Short-duration γ-ray bursts (SGRBs) are energetic explosions with isotropic energy scales of order ∼10 51 erg and are detected to z≈2 (Narayan et al. 1992;Gehrels et al. 2008;Berger 2014;Lien et al. 2016;Paterson et al. 2020). They have prompt γ-ray emission (T 90 <2 s; Kouveliotou et al. 1993;Nakar 2007) and broadband, synchrotron afterglow emission at radio to X-ray wavelengths as a result of collimated, relativistic material interacting with the circumburst environment (Sari & Piran 1995;Meszaros & Rees 1997). In the context of their likely binary neutron star (BNS) merger progenitors (Berger 2014; Abbott et al. 2017), the nonthermal afterglows of SGRBs are expected to be accompanied by a thermal r-process kilonova (Li & Paczyński 1998;Metzger et al. 2010) powered by the radioactive decay of neutron-rich material synthesized in the merger. For SGRBs where the collimated outflow is viewed on-axis, the afterglow is expected to outshine the kilonova emission at optical wavelengths on 1-day timescales. On 1day timescales, the kilonova emission may dominate the observed optical and near-infrared (NIR) light, depending on the precise explosion properties of the afterglow (e.g., the kinetic energy, jet geometry) and the circumburst medium, as well as the mass, composition, and geometry of the kilonova ejecta (e.g., Barnes & Kasen 2013;Wollaeger et al. 2018;Metzger 2019). Indeed, the four kilonova candidates associated with SGRBs have all been detected on timescales of 1 day (Berger et al. 2013a;Tanvir et al. 2013;Jin et al. 2016;Troja et al. 2018Troja et al. , 2019;;Lamb et al. 2019). The optical and NIR emission of SGRBs and BNS mergers is thus a complex interplay between the nonthermal (potentially) jetted synchrotron emission and the thermal kilonova that results from heavyelement nucleosynthesis.
In general, the radio band is observationally more straightforward for SGRBs, as the primary expected emission component is from the afterglow forward shock (FS). However, despite routine, rapid follow-up observations, only seven SGRBs discovered by the Neil Gehrels Swift Observatory (Swift; Gehrels et al. 2004) have detected radio afterglows (Fong et al. 2015),o r≈5% of the entire Swift SGRB sample (Lien et al. 2016). Rapid-response, radio observations at 1 day have enabled the detection of early excess emission compared to expectations from the FS model, interpreted as reverse shock (RS) emission in two events, GRBs 051221A and 160821B (Soderberg et al. 2006;Lloyd-Ronning 2018;Lamb et al. 2019). As a population, the lack of optical and radio afterglow emission for a majority of SGRBs is a direct reflection of their low beaming-corrected kinetic energy scales (≈10 49 erg, two orders of magnitude lower than long-duration GRBs; Panaitescu 2006;Gehrels et al. 2008) and their low circumburst densities of ≈10 -3 to 10 -2 cm -3 (Panaitescu et al. 2001;Soderberg et al. 2006;Fong et al. 2015;O'Connor et al. 2020a).
SGRBs also exhibit an extended spatial distribution with respect to their host galaxies, as well as to their host light distributions (Berger 2010;Fong et al. 2010;Fong & Berger 2013;Tunnicliffe et al. 2014). Their hosts have a range of stellar population ages of ≈0. 5-8 Gyr (Leibler & Berger 2010;Nugent et al. 2020), which can naturally be explained by the wide expected range of delay times for their BNS merger progenitors (Belczynski et al. 2006;Paterson et al. 2020). The low densities, weak correlation with host stellar mass or star formation, and origin from a diverse range of host galaxies are all hallmarks of the SGRB population (Zheng & Ramirez-Ruiz 2007;Fong & Berger 2013;Fong et al. 2013;Tunnicliffe et al. 2014;Wiggins et al. 2018;O 'Connor et al. 2020a).
The detection of kilonovae associated with SGRBs has been challenging, due to a combination of the faint expected emission and cosmological distances, making sufficient follow-up observations difficult with current resources. The four kilonova candidates associated with SGRBs, as well as the kilonova associated with the BNS merger GW170817, have luminosities and colors that can be explained by standard radioactive heating (Barnes et al. 2016;Kasen et al. 2017). The kilonova of GW170817 has a well-sampled multiband light curve (Andreoni et al. 2017;Arcavi et al. 2017;Cowperthwaite et al. 2017;Chornock et al. 2017;Coulter et al. 2017;Díaz et al. 2017;Drout et al. 2017;Kasliwal et al. 2017;Lipunov et al. 2017;Nicholl et al. 2017;Pian et al. 2017;Smartt et al. 2017;Tanvir et al. 2017;Troja et al. 2017b;Utsumi et al. 2017;Valenti et al. 2017;Villar et al. 2017), providing a benchmark for radioactively powered kilonovae. The remaining SGRB kilonova candidates are more sparsely sampled and have been detected in a variety of rest-frame bands (optical and NIR), but overall they exhibit an evolution from blue to redder colors with time. In addition, the range of observed luminosities for the majority of events is ≈(1-5)×10 41 ergs -1 . If all are in fact kilonovae, this demonstrates the diversity of kilonova emission resulting from BNS mergers (e.g., Gompertz et al. 2018;Ascenzi et al. 2019;Rossi et al. 2020). However, if the SGRB progenitor produces a hypermassive or supramassive neutron star that is at least temporarily stable to collapse, or even an indefinitely stable remnant, a combination of disk winds, neutrino irradiation, and spin-down energy may also be imprinted on the kilonova signal or X-ray emission, resulting in even larger luminosities and bluer colors (Metzger & Fernández 2014;Metzger & Piro 2014;Kasen et al. 2015;Metzger 2019). Variations on the ejecta morphology or sources of heating, independent of the presence of a stable remnant, may have similar effects (Kisaka et al. 2015;Barnes et al. 2016;Rosswog et al. 2017;Metzger 2019;Korobkin et al. 2020). Thus far, there has not been a clear case of an observed kilonova or kilonova candidate that required the existence of a stable neutron star remnant, or major modifications to standard kilonova models.
Here we present X-ray, optical, NIR, and radio observations of the SGRB 200522A and its star-forming host galaxy at z=0.5536. These observations reveal an unusual broadband counterpart that is not easily explained by a single emission component. In Section 2 we present the Swift burst discovery, the discovery of the radio and NIR counterparts with the Very Large Array (VLA) and the Hubble Space Telescope (HST), and observations of the host galaxy with Keck and archival data. In Sections 3 and 4, we introduce two scenarios to explain the peculiar broadband behavior of GRB 200522A: an FS with an NIR excess, or a combination of an FS and RS with a wideangle jet. We present our host galaxy modeling and derived stellar population, morphological, and local properties in Section 5. In Section 6 we introduce radioactively powered and magnetar-boosted kilonova models to explain the NIR excess emission of GRB 200522A, and we compare the NIR luminosity to the landscape of known or candidate kilonovae. In Section 7, we compare GRB 200522A to the population of SGRBs in terms of its transient and host galaxy properties, introduce a radio catalog of SGRB afterglow detections, and discuss implications for detectability. Finally, we conclude and offer a future outlook in Section 8.
Unless otherwise stated, all observations are reported in AB mag and have been corrected for Galactic extinction in the direction of the burst of A V =0.07 mag (Schlafly & Finkbeiner 2011). We employ a standard cosmology of H 0 =69.6 kms -1 Mpc -1 , Ω M =0.286, and Ω vac =0.714 (Bennett et al. 2014).
2.1. Burst Discovery GRB 200522A was discovered by the Burst Alert Telescope (BAT) on board Swift (Gehrels et al. 2004) on 2020 May 22 at 11:41:34 UT (Evans et al. 2020). The BAT position was refined to R.A.=00 h 22 m 40 3, decl.=-00°15′49 9 (J2000) with an uncertainty of 1 59 in radius (90% confidence; Ukwatta et al. 2020). The Swift X-ray Telescope (XRT) began observations of the field of GRB 200522A at δt=83.4 s (where δt is defined as the time since the BAT trigger) and detected an uncataloged X-ray source within the BAT position, later refined to an enhanced position of R.A.=00 h 22 m 43 68, decl.=-00°16′59 4 with a 2 2 radius positional uncertainty (90% confidence; Goad et al. 2007;Evans et al. 2009;Beardmore et al. 2020). The duration of the burst, with T 90 =0.62±0.08 s (15-350 keV), combined with the hardness ratio of 1.46 (fluence ratio, S(50-100) keV/S(25-50) keV), places GRB 200522A solidly in the category of short, hard GRBs (Lien et al. 2016). We measure a Swift/BAT fluence of S γ =(1.04±0.14)×10 -7 ergcm -2 (15-150 keV, 90% confidence), consistent with the results of Ukwatta et al. (2020).
Upon a detailed inspection of the GRB 200522A 64 ms BAT light curve, we find a multipeaked structure in the main GRB pulse. We also note a precursor signal prior to the main pulse between δt=-0.35 s and δt=-0.25 s. Constructing an image over this time interval in the 25-100 keV band, we derive a source significance for the precursor of 3.9σ. The spectrum of the precursor signal is poorly constrained but is consistent with a hard spectrum characterized by photon index Γ γ =0.86±0.70. For GRB 200522A, the power-law (PL) and cutoff power-law (CPL) models provide comparable fits to the T 100 spectrum. Here we employ the CPL model since it provides a constraint on the break energy of the spectrum and therefore a more accurate estimate of the integrated energy. We obtain the best-fit values of G =-
We reanalyze the Swift XRT observations of GRB 200522A to obtain the X-ray light curve spanning δt≈0.006-2.74 days. To perform the X-ray spectral analysis, we obtain the source and background spectra and ancillary and response files for each bin of the light curve as defined by the XRT time-sliced spectra interface (Evans et al. 2009). We reduced the data using the HEASoft software (v. 6.26.1;Blackburn et al. 1999) and caldb files (v. 20190910). We use the methods of Evans et al. (2007Evans et al. ( , 2009) ) for selecting the source and background regions and binning the data, as well as for extracting the counts and producing the spectra.
We first use the Xspec software (v.12.9.0; Arnaud 1996) to fit the spectrum of each bin of the light curve (0.3-10 keV), binning the spectra using grppha to ensure a minimum of one count per bin. We use VERN X-ray cross sections (Verner et al. 1996), WILM abundances (Wilms et al. 2000), and W-statistics for background-subtracted Poisson data (Wachter et al. 1979). We employ a two-component absorption power-law model characterized by photon index (Γ X ), the intrinsic hydrogen column density (N H,int ) at the redshift of the GRB (see Section 2.6), and the Galactic hydrogen column density in the direction of the GRB 200522A(N H,MW =2.94×10 20 cm -2 ; Willingale et al. 2013). Allowing both Γ X and N H,int to vary, we find that the value of N H,int is consistent with zero and that the individual values for Γ X do not exhibit statistically significant changes (to within 1σ) over the course of the observations.
