A nearby long gamma-ray burst from a merger of compact objects

Classification of GRB 211211A

GRBs are classified based on the properties of their prompt gamma-ray phase. The prompt emission of GRB 211211A (Extended Data Fig. 1) displays three different episodes: a weak precursor, a bright multipeaked main burst and a highly variable temporally extended emission. The time intervals for spectral and temporal analysis were selected to characterize them separately. Swift and Fermi data were processed using HEASOFT v.6.30. Spectra were extracted from the Fermi Gamma-ray Burst Monitor data and fitted within XSPEC42. The temporal properties were derived from the Swift BAT light curves using well-established techniques43,44.

The precursor phase has a short duration of 0.15 s, a soft spectrum peaking at ~75 keV, a minimum variability timescale of 21 ± 4 ms, and a positive lag ({tau }_{31}={16}_{-3}^{+4};{rm{ms}}) (68% CL; uncertainties throughout are quoted at 68% CL unless otherwise stated) between the temporal structures observed in the 50–100 keV (band 3) and in the 15–25 keV (band 1) energy bands, respectively. At 346 Mpc, the measured flux of 8 × 10−7 erg cm−2 s−1 (10–1,000 keV) corresponds to a luminosity of only ~1049 erg s−1.

After a 1-s period of quiescence, we detect the onset of the main prompt emission, which consists of multiple overlapping peaks lasting for approximately 10 s. The time-averaged spectrum peaks at 750 ± 10 keV, the minimum variability timescale is 14 ± 5 ms, and the temporal lag is negligible with ({tau }_{31}=-{0.9}_{-2.6}^{+2.8};{rm{ms}}). The total fluence measured during this episode is ~3.7 × 10−4 erg cm−2 (10–1,000 keV), one of the highest ever measured for a GRB. However, at 346 Mpc the total isotropic-equivalent gamma-ray energy Eγ,iso would be ~5 × 1051 erg within the typical GRB range45.

A brief (3-s) period of low-level persistent emission precedes the onset of a long-lasting tail. The time-averaged spectrum of the extended emission has a soft peak of 52 ± 2 keV, the minimum variability timescale is 42 ± 9 ms, and the lag, ({tau }_{31}={7}_{-2}^{+3};{rm{ms}}), is positive. The total fluence is ~5 × 10−5 erg cm−2 (10–1,000 keV), corresponding to Eγ,iso ≈ 7 × 1050 erg.

We compare the properties of the main prompt emission to the population of GRBs using four classifiers: the duration/hardness-ratio diagram46, the lag–luminosity relation43, the variability timescale44 and the Amati correlation45 (Extended Data Fig. 2). Similar to GRB 0606143, GRB 211211A shows characteristics that are intermediate between the two main GRB classes: the traditional classification based on duration and hardness ratio places this event in the class of long GRBs; however, its other properties fit within the class of short bursts. Its hybrid nature does not allow us to unambiguously link it to a progenitor system solely on the basis of its high-energy properties.

The GRB environment and its host galaxy

The GRB environment typically offers stringent, albeit indirect, evidence of its progenitor system. In the case of GRB 211211A, no underlying host galaxy is detected in late-time HST imaging (Fig. 1). By planting artificial sources with an exponential disk profile and different brightness, we derive an upper limit of F814W > 26.5 AB mag and F160W > 27.6 AB mag. Because no coincident galaxy is found, we analyse the GRB field to search for its most probable host. We identify seven galaxies within 10″ from the GRB position (Fig. 1): G1 with r = 19.50± 0.02 mag at an offset of 5.55″ ± 0.03″, G2 with r = 20.88 ± 0.05 mag at an offset of ~10″, and five faint (r > 26 AB mag) extended objects at an offset between 2.5″ and 10″. By using the galaxy’s number counts in the r-band47, we derive a chance alignment Pcc of 1.4% for G1, >10% for G2, and >40% for the other faint galaxies. Therefore, probabilistic arguments favour the association between GRB 211211A and G1. We note that the probability threshold adopted to associate a galaxy with a GRB is generally >1%, meaning that G1 with Pcc ≈ 1.4% would be considered as the most probable host by any previous studies of GRB galaxies12,47. Moreover, in our spectroscopic observations we find no evidence for any emission lines at the GRB position down to >2 × 10−17 erg cm−2 s−1 Å−1 in the range 4,800–6,100 Å. Using [O ii] 3727 and Hβ as indicators of unobscured star formation48, we place an upper limit on the star-formation rate, SFR < 1M yr−1 for z < 0.65. This corresponds to the median SFR of long GRB hosts49 at z < 1, providing additional constraints on any possible underlying galaxy.

