There is a short gamma-ray burst prompt phase at the beginning of each long one

MNRAS 448, 403–416 (2015) doi:10.1093/mnras/stu2664 There is a short gamma-ray burst prompt phase at the beginning of each long one G. Calderone,1‹ G. Ghirlanda,1 G. Ghisellini,1 M. G. Bernardini,1 S. Campana,1 S. Covino,1 D’Avanzo,1 V. D’Elia,2,3 A. Melandri,1 R. Salvaterra,4 B. Sbarufatti1,5 and G. Tagliaferri1 1 INAF–Osservatorio Astronomico di Brera, via E. Bianchi 46, I-23807 Merate (LC), Italy Data Centre, Via del Politecnico snc, I-00133 Rome, Italy 3 INAF–Osservatorio Astronomico di Roma, via Frascati 33, I-00040 Monteporzio Catone (RM), Italy 4 INAF–IASF Milano, via E. Bassini 15, I-20133, Milano, Italy 5 Department of Astronomy and Astrophysics, Pennsylvania State University, University Park, PA 16802, USA 2 ASI–Science ABSTRACT We compare the prompt intrinsic spectral properties of a sample of short gamma-ray bursts (GRBs) with the first 0.3 s (rest frame) of long GRBs observed by Fermi/GBM (Gamma Burst Monitor). We find that short GRBs and the first part of long GRBs lie on the same Ep –Eiso correlation, that is parallel to the relation for the time-averaged spectra of long GRBs. Moreover, they are indistinguishable in the Ep –Liso plane. This suggests that the emission mechanism is the same for short and for the beginning of long events, and both short and long GRBs are very similar phenomena, occurring on different time-scales. If the central engine of a long GRB would stop after ∼0.3 × (1 + z) s, the resulting event would be spectrally indistinguishable from a short GRB. Key words: gamma-ray burst: general. 1 I N T RO D U C T I O N Gamma-ray bursts (GRBs) are transient emission episodes of radiation detected at high energies. The first emission phase, detected at hard X-rays and γ -rays, lasts for ∼0.01 ms–100 s (prompt phase). Then, the bulk of emitted radiation shifts to lower energies and becomes observable at longer wavelengths, from X-rays to radio, with typical duration of ∼days–months (afterglow phase). The observed duration of the prompt phase is characterized by the T90 parameter, i.e. the time interval during which the central 90 per cent of the counts are recorded by the detector. The distribution of T90 of GRBs observed by the Burst And Transient Source Experiment (BATSE) on board the Compton Gamma Ray Observatory (CGRO) has been found to be bimodal with a separation at ∼2 s in the observer frame (Kouveliotou et al. 1993). According to this finding, GRBs are classified either as short gamma-ray burst (SGRB) if T90 < 2, or as long ones (LGRB) if T90 > 2 s (but see Bromberg et al. 2013). Besides, the prompt phase of SGRBs is characterized by harder spectra (Kouveliotou et al. 1993) and smaller spectral lags between different energy bands (Norris, Marani & Bonnell 2000) with respect to the prompt phase of LGRBs.  E-mail: For bursts with reliable redshift estimates, it has been shown that SGRBs are systematically less energetic than LGRBs, with total X-ray- and γ -ray-emitted energies smaller by a factor ∼10– 100 (Ghirlanda et al. 2009). Also, the afterglows of SGRBs, when detected, are correspondingly dimmer than those of LGRBs, but similar in other respects (Gehrels et al. 2008; Margutti et al. 2013; D’Avanzo et al. 2014). Finally, several nearby (z < 0.5) LGRBs have been associated with explosions of core-collapse supernovae (Hjorth & Bloom 2012), while there is no similar evidence for short bursts (Berger 2013). These findings suggest that SGRBs and LGRBs might originate from different progenitors (Mészáros 2006; Berger 2013). Observationally, the most important difference between SGRBs and LGRBs is their T90 duration. A first attempt to compare the spectral properties of SGRBs and LGRBs detected by CGRO/BATSE showed that (i) the difference in hardness could be due to a harder low energy spectral index of SGRBs rather than a harder peak energy and (ii) that the spectra of SGRBs and the first 1–2 s of LGRBs appear similar (Ghirlanda, Ghisellini & Celotti 2004). These results suggested that the engine might be similar in the two classes, but the activity would last longer in the case of LGRBs (Guiriec et al. 2010). Also, Nakar & Piran (2002) found that the ratio of the shortest pulse duration to the total burst duration for both SGRBs and the first 1–2 s of LGRBs were comparable.  C 2015 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society Accepted 2014 December 13. Received 2014 November 27; in original form 2014 July 23 404 G. Calderone et al. MNRAS 448, 403–416 (2015) well-defined, samples of SGRBs with measured redshifts (e.g. D’Avanzo et al. 2014) allowed us to compare the energetic properties of short and long events in their rest frame. The aim of this work is to further explore the similarities between SGRBs and LGRBs by comparing their intrinsic (i.e. rest-frame) spectral properties estimated on the same rest-frame time-scales. The average T90 /(1 + z) duration of the SGRBs with reliable (spectroscopic) redshifts and without X-ray extended emission in the D’Avanzo et al. (2014) sample is 0.3 s (10 bursts). This will be our reference time-scale to perform spectral analysis of the first part of LGRBs, and compare the results with those of SGRBs. Throughout the paper, we assume a  cold dark matter cosmology with H0 = 71 km s−1 Mpc−1 , m = 0.27,  = 0.73. 2 THE SAMPLE Since we aim to study the prompt emission spectral properties and energetic/luminosity of GRBs, we need a broad energy coverage in order to determine where the peak energy is. While Swift/ Burst Alert Telescope (BAT) has a limited energy range (15–150keV) which is not suited for GRB prompt emission spectral characterization, the GBM instrument on board Fermi covers almost two orders of magnitude in energy with the NaI detectors (8 keV–1 MeV) and can extend this energy range to a few tens of MeV with the inclusion of the data of the BGO detectors. Hence, we selected all GRBs observed by Fermi/GBM up to 2013 December with a redshift estimate. This amounts to 64 LGRBs and 7 SGRBs. Among the long ones we discarded: 2 GRBs with missing response matrix files; 2 GRBs observed with a non-standard low-level threshold;1 3 GRBs whose first part was missed by the GBM; 12 GRBs for which we could not constrain either the low energy spectral index or the peak energy (Section 3). The final LGRB sample comprises 45 long bursts. Fermi/GBM observed seven SGRBs with known redshift. To this sample, we added the SGRB flux-limited sample of 12 sources with redshift discussed in D’Avanzo et al. (2014, hereafter D14 sample), but discarded: GRB 080905A since its redshift is likely not accurate, GRB 090426 and GRB 100816A since their classification as SGRB is debated. Four GRBs in the D14 sample were also in the GBM sample: for these bursts we considered the results reported in D14. The final SGRB sample comprises three GRB observed with Fermi/GBM and nine from D14. The SGRB sample, although relatively small, stems from a fluxlimited s (...truncated)