A multiwavelength analysis of a collection of short-duration GRBs observed between 2012 and 2015

Abstract

1 INTRODUCTION

Short-duration gamma-ray bursts (sGRBs) were originally classified using the Konus catalogue (Mazets et al. 1981) that preceded the wider realization that sGRBs likely are binary compact mergers (Narayan, Paczynski & Piran 1992; Nakar 2007) based on various observed properties like duration, fluence etc. as described in Kouveliotou et al. (1993) and Bromberg et al. (2013). During the era of the Neil Gehrels Swift observatory, arcsec X-ray Telescope (XRT) localizations enabled the discovery of the first afterglow of sGRB 050509B (Castro-Tirado et al. 2005; Gehrels et al. 2005) and subsequently other observed features like extended emission (EE) at Swift Burst Alert Telescope (BAT) energies, temporally extended variable X-ray emission suggesting late time central engine activity due to either merger of two neutron stars (NS–NS) or a neutron star and a stellar mass black hole (NS–BH) as possible progenitors (Eichler et al. 1989; Narayan, Paczynski & Piran 1992; Usov 1992; Zhang & Meszaros 2001; Troja et al. 2007; Rowlinson, et al. 2013; D’Avanzo et al. 2014; Gibson et al. 2017; Desai, Metzger & Foucart 2018). The physical nature of the EE, observed in some of the sGRBs, is not yet resolved. It could be connected with the beginning of the afterglow phase (Minaev, Pozanenko & Loznikov 2010), the activity of a magnetar, formed during merger process (Metzger, Piro & Quataert 2008) or viewing angle effects (Barkov & Pozanenko 2011). The prompt emission properties of sGRBs, such as relatively harder spectra (higher E_peak) and nearly zero spectral lag (Gehrels et al. 2006; Zhang et al. 2009), discriminate sGRBs from long GRBs (lGRBs). sGRBs have also been speculated as a potential key to understand gravitational wave (GW) sources and the nucleosynthesis of elements over the history of the Universe (Berger 2014; Kumar & Zhang 2015; Abbott et al. 2017a,b).

More than 90 afterglows of sGRBs have been detected at various wavelengths¹ exhibiting diverse properties (Lee & Ramirez-Ruiz 2007; Gehrels, Ramirez-Ruiz & Fox 2009; Berger 2014). Afterglows of sGRBs are in general less luminous and less energetic and favour typically lower circumburst densities than those seen in the case of lGRBs (Kann, Klose & Zhang 2011; Nicuesa et al. 2012; Berger 2014). Despite intensive efforts, this leads to a lower detection rate for sGRBs: ∼75 per cent in X-rays, ∼33 per cent in optica-NIR, and only a handful in the radio (Berger 2014). In comparison to long ones, sGRBs are observed to occur at over a lower and narrower redshift range (z ∼ 0.1–1.5) and both early and late-type galaxies have been identified as hosts (Fong et al. 2013). Afterglow observations of sGRBs also indicate that these bursts have a range of jet-opening angles (Burrows et al. 2006; Kann et al. 2011; Nicuesa Guelbenzu et al. 2012; Fong et al. 2013; Zhang et al. 2015; Troja et al. 2016; Lamb & Kobayashi 2018; Margutti et al. 2018) and have systematically larger radial offsets from the host galaxies (Fong et al. 2013; Tunnicliffe et al. 2014) in turn supporting compact binary merger as possible progenitors (Bloom, Kulkarni & Djorgovski 2002; Zhang, Liang & Zhang 2007; Troja et al. 2008; Zhang et al. 2009; Salvaterra et al. 2010). Optical afterglows of sGRBs are generally fainter in comparison to those observed in the case of lGRBs, implying the need for fast and deep afterglow observations using moderate- to large-size telescopes.

Study of sGRBs now extends beyond understanding just about their explosion mechanisms, progenitors and environments. These explosions are now key to improve our understanding about multimessenger astronomy and to search for new compact binary mergers as GW sources. It has been proposed that during the compact binary merger process, radioactive decay of heavy elements could give rise to a supernova-like feature, termed ‘macronovae’ or ‘kilonovae’ (Li & Paczynski 1998; Kulkarni 2005; Hotokezaka et al. 2013; Kasen, Fernandez & Metzger 2015) having a component of thermal emission caused by radioactive decay of elements through r-process nucleosynthesis. So far, tentative ‘kilonova’ like signatures have been identified in only a few cases including sGRB 050709 (Zhi-Ping et al. 2016), sGRB 060614 (Yang et al. 2015), sGRB 080503A (Perley et al. 2009), sGRB 130603B (Hotokezaka et al. 2013; Tanvir et al. 2013), sGRB 150101B (Fong et al. 2016; Troja et al. 2018), sGRB 160821B (Kasliwal et al. 2017), and recently sGRB 170817A/GW170817/AT 2017gfo (Abbott et al. 2017a,b). Discovery of the ground-breaking event called sGRB 170817A/GW170817/AT 2017gfo has opened new windows in the understanding of GWs: their electromagnetic counterparts (Abbott et al. 2017a; Albert et al. 2017), and their likely contribution to heavy element nucleosynthesis in the nearby Universe (Lattimer & Schramm 1974; Piran, Nakar & Rosswog 2013; Pian et al. 2017).

Multiwavelength observations of a larger sample of nearby sGRBs and ‘kilonovae’ features like GW170817/sGRB 170817A/AT 2017gfo are crucial to establish whether compact binary mergers are the progenitors (Kasen et al. 2015) for all such events (Abbott et al. 2017a,b) and to put a constraint on the electromagnetic counterparts and number density of GW sources in near future (Li & Paczynski 1998; Shibata & Taniguchi 2011; Loeb 2016).

