Precursors of short gamma-ray bursts
aa r X i v : . [ a s t r o - ph . H E ] S e p Submitted to ApJ 2010-06-07; accepted 2010-09-07
Preprint typeset using L A TEX style emulateapj v. 04/21/05
PRECURSORS OF SHORT GAMMA-RAY BURSTS
E. Troja , S. Rosswog , N. Gehrels Submitted 2010 June 7; accepted 2010 September 7
ABSTRACTWe carried out a systematic search of precursors on the sample of short GRBs observed by
Swift . Wefound that ∼ ∼
13 s and the latter ∼ Subject headings: stars: neutron; gamma rays: bursts; gamma-ray bursts: individual: 090510 INTRODUCTION
The main gamma-ray event in GRBs is occasionallyanticipated by a less intense episode of emission, calleda precursor. With the exception of a few cases, pre-cursors show non-thermal spectra and have been mainlyobserved in long duration GRBs (e.g. Lazzati 2005;Burlon et al. 2008). The first evidence of preburst ac-tivity has been observed by the Ginga satellite in thelong GRB 900126 (Murakami et al. 1991), where a softX-ray peak precedes the burst onset by ∼ γ -ray prompt emission. By us-ing a different search criterion, Lazzati (2005) found in-stead that ∼
20% of long GRBs are preceded by an earlyemission episode, which is much weaker and spectrallysofter than the proper GRB. This result was also con-firmed by Beppo-SAX (Piro et al. 2005) and HETE-2(Vanderspek et al. 2004) observations, yet it is unclearwhether it still holds for the precursor activity detectedin
Swift
GRBs. Hints of different properties (e.g. hard-ness ratio, spectral lag) between the precursor and theprompt emission have been reported in the study of sin-gle bursts (e.g. GRB061121; Page et al. 2007). Howeverthe recent work of Burlon et al. (2008), based on a largesample of long GRBs observed by
Swift , did not find ev-idence for such spectral distinction.An ubiquitous feature, emerging from all the pre-vious studies, is the distribution of delay times be-tween the precursor and the prompt emission, whichextends up to hundreds of seconds. This representsone of the main challenges to the current theoreticalmodels. If the precursor marks the start of the cen- NASA, Goddard Space Flight Center, Greenbelt, MD 20771,USA NASA Postdoctoral Program Fellow School of Engineering and Science, Jacobs University Bremen,Campus Ring 1, 28759 Bremen, Germany tral engine activity (Nakamura 2000), the observed qui-escent time would require a fine tuning of the ejectaLorentz factors or an effective turn-off of the GRB energysource (Ramirez-Ruiz et al. 2001). However, whetherthis early emission physically differs from the burst it-self or is part of the same event remains a contro-versial point. Precursors as a separate phenomenonhave also been discussed in several theoretical scenar-ios (e.g. Lyutikov & Usov 2000; M´esz´aros et al. 2001;Waxman & M´esz´aros 2003; Lyutikov & Blandford 2003;Umeda et al. 2005; Wang & M´esz´aros 2007). A setof models explain the precursor emission within thestandard fireball scenario, commonly invoked to in-terpret the prompt and afterglow emission of GRBs.Within this framework the precursor is associatedwith the transition of the fireball to the opticallythin regime, which produces a photospheric black-body emission (Paczynski 1986a; M´esz´aros et al. 2001;Daigne & Mochkovitch 2002; Ruffini et al. 2008), whilethe GRB is associated with the later formation of shocksat larger radii (Rees & Meszaros 1994). According tothis interpretation, precursors happen relatively close intime to the GRB and should be observable in both longand short bursts.Another class of models interpret the precursor withinthe collapsar scenario and link its origin to the jet break-out from the stellar surface (e.g. Ramirez-Ruiz et al.2002; Waxman & M´esz´aros 2003; Zhang et al. 2003;Lazzati & Begelman 2005). The temporal delay betweenthe precursor and the GRB is explained as an appar-ent period of quiescence due to viewing angle effects(e.g. Morsony et al. 2007) or, alternatively, as an in-trinsic property of the central engine, which might un-dergo a second collapse (Wang & M´esz´aros 2007). In thiscase, precursors might be observed tens of seconds beforethe GRB emission and should occur exclusively in longGRBs. Indeed most of the observational and theoreticaleffort so far has been focused on long duration GRBs.Little attention has been paid to the occurrence ofprecursors in short GRBs, and the possibility of anearly precursor emission originated in the last mo-ments of a compact binary merger has been sporadicallydiscussed in literature (e.g. Hansen & Lyutikov 2001;Rosswog & Ramirez-Ruiz 2002). In this paper we ex-plore in detail the observational evidence for precursors E. Troja et al.in short GRBs and discuss the implication. The paper isorganized as follows: in § §
3; in § § § DATA ANALYSIS
Up to January 2010
Swift detected 38 GRBs classifiedas short bursts ( T . Swift /BATGRBs for signal preceding the main gamma-ray event.We define precursors as those events which fulfill the fol-lowing requirements: 1) the peak flux is smaller than themain event; 2) the flux returns to the background levelbefore the start of the main event; 3) the event locationin the sky corresponds to the GRB position.The
Swift data were retrieved from the public archive and processed with the standard Swift analysis soft-ware (v3.5) included in the NASA’s HEASARC software(HEASOFT, ver. 6.8) and the relevant calibration files.Our analysis has been performed in the 15-150 keV en-ergy band.
Temporal analysis
As a first step we inspected the GRB temporal profilessearching for precursor events, as defined in 1) and 2).For each GRB in the sample we created a light curvewith a time bin of 0.128 ms, as shorter time scales aresubject to noise fluctuations. In the case of
Swift /BATon-board triggers ∼
300 sec of event data are usually col-lected before the GRB trigger. We excluded from ouranalysis those time intervals during which the spacecraftwas slewing, starting our search on average ∼
240 s beforethe GRB. In order to identify the presence of weak emis-sion in the GRB light curves we used a detection algo-rithm based on wavelet transforms (Torrence & Compo1998) with a Morlet mother function. The wavelet algo-rithm performs a multi scale analysis which is well suitedfor detecting a transient event, such as a precursor, whoseduration is a priori unknown. We sampled 13 differenttime scales with a base-two logarithmic spacing, wherethe smallest resolvable scale s is set by the light curvetemporal resolution δt ( s = 2 δt ) and the maximum scalewas arbitrarily set to 4 s.As the count rates derived from mask-weighting proce-dures are already background subtracted, the pre-burstlight curves are dominated by a white Gaussian noisedue to statistical fluctuations (Rizzuto et al. 2007). Inthis particular case the wavelet coefficients are normallydistributed (Lazzati et al. 1999) and their power spec-tra follow a chi-square distribution with two degrees offreedom. Because of this property the significance lev-els of each peak in the wavelet power spectrum can be http://heasarc.gsfc.nasa.gov/docs/swift/archive/ analytically derived (see e.g. Torrence & Compo 1998).We examined the global wavelet spectrum. i. e. thetime-average over all the local wavelet spectra, for peakexceeding the noise spectrum level and set a minimumthreshold of 99.7% significance (corresponding to a 3 σ for a two-sided Gaussian distribution). In three cases(GRB 070406, GRB 080121 and GRB 091117) out of49, the burst has been discovered in ground analysis andonly ∼