The spectrum of G1 shows several emission lines including Hα, [N ii], and [S ii] at a common redshift of z = 0.0762 ± 0.0003, consistent with a previous report9 based on data from the Nordic Optical Telescope (NOT). Assuming a ΛCDM cosmology50 with a Hubble constant of H0 = 69.8 km Mpc−1 s−1, we find a luminosity distance dL = 346 Mpc, and a distance modulus μ = −37.7 mag. Using the host galaxy photometry (Supplementary Table 1), we compute a rest-frame absolute B-band magnitude of MB ≈ −17.6 AB mag, corresponding to LB ≈ 0.1L⁎ (L⁎, characteristic luminosity of the Schechter function) when compared to the galaxy luminosity function51 at a similar redshift (0.05 < z < 0.2).

The brightness (L ≈ 1040 erg s−1) and relative ratio of these lines (log([N ii]/Hα) ≈ −0.7) point to a star-forming galaxy with SFR ≈ 0.05M yr−1 and sub-solar metallicity 12 + log(O/H) ≈ 8.4. We also find evidence for weak [Mg i λ5175Å] absorption at ~5,567 Å, indicative of an evolved stellar population, although this feature is affected by a nearby skyline.

We model the galaxy’s surface brightness using GALFIT52. A good description (({chi }_{nu }^{2}approx 1.03)) of its morphology is obtained by including two Sersic profiles with index n = 1, one with half-light radius Re,1 ≈ 2.15 arcsec (F814W; ~3.1 kpc at z = 0.076) and one with Re,2 ≈ 0.5 arcsec (F814W; ~0.7 kpc at z = 0.076) to model the central bar. Similar results are obtained on the F160W image with Re,1 ≈ 2.34 arcsec and Re,2 ≈ 0.64 arcsec. The half-light radius r50 ≈ 1.1 arcsec obtained through Source Extractor is given by the weighted average of these two components.

The galaxy’s global properties were determined by modelling its SED (Supplementary Table 1) with Prospector53, adopting the same settings used for GRB host galaxies12,54. We derived a stellar mass of (M={0.9}_{-0.4}^{+0.2}times 1{0}^{9}{M}_{odot }), a star-formation rate SFR = (0.06 ± 0.02)M yr−1, a low dust content ({A}_{V}={0.09}_{-0.06}^{+0.08};{rm{mag}}), and a mass-weighted stellar age (tau ={5}_{-3}^{+2};{rm{Gyr}}). When compared to the sample of long GRBs, the properties of the host of GRB 211211A are not unprecedented but extremely uncommon. The inferred SFR lies in the bottom 10% of the observed distribution, leading to an unusually low specific SFR, sSFR ≈ 0.06 Gyr−1. This value is below the main sequence of star-forming galaxies55, indicating that G1 may be migrating to a quiescent phase. This differs from the typical environment of long GRBs at both high and low redshifts: for comparison, nearby events such as GRB 060218 and GRB 100316D were associated with sSFR ≈ 4 Gyr−1 and sSFR ≈ 0.2 Gyr−1, respectively56,57. Dissimilarities with the class of short GRBs also exist: the stellar mass lies at the bottom 10% of both short GRB and supernova type-Ia host galaxies58,59, as for GRB 060614, which was hosted by a dwarf galaxy5.


The SED of the GRB counterpart at different times is shown in Fig. 2. These epochs were selected to maximize simultaneous multiwavelength coverage. When needed, the data were rescaled to a common epoch using the best-fit temporal model.

In the first epoch at T0 + 100 s, the X-ray emission is characterized by a flat spectral index βX = 0.00 ± 0.03. A spectral break is required above ~10 keV to account for the lower flux and soft spectral index, βBAT ≈ 2, measured in the hard X-ray band. In addition, the high X-ray-to-optical flux ratio, FX/FO ≈ 100, requires a turn-over to a steep spectrum between the X-ray and optical band. These properties are consistent with self-absorbed synchrotron radiation in the fast-cooling regime. The location of a self-absorption frequency, νa ≈ 10 eV, indicates a compact emitting region60 with radius R ≈ 1013(Γ/300)3/4 cm, where Γ is the outflow bulk Lorentz factor. This radius is typical of dissipation processes within the GRB outflow, indicating that at ~T0 + 100 s the prompt phase is still dominant at both X-ray and optical wavelengths.