In this paper, we present results based on prompt emission data from INTEGRAL, Swift, Fermi, and multiwavelength follow-up afterglow observations of nine sGRBs. The data sets were mostly not published yet and were observed by various different-size optical and NIR ground-based telescopes including the 10.4 m Gran Canarias Telescope (GTC). Observations of these nine bursts including sGRB 170817A were collected during 2012–2018 as a part of a large multiwavelength collaboration. Our analysis of new data for the subset of sGRBs mainly focused towards constraining prompt emission, afterglow, and host galaxy properties and adding value towards known physics behind these cosmic explosions. We also attempt to compare the observed properties of the subset of sGRBs with a new class of less-studied but associated events called ‘Kilonovae’. The paper is organized as follows: in Sections 2 and 3 we present our own temporal and spectral analysis of the afterglow and host galaxy data of GRB 130603B alongside the published ones; in Section 4 and in Appendix ‘A’ we discuss the results of prompt emission and multiband afterglow observations of the other eight sGRBs, and in Section 5 we present late-time GTC observations of sGRB 170817A/GW170817/AT 2017gfo and compare the observed properties with the subset of the bursts presently discussed. Finally, in Section 6 we summarize the conclusions drawn from the analysis of all the sGRBs. In this paper, the notation F_ν(t) ∝ t^−αν^−β is used, where α is the flux temporal decay index and β is the spectral index. Throughout the paper, we use the standard cosmological parameters, |$H_0=71\,\rm km\,s^{-1}Mpc^{-1}$|⁠, Ω_M = 0.27, Ω_Λ = 0.73.

2 SGRB 130603B, MULTIWAVELENGTH OBSERVATIONS

sGRB 130603B was discovered on 2013 June 3 at 15:49:14 ut by Swift-BAT (Barthelmy et al. 2013; Melandri et al. 2013), and by Konus − Wind (Golenetskii et al. 2013). The γ-ray light curve of GRB 130603B consists of a single group of pulses with a duration of T₉₀ = 0.18 ± 0.02 s (15–350 keV; Barthelmy et al. 2013). The Konus − Wind fluence of the burst is (6.6|$\pm 0.7)\times 10^{-6}\rm \,erg \, cm^{-2}\,$| (20–10⁴ keV), with a peak energy of 660 ± 100 keV (Golenetskii et al. 2013). The reported measured value of E_{iso, γ} ∼ 2.1 × 10⁵¹ erg, places the burst well above the E_peak-E_iso locus for long GRBs in the Amati diagram (Amati et al. 2008, also fig. 6). Such behaviour is often observed for short bursts (Minaev & Pozanenko 2019).

sGRB 130603B shows negligible spectral lag (Norris et al. 2013), typical for short bursts. Many authors (e.g. Hakkila & Preece, 2014; Minaev et al. 2014) have found a strong correlation between pulse duration and spectral lag: longer pulses have larger lags. The correlation is similar for both sGRBs and lGRBs. As sGRBs typically consist of shorter pulses than long ones, they have less significant spectral lags in general. GRB light curves often consist of several pulses including highly overlapping ones: spectral and temporal properties of individual pulses may be not adequately resolved (Chernenko 2011). By performing spectral lag analysis via the superposition of several overlapping pulses, one can obtain an unpredictable result because each pulse has unique spectral and temporal properties (Minaev et al. 2014). As a result, one can find negligible or negative lag under certain conditions even if each pulse has a positive (but unique) lag (for details, see Minaev et al. 2014). sGRB 130603B consists of several very short and overlapped pulses, so its negligible spectral lag may be connected with short duration of pulses while performing spectral lag analysis for superposition of several pulses.

2.1 SPI-ACS INTEGRAL observations

sGRB 130603B was also triggered by the INTEGRAL Burst Alert System (IBAS) system operating with spectrometer for INTEGRAL- anticoincidence system (SPI-ACS) (Fig. 1). SPI-ACS INTEGRAL has very high effective area (up to 0.3 m²) in energy range > 100 keV and stable background at time-scales of hundreds of seconds (Minaev et al. 2010), which makes SPI-ACS a suitable instrument to study light curves of short-hard GRBs and especially to search for weak signals from their precursors and EE components. The off-axis angle of sGRB 130603B to the SPI-ACS axis is 103 deg, which is almost optimal for detection, making sGRB 130603B one of the brightest short bursts ever registered by SPI-ACS. Nevertheless, we do not find statistically significant EE in the SPI-ACS data (Inset in Fig. 1, in terms of peak flux at 50 ms time scale), which is in agreement with results obtained from Swift-BAT in the softer energy range of 15–150 keV (Norris et al. 2013). There is also no evidence for a precursor in SPI-ACS data during time-scales from 0.01s up to 5s, in agreement with the previous results (Troja, Rosswog & Gehrels 2010; Minaev & Pozanenko 2017; Minaev, Pozanenko & Molkov 2018).

Figure 1.

Background subtracted light curve of sGRB 130603B of INTEGRAL SPI-ACS in the energy range 0.1–10 MeV with 50 ms time resolution. The x-axis shows time since BAT trigger. Inset: light curve with time resolution of 100 s.

