A HOT COCOON IN THE ULTRALONG GRB 130925A: HINTS OF A POPIII-LIKE PROGENITOR IN A LOW-DENSITY WIND ENVIRONMENT

GRB 130925A triggered the Swift Burst Alert Telescope (BAT) on 2013 September 25 at 04:11:24 UT, which we refer to as T₀. Its prompt gamma-ray phase was also detected at earlier times by International Gamma-Ray Astrophysics Laboratory at T₀−718 s (Savchenko et al. 2013), by the Fermi Gamma-Ray Burst Monitor at T₀−15 m (Fitzpatrick 2013), and by Konus-Wind (Golenetskii et al. 2013). The GRB emission showed several strong bursts visible by BAT up to ≈7 ks, as observed in other ultralong GRBs. A comparison between GRB 130925A, ultralong GRBs, and long GRBs is shown in Figure 1.

Zoom In Zoom Out Reset image size

Pointed observations with the Swift X-Ray Telescope (XRT) began at T₀+151 s, revealing a bright and highly variable X-ray afterglow. Strong flaring episodes were detected until T₀+6 hr, after which the afterglow exhibited a smooth power-law decay (Figure 1). Follow-up observations with the XRT lasted for ∼6 months, for a total net exposure of 425 ks in Photon Counting mode. XRT data were processed using the XRT Data Analysis Software (XRTDAS; version 12.9.3) distributed within HEASOFT. We used the latest release of the XRT Calibration Database and followed a standard reduction procedure (see Evans et al. 2009). We also analyze a Chandra observation (PI: E. Bellm; see Bellm et al. 2014), performed at T₀+11 days, for a total net exposure of 44 ks. Chandra data were reduced with the CIAO version 4.6 and the relevant calibration files.

In order to characterize the late-time afterglow evolution, a dedicated target of opportunity (ToO) observation with XMM-Newton (PI: L. Piro) was carried out on December 26 (T₀ + 3 months) for 100 ks. XMM data were reduced using SAS version 13.5.0. After applying standard filtering criteria and removing time intervals with high flaring background activity, the total net exposure is 85 ks.

The optical/IR counterpart of GRB 130925A was detected by GROND (Sudilovsky et al. 2013), and later localized by the Hubble Space Telescope (HST) to lie 012 from the galaxy nucleus (Tanvir et al. 2013). The red and relatively faint afterglow suggests that GRB 130925A was a highly extinguished event. Spectroscopic observations of the underlying host galaxy measured a redshift z = 0.347 (Vreeswijk et al. 2013).

Radio observations with the Australia Telescope Compact Array (ATCA) detected a source at a position consistent with the X-ray and optical localizations. These observations were carried out into three campaigns: one in 2013 October (Bannister et al. 2013), one in 2014 January, and one in 2014 February (PI: L. Piro). Radio data were calibrated and imaged using standard procedures within the MIRIAD data reduction package (Sault et al. 1995).

A detailed analysis of the early X-ray flares is presented in Evans et al. (2014); here we focus on the late (>20 ks) afterglow emission. The X-ray emission in the 0.3–10 keV energy band decays as a simple power-law function, f_X∝t^−α, with slope α = 0.82, steepening to α = 1.32 after ∼300 ks. The hardness ratio light curve displays significant variations up to late times, which is unusual for standard GRB afterglows. In order to quantify the spectral evolution, we performed a time-resolved spectral analysis using XSPEC version 12.8.1 (Arnaud 1996). Our results are summarized in Table 1 (spectra A1–A6). The X-ray spectra were described by an absorbed power-law model. The Galactic absorption component was kept fixed at the value of 1.66 × 10²⁰ cm⁻². A redshifted absorption component, modeling the host intrinsic absorption, was initially left free to vary. The afterglow spectrum shows a hard-to-soft (from Γ ∼ 3.5 to ∼4.6) followed by a soft-to-hard (from Γ ∼ 4.6 to ∼2.5) evolution. The spectral fits, although acceptable, yield unphysical variations of the absorbing column, correlated with the evolution of the power-law index (Table 1, Columns 2–3). The intrinsic absorption was therefore linked between the different spectra (Table 1, Columns 4–6). The resulting fit is poor (χ² = 552 for 429 dof), mainly because the soft power-law spectrum underestimates the flux above ∼3 keV.