Since the parameter values for the individual observations are poorly constrained, we use Xspec to jointly fit the entire data set and find best-fit values of G = -+ 1.47 X 0.19 0.24 (1σcon®dence) and N H,int <5.51×10 21 cm -2 (3σ). Fixing the spectral parameters to the best-fit values and freezing N H,int =0cm 2 , we calculate the unabsorbed X-ray fluxes utilizing the cflux model within the 0.3-10 keV energy range. Finally, we determine the X-ray afterglow flux densities, F ν,X at ν X =1 keV, using the spectral index, β X (β X ≡1-Γ X ), which has a value of b =- -+ 0.47 X 0.19 0.24 across all observations. For the last observation at δt=2.74 days, we determine the 3σcount-rate upper limit using the four source photons detected in ∼4.8 ks using Poissonian statistics following Gehrels (1986). Applying the best-fit spectral parameters using WebPIMMS,foot_1 we calculate the unabsorbed X-ray flux and resulting upper limit on F ν,X . The observational details, 1 keV flux densities, and 1σuncertainties for the entire X-ray afterglow light curve are listed in Table 1. These results are consistent within 1σuncertainties to the Swift time-sliced interface results (Evans et al. 2009) under the same assumptions in spectral binning.
The UltraViolet and Optical Telescope (UVOT) on board Swift began observations of GRB 200522A at δt=448 s and obtained preliminary 3σupper limits of >19.5 mag in the white filter (Kuin et al. 2020). Additional observations were taken with the Yock-Allen BOOTES-3 telescope starting at δt≈6.8 hr (Hu et al. 2020), with an upper limit of >18.1 mag in the clear filter.
We initiated R-and I-band observations with the Sinistro instrument mounted on the Las Cumbres Observatory Global Telescope network (LCOGT) 1 m telescope at the South African Astronomical Observatory for a total of 900 s of exposure time in each filter at midtimes of δt=0.65 and 0.69days, respectively. These observations were first reported in Strausbaugh & Cucchiara (2020a, 2020b), and the following analyses supersede those reported in the circulars. We reduce the data with the BANZAIfoot_2 data reduction pipeline, which performs bad-pixel masking, bias subtraction, dark subtraction, flat-field correction, source extraction (using SEP, the Python and C library for Source Extraction and Photometry), and astrometric calibration (using astrometry.net). We align the frames and co-add the individual images using Python/ astroalign, and we perform astrometry relative to the USNO-B1 catalog.
Within the XRT position, we detect a single, clear source in the images, consistent with the position of the Sloan Digital Sky Survey (SDSS) cataloged galaxy SDSS J002243.71-001657.5 (Alam et al. 2015), first reported as the potential host galaxy in Fong et al. (2020c). Performing photometry with SExtractor relative to USNO-B1.0, we calculate a magnitude of R=21.27±0.17 mag, consistent with the archival SDSS magnitude of r=21.17±0.07 mag, and an upper limit of I20.39 mag within the XRT position (Table 1).
We obtained a second, deeper set of LCO R-band observations at δt≈32.6 days. Performing image subtraction between the two LCO epochs using the HOTPANTS software package (Becker 2015), we do not find any significant residuals. We thus measure a 3σupper limit on optical afterglow emission of R22.1 mag at δt≈0.65 days. The details of our observations are listed in Table 1. We note that reported observations taken with the Gemini Multi-Object Spectrograph (GMOS) mounted on the Gemini-North telescope also place a comparable limit on emission outside of the host galaxy but within the XRT position of r>22.2 mag (Dichiara et al. 2020).
We initiated observations with the VLA (Program 19B-217;PI: Fong;reported in Schroeder et al. 2020a) at a central frequency of 6.05 GHz (C band). The observations occurred at a midtime of δt=0.23 days for a total of 1 hr, including time for flux density and phase calibration. We centered the upper and lower sidebands at 5.0 and 7.2 GHz, respectively, and used 3C 147 for flux calibration and J0022+0014 for gain calibration. We excised the effects of radio frequency interference (RFI) from the data and employed standard interferometric calibration techniques for data calibration and analysis within the Common Astronomy Software Applications (CASA;McMullin et al. 2007). We used CASA/tclean to image the field, employing Briggs weighting with a robust parameter of 0 (to minimize sidelobe contamination from neighboring sources) and two Taylor terms (nterms=2). Toward the northeast edge of the 90% XRT error circle, we detect a single radio source (Figure 1). Using a point-source model within CASA/imfit, we measure a source flux density of F ν,6 GHz =33.4±8.2 μJy.
We obtained a second 6.05 GHz epoch at δt=2.19 days, in which the source is still detected with F ν,6 GHz =27.1±7.2 μJy, consistent with a constant flux density within the 1σerrors. In addition, we obtained contemporaneous observations at a mean frequency of 9.77 GHz and do not detect any significant emission within the X-ray error circle to a 3σlimit of F ν,9.7 GHz 23.7 μJy. To assess the nature of the source at 6.05 GHz within the XRT error circle, we obtained a final series of deeper observations at 6.05 GHz at δt≈6.15 and 11.15days. The source is no longer detected to 3σlimits of F ν,6 GHz 18.6 and 14.1 μJy, respectively. We use CASA/concat to combine the exposures of the first two C-band epochs, and we derive a position of R.A.=00 h 22 m 43 706, decl.=-00°16′57 97 (J2000) with 1σpositional uncertainties of ΔR.A.=0 23 and Δdecl.=0 27, with a flux density of F ν,6 GHz =29.7±5.3 μJy. Combining the final two observations in the same manner, we determine a deep limit of F ν,6 GHz 10.9 μJy (3σ). Due to the spatial coincidence with the XRT and HST NIR counterpart positions (see Section 2.5), along with clear fading behavior of the source, we consider this to be the radio afterglow of GRB 200522A. The individual epochs and combined images are displayed in Figure 1, and the details of our observations are summarized in Table 1.
We initiated observations with the HST (PI: Berger, Program 15964) using the Wide Field Camera3 (WFC3) IR channel (previously reported in Fong et al. 2020aFong et al. , 2020b;;O 'Connor et al. 2020b). We obtained observations in the F125W and F160W bands for a total of 5223.5 s in each filter at midtimes of δt=3.52 and 3.66days, respectively. We used the astrodrizzle package to combine the images in each filter, employing combine_type=median, wht_ty-pe=EXP, pixscale=0 0642 pixel -1 (half of the native WFC3/IR pixel scale), and pixfrac=0.8. The images are shown in Figure 2. We performed absolute astrometry on the F125W filter image relative to SDSS DR12 (Alam et al. 2015), with an astrometric tie uncertainty of 0 048 (1σ). The host galaxy (Section 2.6) is clearly detected at a position of R.A.=00 h 22 m 43 717, decl.=-00°16′57 46, along with an additional fainter, extended source within the XRT error circle to the southeast at R.A.=00 h 22 m 43 813, decl.=-00°16′ 59 52. We also note the presence of a point source ≈1 43 to the east of the host galaxy (Figure 2; Fong et al. 2020b).
We obtained two additional sets of observations in the F125W filter at midtimes of δt=16.38 and 55.24days (Kilpatrick et al. 2020) and one additional set in the F160W filter at 55.37days, which we treat in the same manner as the first epoch. For each observation, we used IRAF/ccmap and ccsetwcs to perform astrometry relative to the first epoch of F125W observations (which itself is tied to SDSS), with an average relative astrometric uncertainty of ≈0 01.
Using the observations at δt≈55 days as a template for each filter, we performed image subtraction using the HOTPANTS software package (Becker 2015) between each of the earlier epochs and the template in the relevant filter. The difference images at δt≈3.6 days reveal a point source present at the northeast edge of the XRT position, consistent with the radio afterglow position with R.A.=00 h 22 m 43 727, decl.=-00°1 6′57 43 (Figure 2) in both filters. This source subsequently fades in F125W imaging by 16.4days (also found in independent analysis reported in Tanvir et al. 2015). Given the fading behavior and coincidence with the X-ray and radio positions, we consider this source to be the NIR counterpart to GRB 200522A.
The lack of residuals in the difference image between the latter two F125W epochs signifies a negligible amount of transient emission at δt=16.38 days. Thus, we use astrodrizzle to create a "combined," deep F125W template. The results of the image subtraction between the first epoch and the deep template are shown in Figure 2, exhibiting a highsignificance detection of the NIR counterpart, on which we base our subsequent photometry.
The difference images all exhibit contamination coincident with the core of the host galaxy. Each subframe in the first set of observations has EXPTIME=602.93 s with peak counts near the center of the galaxy of ≈4200e -. This means that the center of the galaxy is nonlinear at the 0.1% level,foot_3 and that even with nonlinearity corrections this will result in an imperfect subtraction at the host centroid.
To obtain reliable photometry and uncertainties of the NIR counterpart, we pursue three independent methods: (1) aperture photometry using a small aperture with an encircled energy (EE) correction, (2) point-spread function (PSF) photometry with width fixed to the in-band WFC3/IR PSF, and (3) PSF photometry with an empirically determined value. First, using the IRAF/phot package, we perform aperture photometry of the source using a small, 0 2-radius aperture fixed at the position of the counterpart. We then apply tabulated encircled energy corrections to correct the small apertures to infinity,foot_4 with corrections of 0.29 (F125W) and 0.34 mag (F160W). For the second method, we use the tabulated values of the FWHM WFC3/IR PSF (Windhorst et al. 2011) of 0 136 for the F125W filter and 0 150 for F160W. We then construct a fixedwidth Gaussian PSF using photutils and apply it in a 0 5 aperture at the location of the residual in our F125W and F160W difference images, fitting for the integrated flux and centroid position of the source in both images. We derive our uncertainties on flux by changing the best-fitting centroid and fixed-width FWHM to within 10% of the input values and measuring the standard deviation in the implied flux. For the third method, we use daophot to empirically determine the best-fit PSF size and shape from isolated stars in the epoch one images. With the resulting PSF model, we then fit for the integrated flux and centroid of the residual in the difference images. Taking the average flux and statistical uncertainty of the results from the three methods in flux space, we find that the NIR counterpart brightness is m F125W = 24.55 ± 0.15 mag and m F160W = 24.62 ± 0.15 mag, in which the dominant source of uncertainty is the difference in methods (with individual measurement uncertainties of 0.05 mag).