Open in new tabDownload slide

In Vigano & Mereghetti (2009), it was shown that one SPI-ACS count corresponds on average to ∼10⁻¹⁰ erg cm⁻² in the (75, 1000) keV range, for directions orthogonal to the satellite pointing axis. Using the conversion factor, we can roughly estimate the flux values in the (75, 1000) keV range for GRBs observed by SPI-ACS. The fluence estimation of sGRB 130603B in SPI-ACS is ∼31 000 counts or S_EE ∼ 3.1 × 10⁻⁶ erg cm⁻² in the (75, 1000) keV range, which is in agreement with Konus–Wind observations (Golenetskii et al. 2013). At a time-scale of 50s, the upper limit on EE activity for sGRB 130603B is ∼7100 counts (S_EE ∼ 7 × 10⁻⁷ erg cm⁻²) at the 3σ significance level, the corresponding upper limit on precursor activity at a time-scale of 1s, is ∼1000 counts (S_EE ∼ 10⁻⁷ erg cm⁻²), both are in the (75, 1000) keV range.

2.2 Optical-IR photometric observations

As a part of this collaboration, photometric observations of the optical-IR afterglow and the host galaxy were performed using several facilities worldwide, including 1.0 m telescope at the Tubitak National Observatory (Antalya, Turkey); the 1.5 m telescope at Observatorio de Sierra Nevada (Granada, Spain); the AS-32 0.7 m telescope at Abastumani Astrophysical Observatory Georgia; the Reionization And Transients Infra-Red RATIR camera at the 1.5 m telescope of the San Pedro Martir observatory; the 2.0 m Liverpool telescope at La Palma; AZT-22 1.5 m at the Maidanak observatory Uzbekistan; the Centro Astronómico Hispano-Alemán (CAHA) 3.5 m located in Almeria (Spain); the newly commissioned 3.6 m Devasthal Optical Telescope (DOT) at Aryabhatta Research Institute of Observational Sciences (ARIES) Nainital, India, and with the 10.4 m Gran Telescopio Canarias (GTC), located at the observatory of Roque de los Muchachos in La Palma (Canary Islands, Spain), equipped with the Optical System for Imaging and low-Intermediate-Resolution Integrated Spectroscopy (OSIRIS) instrument. Our observations by the 1.0 m telescope at the Tubitak, starting ∼0.122d after the burst are the earliest reported ground-based observations so far for sGRB 130603B. All optical-NIR data were processed using DAOPHOT software of NOAO’s iraf package,² a general purpose software system for the reduction and analysis of astronomical data. The photometry was performed in comparison to nearby standard stars and image subtraction was applied whenever it was required to subtract the host galaxy contribution as exaplained in Alard & Lupton (1998). The unfiltered observations made with the AbAO AS-32 telescope have been considered equivalent to r-band as the quantum efficiency of the detector is at a maximum around r-band frequencies. The final AB magnitudes of the afterglow and the host galaxy in different pass-bands as a part of the present analysis are listed in Table 1.

2.3 Spectroscopic observations

A spectroscopic redshift at the location of the afterglow was obtained by several groups, including Xu et al. (2013), Foley et al. (2013), de Ugarte Postigo et al. (2013), and Cucchiara, Perley & Cenko (2013). As a part of the present study, spectroscopic observations were performed to measure the redshift of sGRB 130603B independently and are reported in Sánchez-Ramírez et al. (2013).

We obtained optical spectra with the GTC(+ OSIRIS) starting at 23:58 h. Observations consisted of two 450 s exposures, one with each of the R1000B and R500R grisms, using a slit of width 1.2 arcsec. Data reduction was performed using standard routines from the Image Reduction and Analysis Facility (iraf). The afterglow spectrum shows Ca II in absorption, and we detect a significant contribution from the underlying host galaxy (e.g. [O ii], [O iii], Hβ and Hα emission lines about 1’ offset), together implying a redshift of z = 0.356 ± 0.002, consistent with the values provided by de Ugarte Postigo et al. (2013) and Foley et al. (2013). The reduced spectrum obtained at the location of the afterglow along with the lines identified are shown in Fig. 2. Using our redshift value and the fluence published by Golenetskii et al. (2013), the isotropic-equivalent gamma-ray energy is E_{iso, γ} ∼ 2.1 × 10⁵¹ erg (20– 10⁴ keV, rest-frame).

Figure 2.

Spectroscopic observations of the sGRB 130603B at the location of the afterglow taken by the 10.4 m GTC (+ OSIRIS) using grisms R1000B and R500R starting ∼8 h after the burst (Sánchez-Ramírez et al. 2013). Telluric absorption bands are marked as cyan.

Open in new tabDownload slide

2.4 mm-wavelength observations

The afterglow of sGRB 130603B was observed with the Plateau de Bure Interferometer (Guilloteau et al. 1992), one of the largest observatories in the Northern Hemisphere operating at millimetre wavelengths (1, 2, and 3 mm). Observations were performed in a four-antenna extended configuration for the first epoch whereas a five-antenna configuration on the consecutive dates as listed in Table 2. The data reduction was done with the standard clic and mapping software distributed by the Grenoble GILDAS group. Flux calibration includes a correction for atmospheric decorrelation that has been determined with a UV plane point source fit to the phase calibration quasar 1156 + 295. The carbon star MWC349 was used as the primary flux calibrator due to its well-known millimetre spectral properties (see e.g. Schwarz 1978). The burst location was also followed up using the RT-22 radio telescope of CrAO (Crimea) at 36 GHz and the data reduced using the standard software routines (Villata et al. 2006) and used modulated radiometers in combination with the registration regime ‘ON–ON’ for collecting data from the telescope (Nesterov, Volvach & Strepka 2000). The upper limits based on these observations are also given in Table 2. As a part of the present analysis, upper limits (1σ) based on IRAM Plateau de Bure Interferometer observations of sGRB 140606A, sGRB 140622A, and sGRB 140903A using the carbon star MWC349 as the primary flux calibrator are also tabulated in Table 2.

