBs for this purpose is that gamma-rays are not absorbed by interstellar dust, either within the host galaxy of the GRB or in the intergalac- tic medium along the line of sight to the host galaxy, and so dust extinction will not modify our view of the distant Universe observed using gamma rays. However, there is a problem, as in order to determine a redshift for a GRB the spectrum of the associated burst of optical transient radiation must be detected. If the burst is heavily enshrouded in dust,thenit would bedifficulttodetect suchatransient and to obtain its spectrum. In Section2 we briefly review the differences between the histories of star formation derived from optical/near- infraredandfar-infrared/submillimetreobservations.InSec- Figure 1. A summaryof the current state of knowledge of the tion3 we discuss and predict the associated redshift distri- star formation history of the Universe. The data points plotted bution of GRBs. In Section4 we discuss the consequences are described in Blain et al. (1999a) and Steidel et al. (1999). for determining the history of star formation using obser- Thethickandthindashedcurvesdescribemodelsthatrepresent vations of GRBs. We assume an Einstein–de Sitter world thedust-corrected andnon-dust-corrected historiesderivedfrom model with Hubble’sconstant H0 =50kms−1Mpc−1. optical andnear-infraredobservations, whicharerepresented by data points. Where corrections have been made to the optical data to account for the estimated effects of dust extinction, the dataisrepresentedbyemptysymbols,andthehigherpairofhigh- 2 THE HISTORY OF STAR FORMATION redshiftdiagonalcrosses.Wherenocorrectionshavebeenapplied, the data is represented by filled symbols and by the lower pair In Fig.1we compare fivecurrently plausible star formation of high-redshift diagonal crosses. The solid lines represent mod- histories (Blain et al. 1999a,c), two of which are based on els derived from far-infrared and submillimetre-wave emission: observations made in the optical and near-infrared wave- a ‘Gaussian model’ (Blain et al. 1999c), a ‘modified Gaussian bands and three of which are based on observations in the model’ (Barger et al. 1999b) and an ‘hierarchical model’ (Blain etal.1999a), inorderofincreasingthickness. far-infrared and submillimetre wavebands. A wide variety of observational data that has been gathered in the optical andnear-infraredwavebandsisalsoplotted.Thedataisde- calmodel’(thicksolidline)thepopulationofsubmillimetre- scribed in more detail in the caption of fig.1 of Blain et al. luminous galaxies is described in terms of short-lived lumi- (1999a) and bySteidel et al. (1999). nousburstsinducedbygalaxy mergers(Blain etal.1999a). Inthefirstopticallyderivedmodel(thindashedline)it This model provides a reasonable fit to the Barger et al. is assumed that no dust absorption takes place within the (1999b) redshift distribution if the dust temperature is as- galaxiesdetectedindeepopticalimages,andthatthereisno sumed to be 35K. population of strongly obscured objects missing from these samples. This model closely follows the form of the history of star formation derived by Madau et al. (1996) from an 3 THE REDSHIFT DISTRIBUTION OF GRBS analysis of the Hubble Deep Field at z >∼2, and from obser- vationsoftheCanada–FranceRedshiftSurveyfieldsatz<∼1 Ahistogramshowingtheeightspectroscopicredshiftsofop- byLilly etal. (1996). Inthesecond optically derivedmodel ticaltransientsassociatedwithGRBs(Greiner1999)isplot- (thickdashedline),dustextinctionisassumedtobepresent. tedinFig.2(solid histogram; Metzgeretal. 1997; Djorgov- Themodelfitsthedatathathasbeencorrectedempirically ski et al. 1998a,b,c, 1999a,b; Kulkarni et al. 1998; Galama to take account of the effects of dust, using radio and ISO et al. 1999; Vreeswijk et al. 1999). There are indications satellite observationsat z<∼1(Floreset al.1999) andusing of the redshifts for three more GRBs, which are included estimates of extinction in high-redshift galaxies estimated in the derivation of the dotted histogram shown in Fig.2, from near-infrared spectroscopy (Pettinietal. 