10 seconds of event data around the trigger timeare available. Our search has therefore been restricted tothat interval. Because of border effects due to the veryshort time interval, we did not apply the wavelet detec-tion method and the light curves of these three burstswere simply inspected by eye.We found that 4 out of 38 short bursts and 1 out of 11bursts with extended emission show a possible precursoractivity. Among them, one burst (GRB 090510) showstwo precursors events, the former ∼
13 s and the latter,already known in literature (Abdo et al. 2009), ∼ ∼ Swift /BATlight curves (reported in each upper panel) with the
Fermi /GBM (GRB 090510 and GRB 081024A) andSuzaku/WAM (GRB 091117) light curves (reported inthe bottom panel). Times are always given relativeto the BAT trigger time. The cross-check of the lightcurves shows that the precursors in GRB 081024A andGRB 091117 can be confidently considered real, as a si-multaneous episode of emission has been observed by
Fermi /GBM and
Suzaku respectively. Instead no signif-icant emission above the background level is observedin correspondence of the first precursor at T -13 s inGRB 090510, while the second precursor at T -0.5 sis clearly detected by the Fermi /GBM. GRB 090510also triggered
Suzaku and Konus-Wind, unfortunately notime-resolved events are available during the interval ofthe first precursor at T −
13 s and any short time scalevariability is hard to detect. Indeed Suzaku and Konus-Wind light curves (with a resolution of 1 s and 2.9 srespectively) do not show any significant excess at suchearly times (K. Yamaoka, V. Pal’shin; private communi-cations). This non-detection does not necessarily implyrecursors of short GRBs 3 −100 −50 0 . C t s / s / de t [ − k e V ] Time since BAT trigger [s] −110 −105 − . . . −100 −50 0 . . C t s / s / de t [ − k e V ] Time since BAT trigger [s] −142 −140 −138 . . . . C t s / s / de t [ − k e V ] Swift/BAT −4 −2 0 2 4 C t s / s [ − k e V ] Time since BAT trigger [s]Fermi/GBM
NaI +NaI . C t s / s / de t [ − k e V ] Swift/BAT −10 −5 0 C t s / s [ − k e V ] Time since BAT trigger [s]Fermi/GBM
NaI +NaI +NaI . C t s / s / de t [ − k e V ] Swift/BAT −2 0 2 4 C t s / s Time since BAT trigger [s]
Suzaku/WAM
Fig. 1.—
Swift /BAT mask-weighted light curves (15–150 keV) of short GRBs with possible precursor activity. Dashed vertical linesmark the precursor duration. The precursors of GRB080702A and GRB050724 are shown in greater detail in the insets. For comparison,we also show the background-subtracted light curves of Fermi/GBM (090510 and 081024A) and Suzaku/WAM (091117). that the feature is spurious. Possible explanations arethe smaller effective areas compared to BAT, or a pre-cursor with a soft spectrum, e. g. peaking in the BATenergy range, as also expected on theoretical grounds.
Imaging analysis
In order to further check whether the excess in the lightcurve is related to the GRB, we produced a background-subtracted sky image in the interval of the candidate pre-cursor and searched for a source at the GRB position. E. Troja et al.
TABLE 1Image significance of the candidate precursors.
GRB T i T f Significance Probability a Others[s] [s] [ σ ]050724 (EE) . . . . . . . . -108.5 -107.5 3.7 5 × − –080702A . . . . . . . . . . . . -140.6 -139.5 3.2 3 × − –081024A . . . . . . . . . . . . -1.70 -1.45 5.5 < − Fermi090510 . . . . . . . . . . . . . -13.0 -12.6 5.2 < − –-0.55 -0.5 4.6 10 − Fermi091117. . . . . . . . . . . . . . -2.75 -2.65 1.8 6 × − Suzaku a Probability of a spurious detection with equal or higher significance. Derived fromMontecarlo simulations.