Observations at mm-wavelengths are very important as they suffer negligible absorption or interstellar scintillation effects, so sGRBs at high redshifts or highly extinguished bursts could be observed. It is expected that emission at mm-wavelengths is normally above the self-absorption frequency and lies around peak of the GRB synchrotron spectrum, allowing to probe for possible reverse shock emission at early epochs and to constrain afterglow models observed recently in case of many lGRBs (de Ugarte Postigo et al. 2012; Perley et al. 2014).

In Fig. 3, observed mm-wavelength upper limits of four sGRBs presented in Table 2 were plotted along with previous observations of another five sGRBs (namely sGRB 020531, sGRB 050509B, sGRB 051105A, sGRB 060801, and sGRB 080426, data taken from Castro-Tirado et al. 2019) and were compared with the afterglow light curve of a well-known nearby and bright lGRB 130427A observed at 3 mm (Perley et al. 2014). It is clear from Fig. 3 that using PdBI, we have been able to observe nine sGRBs so far but none was detected at mm-wavelengths in contrast with lGRBs that have been detected in many cases constraining various physical models (de Ugarte Postigo et al. 2012; Perley et al. 2014). Out of these nine sGRBs, only sGRB 130603B (Fong et al. 2014a) and sGRB 140903A (Troja et al. 2016) were detected at VLA radio frequencies so far. However, as discussed further in this work, the observed 3-mm PdBI 1σ upper limits for these two bursts are consistent with those predicted by the forward shock afterglow models. The gamma-ray fluence and observed X-ray flux values for these nine sGRBs are similar to those observed in case of other sGRBs. Non-detections of these nine sGRBs at 3 mm in the last decade using PdBI and other mm-wavelength facilities globally are helpful to constrain underlying physics behind these energetic sources and demand for more sensitive and deeper follow-up observations.

Figure 3.

Comparison of the 3-mm afterglow light curve of nearby lGRB 130427A (Perley et al. 2014) to the present set of mm-wavelength upper limits (1σ) of four sGRBs (from Table 2) along with another set of upper limits of 5 sGRBs as discussed in Castro-Tirado et al. (2019) placed at a common redhsift of z = 0.34.

Open in new tabDownload slide

3 PROPERTIES OF SGRB 130603B

3.1 Afterglow light curves and comparison to models

Fig. 4 shows the r and i pass-band light curves of the sGRB 130603B afterglow including data from the present analysis and those published in the literature (Berger, Fong & Chornock 2013; Cucchiara et al. 2013a; Tanvir et al. 2013; de Ugarte Postigo, Thone & Rowlinson 2014). To plot the light curves along with those published in the literature, the data were scaled to respective AB magnitudes in SDSS r and i bands (see Fig. 4). The R_c band data taken at ∼0.122d comprise of the earliest reported ground-based detection and the remaining data fill the temporal gap in the light curve for this interesting short-duration burst. From the present analysis, the number of new data points in both r and i bands are four each spread up to ∼2.3d post-burst. Careful image-subtraction and calibration of the afterglow data < 0.23d post-burst indicates possible deviations from smooth power-law behaviour during the first few hours.

Figure 4.

Afterglow optical r (pink) and i (blue) pass-band light curves of the sGRB 130603B. The solid red curves are the best-fit broken power-law model to the r-band light curves as described above. The red dashed line is the model over-plotted on the i-band light curve to guide eyes. The green triangle in the right bottom corner is the single point detection of the underlying ‘kilonova’ detection as described in Tanvir et al. (2013). The green dashed lines are the H-band ‘kilonova’ models at the redshift of ∼0.36 as taken from Tanaka et al. (2014). The black triangles are the H-band light curve (at redshift z = 0.36) of the electromagnetic counterpart of the recently discovered GW170817 (sGRB 170817A/AT 2017gfo) for comparisons as compiled in Villar et al. (2017a).

Open in new tabDownload slide

To determine the temporal flux decay slopes and the break time, we fitted an empirical function representing a broken power law, |$F_\nu = A[(t/t_b)^{s\alpha _1} +(t/t_b)^{s\alpha _2}]^{-1/s}$| (Beuermann et al. 1999) to the r band combined light curve. The quantities α₁ and α₂ are asymptotic power-law flux decay slopes at early and late times with α₁ < α₂. The parameter s > 0 controls the sharpness of the break and t_b is the break time. The best fit of this broken power-law function to the r-band data including the very first data point taken at ∼0.122d gives α₁ = 0.81 ± 0.14; α₂ = 2.75 ± 0.28 and t_b = 0.41 ± 0.04 with |$\tilde{\chi }^2/dof = 2.22$| for a value of the smoothing parameter s = 4. The values of t_b and α₂ are similar to those derived by Fong et al. (2014a). Although the data from Swift XRT is consistent with a break occurring around 0.3 days, the later XMM-Newton observations suggest no turnover at X-ray frequencies and a continuing power law instead (this ‘X-ray excess’ is also discussed by Fong et al. (2014a)). The present analysis also helped to constrain the value of α₁ using a single-band light curve and found to be shallower in comparison to that derived by Fong et al. (2014a).