Table 1. Results of the Spectral Fits^a

Time Interval	Power Law		Power Law (NH Linked)			Power Law (NH and Γ Linked)+ Blackbody
	NH	Γ	NH	Γ	χ2/dof	NH	Γ	kTBB	$L_{44}^{{\rm BB}}$	$R_{11}^{{\rm BB}}$	χ2/dof
E1: 150–500 s	$1.36^{+0.05}_{-0.05}$	$1.72^{+0.03}_{-0.03}$	$1.38^{+0.05}_{-0.05}$	$1.73^{+0.03}_{-0.03}$	852/820	$1.80^{+0.15}_{-0.15}$	$1.73^{+0.05}_{-0.08}$	$1.4^{+0.3}_{-0.3}$	$2.4^{+0.9}_{-0.9}\;{\times}$ 103	1.2 ± 0.6	806/817
E2: 500–700 s	$1.50^{+0.10}_{-0.10}$	$1.71^{+0.06}_{-0.06}$	...	$1.67^{+0.04}_{-0.04}$	...	...	...	$1.30^{+0.15}_{-0.15}$	$2.6^{+0.6}_{-0.6}\; \times$ 103	1.4 ± 0.3	...
E3: 1150–1340 s	$1.40^{+0.10}_{-0.10}$	$1.9^{+0.10}_{-0.10}$	$1.40^{+0.10}_{-0.10}$	$1.9^{+0.10}_{-0.10}$	71/83	$1.4^{+0.2}_{-0.2}$	$2.1^{+0.3}_{-0.3}$	$1.5^{+0.6}_{-0.6}$	$7^{+7}_{-7}\; {\times}$ 102	0.6  ±  0.6	65/81
A1: 20–300 ks	$2.30^{+0.10}_{-0.10}$	$3.5^{+0.06}_{-0.06}$	$2.10^{+0.10}_{-0.10}$	$3.4^{+0.10}_{-0.10}$	552/429	$1.40^{+0.10}_{-0.10}$	$2.4^{+0.2}_{-0.2}$	$0.5^{+0.03}_{-0.03}$	23 ± 3	1.00 ± 0.10	471/422
A2: 300–700 ks	$2.7^{+0.2}_{-0.2}$	$4.6^{+0.2}_{-0.2}$	...	$4.10^{+0.10}_{-0.10}$	...	...	...	$0.45^{+0.03}_{-0.03}$	10.0 ± 1.0	0.90 ± 0.10	...
A3: 0.7–2 Ms	$1.9^{+0.2}_{-0.2}$	$4.0^{+0.2}_{-0.2}$	...	$4.2^{+0.2}_{-0.2}$	...	...	...	$0.34^{+0.04}_{-0.04}$	4.2 ± 0.8	0.9 ± 0.2	...
A4b: 0.95–1 Ms	$1.90^{+0.10}_{-0.10}$	$3.80^{+0.10}_{-0.10}$	...	$4.0^{+0.10}_{-0.10}$	...	...	...	$0.35^{+0.03}_{-0.03}$	3.4 ± 0.4	0.80 ± 0.10	...
A5: 3–8 Ms	$1.3^{+0.4}_{-0.4}$	$3.0^{+0.4}_{-0.4}$	...	$3.8^{+0.5}_{-0.5}$	...	...	...	$0.30^{+0.10}_{-0.10}$	0.3 ± 0.2	$0.4^{+0.6}_{-0.3}$	...
A6c: 0.78–8 Ms	$1.10^{+0.14}_{-0.14}$	$2.50^{+0.10}_{-0.10}$	...	$3.5^{+0.2}_{-0.2}$	...	...	...	$0.23^{+0.05}_{-0.05}$	0.17 ± 0.10	0.4 ± 0.3	...

Notes. ^aFor the power-law model, we report the absorbing column N_H in units of 10²² cm⁻², and the photon index Γ. For the blackbody model, we report the blackbody temperature kT_BB in units of keV, luminosity L_BB in units of 10⁴⁴ erg s⁻¹, and radius R_BB in units of 10¹¹ cm. ^bChandra. ^cXMM-Newton.

Download table as:  ASCIITypeset image

A steep non-thermal spectrum (3.3 <Γ < 4.4) is atypical in GRB afterglows, and is generally seen as an indication of a thermal component which, over the limited 0.3–10 keV energy bandpass, cannot be fully resolved. Each spectrum was then fit by adding a blackbody component to the simple non-thermal power-law. The improvement of χ²is highly significant. However, despite the good statistics of the Swift and Chandra spectra, the spectral parameters were not well constrained, with an error range that included the blackbody temperatures and spectral indices found by Bellm et al. (2014). This is mainly because the flux of the blackbody component is comparable to, or even larger than, the power-law flux and dominates the emission below 3 keV.

In this respect, the late XMM observation was crucial to constrain the model. The high throughput of the telescope allowed us to gather enough counts at late times, when the thermal component had shifted to lower temperatures, and the two emission components could be better disentangled. The XMM spectrum can be described by a power law with Γ = 2.50 ± 0.10, rather common in GRB afterglows. The blackbody component is still significantly detected (Δχ² = 15.3 corresponding to a chance probability of 6.8 × 10⁻⁴) with temperature kT = 0.25 ± 0.04 keV.