This step allows a better characterization of the back-ground level, as the contribution of other nearby sourcesis properly removed. Results are reported in Table 1,which lists the GRB name, the precursor time interval,and the significance of the source in the image domain ascalculated by the tool batcelldetect . If the precursorhas been detected by other instruments (see § σ is usually adopted to confidently assess that asource is real. Indeed the Subthreshold experiment car-ried out by the Swift team showed that lowering thisthreshold significantly increases the numbers of false de-tections ( ∼
96% of false positives). However our searchwas not performed on the whole image, as the source po-sition was a priori known. This reduces the number oftrials by a factor of ∼ × , i.e. the number of inde-pendent pixels in a BAT image, with respect to a blindsearch and the 6.5 σ threshold poses therefore a too re-strictive cut.We determined the probability to have a spurious N σ detection at a fixed position through Montecarlo simula-tions. An inspection of the detector plane images (DPIs)shows no noisy detectors during the selected time inter-vals, and therefore statistical fluctuations are the dom-inant source of noise. By assuming a Poissonian distri-bution with a mean count rate of ∼ − det − ,we simulated 10 source-free DPIs and derived the cor-responding sky images. On each simulated image we ranthe tool batcelldetect searching for a source at theGRB position. The probability that the detected sourceis due to background fluctuations is then calculated asthe ratio between the total number of fictitious detec-tions with significance equal or greater than that of theprecursor (Tab. 1, column 4) and the number of simu-lated images. The resulting values are listed in Tab. 1(column 5). RESULTS
The results of our analysis are summarized in Tab. 1.We found evidence of possible precursor activity in 4short GRBs, out of a sample of 38 events, and only in1 GRBs with extended emission (EE), out of a sampleof 11 events. One burst (GRB 090510) shows two pre-cursors, at ∼ T −
13 s and ∼ T − γ -ray light curves of these bursts have been shown inFig. 1. Our definition of precursor, detailed in §
2, dif-fers from those given in previous systematic studies (e.g http://gcn.gsfc.nasa.gov/subthreshold.html Koshut et al. 1995; Lazzati 2005), as it does not imposeany particular constraint on the quiescence time or theinstrumental trigger (e.g. Burlon et al. 2008). In onlyone case the classification of the event as precursor de-pends on our operational definition: the latter precursorin GRB 090510 does not satisfy either the conditions ofKoshut et al. (1995), because of its short delay time fromthe main GRB, and of Lazzati (2005), as the precursorevent triggered the
Fermi /GBM.Because our wavelet analysis ( § ∼ Swift and other satellites confirms that atleast three of our candidate precursors are real (Tab. 1,col. 5), namely the cases of GRB 081024A, GRB090510(2nd precursor), and GRB 091117.We verified that the detected excess in the light curvecorresponds to a point source at the GRB position inthe image domain ( § > σ signif-icance. Montecarlo simulations showed that the prob-ability of being a background fluctuation is very low( < − ), in agreement with the high significance of thedetection. Two cases remain controversial. The pre-cursors in GRB 080702A and GRB 050724 are detectedat a significance of 3.2 and 3.7 σ respectively, havinga ≈ − probability of being spurious. They are alsonot been seen by other instruments. These two precur-sors are very intriguing, as they show the longest delaytimes from the GRB triggers ( &
100 s) similar to thoseobserved in some long GRBs. However, in the presentstudy we are unable to confidently determine whetherthey are real features or not. In this context, it is worthnoting that the only short GRB (BATSE trigger 2614)in the sample of Koshut et al. (1995) shows a precur-sor ∼
75 s before the main burst. This strengthens theidea that long delay times are possible in short GRBprecursors, as we will discuss in § P r e c u r s o r H a r dne ss r a t i o GRB Hardness ratio
Fig. 2.—
Hardness ratio of the precursor vs. hardness ratio ofthe main GRB event. Error bars are at 1 σ confidence level. Thedashed line shows where the two ratios are equal. The hardnessratio is defined as the count rate in the 50-150 keV band over thecount rate in the 15-50 keV band. We also report the precursors ofGRB 050724 and GRB 080702A, albeit we are unable to confidentlydetermine whether they are real features, as explained in the text. spectral difference by comparing the precursor hardnessratio (HR) to that of the main GRB event. This is shownin Fig. 2: all the precursors appear consistent with themain GRB properties, in agreement with the findingsof Burlon et al. (2008). This result however might bepartially a consequence of the Swift /BAT narrow band-pass. For instance, the broadband
Fermi light curves ofGRB 090510 (see Abdo et al. 2009, Fig. 1) clearly showsthat the main GRB event has an extremely hard spec-trum, peaking in the MeV range, while the precursor at T − . T −
13 s, found in the
Swift /BAT light curves, is evensofter, peaking in the 15-50 keV energy band (Fig. 2,star symbol). On the other side, most theoretical modelspredict the peak of the precursor emission in the X-rayenergy range, i.e. at the lower end of the BAT energythreshold. Therefore, independently of the BAT band-pass, this should be reflected in our Fig. 2 by precursorsoccupying the region with HR .