The present data set has also been used to constrain the spectral energy distribution (SED) of the afterglow. The RATIR data taken simultaneously at ∼0.52d post-burst (see Table 1) require an optical-NIR spectral index β_opt ∼ 0.7 once corrected for Galactic and considerable host extinction, similar to those measured by de Ugarte Postigo et al. (2014) at ∼0.35d and by Fong et al. (2014a) at ∼0.6d post-burst. The optical-NIR spectral index, together with the published value of the XRT spectral index β_X = 1.2 ± 0.1, is consistent with Δβ = β_X − β_opt = 0.5, as expected in the case of a slow-cooling synchrotron spectrum (Sari, Piran & Narayan 1998) where the optical and XRT frequencies lie in two different spectral regimes.

Additionally, the derived values of the temporal slope α₁ and the spectral slope β_opt above are consistent with the closure relation β = 3α/2 in the case of adiabatic deceleration in the interstellar medium ISM afterglow model for the spectral regime ν_m < ν < ν_c, where ν_m is the break frequency corresponding to the minimum electron energy and ν_c is the cooling break frequency. The temporal flux decay index α₂ = 2.75 ± 0.28, the break-time t_b = 0.41 ± 0.04 and estimated slopes of the SEDs using the optical-NIR and XRT data are broadly consistent with the scenario described by Rhoads (1999) where the edge of the relativistic outflow causes a steepening (jet-break) in the observed light curve by t^−p (Sari, Piran & Halpern 1999), where p is the electron energy index. Also, for the observed XRT frequencies that lie above ν_c, the temporal and spectral indices are consistent with the predictions made by the ISM model in case of the adiabatic deceleration for the data up to 1 day post-burst (de Ugarte Postigo et al. 2014; Fong et al. 2014a).

Present afterglow data have made it possible to construct a single-band afterglow light curve and do the temporal fitting to derive parameters like temporal indices and jet-break time. The optical afterglow data in r and i bands from the present analysis have allowed us to construct a better-sampled light curve of the sGRB 130603B and to constrain the value of the pre-jet-break temporal decay index α₁ for the first time using data from a single band. This overall analysis supports the scenario that the observed steepening in the optical light curves is a jet-break as predicted theoretically by Sari et al. (1999) and Rhoads (1999). However, the observed X-ray excess emission (Fong et al. 2014a) for epochs > 1d is not supported by the afterglow model.

3.2 Afterglow SED at the epoch of mm observations

Based on the present analysis and using the afterglow data in X-ray, r, i bands and the results published by de Ugarte Postigo et al. (2014) and Fong et al. (2014a), an afterglow SED was constructed for the epoch of our earliest millimeter observations i.e. 0.22d after the burst (see Fig. 5). We first built a time-sliced X-ray spectrum from the Leicester XRT webpages,³ extracting data in the range 10ks–18ks after the trigger. This tool provides the appropriate spectral and response files that are compatible for use with the spectral fitting package xspec. The source spectral file was normalized so that it has the same count rate as a single epoch spectrum measured at 0.22d (see Schady et al. 2010 for details). For the optical data, we created appropriate spectral and response files for each filter. The flux values at 0.22d for each spectral file were determined from an extrapolation/interpolation the data between 10ks and 30ks by fitting a power law and fixing the slope as 0.81. This is the decay index found for the first segment of the broken power-law fit to the r-band data. The optical errors were estimated by taking the average error of the data between 10 and 30ks and adding a 5 per cent systematic error in quadrature.

Figure 5.

X-ray and optical SED of sGRB 130603B at the epoch of first millimetre observations i.e. 0.22d after the burst. We plot the best-fitting absorption and extinction corrected spectral model (solid red lines, broken power-law model), as well as the host galaxy absorbed and extinguished spectral model (orange dash lines) and the data (black circles) using the method described in Schady et al. (2010).

Open in new tabDownload slide

The SEDs were fitted using XSPEC, following the procedure outlined in Schady et al. (2010, 2007). We fit two different models, a power law and broken power law, which include Galactic and host galaxy absorption and extinction components (phabs, zphabs, and zdust). The best-fitting results obtained using the procedure mentioned above are plotted in Fig. 5, which supports broken power-law model for Milky Way (MW) type of host extinction. Values of the best-fitting borken power-law model and MW type of host extinction are consistent with those derived by de Ugarte Postigo et al. (2014). Assuming ν_m around mm-wavelengths, 86.7 GHz upper limits of the sGRB 130603B at 0.22d post-burst (see Table 2) are also consistent with the extrapolated modeled flux values.

3.3 Broad-band modeling of sGRB 130603B afterglow

The multiband afterglow data of sGRB 130603B discussed above along with those published in Fong et al. (2014a) were used to fit numerical-simulation-based model to constrain physical parameters of the jetted emission as described in Zhang et al. (2015). The numerical modeling (Zhang et al. 2015) calculates the flux density at any frequency and observer time. The Monte Carlo method is used to determine the best parameter values (i.e. with the smallest χ² value) utilizing the MultiNest algorithm from Feroz, Hobson & Bridges (2009). The optical-NIR data were corrected for the Galactic and host extinction values as constrained in Fong et al. (2014a). The XRT data were also corrected for absorption effects. Based on the literature, it was decided to utilize the data 1000s after the burst for the modeling to avoid possible prompt emission effects at early epochs as described in Zhang et al. (2015).