1998; Steidel while for about another twenty GRBs no optical transient et al. 1999). has been detected, in some cases despite sensitive searches. Thethreedifferentfar-infrared/submillimetremodelsof These numberswere compiled on 1999 November16. thehistory of starformation fitall of theavailable datade- In Fig.2 we also present the expected redshift distri- scribing the background radiation intensity and the counts butions of GRBs for each of the five star formation histo- ofdustygalaxies inthesewavebands.The‘Gaussian model’ riesshowninFig.1.Theseredshiftdistributionsarederived (thinsolidline)wasderivedbyBlainetal.(1999c),andthe by integrating the function that describes the evolution of ‘ModifiedGaussianmodel’(mediumsolidline)waschanged the global star-formation rate along a radial section of the slightly in order to fit the median redshift of plausible Universe (Totani 1997; Wijers et al. 1998) and have been counterpartstosubmillimetre-selectedgalaxies(Smailetal. normalised tounity. 1998)determinedbyBargeretal.(1999b).Inthe‘Hierarchi- InthiscalculationweassumethatatypicalGRBwould (cid:13)c 0000RAS,MNRAS000,000–000 GRBs and star formation 3 beobscured.Therefore, theinferred GRBredshift distribution at present might in fact be biased to lower redshifts, given the large number of GRBs that do not have detected optical transient counterparts. This selection effect for the detection of optical transients will depend on the geometry and environment of the host galaxy, and so is likely to be difficult to investigate reliably until a much larger sample ofhigh-qualityfollow-upobservationsofGRBshasbeenas- sembled. It is interesting to note that the predicted redshift distribution of GRBs that was derived from unobscured optical observations of the star formation history, as shown by the thin dashed curve in Fig.2, appears to provide the bestagreementwiththeobservedredshiftdistributionofthe opticaltransientsthathasbeendeterminedsofar.Thissug- gests that there could be a common extinction bias against thedetectionofbothdust-enshroudedstarformation activ- Figure 2. The curves represent the GRB redshift distributions ityin opticalgalaxysurveysandoftheopticaltransientsof expectedinthefivestarformationhistoriesshownbythecurves GRBs. in Fig.1 (the line styles are the same as those used in Fig.1). Since systematic, rapid, deep and reliable searches for The solid histogram shows the redshift distribution of the eight GRBs for which optical transients have confirmed spectroscopic opticaltransientsbeganin1998March(Akerlofetal.1999), redshifts (Greiner 1999). It is plausible that a further three ninegamma-rayburstshavebeendetectedwithopticaltran- GRBs,GRB970228, GRB970828 andGRB980329 have redshifts sients, four without and a further thirteen unreported, on 1.3<z<2.5, z≃0.33and z∼5respectively (Djorgovski etal. 1999 November 16 (Greiner 1999). If those GRBs without 1998b;Greiner1999;Fruchter1999).GRB980329istheonlyGRB optical transients are in dusty regions, then this lends sup- that has currently been detected at a submillimetre wavelength porttotheideathatcomparableamountsofhigh-massstar of 850µm (Smith et al. 1999). These three sources are included formation occurs in heavily and lightly dust-enshroudedre- in the dotted histogram. GRB980424, which has been identified gions. OneGRB(GRB980329) hasso far beendetected us- withalow-redshiftsupernova,hasnotbeenincludedinthefigure. ingSCUBAatasubmillimetrewavelengthof850µm(Smith etal.1999),althoughitisunclearwhetherthereisanycom- be detectable at any redshift. If, in fact, there is a redshift- ponentofthermaldustemission involved,oriftheemission dependent selection function for GRBs, then the observed is entirely attributable to synchrotron radiation from the redshift distribution of burstswould be expected to be sys- shocked interstellar medium. See Taylor et al. (1998, 1999) tematically lower as compared with the curves