1. Indeed the points seemto follow this trend.We further investigated whether short GRBs with pre-cursors differ from the other short GRBs in the sample,either in the prompt or afterglow emission. We comparedthe distributions of their observed properties, such as γ -ray fluence, duration (T ) and afterglow brightness,and ran a Kolmogorov-Smirnov (KS) test between thethe two samples (short bursts with and without precur-sors). The probability that they belong to the same GRBpopulation is 36%, based on the distribution of their flu-ences (in the 15-150 keV band), and 34%, based on thedistribution of their durations. Similarly a comparisonof the X-ray (0.3-10 keV) afterglow fluxes distributions,observed at 100 s and 1000 s, shows no substantial differ-ence (KS test probability of 96% and 68% respectively). DISCUSSION
Precursor activity has been so far associated to longGRBs. Previous systematic studies have in fact been fo-cused on the class of long GRBs, such as in the case ofLazzati (2005) who excluded those bursts with a durationT ≤ T -8 s in the SN-less burst GRB 060505,considering this as a further dissimilarity with the classof short GRBs. Our analysis showed instead that shortGRBs are also preceded by a precursor event, thoughless frequently than long GRBs (10% vs. 20% of longGRBs). The precursors in our sample are charaterizedby short durations, never exceeding the GRB T . Thisdisfavors the sub-jets model of Nakamura (2000), accord-ing to which the precursor duration is longer than thatof the main burst.Only one burst with extended emission (GRB 050724)shows a possible precursor ∼
100 s before the onset ofthe main GRB. However, as discussed in §
3, we cannot confirm in the present study whether it is a realevent and therefore whether precursors are also presentin bursts with EE. It has been suggested that burstswith EE may be originated by a different progenitor sys-tem (Troja et al. 2008; Metzger et al. 2008). The currentsample of short bursts with EE is still too small to drawany conclusion, but the absence of precursors in this sub-set of bursts, if confirmed by future observations, couldprovide a further evidence of their different nature.The association between precursors and long GRBs hasalso driven most of the theoretical work, which has oftenrelated the precursor to the interaction of the jet with themassive star progenitor (e.g. Ramirez-Ruiz et al. 2002;Lazzati & Begelman 2005). The presence of precursorsin either long and short GRBs might represent a chal-lenge for such interpretation. Given the fact that inthe internal shock model (Rees & Meszaros 1994; Piran1999) the GRB production is rather decoupled from thedetails of the central engine and both short and longbursts exhibit precursors, one my speculate that the pre-cursor production is related to the fireball rather than thecentral engine itself. An obvious idea to test is whetherthe precursor could be caused by a fireball becoming opti-cally thin (e. g. Paczynski 1986b) prior to the productionof the prompt GRB emission. If we consider an “iso-lated” fireball (i.e. on becoming transparent the photonsare not released into a surrounding, possibly intranspar-ent environment), an observed duration of the precur-sor ∆ t would imply a fireball radius (at this stage) of R ∼ c ∆ t , where Γ is the fireball bulk Lorentz factor.If we assume a saturated fireball with Γ ≈ η ≡ E/M c ,where E is the fireball energy and M its baryonic massloading, and equate the above radius to the one wherethe fireball should become transparent to its own photons(Abramowicz et al. 1991; Piran 1999), we find a relationbetween Γ and E for an observed duration ∆ t :Γ ≈ E / (cid:18) t (cid:19) / (1)If, as indicated by recent Fermi results(Ackermann et al. 2010), short GRBs do indeedpossess Lorentz factors in excess of 10 , a precursor E. Troja et al.origin related to a fireball becoming optically thin,would require a large fireball energy, E > erg.Moreover, if we assume that the main GRB signal isproduced by internal shocks, then the variability timescale δt var can be restricted to δt var & ∆ T (Lazzati 2005),where ∆ T is the time interval between the precursor andthe main prompt emission. The observed delays wouldsuggest implausibly long variability time scales, longerthan the main GRB duration itself. Thus, at least if theprompt emission is caused by internal shocks, we considerit unlikely that the observed precursors are produced byfireballs becoming optically thin. This conclusion couldneed to be modified if the fireball is released into an op-tically thick surrounding, say from a previously ejectedwind (see § Central engine-related mechanisms
Mergers of compact binaries, either in the form ofa double neutron star (DNS) (Blinnikov et al. 1984;Paczynski 1986b; Goodman 1986; Eichler et al. 1989)or a neutron star-black hole system (NS-BH)(Paczynski1991; Narayan et al. 1992), are still arguably the mostlikely central engines of short GRBs. In the following wewill focus on how such systems may produce an electro-magnetic transient prior to the main GRB.