In the second epoch at T0 + 1 h, the GRB counterpart displays blue colours with a spectral index βO = 0.23 ± 0.10 in the UV and optical bands. At X-ray energies the spectrum, extracted between 3 ks and 5 ks, has a slope of βX = 0.50 ± 0.05. This index points to synchrotron radiation in the slow cooling regime, in which the cooling frequency is νc > 10 keV and the synchrotron frequency is νm 1 eV. In this case, the X-ray spectral slope is related to the energy distribution of the emitting electrons, N(E) Ep with p = 2βX + 1 = 2.0 ± 0.1. This is a fundamental constraint to the long-term afterglow evolution. The steepest spectral slope explained by this model is p/2 ≈ 1.05, and only for energies above νc. Therefore, the UVOIR and X-ray non-thermal afterglows are bound to remain on the same spectral segment over the time span of our observations.

Starting from ~T0 + 5 h, a simple non-thermal spectrum can no longer reproduce the broadband emission. An UVOIR excess is detected at all epochs. It is characterized by a narrow spectral shape peaking in the u band, well described by a blackbody function with temperature T ≈ 16,000 K (rest frame) and a luminosity Lbol ≈ (3.5 ± 2.0) × 1042 erg s−1. We therefore fit each SED epoch with a blackbody (UVOIR) plus power-law (X-ray) model, and derive the total integrated blackbody luminosity, its temperature and radius as a function of time (Fig. 2 and Extended Data Table 1). The luminosity is better constrained in our second epoch at T0 + 10 h, Lbol = (1.90 ± 0.15) × 1042 erg s−1 and is seen to decrease in time following a power-law t−0.95, consistent with the evolution of AT2017gfo27.

GRB distance scale

We investigate the joint X-ray/UV/optical SED at 1 h to place a direct upper limit on the GRB distance scale. UVOT spectra were created with the tool uvot2pha using the same source and background regions selected for photometry. We adopt a power-law model and include the effects of absorption, dust reddening and intergalactic medium attenuation as implemented in the XSPEC models zphabs, zdust and zigm. The Galactic absorption was fixed to NH = 1.76 × 1020 cm−2 and the reddening at E(B − V) = 0.015 mag. All other parameters were left free to vary. We increase the redshift from 0 to 2.5 in steps of 0.1 and find the best-fit model by minimizing the Cash statistics, recording its value at each step. On the basis of the variations of the test statistics, we derive an upper limit of z < 2.3 (99.9% CL) from the UV/optical data, and z < 1.5 (99.9% CL) from the joint X-ray/UV/optical fit. By imposing the redshift of the putative host galaxy, z ≈ 0.0762, we find no evidence for any dust extinction or absorption at the GRB site with 3σ upper limits of E(B − V)z < 0.005 mag and NH,z < 9 × 1019 cm−2, respectively. This is consistent with the location of the GRB, well outside the galaxy’s light.

Origin of the X-ray afterglow

Swift observations show a rapidly fading X-ray afterglow followed by a shallower decline FXtα with (alpha ={1.11}_{-0.07}^{+0.08}) between 1 ks and 40 ks, and a final steep decay with α = 3 ± 0.5 after 40 ks. On the basis of this model, we infer an X-ray flux of ~4 × 10−12 erg cm−2 s−1 at 11 h. This corresponds to a luminosity LX ≈ 6 × 1043 erg s−1 at 346 Mpc, nearly two orders of magnitude below the typical X-ray luminosity of cosmological GRB afterglows at this epoch (see figure 7 of ref. 23). The low ratio between the observed X-ray flux and the emitted gamma-ray fluence, logfX,11hr/Fγ ≈ −7.9, is indicative of atypical properties for this explosion (compare with figure 17 of ref. 12).