shown in foradiscussionofthepropertiesofGRBsintheradiowave- Fig.2. At present, there are too few redshifts with which band,where they are not subject to obscuration by dust. to estimate the possible size of this effect; however, the de- The number of spectroscopic redshifts for the optical tection of two GRBs at z > 3 tends to argue against the transientsofGRBsisgrowing steadily,withanewdetermi- strong anti-selection of high-redshift bursts. nation being reported every two to three months (Greiner 1999). The number of redshifts for GRBs already exceeds the number of redshifts that have been obtained for galaxies detected in submillimetre-wave surveys that have reli- 4 DISCUSSION ableidentifications.Onceasampleofabout100GRBshave It is clear from thehistograms shown in Fig.2 that too few redshiftsthenit should bepossible todiscriminate between redshiftsarecurrentlyknowntoallowustodiscriminatebe- thedifferentmodels. Asthepredictedredshift distributions tweenthedifferentmodelsofthestarformationhistory.Nor shown in Fig.2 are significantly different,theGRB redshift is the efficiency of generating GRBs from massive star for- distributionmayprovideaverysignificantconstrainttothe mationactivityknowninsufficientdetailtoallowustodis- history of star formation activity at all epochs. criminatebetweenthedifferentmodelsonthegroundsofthe absolutenumberofburstsobserved.Asanadditionalcaveat, 4.1 Dust-enshrouded infrared transients it seems likely that about 20per cent of the submillimetre- selectedgalaxiesarepoweredbygravitationalaccretiononto ItwillbeimportanttopaycloseattentiontotheGRBswith- active galactic nuclei (AGNs; Almaini, Lawrence & Boyle outopticaltransients.Iftheopticalandultravioletradiation 1999), and assuchshould not beassociated with GRBsde- released by the GRB is obscured by dust, then since these rived from exploding high-mass stars, unless a significant heateddustgrainshavealowheatcapacity,thereprocessed amount of high-mass starformation activity is takingplace thermal emission could be detected directly as a transient coevally with AGN fueling. signalatsubmillimetretonear-infraredwavelengths.Intense If much of thehigh-mass star formation activity in the shocking and heating of the gas in the interstellar medium Universe does indeed take place in dust-enshrouded galax- of the host galaxy could also lead to detectable emission ies, thenthe optical transients of GRBs that occur in these offar-infraredandsubmillimetre-wavefine-structureatomic galaxies would be less likely to be detected than those in line radiation, and molecular rotational line radiation (see dust-free galaxies, because while the GRB gamma-ray sig- alsoPerna,Raymond&Loeb2000).Becausethehostgalaxy nalcanescape,theassociatedopticaltransientemissionwill of theGRB is likely to be optically thin at far-infrared and (cid:13)c 0000RAS,MNRAS000,000–000 4 A.W. Blain and Priyamvada Natarajan submillimetre wavelengths, the detection of this radiation Blain A.W., Ivison R.J., Kneib J.-P., Smail I., 2000, in couldprovidearedshiftforaGRB,evenintheabsenceofan BunkerA.J.,vanBreughelW.J.M.eds,TheHy-redshiftuni- opticaltransientsourceiftheopacityatopticalwavelengths verse.ASP Conf.Ser.Vol.193, Astron.Soc. Pac.,San Fran- is very high. In the future, the large collecting area, excel- cisco,inpress(astro-ph/9908024) Djorgovski S.G., Kulkarni S.R., Goodrich R., Frail D.A., lent sensitivity and subarcsecond angular resolution of the BloomJ.S.,1998a,GCNnotice: 137 Atacama Large Millimeter Array (ALMA; Wootten 2000) Djorgovski S.G., Kulkarni S.R., Goodrich R., Frail D.A., willmakeittheidealinstrumenttoconductobservationsto BloomJ.S.,1998b,GCNnotice:139 search for any transient line and continuum radiation from Djorgovski S.G., Kulkarni S.R., Bloom J.S., Frail D.A., Chaf- GRBsinthesubmillimetrewaveband.Recently,Waxman& feeF.,GoodrichR.,1998c, GCNnotice:189 Draine (2000) discussed the effects of a GRB on the subli- DjorgovskiS.G.,KulkarniS.R.,BloomJ.S.,NeugebauerG.,Ko- mation of dust in the