Interaction of neutron star magnetospheres
Hansen & Lyutikov (2001) model the electromagneticsignatures that result from the interaction of two NSmagnetospheres prior to a double neutron star merger.The main prediction of their model is an X-ray transientpreceding the merger on a time scale of a few seconds.This signal could also be accompanied by a radio pulse.Hansen & Lyutikov (2001) consider a binary systemconsisting of an old, recycled pulsar that is rapidlyspinning ( P ≈ −
100 ms) and possesses a magneticfield of moderate strength ( B ∼ − G) and ayounger, slowly rotating ( P ≈ − B ∼ − G) neutron star (possibly a magne-tar), a combination that can be expected on evolution-ary grounds. If the magnetar birth rate is about 10%of the “ordinary pulsar” birth rate, a decent fraction ofdouble neutron stars should contain magnetars, at leastinitially. The recycled pulsar is considered as a per-fectly conducting sphere that passes through the externalfield prescribed by the magnetar. In this way a dipolarmagnetic field is induced whose magnetic dipole is di-rected against the external magnetic field. The motionof the pulsar through the external field induces surfacecharges that in turn produce electric fields with a com-ponent along the total magnetic field which acceleratecharges in an attempt to short out this parallel electricfield component. Once energetic enough, the latter pro-duce curvature photons together with a dense populationof electron-positron pairs. Pair plasma released into re-gions of increasing magnetic field strength are likely to betrapped in a optically thick cloud, while those releasedinto regions of decreasing field strength result in a rela-tivistically expanding wind of pairs and photons.The strongest prediction of this model is the pres-ence of an early precursor produced by the relativisticwind. The precursor spectrum should be close to ther-mal and hardening as the stars are driven towards coa-lescence. Interestingly, GRB 090510 has two precursor signals, where the first one peaks in the 15-50 keV en-ergy band while the second peaks around 300 keV. Suchbehavior would be consistent with the predictions of theHansen-Lyutikov model. The maximum luminosity thatthe precursor can reach is of the order of: L ≈ × erg s − (cid:18) B G (cid:19) (cid:16) a cm (cid:17) − (2)It follows that in order to match with the observed prop-erties of short GRB precursors a NS with a magnetar-like field ( B > G) is required. Such strong magneticfields likely decay on much shorter time scales ( ∼ -10 yrs; Heyl & Kulkarni 1998; Harding & Lai 2006, and ref-erences therein) than the merger lifetime, and a NS witha moderate magnetic field B ∼ -10 G looks a moreplausible configuration. Some population synthesis mod-els (Belczynski et al. 2002, 2006) however predict that asizable fraction of DNS mergers has much shorter inspi-ral times. This short-lived channel peaks at an inspiraltime of ∼ × years and after this time the magneticfield should have decayed by only a factor of a few (seeFig. 1 of Heyl & Kulkarni 1998). Neutron star flares induced by tidal crust-cracking
As a compact binary system secularly spirals in, theneutron star(s) become(s) vulnerable to tidal distortion.At a separation a the companion induces an ellipticityof ǫ ∼ δR /R ∼ m m (cid:0) R ns a (cid:1) , where m and m are theNSs masses and R ns is the NS radius. Once the elliptic-ity exceeds a critical value, the neutron star crust cracksand likely triggers a violent restructuring of the mag-netic field that may go along with a reconnection flare,similar to what is thought to happen in a magnetar gi-ant flare (Thompson & Duncan 1995; Palmer et al. 2005;Hurley et al. 2005). Once the crust has been cracked forthe first time, the neutron star enters a “tidal grindingphase” in which the tides exert a constant restructuringof crust, likely going along with further magnetic fieldreconfiguration and dissipation.The exact numerical value of the critical ellipticitythat the crust can still sustain, ǫ c , is not well-known,but recent studies based molecular dynamics simula-tions (Horowitz & Kadau 2009) suggest that neutronstar crusts can sustain strains up to a breaking valueof σ max ≈ .