Our SED analysis (Fig. 2) demonstrates that the X-ray counterpart is dominated by non-thermal emission consistent with synchrotron radiation. Although we interpret the early (<300 s) X-ray emission as the tail of the prompt phase, at later times (>1,000 s) the most common origin of non-thermal afterglow radiation is the interaction between the ambient medium and the GRB jet occurring at large distances (>1017 cm) from the central source. In this external-shock model61, a flux decay rate of 2 or faster is explained by geometrical factors owing to the collimation of the GRB outflow62. The time tj at which the light curve steepens, the so-called jet break, increases with the jet opening angle θc. A jet break at 40 ks would require a very narrow jet, and then can only achieve a decay of α = p ≈ 2.1, in mild tension with the observations. We tested the hypothesis of an early jet break by modelling the X-ray and early (~T0 + 1 h) UVOT data with afterglowpy63 assuming a uniform external environment and both a top-hat and a Gaussian lateral structure for the jet. Despite the dataset being limited, it provides tight constraints to the model: the flat UVOT SED at T0 + 1 h (Fig. 2) requires the synchrotron peak to lie close to the optical range, constraining the value of the synchrotron frequency νm and the peak flux Fpk; the X-ray spectrum places the cooling frequency at νc > 10 keV and provides a measurement of p ≈ 2.0–2.1, and the X-ray light curve constrains the jet opening angle θc and the viewing angle θv. We performed Bayesian parameter estimation with emcee64 and nine free parameters: npEK,iso, θc, θv, an outer jet truncation angle θw, shock microphysical parameters εe and εB, and the participation fraction ξN. The best fit has a reduced chi-squared ({chi }_{nu }^{2}approx 1.8); fits with ξN frozen at 1 found a similar ({chi }_{nu }^{2}) but required unphysical shock parameters εe ≈ εB ≈ 1. The parameter estimation reports a jet of energy EK,iso ≈ (0.8–17) × 1051 erg, width θc ≈ 1.9–5.7°, viewed at θv ≈ 1.1–5.4° from the jet axis. The external density is n ≈ 0.016–12 cm−3. The shock parameters are p ≈ 2.1–2.2, εe ≈ 0.05–0.77, εB ≈ (0.1–6.0) × 10−4, and ξN ≈ (0.8–9.6) × 10−2. The beaming-corrected kinetic energy of the jet in this scenario is (0.4–4.4) × 1049 erg. Assuming that the angular size corrections between the afterglow and prompt emissions are similar, this scenario gives ~65% probability to an unphysical gamma-ray efficiency ηγ = Eγ,iso/EK,iso > 100% and a 90% probability ηγ > 15%. In combination with the poor reduced chi-squared of 1.8 we conclude it is challenging for an external shock to simultaneously reproduce the salient features of the GRB afterglow—a flat UV/optical spectrum at T0 + 1 h, an X-ray spectrum βX ≈ 0.5, and a steep decay of the X-ray flux after 40 ks—while remaining within the energetic limits of the prompt emission. This tension may be alleviated when considering the effects of inverse Compton cooling. In the limit of Thompson-scattering-dominated inverse Compton cooling65, we estimate that the required isotropic energy would increase by a factor of ~100, and the density decreased by a factor of ~1,000. However, the jet opening and viewing angles would have to decrease down to 0.5° to reproduce the final steep decay.

If not caused by a jet break, a rapid drop in brightness is difficult to produce, owing to the relativistic and extended nature of the GRB outflow. Owing to the curvature effect13, any rapid decrease in brightness in the lab frame of the GRB will be smeared out in the observer frame as a result of the different arrival times of the photons, producing a decay of α = 2 + βX ≈ 2.5. Nevertheless, this is a steeper slope than that allowed by the jet-break model and may present a better description than the standard external shock. If interpreted as a curvature effect, the steepening at 0.5 d links the observed X-ray emission either to long-lasting activity of the central engine, as in the ‘internal plateau’ model66,67, or to the angular structure of the GRB jet. If a structured jet produces GRB prompt emission in the high-latitude regions (the jet ‘wings’), this emission would be Lorentz-deboosted relative to the core prompt emission and delayed via the curvature effect36. With appropriate jet structures, this can manifest as X-ray emission with a shallow decay followed by a steep declining light curve. This feature, normally hidden by the brighter external shock emission, may become apparent in the case of a ‘naked’ structured GRB exploding in a rarefied medium. This latter model offers a consistent explanation of the X-ray behaviour of GRB 211211A and its physical offset from the galaxy without the requirement of hours-long activity of the central engine.

Despite uncertainty in the physical origin of the afterglow emission, the observed X-ray spectrum is well measured and its extrapolation to the UVOIR bands unambiguously places it below the UV/optical detections after ~T0 + 5 h. The observed UVOIR excess was measured by subtracting this extrapolated non-thermal component. This procedure does not require a physical interpretation of the non-thermal emission and provides an upper bound on the non-thermal contribution in the UVOIR bands. Thus the identification of the UVOIR excess does not depend on the specific physical interpretation of GRB 211211A’s non-thermal emission.