Based on these results, we tested whether the observed X-ray emission could be described by an underlying non-thermal afterglow, and a dominating, highly variable thermal component. The spectra were simultaneously fit by linking the absorbing column and the power-law photon index, and by letting the black-body parameters free to vary (Table 1, Columns 7–12). Compared to the reference model (Columns 4–6), the addition of the blackbody improves the fit at a very high level, yielding a Δχ² = 81 for 7 additional parameters corresponding to a chance probability of ≪10⁻⁸.

Motivated by the results derived at t ≳ 20 ks, we also searched for a thermal component at earlier times, excluding the periods dominated by the X-ray flares. The results are reported in Table 1 (spectra E1–E3). The early-time spectra are dominated by the non-thermal emission, but consistent with the presence of a blackbody component with temperature kT_BB ≈ 1 keV.

The full set of ATCA measurements is listed in Table 2. Flux densities and corresponding 1σ errors were obtained from the task maxfit in MIRIAD. In the first two campaigns (2013 October, and 2014 January) the source was localized with typical uncertainty of 0.9–13. To further improve the positional accuracy of the target, another 17 and 19 GHz follow-up was carried out in 2014 February with the ATCA in its most extended array configuration (6D: maximum baseline length 6 km). Notwithstanding poor observing conditions, we were able to obtain a weak (S/N = 3.3) detection of the target at 17 GHz. The best positional accuracy was obtained by fitting a two-dimensional Gaussian model to the target image at 17 GHz with the MIRIAD task imfit. The resulting position is: R.A. = 02:44:42.949, decl. = −26:09:11.090 with (1σ) errors ΔR.A. = 0106, and Δdecl. = 0621.

The radio source remains nearly constant in brightness (within the uncertainties) over a period of ∼4 months. A comparison with the HST images, which we downloaded from the public archive, shows that the radio position is consistent with the faint transient reported by Tanvir et al. (2013) and offset from the galaxy nucleus. The probability of a chance alignment for a source this brightness is negligible (P ≈ 3 × 10⁻⁵), and we conclude that the radio source is the GRB afterglow. Its relatively constant flux suggests that the blastwave is expanding into a circumburst medium with a wind-like density profile, ρ(r)∝r⁻².

The non-thermal emission is well-described by a power-law with spectral slope β = Γ − 1 = 1.4 ± 0.2, consistent with the spectral indices observed in GRB afterglows (de Pasquale et al. 2006; Willingale et al. 2007). The X-ray light curve above 3 keV (inset of Figure 2), where the non-thermal component dominates, shows a power-law temporal decay with α = 1.20 ± 0.05, consistent with the closure relation α = (3β − 1)/2 for ν_X > ν_c. This evidence motivated us in modeling the X-ray non-thermal component and the radio data with a standard forward shock model (Granot & Sari 2002). The broadband spectra from radio to X-rays at three epochs were fitted simultaneously, yielding a good agreement with a forward shock expanding in a wind-like environment. The best fit model and parameters are presented in Figure 2. The blastwave energy E_k ≈ 10⁵³ erg is comparable to the observed gamma-ray energy E_{γ, iso} ≈ 1.5 × 10⁵³ erg, which implies a high radiative efficiency of the prompt emission mechanism. The tenuous, wind-driven medium derived from the fit implies a low mass loss rate of $\dot{M}\approx 3.6 \times 10^{-8}$ M_☉ yr⁻¹ for a wind velocity of v_w = 10³ km s⁻¹, consistent with a very low-metallicity BSG progenitor (Vink et al. 2001; Kudritzki 2002). From the lack of jet-break signature in the 3–10 keV light curve, we derive a lower limit on the jet-break time t_j ≳ 90 days, and a jet opening angle θ_j ≳ 2 deg.

Zoom In Zoom Out Reset image size

The luminosity, temperature, and apparent radius of the blackbody component are plotted in Figure 3 as functions of time. The luminosity declines as L_BB∝t^−0.9, while the temperature exhibits a slower decreasing trend, T∝t^−0.2. The apparent radius is R_BB ≈ 10¹¹ cm, showing that, within the uncertainties, the size of the emitting source remained remarkably constant in spite of the large variation in luminosity.

Zoom In Zoom Out Reset image size

Three main mechanisms can produce a thermal component in GRBs, namely a shock break out, the jet photosphere, and a hot cocoon. Shock breakouts are characterized by durations ≪10,000 s and peak X-ray luminosities from ≈10⁴⁴ erg s⁻¹ to 10⁴⁶ erg s⁻¹ (Ensman & Burrows 1992; Campana et al. 2006), not consistent with the long timescale and large luminosity observed in GRB 130925A.

