THE AFTERGLOW AND EARLY-TYPE HOST GALAXY OF THE SHORT GRB 150101B AT z = 0.1343

2.1.1. Optical Afterglow Discovery

We obtained r-band observations of GRB 150101B with the Inamori Magellan Areal Camera and Spectrograph (IMACS) mounted on the Magellan/Baade 6.5 m telescope at a mid-time of 2015 January 03.297 UT (δt = 1.66 days, where δt is the time after the BAT trigger in the observer frame). The observations cover the entire BAT position (Figure 1 and Table 1). To assess variability of any optical sources within the BAT position, we obtained a second set of IMACS observations at δt ≈ 2.65 days. Performing digital image subtraction between the two sets of observations using the ISIS software package (Alard 2000), we find a faint point-like residual ≈3'' southeast of a bright galaxy (Figure 1), indicative of a fading optical source.

Zoom In Zoom Out Reset image size

Table 1.  GRB 150101B Optical and Near-IR Afterglow and Host Galaxy Observations

Date	δt	Telescope	Instrument	Band	Exposures	Afterglow	${F}_{\nu ,\mathrm{opt}}$	Host	${A}_{\lambda }$
(UT)	(days)				(s)	(AB mag)	(μJy)	(AB mag)	(AB mag)
2015 Jan 3.297	1.66	Magellan/Baade	IMACS	r	8 × 150	23.01 ± 0.17	2.27 ± 0.36	⋯	0.094
2015 Jan 4.289	2.65	Magellan/Baade	IMACS	r	10 × 120	23.53 ± 0.26	1.41 ± 0.34	⋯	0.094
2015 Jan 4.298	2.66	VLT	HAWK-I	J	12 × 60	>22.3	<4.37	⋯	0.029
2015 Jan 4.310	2.67	VLT	HAWK-I	H	12 × 60	>21.4	<10.0	⋯	0.018
2015 Jan 4.322	2.68	VLT	HAWK-I	K	12 × 60	>21.5	<9.12	⋯	0.012
2015 Jan 4.359	2.72	VLT	HAWK-I	Y	22 × 60	>20.5	<22.9	⋯	0.043
2015 Jan 11.261	9.62	TNGa	NICS	J	⋯	>21.7	<7.59	⋯	0.029
2015 Jan 12.341	10.70	Gemini-South	GMOS	r	19 × 90	>24.2	<0.76	⋯	0.094
2015 Jan 16.201	14.56	TNGa	NICS	J	⋯	>22.4	<3.98	⋯	0.029
2015 Jan 16.307	14.67	VLT	HAWK-I	H	22 × 60	>23.4	<1.58	⋯	0.018
2015 Jan 16.331	14.69	VLT	HAWK-I	Y	22 × 60	>23.6	<1.32	⋯	0.043
2015 Jan 21.533	19.89	UKIRT	WFCAM	J	12 × 200	>22.4	<3.98	15.47 ± 0.05	0.029
2015 Jan 21.569	19.93	UKIRT	WFCAM	K	12 × 200	>22.2	<4.79	15.11 ± 0.05	0.012
2015 Jan 30.508	28.87	UKIRT	WFCAM	J	12 × 200	>23.0	<2.29	15.55 ± 0.05	0.029
2015 Jan 30.544	28.90	UKIRT	WFCAM	K	12 × 200	>22.4	<3.98	15.20 ± 0.06	0.012
2015 Feb 11.320	40.70	HST	WFC3/IR	F160W	3 × 250	>25.3	<0.28	15.09 ± 0.01	0.021
2015 Feb 11.385	40.74	HST	WFC3/UVIS	F606W	4 × 350	>25.1	<0.33	16.57 ± 0.01	0.102
2015 Feb 14.274	43.63	VLT	HAWK-I	H	22 × 60	>23.9	<1.00	⋯	0.018
2015 Feb 19.036	48.39	Magellan/Baade	IMACS	r	8 × 150	>24.6	<0.52	⋯	0.094
2015 Feb 19.048	48.41	Magellan/Baade	IMACS	r	3 × 45	⋯	⋯	16.51 ± 0.04	0.094
2015 Feb 19.053	48.41	Magellan/Baade	IMACS	g	3 × 60	⋯	⋯	17.42 ± 0.04	0.136
2015 Feb 19.058	48.42	Magellan/Baade	IMACS	i	3 × 45	⋯	⋯	16.08 ± 0.04	0.070
2015 Feb 19.063	48.42	Magellan/Baade	IMACS	z	3 × 60	⋯	⋯	15.77 ± 0.05	0.052

Note. All upper limits correspond to 3σ confidence, and uncertainties correspond to 1σ confidence. All values are corrected for Galactic extinction in the direction of the burst, ${A}_{\lambda }$ (Schlafly & Finkbeiner 2011).