1, corresponding to critical ellipticities upto ǫ c ≈ × − (Ushomirsky et al. 2000; Owen 2005).Thus, the tidally-induced crust cracking is expected tooccur at a separation of a crit ≈ (cid:18) m m (cid:19) / ǫ − / , − R ns (3)where ǫ − / , − is the ellipticity in units of 10 − . Applyingthe point-mass limit for a circular binary system (ignor-ing the effects of the finite stellar radii; Peters 1964) onefinds for the duration of the tidal grinding phase prior tothe merger: τ tg ≈ c G a m m ( m + m ) ≈
62 min ǫ − / , − q / q (cid:18) m ns . ⊙ (cid:19) − (cid:18) R ns
10 km (cid:19) (4)recursors of short GRBs 7where we defined the mass ratio q = m /m . Thus fora binary system with the most likely parameters one ex-pects the crust restructuring to set in about an hourahead of the burst (where we have assumed the delaybetween coalescence and burst is negligible). Should thecrust be able to sustain substantially larger deformations,say an order of magnitude more, this duration could bebrought down to minutes. Due to the larger total massthe tidal grinding duration for NS-BH binaries is some-what shorter than the above estimate, but for the lowmass black holes that are most interesting for GRBs, e.g.Rosswog (2005), the difference is just a factor of two.Naively, one would expect the major flaring activ-ity to occur coincident with the first crust cracking(Eq. 4), and just by analogy with magnetar giant flares(Palmer et al. 2005), such precursors should have spec-tral properties similar to the properties of the main burst.The elastic energy stored within the deformed NS crustis ∼ ( σ max /0.1) erg (Thompson 2001), and if thisis the main energy source, the corresponding precursorswould not be visible beyond 40-80 Mpc. A relativistic jet ploughing through a pre-ejected,neutrino-driven baryonic wind
Directly after the merger –but possibly before a rel-ativistic jet can be launched– the remnant of a neu-tron star merger consists of a hot, differentially rotat-ing, super-massive neutron star, surrounded by a massive( ∼ . ⊙ ), thick accretion disk of neutron-rich debris(e.g. Ruffert & Janka 2001; Rosswog et al. 2003). In theinner parts of this disk, at a distance r from the centreof the central object, a nucleon is gravitationally boundwith an energy of E grav ≈
35 MeV (cid:16) M co . ⊙ (cid:17) (cid:0)
100 km r (cid:1) .On the other hand, a large fraction of gravitationalbinding energy of the binary system is released in theform of neutrinos, with average energies of h E ν e i ≈
10 MeV, h E ¯ ν e i ≈
15 MeV and h E ν X i ≈
20 MeV(Ruffert & Janka 2001; Rosswog & Liebend¨orfer 2003)where the index X refers collectively to the heavy leptonneutrinos. It had been realized early on (Ruffert et al.1997; Rosswog & Ramirez-Ruiz 2002) that such a con-figuration should ablate a substantial fraction of the de-bris in a neutrino-driven, baryonic wind. A quantita-tive calculation beyond order of magnitude estimates hasonly recently become possible (Dessart et al. 2009). Thisstudy found that a bi-polar, non-relativistic ( v ∼ . c )wind of ˙ M ∼ − M ⊙ /s heavily pollutes the polar re-gions and prevents the formation of ultra-relativistic out-flow. Thus, for at least as long as the central object hasnot collapsed into a black hole, it seems impossible toproduce a GRB.After a (likely, but not necessarily guaranteed) col-lapse, one is left with the “standard” central engine, ablack hole-disk system. Once this happens, the neutrino-driven mass loss will be seriously reduced and jet forma-tion seems likely. It may be speculated that the emerg-ing relativistic