Using the model and initial guess values, following set of parameter values were determined: the blast wave total energy E_{iso, 53} (in the units of 10⁵³ ergs), the ambient number density n, the electron energy density fraction ϵ_e , the magnetic field energy density fraction ϵ_B, the electron energy index p, and values of jet opening angle θ_jet and the observed angle θ_obs. The best-fitting light curves obtained at different wavelengths are plotted in Fig. 6, the Monte Carlo parameter distributions are plotted in Fig. 7, and the resulting best-fit parameters and their uncertainties are listed in Table 3. A cross-check using an updated version of the scalefit package (van Eerten & MacFadyen 2012; Ryan et al. 2015) produces a similar jet opening angle and inferred energy.

Figure 6.

The best-fitting modeled multiband light curves determined from the numerical simulations as described above (Zhang et al. 2015). The corresponding frequency is marked on the right corner in each panel in the unit of Hz. The x-axis is the time since trigger in units of seconds. The observed flux density of each instrument is on the y-axis in units of mJy. All data were corrected for MW and host galaxy absorption and extinction effects before modeling. Red solid lines represent the modeled light curves.

Open in new tabDownload slide

Figure 7.

Triangle plot of the Monte Carlo fitting to our simulation-based model as described above (Zhang et al. 2015). It shows the posterior distribution and the correlation between the parameters.

Open in new tabDownload slide

Using the new data set discussed in this work, derived values of the physical parameters using present modeling method (Table 3) are constrained better than those reported by Fong et al. (2014a). The derived value of observed jet opening angle, θ_obs, is ∼3.2 deg. This value of θ_jet gives rise to the beaming corrected E_{iso, 53} is ∼1.4 × 10⁴⁹ erg. It is also clear from the present modeling that the best-fitting model was unable to reproduce the very late time X-ray emission observed in case of sGRB 130603B as noticed by using Chandra observations (Fong et al. 2014a). It is also noted that values of the isotropic-equivalent gamma-ray energy is E_iso and the blast wave energy E_{iso, γ} are comparable, which in turn indicates the GRB radiative efficiency η_γ to be ∼23 per cent (with an uncertainty of ∼4 per cent), one of the highest among the known sample of sGRBs (Lloyd-Ronning & Zhang 2004; Wang et al. 2015).

3.4 sGRB 130603B and ‘kilonovae’ connection

The ‘kilonova’ or ‘macronova’ events are electromagnetic transients powered by the radioactive decay of r-process elements synthesized in dynamical ejecta, and in the accretion disc winds during compact binary mergers where at least one component is a neutron star (Li & Paczynski 1998; Kulkarni 2005; Rosswog 2005). Compact binary mergers are also expected to be sources of GWs (Metgzer & Berger 2012; Nissanke, Mansi & Georgieva 2013; Tanaka & Hotokezaka 2013; Seigel & Ciolfi 2016a; Abbott et al. 2017a,b). For ‘kilonovae’, ejection of radioactive material during the merging process of the compact binaries could lead to an excess emission at optical-infrared or ultra-voilet frequencies. The brightness, duration, and spectrum of such emission is a function of the opacity, velocity, ejecta mass, and viewing angle (Metgzer et al. 2010; Bernes & Kasen 2013; Piran et al. 2013; Rosswog et al. 2014; Tanaka et al. 2014; Mooley et al. 2018; Radice et al. 2018). In turn, the opacity depends crucially on the neutron richness of the ejecta, which determines how far any r-process nucleosynthesis proceeds. The high-masst lanthanides, in particular, create heavy line-blanketing tha is expected to largely block out light in the optical bands. Recently, hydrodynamical modeling of such processes (Metzger & Fernandez 2014; Kasen et al. 2015) has predicted a brief early blue emission component produced in the outer lanthanide-free ejecta and a rather longer infrared transient produced in the inner lanthanide-blanketed regions at later epochs (Bulla et al. 2019). Using their disc-wind model for a case with a non-spinning black hole (Kasen et al. 2015), the optical bump observed in the case of sGRB 080503 (Perley et al. 2009) was interpreted in terms of an underlying ‘kilonova’ emission for an assumed redshift of z = 0.25. Their (Kasen et al. 2015) models were, however, unable to explain the observed infrared excess in sGRB 130603B, which required higher accretion disc mass and perhaps a rapidly spinning black hole (Fan, Yu & Xu 2013; Tanaka et al. 2014; Just et al. 2015). In this section, we attempt to place some constraints on the possible blue component of associated ‘kilonova’ based on the observed prompt emission and afterglow observations in bluer wavelengths for sGRB 130603B and their comparison with theoretical models.

It has been proposed by Barkov & Pozanenko (2011) that one should observe extended prompt emission in the case of sGRBs initiating Blandford–Znajek (BZ) jets (Blandford & Znajek 1977) due to large accretion disc mass and high accretion rate. However, in the case of sGRB 130603B EE was not detected (see Section 2 and Fig. 1). The absence of observable EE may indicate either that the observer is located off-axis with respect to the narrow BZ-jet, or that the accretion disc mass is small. In general, accretion disc mass should correlate with the ejected mass and the presence of EE could be an indicator of the emerging ‘kilonovae’ in sGRBs. Indeed, the plateau phase in X-ray emission observed in sGRB 130603B cannot be explained by a BZ-jet model (Kisaka & Ioka 2015) if we assume a small accretion disc mass. The absence of the EE and the presence of a plateau phase could be explained by a low accretion rate that has still initiated BZ jet but with moderate bulk relativistic gamma-factor. Alternatively, the magnetar model could explain the plateau phase of sGRB 130603B and ‘kilonovae’ features (Fan et al. 2013; Metzger & Piro 2014). Observing EE during the burst phase, along with the presence/absence of an early time X-ray plateau during afterglow phase for a larger sample of sGRBs, would allow discriminating among the possible progenitors as a subclass of compact-binary mergers producing magnetars (Zhang & Meszaros 2011; Rowlinson et al. 2013; Siegel & Ciolfi 2016a,b) but would also allow predicting some of them as potential candidates like GW170817.