Finally, to obtain an upper limit in the 16.38-day observation, we use dolphot to inject fake sources of known brightness (m F125W =24-28.5 mag) at and near the counterpart location in the difference image. These sources have a shape matched to the WFC3/IR F125W instrumental PSF. We then recover these sources using dolphot and change the brightness in increments of 0.1 mag until we find the threshold at which >99.7% of sources are recovered at a signal-to-noise ratio (S/N) of >3, from which we derive m F125W 27.5 mag (3σ) at δt≈16.38 days.
To quantify the probability that SDSS J002243.71-001657.5 is the host galaxy of GRB 200522A, we calculate the angular offsets between the NIR counterpart and the host galaxy centroid derived in HST imaging. We use the final observations at δt≈55 days, as the host centroid determination in earlier epochs will be contaminated by the transient emission. We consider three sources of uncertainty in the offset calculation: the counterpart positional uncertainty (σ HST =0 0012), the host positional uncertainty (σ host,F125W =0 052, σ host,F160W =0 0007), and the relative astrometric uncertainties between HST observations (σ tie,F125W =0 029, σ tie,F160W =0 013). We measure projected angular offsets of δR=0.155±0 054 (F125W) and 0.143±0 029 (F160W).
Using the angular offsets and R-band magnitude of the host galaxy (Table 1), we calculate a low probability of chance coincidence of P cc =3.5×10 -5 following the methods of Bloom et al. (2002). There are only two other cataloged galaxies within 0 5, both of which have significantly higher values of P cc =0.25-0.4. Repeating the same exercise based on the VLA position, and taking into account the absolute astrometric uncertainty between the F125W observations and SDSS DR12, we calculate a similarly low P cc =4.8×10 -4 . We thus confirm SDSS J002243.71-001657.5 as the host galaxy of GRB 200522A.
To further characterize the host galaxy, we used the Low Resolution Imaging Spectrometer (LRIS) mounted on the 10 m Keck I telescope (PI: Blanchard; Program O287) to obtain Gand R-band imaging on 2020 June 21 UT at a midtime of δt≈30.1 days (Table 1). We apply bias and flat-field corrections using the photpipe image reduction and processing software (Rest et al. 2005;Kilpatrick et al. 2018). We perform relative alignment of the individual frames and stack them with the SWarp software package (Bertin 2010). For the final stacked frames, we use IRAF tasks ccmap and ccsetwcs to align the images to SDSS DR12.
We also obtained I-, Z-, and V-band imaging of the host galaxy with the DEep Imaging Multi-Object Spectrograph (DEIMOS) mounted on the 10 m Keck II telescope on 2020 July 17 UT at a midtime of δt≈56.1 days (Table 1; PI: Blanchard). We apply bias and flat-field corrections and align and stack the individual images using a custom pipeline. 19 We perform aperture photometry using phot, employing source apertures of 2 5, chosen to fully encompass the host galaxy. After calibrating each image to the SDSS DR12 catalog and converting to the AB system using the relevant relations from Chonis & Gaskell (2008), we obtain host galaxy magnitudes in the GRIZ filters and an upper limit in the V-filter; the results are listed in Table 1. From our HST imaging (Section 2.5), we use IRAF/phot to measure host magnitudes of m F125W = 20.84 ± 0.01 mag and m F160W = 20.65 ± 0.01 mag (Table 1).
We supplement these data with available photometry in other bands based on archival imaging in the SDSS DR12, Pan-STARRS1 (PS1), and Spitzer Space Telescope imaging as part of the Stripe 82 survey (Program 90053, PI: Richards; Werner et al. 2004;Alam et al. 2015;Chambers et al. 2016;Timlin et al. 2016;Papovich et al. 2016). For SDSS DR12, the host galaxy is cataloged, and we use the available u-band photometry to supplement the Keck photometry. The host galaxy is weakly detected in the PS1 3π y-band stacks and in the Spitzer 3.5 μm and 4.6 μm imaging but is not cataloged. Thus, we download the imaging and perform aperture photometry of the host. The Spitzer photometry is complicated by a varying background due to nearby sources, which we ameliorate by selecting roughly five source-free, background regions in the vicinity of the host, and we report the variance in the derived flux density as the uncertainty. Our host galaxy photometry based on archival imaging is also listed in Table 1.
In addition, we obtained Keck/LRIS spectroscopy on 2020 June 21 UT for a total of 3×900 s with the blue camera and 3×860 s with the red camera, with a fixed dichroic wavelength of 5600 Å. The spectrum was taken with a 1 0long slit, 400/3400 grism (blue), and the 400/8500 grating (red), with a central wavelength of 7830 Å. The resulting spectrum spans a continuous range of ≈3200-10280 Å with a spectral resolution of ∼7 Å in both arms. We use standard IRAF tasks to subtract the overscan, apply flat-field corrections, model the sky background, and subtract it for the individual frames. We also perform wavelength calibration using HeNeArCdZn arc lamp spectra, and spectrophotometric flux calibration using the standard star Feige 110 taken at a similar air mass on the same night. We use apall to extract the 1D spectra, which we then co-add. We determine the error spectrum by performing the same reduction steps but on spectra without sky subtraction and performing standard error propagation in the combination. The resulting spectrum is shown in Figure 3.
The spectrum overall exhibits a blue continuum, with a 4000 Å break at ∼5800 Å. We detect several emission lines:
In the following two sections (Sections 3 and 4), we present our afterglow modeling and two interpretations of the broadband data set (termed Scenarios I and II, respectively).
Here we first interpret the radio, NIR, and X-ray observations of GRB 200522A in the context of synchrotron emission from an FS produced by the interaction of the GRB jet with the ambient environment (Sari et al. 1998;Granot & Sari 2002). The parameters of the model are the isotropic-equivalent kinetic energy (E K,iso ) of the jet, the particle density of the circumburst environment (n 0 ), the power-law index of accelerated electrons (p), the opening angle of the outflow (θ jet ), and the fractions of the FS energy imparted to electrons (ò e ) and magnetic fields (ò B ). The resulting synchrotron spectrum is characterized by three break frequencies: the synchrotron self-absorption frequency (ν a ), the characteristic synchrotron frequency (ν m ), and the cooling frequency (ν c ).We use the convention F ν ∝t α ν β throughout.
We assume negligible intrinsic extinction, which is supported by the observed Balmer decrement in the Keck spectrum as consistent with the theoretical value, and the broadband SED modeling of the host galaxy (Section 5.1). We also assume a uniform-density profile characteristic of the interstellar medium (ISM), as expected for SGRBs.
At high electron Lorentz factors, inverse Compton (IC) cooling (with a strength determined by the Compton-Y parameter) modifies the electron distribution and the resulting synchrotron radiation. Whereas IC cooling can be significant for long-duration GRBs (Sari & Esin 2001;Laskar et al. 2015), for the typical parameters of SGRBs (E K,iso ≈10 51 erg, n 0 ≈10 -2 cm -3 ; Fong et al. 2015), the Klein-Nishina (KN) effect limits Y<Y max ≈0.2 (assuming p≈2.2 and ò e ≈0.1). 20 In this regime, the synchrotron spectrum is better approximated by ignoring IC cooling effects (Nakar et al. 2009). We therefore ignore IC cooling in our modeling and subsequently verify whether the KN limit indeed applies to the derived parameters.
From the XRT data, we measure b =- -+ 0.47
(Section 2) and α X =-0.67±0.10 (1σ) over δt≈6×10 -3 days to 0.6days. For the radio band, we measure a fairly shallow radio evolution of α R =-0.1±0.2 between δt=0.23 and 2.19days, followed by a decline of α R -0.4 at δt>2.2 days. The faintness of the radio detection precludes a meaningful in-band spectral index. The nondetection at 9.77 GHz implies that the radio emission is optically thin (β radio -0.3 at δt≈2.2 days), with ν m 6 GHz at δt≈2.2 days. Finally, from the NIR F125W observations, we measure a decline rate of α NIR -1.7 between δt=3.6 and 16.4days. Next, we use the α-β closure relations (Granot & Sari 2002) to infer the location of the cooling frequency, ν c , relative to the X-ray band. We calculate the value of p from both the spectral and temporal indices of the XRT data for two scenarios: ν m <ν X <ν c and ν X >ν c , requiring the value of p to be in agreement within each scenario. We find consistency between the observed X-ray light-curve spectrum and decline rate for ν X <ν c , with p=1.90±0.13 from α X and p=1.94±0.40 from β X , with a weighted mean and 1σ uncertainty of 〈p〉=1.90±0.13.
We now demonstrate that the NIR observations cannot be reconciled with the X-ray and radio observations in a simple FS model. The shallow radio light curve between δt=0.23 and 2.19days followed by a decline, together with the shallow radio spectral index at δt≈2.2 days, suggests that ν m passes through the radio band between the first two radio observations. Taking ν m ≈6 GHz at δt≈1 day, we require F ν,max ≈F ν,radio ≈25 μJy. At the time of the HST observations at δt=3.5 days, we thus expect ν m ≈0.9 GHz. For a maximally shallow spectral index of β radio-NIR ≈-0.5, this gives a predicted NIR flux of F ν,F125W ≈0.049 μJy. Even in this optimistic case, the predicted flux is ≈10 fainter than the observed value of F ν,F125W ≈0.55 μJy.
In fact, the observed spectral index between the predicted radio and observed NIR fluxes at δt=3.5 days is extremely shallow, with β R-NIR ≈-0.3, which cannot be explained in the context of an FS model. We find that any model that fits the X-ray and radio behavior will underestimate the observed NIR flux by factors of 5-10 and requires an NIR excess. The NIR excess flux, relative to representative afterglow light curves, and spectral energy distribution (SED) models are shown in Figure 4. In this first scenario (Scenario I), we subsequently model the X-ray and radio afterglows with an FS model and address the NIR excess emission separately in Section 6.W e present an alternative scenario to explain the entire broadband data set (Scenario II) in Section 4.
Setting aside the NIR emission as arising from an additional component, we now outline the available constraints and priors from the radio and X-ray observations and use Markov Chain Monte Carlo (MCMC) analysis to determine the median values and posteriors in the burst explosion properties. We find that for typical parameters the self-absorption frequency ν a ≈(0.8GHz)E K,iso 1/5 n 0 3/5 <ν R . In this regime (the ν 1/3 power-law segment), the radio flux density is sensitive to a combination of kinetic energy and circumburst density (F ν,R ∝E K,iso 5/6 n 0 1/2 ). For the X-ray band, our inference that ν m <ν X <ν c provides an additional constraint on the combination of energy and density (F ν,X ∝E K,iso (3+p)/4 n 0 1/2 ). Since the flux densities in both observing bands depend on n 0 in the same way, the density is expected to be very weakly constrained for this burst. In this regime, the X-ray and radio observations, together with the constraint that ν c >ν X , require ò B 6×10 -2 for ò e ≈0.1 and p≈2.05.