surrounding interstellar medium. We reskoC.etal.1999a, GCNnotice: 256 arecurrentlyinvestigatingtheobservabilityofinfraredtran- Djorgovski S.G., Kulkarni S.R.,Bloom J.S.,Frail D.A.,1999b, sients from dust-enshrouded GRBs (Venemans & Blain, in GCNnotice:289 preparation). Eales S.A.,LillyS.J., Gear W.K.,Dunne L.,Bond J.R., Ham- merF.,LeFèvreO.,CramptonD.,1999, ApJ,515,518 GalamaT.J.etal.1999,GCNnotice:388 GreinerJ.,1999,http://www.aip.de/˜jcg/grbgen.html 5 CONCLUSIONS FloresH.etal.,1999,ApJ,517,148 FruchterA.S.,1999,ApJ,512,L1 Alarge statistical sampleof redshiftsforgamma-ray bursts HoggD.W.,Fruchter A.S.,1999,ApJ,520,54 HollandS.,HjorthJ.,1999,A&A,344,L67 (GRBs) will allow the star formation history of the Uni- HughesD.H.etal.,1998,Nat,394,241 versetobeprobedinmoredetailthaniscurrentlypossible. KrumhotzM.R.,ThorsettS.E.,HarrisonF.A.,1999,ApJ,5066, Once about a hundred examples are known, the fraction of L81 dust-enshroudedstar formation activity that takes place as Kulkarni S.R., Adelberger K.L., Bloom J.S., Kundic T., Lu- a function of redshift can be addressed by an analysis of binL.,1998, GCNnotice: 29 both the redshift distribution of the optical transients and Lilly S.J., Le Fèvre O., Hammer F., Crampton D., 1996, ApJ, the fraction of GRBs for which an optical transient is de- 460,L1 tected. The results will also allow the fraction of AGN in Madau P., Ferguson H.C., Dickinson M.E., Giavalisco M., Stei- submillimetre-selected samples and the form of their evo- delC.C.,FruchterA.,1996,MNRAS,283,1388 lution to be estimated in a new way. The commissioning MaoS.,MoH.J.,1999,A&A,339,L1 MetzgerM.,DjorgovskiS.G.,SteidelC.C.,KulkarniS.R.,Adel- of the ALMA interferometer array will hopefully allow the bergerK.L.,FrailD. A.,1997,Nat,387,879 absorbedopticalandultravioletlightfromdust-enshrouded PaczynskiB.,1986,ApJ,308,L43 GRBs to be detected in the form of a far-infrared transient PaczynskiB.,1998,ApJ,494,L45 signal,potentiallyrevealingtheredshiftoftheGRBwithout Perna R., Raymond J., Loeb A., 2000, ApJ, submitted (astro- requiring observations in theoptical waveband. ph/9904181) Pettini M., Kellogg M., Steidel C.C., Dickinson M., Adel- bergerK.L.,GiavaliscoM.,1998,ApJ,508,539 PiranT.,1999a,preprint(astro-ph/9907392) ACKNOWLEDGEMENTS PiranT.,1999b,PhysicsReports,314,575 SmailI.,IvisonR.J.,BlainA.W.,1997,ApJ,490,L5 AWB, Raymond & Beverly Sackler Foundation Research Smail I., Ivison R.J., Blain A.W., Kneib J.-P., 1998, ApJ, 507, Fellow, gratefully acknowledges support from the Founda- L21 tion as part of the Deep Sky Initiative programme at the SmithI.A.etal.,1999,A&A,347,92 IoA. We thank Brad Hansen, Kate Quirk and an anony- SteidelC.C.,AdelbergerK.L.,DickinsonM.,GiavaliscoM.,Pet- mous referee for helpful comments on the manuscript, and tiniM.,1999,ApJ,519,1 Andrew Fruchter, George Djorgovski and Shri Kulkarni for Taylor G.B., Frail D.A., Kulkarni S.R., Shepherd D.S., Fe- useful conversations. rociM.,FronteraF.,1998,ApJ,502,L115 Taylor G.B., Beasley A.J., Frail D.A., Kulkarni S.R., ReynoldsJ.E.,1999, A&AS,138,445 TotaniT.,1997,ApJ,486,L71 VenemansB.,BlainA.W.,inpreparation REFERENCES VreeswijkP.M.etal.,1999,GCNnotice:324 AkerlofC.etal.,1999,Nat,398,400 Waxman E., Draine B.T., 2000, ApJ, submitted (astro- AlmainiO.,LawrenceA.,BoyleB.J.,1999,MNRAS,305,L59 ph/9909020) BargerA.J.,CowieL.L.,SandersD.B.,FultonE.,TaniguchiY., WoosleyS.E.,1993,ApJ,405,273 SatoY.,KawaraK.,OkudaH.,1998,Nat,394,248 Wootten A., 2000, Science with the Atacama Large Millimeter BargerA.J.,CowieL.L.,SandersD.B.,1999a,ApJ,518,L59 Array.Astron.Soc.Pac.,SanFrancisco,inpress Barger A.J., Cowie L.L., Smail I., Ivison R.J., Blain A.W., Wijers R.A.M.W., Bloom J.S., Bagla J., Natarajan P., 1998, KneibJ.-P.,1999b, AJ,117,2656 MNRAS,294,L13 Blain A.W., Jameson A., Smail I., Longair M.S., Kneib J.-P., IvisonR.J.,1999a, MNRAS,309,715 BlainA.W.,KneibJ.-P.,IvisonR.J.,SmailI.,1999b, ApJ,512, L87 Blain A.W., Smail I., Ivison R.J., Kneib J.-P.,1999c, MNRAS, 302,632 (cid:13)c 0000RAS,MNRAS000,000–000