jet has to plough through the pre-ejectedneutrino-driven baryonic cloud, possibly producing a pre-cursor signal. The details of such a jet-cloud interactionare very likely rather involved and a quantitative inves-tigation of this issue is beyond the scope of this paper.We note however that as in this case the precursor marksthe start of the central engine activity, the merging pro- cess could happen significantly before the observed mainburst, stretching the temporal window over which a grav-itational waves signal must be searched (Abbott et al.2010). Constraints on quantum gravity
GRB 090510 is accompanied by high-energy ( > z =0.903; Rau et al. 2009) led to very tightconstraints on the quantum gravity mass M QG , exclud-ing a possible linear energy dependence of the propa-gation speed of light (Abdo et al. 2009). The authorsadopt two different approaches to constrain Lorentz in-variance violation effects: the former, based on themethod outlined in Scargle et al. (2008), derives a limiton the quantum gravity mass of M QG > P l , whereM
P l =1.2 × GeV/c is the Planck mass. This limitremains unchanged by the present findings.As the photon emission time and location are un-known, the latter approach conservatively assumes thatthe observed 31 GeV photon has not been emitted be-fore the onset of the low energy emission. Under thisassumption, the limit on the quantum gravity mass isM QG > P l . In their calculation Abdo et al. (2009)considered the precursor at T -0.5 s, also detected by the Fermi /GBM (see Fig. 1), as the earliest possible emissiontime. The detection of an earlier precursor, presented inthis work, shows that emission started well before theGRB, implying a maximum delay of ∼ QG > P l . CONCLUSION
We carried out a systematic search of precursors onthe sample of short GRBs observed by
Swift . We foundthat ∼ T ≤
13 s). In our sample we found some evidence,though not yet conclusive, that the observed delay canbe as long as ∼
100 s. The spectral properties of theseprecursors do not substantially differ from the promptemission. This result however might be partially a con-sequence of the
Swift /BAT narrow bandpass.We consider it unlikely that the observed precursors areproduced by fireballs becoming optically thin, and argueinstead that the preburst activity in short GRBs is re-lated to their progenitors, i. e. compact objects mergers.We discuss three possible central engine-related mecha-nisms: the interaction of neutron star magnetospheres(Hansen & Lyutikov 2001), which requires one of the twocompact objects to be a magnetar; flares from NS crustcracking, which predicts very long delays between theprecursor and the main burst. Finally, analogously tolong GRB precursors which are associated to the inter-action of the relativistic jet with the stellar envelope, theprecursors in short GRBs might be produced by the jetinteracting with a pre-ejected neutrino-driven baryonicwind. In this last case, the precursor is produced af-ter the merger and marks the start of the central engineactivity. E. Troja et al.We thank G. Skinner and C. Markwardt for discus-sions and useful suggestions on the
Swift /BAT data anal-ysis. This research was supported by an appointment to the NASA Postdoctoral Program at the Goddard SpaceFlight Center, administered by Oak Ridge AssociatedUniversities through a contract with NASA./BAT data anal-ysis. This research was supported by an appointment to the NASA Postdoctoral Program at the Goddard SpaceFlight Center, administered by Oak Ridge AssociatedUniversities through a contract with NASA.