We therefore consider two values of ò B =10 -2 and 10 -3 , selected to be consistent with the above-derived constraint, and also matched to the few values of ò B that have been derived for SGRBs (Fong et al. 2015), to estimate E K,iso and n 0 . We follow the methods outlined in Fong et al. (2015), which use the afterglow flux densities to map to an allowed parameter space for kinetic energy and density. Using the 6.05 GHz observation at δt=0.23 days of F ν,R =33.4±8.2 μJy and the first XRT detection at δt=0.006 days of F ν,X =0.33±0.08 μJy, we determine the respective solutions in the allowed E K,iso -n parameter space. Since the radio and X-ray bands are on different spectral segments, they each provide a unique solution. Taking advantage of the fact that ν c >ν X , we also include an upper limit constraint on the location of the cooling frequency assuming a minimum value at the upper edge of the X-ray band of ν c,min =2.4×10 18 Hz (corresponding to 10 keV). We combine the probability distributions from the two solutions and constraints to obtain a 2D solution and marginalize over the parameter space to obtain 1D solutions: 20 This limit,
log (E K,iso /erg)=51.09±0.22 and log (n 0 /cm -3 )=-1.6±0.50 for ò B =10 -2 and log (E K,iso /erg)=52.06±0.24 and log (n 0 /cm -3 )=-2.54±0.54 for ò B =10 -3 . We use these probability distributions of E K,iso and n 0 in our subsequent multiwavelength modeling as lognormal priors on the corresponding parameters, together with a uniform prior on pä[2.001, 3.01].Wefix ò e =0.1 (Panaitescu & Kumar 2002;Sironi & Spitkovsky 2011) and carry out the modeling using both representative values of ò B . Our priors and assumptions for fixed values are listed in Table 2.
We now explore the parameter space of n 0 , E K,iso , and p in this scenario, using the modeling framework described in Laskar et al. (2014). We incorporate upper limits into the loglikelihood assuming a Gaussian error function. We run 10,000 MCMC iterations, discarding the first few steps as burn-in, after which the log-likelihood and parameter distributions appear stationary. We thin the output samples by a factor of 10 and plot correlation contours and histograms of the results in Figure 5. We list the median parameters derived from the MCMC fit for both values of ò B in Table 2. As expected, the energy and density are poorly constrained, and the output Notes.Afterglow priors and posteriors for two scenarios: (I) a spherical, FS model to fit the radio and X-ray bands, leaving an NIR excess, and (II) a joint FS and RS model with a jet break to explain the broadband data set. a Fixed parameters. b Derived from preliminary considerations, and used as priors for the MCMC. c Lower limit is set by the constraint on the jet opening angle, while the upper limit is set by the isotropic-equivalent value.
posterior is very similar to the input priors. We do, however, probe the joint density between the two parameters, and we find that the major axis of the correlation is aligned along the direction given by E K,iso ∝n 0
. This relation is consistent with the expected degeneracy when ν c is unknown (Laskar et al. 2014), indicating that ν c >ν X provides the dominant source of the correlation between these parameters.
We plot synchrotron light curves for a representative model in Figure 4. For the median parameters, we calculate Y max ≈0.3 and Y max ≈0.7 (at δt≈0.1 days) for ò B =10 -2 and ò B =10 -3 , respectively, confirming that IC cooling occurs deep in the KN regime and does not modify the synchrotron cooling frequency significantly, thus validating our previous assumption regarding IC cooling.
We find that the median parameters of E K,iso ≈(2-20)×10 51 erg and n= (3.4-5.5)×10 -3 cm -3 are close to the median values of cosmological SGRBs for the same values of ò B (Fong et al. 2015). Furthermore, the X-ray and radio data constrain the time of any potential jet break due to collimation effects to δt3.5 days, translating to θ jet 6.5°f or the median values of the E K,iso and n 0 (Sari et al. 1998), comparable to some limits measured for SGRBs. Finally, using the derived range of E K,iso and the value of E γ,iso (15-150 keV)≈8.4×10 49 erg derived in Section 2.1, we calculate a gamma-ray efficiency of η≈0.04.
We can alternatively ameliorate the inconsistency between the radio, NIR, and X-ray observations outlined in Section 3.2 by not requiring the FS to explain the first radio detection at ≈0.2 days. If we extend the ν 1/3 segment to 6 GHz by increasing ν m , the resultant spectrum above ν m can be made to pass through the NIR detection. Now, since α R =0.5 for ν a <ν R <ν m , we would expect F radio,FS (0.2 days)≈9 μJy, which is a factor of ≈3 fainter than the observations. Therefore, we must explain the first radio detection by another component in this model. Early excess flux at radio bands has sometimes been attributed to RS emission in both long and short GRBs (Kulkarni et al. 1999;Soderberg et al. 2006;Laskar et al. 2013;Lamb et al. 2019;Troja et al. 2019). Owing to the limited information available, a variety of RS models are possible. We label this set of models Scenario II.
To derive constraints on the physical parameters in this scenario, we first compare the observed X-ray and NIR behavior to expectations in a standard FS model, as any RS is not expected to contribute significantly in these bands at the times of our observations. The X-ray flux density, extrapolated as a single power law to the time of the first HST observations at δt≈3.5 days, is F ν,X ≈0.0057 μJy. Relative to the observed value of F ν,F125W ≈0.55 μJy, this yields an NIRto-X-ray spectral index of β NIR-X =-0.66±0.06, significantly steeper than the measured β X ≈-0.47. Therefore, simply extending the FS emission as a single β≈-0.5 power law past the NIR would overpredict the X-ray flux by a factor of ≈5 unless an additional spectral break were to be present between the NIR and X-ray bands. If we identify this break as ν c , we expect an X-ray spectral index of β X ≈-1 and a lightcurve decline rate of α X ≈-1. The former is steeper than the observed value of b =- -+ 0.47 X 0.19 0.24 , and the latter is steeper than the observed value of α X ≈-0.67. The shallow X-ray spectrum cannot be easily reconciled and remains a concern for any model attempting to explain the X-ray and NIR observations as arising from a synchrotron FS emission.
On the other hand, we note that fitting the X-ray light curve at δt4×10 -2 days yields a steeper power law, α X =-0.85±0.15, than that obtained from fitting the entire X-ray light curve, and that the latter value is consistent with the expected decline of α≈-1 for the regime ν m <ν NIR <ν c <ν X . Naturally, extrapolating this slope back in time overpredicts the first X-ray detection at δt≈6×10 -3 days, which is one of the shortcomings of this model. One possible solution to this is a continuous injection of energy into the FS between 6×10 -3 days and 5×10 -2 days, such that the FS energy increases by a factor of ≈4 during this period. Similar injection episodes have been inferred for long-duration GRBs in the past (Rees & Meszaros 1998;Björnsson et al. 2004;Laskar et al. 2015). A similar effective energy injection could also be attributable to a slightly off-axis viewing geometry of the jet core at 4×10 -2 days. However, given the paucity of data, it is not possible to obtain meaningful constraints on either effect, and we therefore do not attempt it here. We ignore the first X-ray data point at 6×10 -3 days in our subsequent analysis under Scenario II.
In this scenario, the NIR detection at δt≈3.5 days arises from FS synchrotron emission in the regime ν m <ν NIR <ν c . From the X-ray light curve, we have inferred that p≈2. This Posterior probability density functions and parameter correlations from MCMC fitting for ScenarioI (NIR excess), of all available X-ray and radio afterglow observations of GRB 200522A for ò B =10 -2 . In each posterior distribution, vertical lines denote the median and 68% confidence intervals, while contours in the parameter correlation plots correspond to 1σ, 2σ, and 3σsolutions, respectively. We have fixed ò e =0.1 and employed uniform priors on pä[2.001, 3.01]. For n 0 and E K,iso , we used the constraints derived from the radio and X-ray detections as lognormal priors (Table 2; the derived correlation between these parameters is consistent with the expectation for when ν c is unconstrained, E K,iso ∝n 0 -1/3 ).
implies an NIR decay rate of α NIR ≈-0.75. However, the F125W upper limit at δt≈16.4 days implies a much steeper decline of α NIR <-1.8 at δt3.5 days. GRB jets are expected to be collimated outflows, and the signature of ejecta collimation has previously been observed in SGRB light curves (Nysewander et al. 2009;Fong et al. 2015). One possibility that could explain the steep NIR light curve is that a jet break occurs at 3.5 dayst jet 16.4 days, and we include the possibility of a jet break in our MCMC modeling within Scenario II in the next section.
We now consider constraints imposed on the physical parameters by this RS+FS model. Requiring ν opt <ν c <ν X , taking p≈2.05, and matching the observed radio flux density at ≈2.2 days and the X-ray flux density at ≈0.05 days, we find that no solutions are possible for ò B <1, unless ò e 0.3.
Taking ò e ≈0.3, we find ò B 0.3, n 0 2×10 -2 , and E K,iso 2×10 50 erg. Once again following the methods of Fong et al.
(2015) in the regime ν a <ν R <ν m <ν opt <ν c <ν X and including the constraint ν c <ν X , we obtain log (E K,iso /erg)=50.16±0.22 and log (n 0 /cm -3 )=-1.30±0.21. We use these probability distributions of E K,iso and n 0 in multiwavelength modeling as lognormal priors on the corresponding parameters. We fix ò e =ò B =0.3 and leave p and t jet as additional free parameters.
We do not include the radio point at ≈0.2 days (dominated by the RS in this scenario) and the first X-ray point at ≈6×10 -3 days (as this cannot be explained in this model). We run and process MCMC iterations in a similar fashion to that for ScenarioI. We plot a representative model from our fits in Figure 6. Since ν X <ν c in this scenario, the X-ray band is sensitive to E K,iso , and so this parameter (and, consequently, also n 0 ) is slightly better constrained than in Scenario I. Interpreting the NIR steepening as a jet break allows us to constrain t jet ≈3.4 days, around the time of the NIR detection, which yields a fairly wide opening angle of ≈14°. We follow Sari et al. (1999) to calculate θ jet from t jet , E K,iso , and n 0 and calculate the beaming-corrected kinetic energy (E K ) for each sample. We plot correlation contours between the parameters from the fit in Figure 7 and list summary statistics from the marginalized posterior density functions in Table 2.