Origin of the UVOIR excess

Collapsar model

We first examine the most common case of a long GRB produced by the collapse of a rapidly rotating massive star (collapsar). The emergence of the supernova blast wave can produce a luminous blue emission in excess of the standard afterglow25, and we test whether this is consistent with the observed UVOIR excess in GRB 211211A. Collapsars arise from compact stellar cores and produce energetic and long-lived type-Ic supernovae or hypernovae. However, if the collapsar engine does not produce considerable 56Ni (for example, from a fallback collapsar), the blast wave produces a short-lived supernova light curve that dies out in the first 10 d. To test this model, we ran a series of hypernova explosions, varying the mass ((2.5–40)M) and density profile (varying the slope in the density of the core and envelope) of the progenitor star as well as the explosion energy (spherically 1051–1052 erg). Although we can reproduce the evolution of the bolometric luminosity (Extended Data Table 1), the early-time emission in our best-fit model is too energetic (in the UV and extreme UV). As the ejecta cools, the emission peaks in the infrared at late times, but the luminosity is several orders of magnitude too dim to explain the observations. To account for the optical and infrared emission, the photosphere of the rapidly expanding supernova must uncover the collapsar accretion disk and wind ejecta from this disk must have similar-enough properties to neutron star merger disks68,69 to produce a kilonova-like transient. However, even in this case, the large mass reservoir of a collapsar would power a long-lived late-peaking transient, not consistent with the observations.

For the collapsar model to work, we must also explain the offset of the GRB from its host galaxy. O/B stars in binaries can be unbound during the supernova explosion of the primary star, imparting a ‘kick’ of up to 200 km s−1 onto the O/B companion70. This proper motion could move the companion O star well beyond its star forming region (~1 kpc in 5 Myr), but it is unlikely that this kick is sufficient to explain the large offset of this burst. In summary, a massive star progenitor for GRB 211211A would naturally account for its long duration but requires a combination of unusual circumstances (a low 56Ni yield explosion, a low-mass neutron-rich disk outflow, and an extreme kick velocity) to explain the entire set of observations.

Compact binary merger model

The observed excess emission is much better fit by the ejecta from a compact binary merger, composed either of two neutron stars or a neutron star and a stellar mass black hole. Figure 3 shows the range of model predictions consistent with the observations: only a small subset of light curves (4 out of 900 in the ‘on-axis’ angular bin; θv ≈ 0–16°) match the observing constraints. The near-infrared luminosities are well described by dynamical ejecta of mass Md ≈ (0.01–0.03)M, lower than the value inferred for GRB 0606147,8. The bright UV/optical counterpart suggests a massive (>0.01M) wind component to the kilonova ejecta. However, the time-dependent spectra from the Los Alamos National Laboratory (LANL) grid of kilonova models28 produce light curves that are too dim to match the observed UV/optical luminosities or require too large an ejecta mass (~0.1M). Models with large ejecta mass (Mw ≈ 0.1M) better fit the early time data but overpredict the fluxes at later times; by contrast, the model with lower ejecta mass (Mw ≈ 0.01M) provides a good description of the dataset only after ~11 h. All consistent models adopt a toroidal morphology for the high-opacity ejecta and a polar outflow of low-opacity ejecta and high expansion velocity vw ≈ 0.3c.

It is probable that a number of alterations to the kilonova ejecta mechanism can help explain the early excess emission. For example, we have not conducted a detailed study varying the composition that changes both the opacity and the radioactive heating. Uncertainties in radioactive energy deposition71 and in the properties of the disk-wind ejecta allow for a wide range of behaviours and our study here only touches the surface of all possibilities. However, in its simplest form, a radioactive-powered kilonova captures the late-time evolution of the observed UVOIR transient but has difficulties in reproducing the bright optical emission seen at early times (T0 + 0.2 d).

An alternative way to alleviate the requirement on the ejecta mass is that the kilonova is powered by an additional energy source or affected by the jet–ejecta interactions33. To study the engine-powered models, we used the same method as in previous studies31. For central power sources—either a magnetar or fallback accretion on the central black hole—the energy must transport out from the centre to affect the light curves. In these models31, the central power sources do not alter the emission until ~5 d after the merger for wind mass ~0.01M. However, if the jet is able to evacuate a region above the compact remnant, this delay can be reduced. We mimicked this evacuation by a series of spherically symmetric models, reducing the total wind mass to ~10−7M. Although the signal peaks earlier it is still too late to explain our observations and the resultant spectrum is too high energy (peaking in the extreme UV; Extended Data Fig. 6). Turbulent motion may help to accelerate the UV peak by advecting the energy toward the outer layers more rapidly.

Although we caution that kilonova models are affected by large systematic uncertainties, we find that the majority of engine-driven kilonova models31,72,73 peak several hours or days after the merger, whereas jet–ejecta interactions remain a plausible solution to enhance the early emission.

In summary, we find that a compact binary merger would naturally account for most of the observed features of GRB 211211A, from the onset of its kilonova to its environment and high-energy properties. The main challenge to this model remains the long duration of the prompt gamma-ray emission, requiring an active central engine for up to ~100 s.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button