In addition to the analysis described above, using published early-time afterglow data of sGRB 130603B in Swift-UVOT u and Gemini g′ bands around ∼1.5d post-burst (de Ugarte Postigo et al. 2014), we attempt to constrain the possible early time blue emission contributing to the underlying ‘kilonova’. The observed limiting magnitude in u > 22.3 mag and g′ > 25.7 mag place limits on the corresponding luminosities of L_u < 3.5 × 10²⁷ erg s⁻¹ Hz⁻¹ and |$L_{g^{\prime }} \lt 0.3\times 10^{27}$| erg s⁻¹ Hz⁻¹, respectively. Using the transformation equations (2) and (3) given in Tanaka (2016) (also see equations 7 and 8 in Fernandez & Metzger 2016), we tried to constrain the parameter called ejected mass M_ej. However, these limiting values of luminosities in the two bands are not sufficiently deep to constrain values of the ejected mass meaningfully (>1.5 M_⊙) for the bluer component of ‘kilonova’ at the given epoch for the assumed values of the standard parameters. Considering the WIND models of ‘kilonovae’ with rather lower opacity and expansion velocities (Metzger & Fernandez 2014; Kasen et al. 2015; Tanaka 2016), constraints for the ejected mass M_ej are even weaker i.e. M_ej > a few M_⊙, which is unphysical. We caution that the placed limits on M_ej could be shallower if there were some contribution from the afterglow at the epoch of observations, which is certainly plausible. It is also worth mentioning that the some of parameters in the ‘kilonovae’ models like the range spin of the neutron star, f-parameter, neutron richness have not been well-constrained so far (Metgzer et al. 2010; Kasen et al. 2015), causing large uncertainty when predicting the possible emission at UV, optical, or IR frequencies. On the other hand, in case of recently observed underluminous and nearby event sGRB 170817A/GW170817, lanthanide-poor observed blue components were successfully modeled using a three-component ‘kilonova’ model (Villar et al. 2017a,b) with more realistic value of M_ej ∼ 0.016 M_⊙. So, present constraint on M_ej in case of sGRB 130603B indicates that either blue-component ‘kilonova’ emission was absent/weaker in comparison to the observed blue component in case of GW170817. These constraints further indicate that it could be possible to get a range of blue component of ‘kilonovae’ emission due to possible effects caused by a range of the dynamical ejecta, lifetime, and spin of the promptly formed magnetar/Black Hole, viewing angle effects etc. in case of some of the sGRBs. Early time deeper observations at bluer wavelengths for many such events at various distances are required to determine the range of properties like brightness, duration, and possible diversity among these events.

3.5 Host galaxy SED modeling of sGRB 130603B

Information about the host galaxy, such as the characteristic age of the dominant stellar population and the average internal extinction, was obtained by analyzing its broad-band SED (Table 4) using stellar population synthesis models. The host galaxy of GRB 130603B is a perturbed spiral galaxy as seen in high-resolution HST image (Tanvir et al. 2013) due to interaction with another galaxy. We combined our observational data in filters B, g, r, R_C, i, z, J, H, K_s obtained with GTC, CAHA, and DOT telescopes (see Table 1) and combined them with ultra-violet data in uvw2, uvm2, uvw1, U bands from de Ugarte Postigo et al. (2014) to construct the broad-band SED of the host galaxy. Taking into account a Galactic reddening along the line of sight of E(B − V) = 0.02 mag, and fixing the redshift of z = 0.356, we fitted the host SED using lephare software package (Arnouts et al. 1999; Ilbert et al. 2006). We used the PEGASE2 population synthesis models library (Fioc & Rocca-Volmerange 1997) to obtain the best-fitted SED and the main physical parameters of the galaxy: type, age, mass, star-formation rate (SFR) etc. We tried different reddening laws: MW (Seaton 1979), LMC (Fitzpatrick 1986), SMC (Prévot et al. 1984), and the reddening law for starburst galaxies (Calzetti et al. 2000; Massarotti et al. 2001).

According to the best fit, the host is a type Sd galaxy with absolute magnitude in rest-frame M_B = −20.9, moderate bulk extinction of E(B − V) = 0.2, and MW dust extinction law. It is about 0.7 Gyr old, has a mass of 1.1 × 10¹⁰ M_|$\odot$| and a low SFR of ∼6 M_|$\odot$| yr⁻¹. All the parameters are listed in Table 4. The reduced χ², galaxy morphological type, bulk extinction, absolute rest-frame B magnitude, age, mass, star formation rate, and specific star formation rate (SSFR) per unit galaxy stellar mass are listed for all four tested extinction laws. Fig. 8 represents the best model corresponding to the MW extinction law.

Figure 8.

The SED of the host galaxy of sGRB 130603B fitted by the lephare with fixed redshift z = 0.356. Filled red circles depict, respectively, the data points in the filters uvw2, uvm2, uvw1, U, taken from de Ugarte Postigo et al. (2014, Table 4), and B, g, r, R_C, i, z, J, H, K_s from original observations (see Section 2.2). Data points in B and R_C pass-bands were obtained using the 4K × 4K CCD Imager (Pandey et al. 2018) mounted at the axial port of the recently commissioned 3.6m DOT at Nainital, India (Kumar et al. 2018). Open circles represent model magnitudes for each filter. All magnitudes are in AB system.