In this interpretation, there is only one detection of the putative RS, and thus it is impossible to constrain its properties fully. Under the assumption that ν a,RS <ν m,RS ν R at ≈0.2 days, we require F ν,m,RS ≈80 μJy(ν m,RS /GHz) -0.5 (t/0.2 days) -1. 5 , where we have assumed a spectral index of ≈(1-p)/2≈-0.5 above the RS peak and the time evolution of ν m,RS is appropriate for either a relativistic RS (where it is expected to evolve as t -73/48 ; Kobayashi 2000) or a nonrelativistic RS for the gparameter, g≈2.2 (where it is expected to evolve as t -(15 g+24)/(14 g+7) ∼t -1.5 ; Kobayashi & Sari 2000). For our representative FS model in Scenario II, we have F ν,m,FS ≈80 μJy. Thus, the initial Lorentz factor (assuming equal magnetization of the FS and RS), Γ 0 ≈F ν,m,RS (t dec )/F ν,m,FS , where t dec is the deceleration time (Kobayashi & Sari 2000). This yields In this scenario, the NIR-band temporal evolution is consistent with the FS model with a jet break, but it is steeper than the observed X-rays and underpredicts the early radio emission. In addition, the measured X-ray spectral slope (purple regions, representing 1σconfidence region) is shallower than the predicted slope of β X =-1. In both panels, models and data points are scaled as denoted for clarity. Error bars correspond to 1σand are generally smaller than the size of the symbols, and triangles correspond to 3σupper limits. The NIR model curves pass within 1σ of the NIR data points.
Taking t dec 6×10 -3 days, the time of the first X-ray detection, and ν m,RS 6 GHz at 0.2days, we find a reasonable value for the initial ejecta Lorentz factor, Γ 0 80. We include one such RS model in Figure 6. If, instead, the X-ray detection at δt≈6×10 -3 days is taken as pre-deceleration emission from the FS, we expect an X-ray light curve, F X ∝E(t) 2+p/4 t (2-3 p)/4 ≈t 1-p/2 , as the energy in the FS grows as E∼t during this period (Laskar et al. 2018a). With p≈2.1, this light curve is expected to be essentially flat. In this case, we would infer t dec ≈2.3×10 -2 days and Γ10. Such a low Lorentz factor would likely require a different prompt emission mechanism, such as shock breakout, as postulated for GRB170817A, for instance (Matsumoto et al. 2019). In each posterior distribution, vertical lines denote the median and 68% confidence intervals, while contours in the parameter correlation plots correspond to 1σ,2 σ, and 3σsolutions, respectively. We have fixed ò e =ò B =0.3 and employed uniform priors on pä[2.001, 3.01] and (t jet /days)ä[10 -5 ,1 0 5 ]. For n 0 and E K,iso , we used the constraints derived from the radio, NIR, and X-ray observations as lognormal priors (Table 2). The opening angle (θ jet in degrees) and the beaming-corrected kinetic energy (E K in erg) are derived from the individual Monte Carlo samples.
Using the Pan-STARRS1 Source Types and Redshifts with Machine learning (PS1-STRM) catalog (Beck et al. 2021), the next two closest cataloged galaxies besides the host of GRB 200522A have redshifts of z phot =0.89 and z phot =0.55 at δR=10 3 and 11 9, respectively. While the nearby galaxy at a similar redshift of z phot ≈0.55 could point to an origin in a group, given the star-forming nature of the host coupled with the fairly even photometric redshift distribution of surrounding galaxies, it is unlikely that this burst is part of a low-redshift galaxy cluster.
We model the stellar population properties of the host galaxy of GRB 200522A with Prospector, a Python-based stellar population inference code (Leja et al. 2017). We use Prospector to determine the following stellar population properties and characteristics: stellar mass (M * ), massweighted stellar population age (t m ), dust attenuation (A V ), stellar metallicity (Z * ), and star formation history (SFH) characterized by an e-folding factor τ. We apply a nested sampling routine with dynesty (Speagle 2020) to the observed photometry and spectroscopy and produce model SEDs with Python-fsps (Flexible Stellar Population Synthesis; Conroy et al. 2009;Conroy & Gunn 2010). For our fits, we fix redshift to the value of the spectroscopically determined redshift, z=0.5536 (see Section 2.6), and leave all other parameters free. We jointly fit the observed photometry and spectrum of the host of GRB 200522A weighted by the 1σphotometric uncertainties and error spectrum.
We initialize our stellar population models with a Chabrier initial mass function (IMF; Chabrier 2003) and Milky Way dust extinction law (Cardelli et al. 1989). We use a parametric, delayed-τ SFH, given by
where SFR is star formation rate, M F is the total mass formed from dust to stars over the lifetime of the galaxy, and t represents the age of the galaxy at which star formation commences. Prospector provides posteriors on M F , t, and τ from which we determine the posteriors in M * and massweighted age, t m , using the SFH and analytic conversions from total mass to stellar mass (Equation (2) in Leja et al. 2013, anddetailed on Nugent et al. 2020). We choose t m as the stellar population age metric, to avoid disproportionately weighting contributions from younger, brighter stars (as is the case for simple stellar population ages; Conroy 2013) and to provide a more robust estimate of when the SGRB progenitor could have formed.
We also employ a 10th-order Chebyshev polynomial to fit the spectral continuum. We include a model for nebular emission, characterized by two additional free parameters: log (Z gas /Z e ), which measures gas metallicity, and a parameter for gas ionization. Finally, we impose a 2:1 ratio on the amount of dust attenuation between the younger and older stellar populations, respectively, as young stars in SF regions typically experience twice the amount of dust attenuation as older stars (Calzetti et al. 2000;Price et al. 2014).
We present the resulting posterior distributions of the free parameters in Figure 8 and report the median values and bounds corresponding to 68% credible intervals in Table 3. The observed host galaxy photometry and spectrum, along with the model spectrum and photometry characterized by the Prospector median parameters, are shown in Figure 8. The shape of the spectrum and the locations of the emission lines are well fit by the model. We find that the host is characterized by a young stellar population with t m ≈0.53 Gyr, M * ≈4.5×10 9 M e , A V ≈0, and near-solar stellar metallicity of log (Z * /Z e )≈0.02. The determined log (Z gas /Z e ) is ≈-0.07, approximately the expected value from the M-Z relation at redshifts of 0.07<z<0.7 (Savaglio et al. 2005;Kewley & Ellison 2008). Based on these parameters, we calculate an SFR of ≈2.1 M e yr -1 and a specific SFR per unit mass (sSFR) of 4.7×10 -10 yr -1 .
We measure the flux-weighted centroids and integrated fluxes of the nebular emission lines using a custom Python routine. 21 The derived line centroids and emission-line fluxes and uncertainties are shown in Table 4. The observed Hα-to-Hβ line ratio of ≈2.88 is consistent with the expectation for ionization equilibrium under case B recombination at a typical nebular temperature of 10 4 K and electron density of 10 2 cm -3 (Osterbrock 1989). This indicates no additional extinction (A V 0.1 mag) along the line of sight to star-forming regions within the host, consistent with the results from SED fitting. For the observed Hα line flux (Table 4), we obtain an Hα line luminosity of L(Hα) = (6.21 ± 0.59) × 10 41 erg s -1 . Using standard conversions (Kennicutt 1998;Moustakas et al. 2006), we determine SFR(Hα) = 4.90 ± 0.47 M e yr -1 . This is a factor of ≈2 larger than the SED-derived SFR, although we note that both diagnostics can have systematic uncertainties by factors of ≈2 or more (Moustakas et al. 2006;Theios et al. 2019), and we report both values for completeness. The Hα-derived value gives sSFR ≈ 10.5 × 10 -10 yr -1 .
Using the calibration of Curti et al. (2017), searching over a grid of the metallicities derived from the R 2 , R 3 , R 23 , and O 32 metallicity diagnostics (equally weighted), and using the solar photospheric oxygen abundance from Asplund et al. (2009),we find a gas-phase metallicity of 12+log(O/H)=8.54±0.03, or log (Z gas /Z e )=-0.16±0.03, similar to the value of log (Z gas /Z e )≈-0.1 from SED modeling.
We use the GALFIT software (Peng et al. 2007) to fit the 2D surface brightness profile of the host galaxy of GRB 200522A in each of the F125W and F160W images. For each image, we perform a three-component fit representing the galaxy, the neighboring galaxy to the southeast, and the sky background. We use Sérsic surface brightness profile models for the two galaxies, allowing the centroid, central surface brightness, effective radius (r e ), and Sérsic index n to vary. The resulting best-fit F160W solution is characterized by n=2.3 and r e =0 60 for the host, with c = n 2.2 2 . For F125W, the bestfit solution is n=2.1 and r e =0 60. At the redshift of GRB 200522A, the host effective radius becomes r e =3.90 kpc. Taking into account the size of the host galaxy, we also calculate a host-normalized offset of δR=(0.24±0.04)r e (Table 3). The residual images exhibit a clean subtraction of the neighboring galaxy, an indication that it is well modeled by GALFIT. On the other hand, the residuals for the host galaxy exhibit clear structure in both filters, extending from NW to SE. The galaxy appears to be bulge dominated with a disturbed outer stellar halo, potentially indicative of a fairly recent galaxy merger or interaction with a neighboring galaxy.
We also determine the location of GRB 200522A with respect to its host light distribution, using the "fractional flux" diagnostic (FF; Fruchter et al. 2006). The FF is defined as the fraction of cumulative host light in pixels fainter than brightness level at the counterpart position. It is a complementary diagnostic to probe the burst's location relative to its host galaxy that is independent of host morphology. Using the position of the NIR counterpart, and employing a 1σcutoff to determine the bounds of the host galaxy, we calculate fractional flux values of 0.95-0.96 for the two filters, indicative of a strong correlation with its host stellar mass distribution. The derived morphological properties, offset, and FF values are listed in Table 3.
The total observed NIR luminosity of GRB 200522A is L F125W,tot ≈1.7×10 42 ergs -1 and L F160W,tot ≈1.3×10 42 ergs -1 at a rest-frame time of δt rest ≈2.3 days. This emission may be interpreted as originating from the FS of a GRB synchrotron afterglow (Scenario II in Section 4). However, the broadband observations require an RS to explain the early radio excess and a jet break to explain the steep NIR decline. Moreover, the model predicts a steeper X-ray decline than the observed rate.
In this section, we further consider the implications of Scenario I, in which the radio emission and X-ray emission originate from an FS, with an excess NIR luminosity relative to this model by factors of ≈5-10 (Section 3). We explore viable emission mechanisms that can explain the observed GRB 200522A F125W and F160W luminosities (corresponding to rest-frame i and y bands, respectively). The detected NIR emission observed in GRB 200522A clearly lies in a unique part of parameter space. It is well below the afterglow luminosities of detected SGRBs (Figure 9), albeit with sparser sampling in the relevant bands and on the same timescales. Meanwhile, it is significantly more luminous than any known kilonova in the same rest-frame bands, which on average have νL ν ≈10 41 ergs -1 at similar rest-frame times. The observed luminosities of previous SGRB kilonovae and GW170817 match expectations for kilonovae powered by pure radioactive heating ("radioactively powered"; Figure 9;L i& Paczyński 1998; Metzger et al. 2010;Tanaka et al. 2014). The NIR excess emission of GRB 200522A has a luminosity intermediate to detected on-axis SGRB afterglows and known kilonovae or kilonova candidates. Furthermore, we find that GRB 200522A is significantly bluer than GW170817, which had a color of (i-y)=0.58±0.10 at the same rest-frame time. Compared to GRB 160821B, the only other SGRB kilonova candidate with data adequate for comparison, the NIR counterpart is slightly bluer than GRB 200522A (with (r-i)≈0.10±0.26 and (y-J)≈0.26±0.04 at ≈1.7-3.3 days; Lamb et al. 2019), although consistent within the uncertainties.