Bright thermal emission from the fireball photosphere may emerge during the prompt gamma-ray phase. This high-energy photospheric component is associated to the optically thick plasma of a relativistic jet, and decays in luminosity and temperature as a power law in time (Ryde & Pe'er 2009). It may still be detectable in the soft X-rays a few hundreds seconds after the burst (Sparre & Starling 2012; Friis & Watson 2013) and, in principle, can continue on much longer timescales if the jet continues to be powered. Wong et al. (2014) showed that for a BSG this can be indeed the case: the fall-back of the external layers onto the central black hole yields an accretion rate $\dot{M}$ ≈ 10 t^−5/3 M_☉ yr⁻¹, and a corresponding jet luminosity L_jet ≈ 2 × 10⁴⁹ η₋₁ $t_{3}^{-5/3}$ erg s⁻¹. Here η = 0.1η₋₁ is the mass to energy conversion efficiency, and t = 1000 t₃s. The photospheric radius is $r_{{\rm ph}}=5.8\ 10^{11} L_{{\rm jet},51} \Gamma _2^{-3}$ cm (Abramowicz et al. 1991; Pe'er et al. 2012), where Γ = 100 Γ₂ is the jet Lorentz factor. Therefore, a photospheric emission with constant r_ph ≈ 10¹¹ cm requires Γ₂ ≈ 0.5 $\eta _{-1}^{1/3} t_3^{-5/9}$ from a few hundreds seconds to 10⁷ s. This fine coupling $\Gamma \propto L_{{\rm jet}}^{1/3}$ required to keep a constant photospheric radius seems somewhat contrived, although it cannot be excluded (e.g., Fan et al. 2012)

Let us now discuss the association of the blackbody with a hot plasma cocoon. The cocoon develops inside the star by the interaction of the jet with the stellar layers, and eventually breaks out at the stellar surface when the jet emerges (Lazzati & Begelman 2005). Starling et al. (2012) have proposed that the blackbody component found in a few GRBs during the steep decay phase of the X-ray light curve can be associated, at least in one event, to a relativistically expanding hot plasma cocoon. This component is short-lived (${{<}{{\sim}}}1{,}000$ s) and its radius is rapidly increasing with time. On the contrary, our observations exhibit a long-lasting blackbody emission with a constant radius, indicating that the cocoon does not expand. Thus, some process must confine it as it emerges at the stellar surface. A promising mechanism is magnetic confinement (Komissarov 1999). For instance, the toroidal component of the magnetic field could be advected into the inner part of the cocoon (Levinson & Begelman 2013), suppressing the plasma expansion across the magnetic field lines and confining it around the jet. In this scenario, the transverse size of the cocoon when it emerges at the stellar surface should be similar to the jet opening angle, θ_c ≈ θ_j ≈ R_BB/R_*, where R_* is the radius of the progenitor star. The limit on the jet opening angle derived previously implies $R_*\,{{<}{{\sim}}}\, 3\times 10^{12}$ cm, consistent with the typical radii of a BSG producing an accretion disk after the collapse (Woosley & Heger 2012; Kashiyama et al. 2013). The energy of the baryons entrained in the cocoon can be derived as E_{th, b} ≈ 3 M_c/2m_pkT ≈ 3 × 10⁴⁷ erg. Here M_c is the stellar mass contained within the volume excavated by the jet, that is $V_c \sim \pi R_{{\rm BB}}^2 R_*\approx 10^{35}$ cm³. This energy is several orders of magnitude lower than the energy of the accretion-powered jet, E_jet ≈ 5 × 10⁵³ η₋₁  erg (Wong et al. 2014), thus, if the jet is Poynting-flux dominated, its magnetic field can easily confine the plasma of the cocoon.

The blackbody energetics, E_BB = 1.5× 10⁵¹ erg, are safely below the energy produced by a jet piercing through a BSG, E_c ≈ 10⁵² erg (Kashiyama et al. 2013), and are much larger than the energy of the baryons computed above. The cocoon is therefore radiation dominated, and its temperature at the instant of the break out can be estimated as kT_BB ≲ (E_BB/V_c)^1/4 ≈ 1 keV, consistent with that observed. The cocoon's cooling time, t_c ≈ E_BB/L_BB ≲ 10⁴ s, is however much shorter than the duration of the thermal emission. Thus, the cocoon must be continuously energized by a fraction η_th of the jet energy which, as mentioned above, follows a L_jet∝t^−5/3. The blackbody luminosity can be expressed as L_BB ≈ η_thL_jet. In our case η_th ≈ 0.01. The flatter decay slope of the thermal component can be accounted by a slow increase of η_th when the jet energy decreases.