Open in new tabDownload slide

These results confirm the previous host galaxy studies (Cucchiara et al. 2013a; de Ugarte Postigo et al. 2014; Chrimes et al. 2018) by independent observations and modeling, and adding a new piece of information about the extinction law inside the host galaxy. Our SED modeling results also constrain that SFR and mass of the host galaxy of sGRB 130603B are typical to those observed in case of other short bursts as shown in Fig. 11. However, the resulting SFR is five times higher than that obtained by Chrimes et al. (2018), using different population synthesis libraries.

4 MULTIWAVELENGTH OBSERVATIONS OF EIGHT SGRBS DURING 2012–2015

During 2012–2015, a total of 45 sGRBs were localized by several space missions. Only 23/45 of these sGRBs were seen by Swift-XRT. Out of those 23, only 9 were detected at optical bands, and, for 7 such events redshifts were determined. In this section, details of the prompt emission and multiband observations to detect optical afterglow and host-galaxy of eight events (sGRB 121226A, sGRB 131224A, sGRB 140606A, sGRB 140622A, sGRB 140903A, sGRB 140930B, sGRB 141212A, and sGRB 151228A) besides sGRB 130603B are discussed. Out of these eight sGRBs, three events, namely sGRB 131224A, sGRB 140606A, and sGRB 151228A, were not detected by Swift-XRT. However, sGRB 140606A and sGRB 151228A were seen by Fermi-Gamma-ray Burst Monitor (GBM) continuous Time-Tagged Event (TTE) data having detailed description in Appendix ‘A’. Out of the eight sGRBs from the present sample during 2012–2015, late-time follow-up observations using GTC 10.4 m and Gemini-N 8.0 m could be obtained for 4 Swift-XRT localized bursts i.e. for sGRB 121226A, sGRB 140622A, sGRB 140930B, and sGRB 141212A, useful to constrain late-time afterglow emission, placing limits on possible ‘kilonovae’ emission and host galaxy as described in respective sections of Appendix ‘A’.

The INTEGRAL SPI-ACS having a stable background (see Bisnovatyi-Kogan & Pozanenko 2011 and Minaev et al. 2010 for details) is particularly useful in the search for EE after the prompt emission phase of sGRBs. As a part of the present analysis, prompt emission INTEGRAL SPI-ACS observations of sGRB 121226A, sGRB 130603B, sGRB 140606A, sGRB 140930B, sGRB 141212A, and sGRB 151228A were analyzed and compared with other contemporaneous observations with the Swift-BAT and Fermi-GBM, when available. Details about the gamma-ray and X-ray data analysis are described in respective sub-sections of Appendix ‘A’. The analysis of the sub-set of these events does not show any signature of extended emission except sGRB 121226A and their spectral and temporal properties do not differ from those seen by Swift-BAT. Out of the eight sGRBs, for sGRB 140606A and sGRB 151228A, the characteristic photon peak energy E_peak could be determined using the prompt emission Fermi-GBM data. These two sGRBs along with others discussed in this paper with presumed redshift values allowed us to construct the Amati diagram along with published lGRBs (see Fig. 9). Based on this diagram, nature of these four bursts (namely sGRB 140606A, sGRB 140622A, sGRB 140930B, and sGRB 151228A) are clearly categorized as short bursts.

Figure 9.

Amati diagram – a relation between equivalent isotropic energy emitted in the gamma-ray E_iso versus characteristic photon peak energy E_peak(1 + z) in the rest frame (Amati et al. 2008). The solid straight line indicates a power-law fit to the dependences for the long bursts; the dashed lines bound the 2σ correlation region. The trajectories of sGRB 140930B and sGRB 151228A are plotted as a function of the presumed redshift z. Open circles indicate short bursts (sGRB 140606A, sGRB 140622A, and sGRB 130603B) with measured values of E_peak and redshift. Parameters of sGRB 170817A/GW170817 are also overplotted for comparisons.

Open in new tabDownload slide

Follow-up observations of these eight sGRBs suggest that the afterglows of these events were faint and were located either next to a bright star or embedded within the host galaxy, making the photometry complicated at the epoch of observations. Photometric results regarding the afterglow or host galaxies observed by the GTC 10.4 m and other ground-based telescopes as a part of the present analysis are tabulated in Table 5. Our optical-NIR observations indicate that for sGRB 141212A, the observed host galaxy was relatively bright and had star formation activity. Deeper GTC 10.4 m observations of the sGRB 140622A reveal that the burst could belong to a group of host-less bursts (Tunnicliffe et al. 2014). Follow-up optical observations of sGRB 140903A constrain any underlying ‘kilonovae’ emission down to a limiting magnitude of R > 22 mag at 10d after the burst. Our early- to late-time afterglow observations of sGRB 140930B using William Herschel Telescope (WHT) 4.2 m and Gemini-N 8.0 m observations along with those observed by Swift-XRT are able to constrain the decay nature of the burst and late time 10.4 m GTC observations places a deeper upper limit of r ∼ 24.8 mag for any possible host galaxy. Details about observations of the afterglows, host galaxies and their data analysis, calibrations etc. of each of the eight individual bursts are described in Appendix ‘A’ below. A summary of the observed prompt emission and afterglow properties of all the nine sGRBs is also listed in Table 6.