We first explore the possibility that the luminosity and color (i-y=-0.08±0.21) of the NIR counterpart to GRB 200522A can be explained by pure r-process radioactive decay. The observed NIR luminosity is ∼10 times greater than that of other known kilonovae or candidates at similar epochs (Figure 9). If attributed solely to radioactivity, this implies that the kilonova accompanying GRB 200522A ejected a higher mass than other kilonovae, was heated by radioactivity at a higher specific heating rate rp ( = --
) than is commonly assumed ( rp,typ ), or experienced some combination of these effects, subject to the rough constraint (M ej /M ej,typ ) × » 10 rp rp,typ ( ) . R-process radioactivity is generally divided into two regimes: a heavy or main r-process, and a light r-process. The first occurs in extremely neutron-rich conditions and produces heavy elements (lanthanides and actinides) whose high opacities cause the resulting emission to peak at redder (e.g., NIR) wavelengths (Barnes & Kasen 2013;Kasen et al. 2013;Tanaka & Hotokezaka 2013). In contrast, the latter, a product of relatively neutron-poor outflows, synthesizes a lighter composition with a lower opacity, leading to a transient that generally peaks at bluer (optical) wavelengths. Though GW170817 showed evidence of both a light and a main r-process (Kasen et al. 2017;Villar et al. 2017;Metzger 2019), the bluer color of GRB 200522A suggests that its emission is dominated by a light r-process, low-opacity component. This is not unexpected for kilonovae viewed from the polar direction (Metzger & Bower 2014;Perego et al. 2014;Wanajo et al. 2014;Sekiguchi et al. 2015;Barnes et al. 2016;Kilpatrick et al. 2017), or whose central remnants are long-lived NSs. In the latter case, neutrino irradiation of the accretion disk by the central NS will raise the electron fraction (Y e ; the number of electrons per baryon) of outflowing disk material, inducing a light r-process (Metzger & Fernández 2014;Kasen et al. 2015;Lippuner et al. 2017). Magnetar winds from the NS surface can Rest-frame 0.7-0.95 μm (left; i and z bands) and 0.95-1.3 μm (right; y and J bands) luminosity vs. rest-frame time compilations. The data displayed include GRB 200522A (purple star), SGRB light curves including afterglow emission (blue squares),3 σupper limits (blue triangles), and the kilonovae of GRB 130603B (Berger et al. 2013a;Tanvir et al. 2013), GRB 160821B (Lamb et al. 2019;Troja et al. 2019), and GW170817 (Villar et al. 2017). Compared to the radioactively powered kilonova of GW170817, the NIR counterpart of GRB 200522A is ≈8-17 times more luminous. GRB 200522A is also significantly more luminous than other kilonova candidates in the rest-frame i and z bands and relevant times. We propose that the NIR counterpart is a kilonova with luminosity boosted energy deposition from a magnetar ("magnetar-boosted"; dashed line), or a radioactively powered kilonova with distinct ejecta properties from previously observed kilonovae.
provide additional high-Y e , low-opacity material (Metzger et al. 2018).
The apparent low opacity complicates the question of enhanced r-process heating for GRB 200522A. There is some variability in predictions of r-process heating rates, due to the uncertain physics of the neutron-rich nuclei involved and the diverse astrophysical conditions that may characterize an rprocess event (see, e.g., Barnes et al. 2016;Rosswog et al. 2017). However, these uncertainties are greatest for the heaviest nuclei, while the relatively blue color of the NIR counterpart to GRB 200522A suggests an r-process that failed to fuse many elements with A130 and a light r-process. (The higher temperatures associated with higher specific heating rates could in theory push the thermal SED blueward, reproducing the blue colors without the requirement of low opacity. However, we found that absent an extreme choice of heating rate, this effect was too small to overcome the reddening from from high-opacity lanthanides and actinides if these are present at mass fractions greater than X lan ∼10 -3 .) If the NIR counterpart is to be explained by pure radioactive decay, the observed color seems to require a weak (lowlanthanide) r-process.
As a test case, we consider an outflow with ejecta mass M ej =0.1 M e , an average ejecta velocity v ej =0.15c, and a combined lanthanide and actinide mass fraction of X lan =10 -3 . This could be considered a pure radioactive energy analog to the magnetar-boosted model (Section 6.3). Such a scenario might arise if an NS central remnant survived long enough to neutrino-irradiate its accretion disk and drive the material to a high Y e (e.g., Lippuner et al. 2017), but not long enough to impart its spin-down energy to the ejecta (however, see also Miller et al. 2019, who suggest that a central NS may not be necessary for a high-Y e disk outflow).W e simulate the resulting emission using the radiation transport code Sedona (Kasen et al. 2006), parameterizing the r-process heating rate with a power law,
The power-law index α=1.3 is a standard analytic approximation for r-process heating. It is expected from Fermi's theory of β-decay (Hotokezaka et al. 2017;Kasen & Barnes 2019) and has been shown to be consistent with the results of detailed numerical models of the r-process(e.g., Metzger et al. 2010;Korobkin et al. 2012). Typical values for rp,0 are ∼10 10 erg s -1 g -1 . Here we consider a range of models, = 13 1 0 rp,0 10
(-) ergs -1 g -1 (Figure 10). While not all of the energy released by the r-process is actually available to power the kilonova's electromagnetic emission, due to inefficient thermalization of radioactive energy (Barnes et al. 2016), thermalization is efficient at early times and for more massive and/or slower-moving ejecta. We therefore absorb the effects of thermalization into Equation (3) and assume in our radiation transport calculation that all emitted energy is efficiently absorbed.
Our radioactively powered model is able to reproduce both the color and the observed i-and y-band luminosities of the NIR counterpart of GRB 200522A only for 1.5 10 rp,0 10 erg s -1 g -1 (Figure 10). This is a factor of 1.5 higher than what has typically been assumed. For example, the kilonova models of Kasen et al. (2017) and Chornock et al. (2017) to explain GW170817 had an effective heating rate (including thermalization) approximately equal to 8×10 9 (t/day) -1.3 erg Figure 10. NIR counterpart (F125W: blue star and triangle; F160W: green diamond) of GRB 200522A alongside radioactive models with enhanced heating (pink lines). The four Sedona light-curve models shown assume a power-law heating rate with a range of fixed radioactive heating rate constants, rp,0 (1×10 10 erg s -1 g -1 to 3×10 10 erg s -1 g -1 ; pink lines), a lanthanide and actinide mass fraction of X lan =10 -3 , ejecta mass of M ej =0.1 M e , and ejecta velocity of v ej =0.15c. These parameters have been chosen in attempts to match the luminosity and color of GRB 200522A; all of these models are significantly more luminous than GW170817 (gray diamonds). For these model parameters, the NIR counterpart of GRB 200522A requires rp,0 1.5×10 10 ergs -1 g -1 , a factor of ≈1.9 larger than assumed for GW170817 (Chornock et al. 2017;Kasen et al. 2017).
s -1 g -1 for 0.1t/day5, lower than the model that can explain GRB 200522A by a factor of ∼1.9.
Assuming that β-decays supply most of the radioactivity and that the difference between emitted and thermalized radioactive energy is due only to neutrinos, which carry away ∼1/3 of the energy of a typical β-decay, our results suggest a true r-process heating rate of »´- t 2.3 10 day rp 10 1.3
() erg s -1 g -1 .I n summary, if the NIR emission of GRB 200522A is produced by a radioactively powered kilonova, the properties of this ejecta (e.g., mass, heating, and/or composition) must be different from those inferred for GW170817. Detailed models exploring these properties, coupled to more detailed heating prescriptions, are required to fully understand the NIR counterpart of GRB 200522A in the context of radioactive models, as well as implications for other kilonovae.
As described in the previous section, the NIR emission and color of GRB 200522A are difficult to explain by a radioactive heating alone, under standard assumptions about ejected mass and the specific heating from r-process decay. However, it is possible that deposition of energy from an NS remnant created as a result of the merger can boost the optical and NIR luminosity of the kilonova by up to a factor of ≈100 ("magnetar-boosted" kilonova; Yu et al. 2013;Metzger & Piro 2014; see also Kisaka et al. 2016;Matsumoto et al. 2018, for general "engine-powered" models). Indeed, a small fraction of BNS mergers are expected to produce a supramassive NS remnant that is indefinitely stable to collapse (e.g., Margalit & Metzger 2019). The remnant may acquire large magnetic fields during the merger process and is necessarily spinning near breakup (e.g., Siegel et al. 2013;Kiuchi et al. 2018;Mösta et al. 2020), resulting in a rapidly spinning "magnetar," which provides a reservoir of energy via spin-down that is not available in the scenario of a prompt collapse to a black hole. Since the kilonova ejecta mass is expected to be of order M ej ≈0.01-0.1 M e (Metzger 2019), in this scenario, the rotational energy is deposited behind the ejecta into an expanding nebula with a nonthermal component in the X-ray band and a thermal component peaking at optical and NIR wavelengths. The combined effects of the magnetar in depositing its spin-down energy and accelerating the ejecta provide the significant luminosity boost, as opposed to magnetar remnant models that rely on radioactive heating (see Kawaguchi et al. 2020).
We investigate the feasibility that the NIR excess emission of GRB 200522A can be explained by a magnetar-boosted kilonova. Using the formalism presented in Metzger (2019) (accounting for corrections to the effective engine luminosity from Metzger & Piro 2014),w efix the opacity to κ=1cm 2 g -1 (corresponding to an electron fraction, Y e ≈0.4, in the "blue" regime), as was found to explain the early blue emission of GW170817 (Tanaka et al. 2020).W e employ light-curve models with magnetic field strengths of B=(2.5-3)×10 15 G, initial spin period P 0 =0.7 ms (corresponding roughly to the breakup rate), and a total ejecta mass of M ej =0.1 M e (similar to the disk wind ejecta in the case of a long-lived neutron star; e.g., Metzger & Fernández 2014). The spin-down luminosity (L X,sd ∝t -2 ) provides an energy reservoir, which powers the expanding nebula, and which is thermalized at optical and NIR wavelengths. The nebula is not expected to be transparent to X-rays until the ejecta are ionized (on 1 to few-day timescales). A comparison of our model to the X-ray observations of GRB 200522A demonstrates that the predicted nebular X-ray emission is a factor of ≈2 below the observed values (Figure 11), although it does have a similarly shallow decline rate at 0.4 days of L X,neb ∝t -0.6 . Thus, the observed X-ray emission of GRB 200522A is likely to be dominated by the FS afterglow emission in this model. In addition, the early radio emission cannot easily be explained by the same physical origin as the X-ray nebula, as the synchrotron self-absorption frequency is too large to allow radio emission to escape on these timescales. We note that the NIR photons from the nebula may provide an additional source of cooling for X-ray synchrotron-emitting electrons at the FS. However, for the high Lorentz factor of the FS at the time of the X-ray observations (δt3.5 days; Γ6), this effect is negligible even for the high NIR photon density inferred here (Linial & Sari 2019).
We find that the magnetar model matches the colors and luminosity of the NIR excess emission (Figure 11). For these parameters, the peak of the kilonova SED is significantly bluer than our observing bands: at δt rest ≈2.3 days, the effective temperature is T eff ≈6430-6960 K, corresponding to λ pk ≈0.42-0.45 μm. Thus, our HST observations only 11. X-ray afterglow light curve and NIR excess of GRB 200522A (purple, blue, and green points); triangles represent 3σupper limits. Also shown are the predicted emission contributions of a magnetar model with B=2.5×10 15 G, P=0.7 ms, and κ=1cm 2 g -1 (Metzger & Piro 2014;Metzger 2019). A fraction of the spin-down luminosity (dotted line) powers the nonthermal nebular X-rays (purple solid line), the latter of which is predicted to be subdominant compared to the FS afterglow. The nebular emission is also thermalized into an optical/NIR "magnetar-boosted" kilonova with a peak bolometric luminosity of ≈3×10 43 ergs -1 (dotted-dashed gray curve). The contribution in the F125W and F160W bands (rest-frame i and y bands) are shown as solid lines. The bottom panel illustrates the fraction of luminosity in two HST filters contributing to the bolometric kilonova luminosity, ≈3%-5% at the time of the HST detections. account for ≈3%-5% of the predicted bolometric kilonova luminosity at that time (Figure 11).
In the context of interpreting the NIR excess emission of GRB 200522A as a kilonova, we are thus motivated to directly compare the NIR emission to that of GW170817 and to the landscape of SGRBs with optical or NIR emission (or limits) within ≈10 times the luminosity of GW170817 across all observed bands (Figure 12).
Our comparison sample of relevant SGRBs consists of GRB 050709 (Jin et al. 2016), GRB 130603B (Berger et al. 2013a;Tanvir et al. 2013), GRB 1501010B (Fong et al. 2016b), and GRB 160821B (Lamb et al. 2019;Troja et al. 2019). 22 For GRB 160821B we include only optical detections at 1.75 daysδt rest 5 days and NIR detections at δt rest 1.5 days, where the kilonova emission was found to dominate the afterglow (Lamb et al. 2019;Troja et al. 2019). We also include highly constraining afterglow upper limits (e.g., GRB 050509B, Cenko et al. 2005;GRB 061201, Fong et al. 2015; GRB 160624A) and low-luminosity SGRB afterglows that do not have existing kilonova interpretations (GRBs 050724A, Berger et al. 2005;GRB 080905A, Rowlinson et al. 2010b;GRB 090515, Rowlinson et al. 2010a). Each SGRB has a clear, well-measured redshift that allows us to calculate accurate luminosities. For each of the bursts, we select the most relevant or constraining observations available in the observed grizyJ bands.
For GW170817, we make use of the available multiband light curves compiled in Villar et al. (2017), performing a linear interpolation in 1 hr time bins, transforming them to rest-frame luminosities and times. Similarly, we transform each of the SGRB observations to their rest-frame wavelengths, luminosities, and times. For each SGRB observation, we compute the ratio of luminosities, n = n L (SGRB)/νL ν (GW170817),a t the relevant rest-frame time. We show the ratio versus restframe time. The gray horizontal line represents a 1:1 ratio ( = 1) against which each SGRB observation can be independently compared.
It is clear that the NIR excess observed in GRB 200522A is significantly more luminous than candidate kilonovae and GW170817 (Figure 12). The color evolution from blue to redder bands over time as expected for kilonovae is overall apparent. The NIR counterpart of GRB 200522A at δt rest ≈2.3 days is significantly brighter than GW170817, with » 16.8 and 8.8 in the i and y bands, respectively. These ratios are also significantly higher than » 4 for the candidate kilonovae of GRB 130603B and GRB 150101B (Berger et al. 2013a;Tanvir et al. 2013;Troja et al. 2018). GRB 200522A is ≈9.3 and ≈13.2 times more luminous than GRB 160821B, the only SGRB kilonova candidate for which data exist at similar restframe times and bands. Overall, Figure 12 highlights the diversity of late-time excess emission in SGRBs in terms of luminosities and colors (see also Gompertz et al. 2018;Ascenzi et al. 2019;Rossi et al. 2020). It also highlights the effectiveness of searches traditionally fine-tuned for afterglows in reaching the depths required to detect nearby (z0.3) kilonovae similar to the luminosities of GW170817.
First, we examine the host of GRB 200522A in the context of the SGRB population and field galaxies. GRB 200522A is located at a small projected physical offset of ≈1 kpc, or ≈0.24r e from the center of its host galaxy, closer than 90% of SGRBs (Fong & Berger 2013). The location of GRB 200522A is also indicative of a strong correlation with its host stellar mass distribution, residing at the 95% level in terms of its host rest-frame optical light. However, the low afterglow-inferred circumburst density of ≈10 -3 to 10 -2 cm -3 is somewhat surprising given its placement in its host galaxy (modulo projection effects); indeed, the inferred value is in line with the typical expected densities of SGRBs, the majority of which occur at significantly larger offsets. The host galaxy also exhibits an asymmetric morphology with a bulge and a disturbed disk, potentially indicative of a recent merger or fly-by encounter.
Compared to the host galaxies of other SGRBs, the host of GRB 200522A comprises a fairly young, low-mass stellar population, falling in the lower 38% and 25% of all SGRB host stellar masses and ages that have been derived in a similar manner (Nugent et al. 2020). Compared to the galaxy luminosity function at this redshift, the host galaxy has a luminosity ≈0.5L * (Willmer et al. 2006), on the low end for SGRB hosts. Approximately 70% of SGRB host galaxies have evidence of ongoing star formation (Fong et al. 2013), with a median SFR≈1 M e yr -1 (Berger 2014); in comparison, the host of GRB 200522A is more strongly star-forming than most SGRB hosts, with SFR≈2.1-4.8 M e yr -1 . However, compared to field galaxies of similar stellar mass at 0.5<z<1, the host is consistent with or just below the main locus of starforming galaxies on the main sequence, depending on where in the range the true SFR is (Whitaker et al. 2014;Fang et al. 2018). This means that given its stellar mass, the host of GRB 200522A is forming stars comparable to or at a slightly lower rate than contemporary field galaxies.
Shocks in SGRBs We now place the broadband properties of GRB 200522A and its host galaxy in the context of the SGRB population. The possible presence of γ-ray precursor emission on timescales of <1 s of the main pulse of GRB 200522A is intriguing, given that only ≈10% of Swift/BAT SGRBs have been found to have such emission (Troja et al. 2010). Furthermore, most SGRBs with precursor emission had significantly longer quiescence timescales of tens of seconds between the precursor and the GRB; only one other event, GRB 090510, had a detected precursor within 1s. The physical origin of precursor emission is unknown. Theoretical models include the excitement of tidal resonances between the component neutron stars during the merger (Tsang et al. 2012;Suvorov & Kokkotas 2020) and accretion onto a magnetar central engine (e.g., Bernardini et al. 2013).
Turning to the afterglow emission, the radio afterglow of GRB 200522A represents the eighth radio afterglow detection for an SGRB out of a total of >70 events observed. The lack of radio detections has been attributed to the relatively lower energy scales and circumburst densities (Fong et al. 2015) compared to their long GRB counterparts (Panaitescu & Kumar 2002;Yost et al. 2003;Cenko et al. 2010Cenko et al. , 2011;;Laskar et al. 2014Laskar et al. , 2015)). Using the redshift of GRB 200522A, the radio afterglow luminosity is νL ν =(2.5±0.6)×10 39 ergs -1 at δt rest =0.15 days, and the radio counterpart was detected through δt rest =1.4 days. To compare the luminosity and behavior to those of other radio afterglows, we collect available radio afterglow data taken at 5-10 GHz frequencies for SGRBs with redshifts. For the radio afterglow detections, we gather data for GRB 050724A (Berger et al. 2005), GRB 051221A (Soderberg et al. 2006), GRB 130603B (Fong et al. 2014), GRB 140903A (Troja et al. 2016b), GRB 141212A (Fong et al. 2015), and GRB 160821B (9.8 GHz;Lamb et al. 2019). In addition, we reduce and analyze 9.8 GHz observations for GRB 150424A and 5.0 GHz data for GRB 160821B (Program 15A-235; PI: Berger; Fong 2015; Fong et al. 2016a) and present their fluxes and upper limits here. Finally, we include upper limits for 18 SGRBs with redshifts from Fong et al. (2017). The total sample of SGRB radio afterglows with redshifts comprises 27 events, and their radio luminosity light curves are shown in Figure 13 and listed in Table 5.
For the detections, the redshifts span z=0.16-0.596, tracing the low-redshift end of the distribution of SGRBs (Paterson et al. 2020), which can be attributed to observational selection effects. While GRB 200522A is among the most distant radio afterglow detections, we find that its luminosity is unexceptional, and squarely in the range of those traced by SGRBs, which have νL ν ≈10 39 -10 40 ergs -1 . The one exception is GRB 160821B, whose radio afterglow was an order of magnitude less luminous than the other GRBs; together with its multiwavelength data, that event was interpreted as a slightly off-axis structured jet (Troja et al. 2019) or the result of a narrow jet with an RS (Lamb et al. 2019). Finally, for context, the peak radio luminosity of the off-axis afterglow of GW170817 was ≈8×10 35 ergs -1 at δt≈160 days (Alexander et al. 2018;Dobie et al. 2018;Margutti et al. 2018), well below those of on-axis SGRB afterglows. We also note that the X-ray afterglow of GRB 200522A falls just below the median luminosity for XRT afterglows. Overall, the radio emission and X-ray emission of GRB 200522A seem to exhibit similar behavior to those of on-axis SGRB afterglows.
One of the ways to explain the multiwavelength radio to X-ray light curves of GRB 200522A is through the standard synchrotron FS model, together with an RS and a jet break. RSs are expected in weakly magnetized, baryonic ejecta and provide a means to infer the jet initial Lorentz factor (Γ 0 ) and the relative magnetization (R B ) of the ejecta (Sari & Piran 1999;Harrison & Kobayashi 2013). As the RS peak frequency is suppressed by a factor of Γ 0 2 relative to the FS, the RS is expected to be more easily detectable at radio frequencies (Kobayashi & Sari 2000;Kopac et al. 2015). This has been borne out by observations of long-duration GRBs with the VLA, revealing a wide diversity in initial Lorentz factors (Γ 0 ≈100-300) and magnetization properties (R B ≈0. 5-10;Laskar et al. 2013;Perley et al. 2014;Laskar et al. 2016 2018a, 2018b, 2019a, 2019b).
Similarly, RSs have been used to explain the early-time radio and optical excesses at 1 day in three SGRBs to date. GRB 051221A (Soderberg et al. 2006) and GRB 160821B (Lamb et al. 2019;Troja et al. 2019) each exhibited radio excess emission relative to the FS model, followed by subsequent fading, while for the more recent GRB 180418A, an RS was invoked to explain an excess of optical emission at early times (Becerra et al. 2019). For GRB 200522A, the RS interpretation is driven by the early radio emission.
We interpret the steep NIR decline as post-jet-break behavior with a jet break at t jet ≈3.4 days, leading to a relatively wide opening angle of ≈14°. Two other SGRBs with RS signatures, GRB 051221A and GRB 160821B, also had temporal steepenings in their light curves interpreted as jet breaks, leading to opening angles of ≈7°and ≈2°-8°, respectively (Soderberg et al. 2006;Lamb et al. 2019;Troja et al. 2019).If this interpretation for GRB 200522A is correct, this would be the widest jet measurement that exists for an SGRB, as SGRBs with measured jets have inferred ≈2°-8°(median of 6°±1°; Fong et al. 2015). In addition, only a few events have comparable lower limits indicative of wider jets, including GRB 050709, GRB 050724A, and GRB 120804A with 13°-25° (Grupe et al. 2006;Watson et al. 2006;Berger et al. 2013b).
Implications for Future Detectability Another way to understand the multifrequency light curves and SEDs of GRB 200522A is by interpreting the NIR emission as a luminous kilonova. While the NIR detections of GRB 200522A are fainter than any on-axis afterglow detected to date at these epochs, they are a factor of ≈8-17 times the luminosity of GW170817, and more luminous than any known kilonova or kilonova candidate across all observing bands (Figure 12). Deep observations of SGRBs on the same timescales have ruled out emission with similar luminosities to the NIR counterpart to GRB 200522A for only two other events (Figure 9).W efind that such a luminous NIR counterpart could be driven by heating from the spin-down of a nascent magnetar or through a radioactively powered model with enhanced specific heating rates (a factor of 2 larger than that assumed for GW170817), a low-lanthanide composition, and a fairly high ejecta mass.
If the progenitor of GRB 200522A indeed produced a magnetar that is stable to collapse, synchrotron radio emission resulting from the interaction between the expanding ejecta and the surrounding medium is predicted on a few ≈year timescales (Metzger & Bower 2014;Hotokezaka & Piran 2015;Liu et al. 2020) 2020b) and an inference on the fraction of SGRBs that produce stable magnetars of 50% (Schroeder et al. 2020b). We use the light-curve modeling described in Schroeder et al. (2020b) for an energy deposition of 10 53 erg representing the maximum energy extractable from a stable remnant, as is expected to explain the magnetar-boosted kilonova interpretation for GRB 200522A. We fix the median parameters from the FS model (ò B =0.01). For a fixed ejecta mass of M ej =0.03 M e (0.1 M e ),w efind that the 6 GHz radio emission will peak at δt≈1.5 yr (≈9.9 yr) after the burst with a flux density of F ν ≈180 μJy (≈25.3 μJy). Due to the rising light curve, with a peak corresponding to the deceleration timescale (e.g., Nakar & Piran 2011), the radio emission from GRB 200522A will be detectable with the VLA at much earlier times than the peak, reaching F ν ≈20 μJy at δt≈0.3-6.0 yr depending on the ejecta mass. The detection of radio emission from GRB 200522A would be a "smoking gun" of this scenario and the first possible evidence of a stable magnetar created as a result of an SGRB.
If the NIR counterpart of GRB 200522A is relatively isotropic, the larger luminosity compared to GW170817 has implications for detectability following gravitational wave (GW) events. Most optical searches following GW events reach depths of ≈21-22 mag (e.g., Hosseinzadeh et al. 2019;Lundquist et al. 2019;Kasliwal et al. 2020). Assuming that the required depth of a search is ≈10 times below peak brightness for robust counterpart detection, kilonovae of comparable brightness to GW170817 are detectable to ≈60-100 Mpc. In comparison, high-luminosity (≈10 42 erg s -1 ) counterparts like that of GRB 200522A will be detectable by current GW counterpart search efforts to ≈160-250 Mpc, well matched to the expected GW network reach of BNS mergers in the O4 observing run (Abbott et al. 2018), and to ≈600 Mpc with the Vera Rubin Observatory (VRO; Ivezić et al. 2019). This is well beyond the expected GW detectability of BNS mergers during the O5 observing run. However, only a small fraction of BNS mergers are expected to produce stable magnetars (Margalit & Metzger 2019; see also Schroeder et al. 2020b, for SGRBs), and thus the expected fraction of high-luminosity counterparts may also be low, if indeed the NIR couterpart of GRB 200522A was a result of a stable magnetar.
It remains uncertain whether a stable magnetar is capable of producing an observable relativistic GRB jet (see Dessart et al. 2009;Murguia-Berthier et al. 2014;Ciolfi 2020). However, alternative and relatively unexplored explanations for GRB 200522A, which are independent of a stable remnant, remain, including variations to the radioactive heating rate, or speculative sources of ejecta heating such as disk winds powered by fallback accretion (which could vary depending on the amount of fallback; e.g., Kisaka et al. 2015;Metzger 2019). Moreover, any modifications to radioactive heating prescriptions would necessarily need to be investigated in the context of all detected kilonovae. Future broadband campaigns following low-z SGRBs will help elucidate the nature and prevalence of the unusual emission of GRB 200522A and in turn the implications on detectability following GW events.
We have presented multiwavelength observations of the counterpart of GRB 200522A and its host galaxy using Swift/ XRT, VLA, HST, Keck, LCOGT, and archival data. We present modeling results of the afterglow and host galaxy and propose scenarios to explain the unusual broadband emission of GRB 200522A.
Against the backdrop of 15 yr of Swift SGRB afterglow discoveries, GRB 200522A represents a remarkable example of the diversity of observed behavior in SGRBs. The detected luminosity of the NIR (rest-frame optical) emission on timescales of ≈few days, during which extremely limited information exists for SGRBs, motivates future such searches with HST, James Webb Space Telescope, and upcoming extremely large telescopes. We come to the following conclusions:
1. The joint X-ray, NIR, and radio observations cannot be explained as synchrotron emission from the GRB FS alone. 2. While the radio and X-ray emission can be well fitt oa n FS, this model underpredicts the observed NIR emission by factors of ≈5-10, leaving an "excess" of NIR (restframe optical) emission. 3. The X-ray and radio luminosity and temporal evolution of GRB 200522A is comparable to that of other cosmological SGRBs. However, the NIR counterpart (≈10 42 erg s -1 ) is subluminous in comparison with detected SGRB afterglows and an order of magnitude brighter than any known kilonova or kilonova candidate. 4. We propose that the NIR (rest-frame optical) excess emission could be a kilonova boosted by energy deposition from a stable magnetar remnant, or a radioactively powered kilonova with modified ejecta or heating properties relative to GW170817. 5. An alternative explanation for the broadband emission of GRB 200522A is an FS with a relatively wide jet opening angle of ≈14°. In this model, the predicted X-ray decline rate is steeper than observed, while the early radio emission is underpredicted, the latter of which can be reconciled with the addition of an RS component. 6. GRB 200522A originated in a bright region of its host galaxy, at a projected offset of ≈1 kpc, or ≈0.24r e , from the center (closer than 90% of SGRBs). The host galaxy is a young (≈0.53 Gyr), modestly star-forming (SFR≈2.1-4.8 M e yr -1 ) galaxy with M * ≈4.5×10 9 M e . 7. The detection of the NIR (rest-frame optical) counterpart to GRB 200522A may contribute to the diversity of counterparts observed accompanying GW-detected BNS mergers. Current (upcoming) optical searches following GW events will be sensitive to such counterparts to ≈160-250 Mpc (≈600 Mpc). However, if the emission of GRB 200522A resulted from a magnetar, the fraction of BNS mergers with such high-luminosity counterparts is expected to be low. 8. If the progenitor of GRB 200522A did indeed produce a stable magnetar, late-time synchrotron radio emission is predicted to become observable with the VLA on ∼0.3-6yr timescales and peak at ≈1-10yr, with the range depending on the ejecta and environmental properties.
Our work demonstrates the power of multiepoch afterglow observations for host galaxy association and uncovering the surprising diversity of broadband properties in SGRBs. Early radio observations of SGRB afterglows at 1 day are key to capturing RS signatures and to constraining the composition of their jets. On the other hand, multifrequency observations at 1-10days are vital for constraining the ejecta collimation and deriving the true cosmological rate of compact object mergers in the era of Advanced LIGO. Future late-time 5-10 yr, sensitive (≈1 μJy) radio searches may be used to test for the presence of the radio emission from any magnetar produced in this and other SGRBs. Such observations in the SKA and ngVLA era may routinely be used to probe the parameter space of initial ejecta mass and magnetic field, thereby constraining magnetar formation and spin-down models and yielding further insight into the GRB central engine and progenitor channels.
The Astrophysical Journal, 906:127 (26pp), 2021 January 10 Fong et al.
https://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3pimms/w3pimms.pl
https://github.com/LCOGT/banzai
http://documents.stsci.edu/hst/wfc3/documents/handbooks/
https://www.stsci.edu/hst/instrumentation/wfc3/data-analysis/ photometric-calibration/ir-encircled-energy
https://github.com/CIERA-Transients/Imaging_pipelines/blob/master/ DEIMOS_pipeline.py
https://github.com/CIERA-Transients/MODS_spectroscopy/blob/ master/spec_SFR_metallicity.ipynb
For this study, we limit the sample to the traditional definition of SGRBs with durations of T 90 <2 s. This excludes the photometric kilonova candidate following GRB 060614(Yang et al. 2015), which had a long duration but lacked an associated supernova to deep limits, leaving the progenitor system up for debate(Fynbo et al. 2006;Gehrels et al. 2006). We note that for GRB 0606014, » 25 -for this event on ≈few-day timescales.