^aFrom D'Avanzo et al. (2015).

Download table as:  ASCIITypeset image

To assess the nature of this source and its association with GRB 150101B, we obtained two deeper sets of r-band observations with the Gemini Multi-Object Spectrograph (GMOS) mounted on the Gemini-South 8 m telescope at a mid-time of δt ≈ 10.70 days and with Magellan/IMACS at δt ≈ 48.39 days (Table 1). We perform digital image subtraction between each of the initial epochs and the last set of Magellan observations at δt ≈ 48.39 days, which serves as the deepest template. In the first two subtractions, a point-like residual source is present at the same position, which we consider to be the optical afterglow of GRB 150101B (Figure 1).

To determine the position of the optical afterglow, we perform absolute astrometry between the earliest Magellan observations and the Two Micron All Sky Survey (2MASS) catalog using the IRAF astrometry routines ccmap and ccsetwcs. Using 19 common point sources, the resulting 1σ astrometric tie uncertainty is ${\sigma }_{2\mathrm{MASS}\to \mathrm{GRB}}=0\buildrel{\prime\prime}\over{.} 24$. Since the position of the afterglow is contaminated by galaxy light in our discovery images, we use SExtractor ¹⁸ to determine the afterglow position and uncertainty (${\sigma }_{\mathrm{GRB}}$) in the subtracted images. The subtraction between the first and last epochs (δt = 1.66 and 48.39 days, respectively) yields a positional uncertainty of ${\sigma }_{\mathrm{GRB}}=0\buildrel{\prime\prime}\over{.} 029$. We therefore calculate an absolute optical afterglow position of R.A. = 12^h32^m05094, decl. = −10°56'0300 (J2000), with a total 1σ positional uncertainty of 024.

To measure the brightness of the afterglow in our first two Magellan observations, we perform point-spread function (PSF) photometry using standard PSF-fitting routines in the IRAF/daophot package. We note that the position of the afterglow is contaminated either by flux from the nearby galaxy in the observations or by residuals from the saturated galaxy core in the subtracted images (Figure 1), preventing accurate aperture photometry. We create a model PSF using three to five bright, unsaturated stars in each field out to a radius of $4.5{\theta }_{\mathrm{FWHM}}$ from the center of each star, and then use this PSF to subtract 20 additional stars in this field. The clean subtraction of these stars indicates an accurate model of the PSF for the observation. We then use the model PSF to subtract a point source fixed at the location of the afterglow. We calibrate all of the photometry relative to the Gemini observations at δt = 10.70 days, which were taken in photometric conditions, using the most recent tabulated zero point. Taking into account the uncertainty in the zero point, as well as the Galactic extinction in the direction of the burst (Schlafly & Finkbeiner 2011), we measure an afterglow brightness of ${r}_{\mathrm{AB}}=23.01\pm 0.17$ mag and ${r}_{\mathrm{AB}}=23.53\pm 0.26$ mag at δt = 1.66 and 2.65 days, respectively. Our optical observations and afterglow photometry, supplemented by published values from the literature, are summarized in Table 1.

To obtain upper limits on any transient optical flux from our late-time Gemini and Magellan observations, we inject fake sources of varying brightness into each observation at the position of the optical afterglow using IRAF/addstar. We then perform PSF photometry and measure the uncertainty for each of the fake sources to determine the 3σ upper limit on the afterglow in each image. The resulting limits are listed in Table 1.

2.1.2. Near-IR Upper Limits

2.1.3. X-Ray Afterglow Discovery

Observations with the X-ray Telescope (XRT; Burrows et al. 2005) on board Swift commenced at δt = 139.2 ks and revealed an X-ray source with a UVOT-enhanced position of R.A. = 12^h 32^m 0496 and decl. = −10°56'007 (J2000) and an uncertainty of 18 in radius (90% containment; Evans et al. 2007, 2009; Cummings et al. 2015). However, this X-ray source was found to remain constant in flux over the duration of the XRT observations, $\delta t\approx 139.2\mbox{--}2612$ ks. Furthermore, the position of this X-ray source is coincident with the nucleus of the host galaxy (see Section 2.2), which is a known galaxy 2MASX J12320498-1056010 as cataloged in the NASA/IPAC Extragalactic Database (Helou et al. 1991, pp. 89–106). No other fading source from the Swift/XRT observations was reported.

We analyze X-ray observations obtained at a mid-time of 2015 January 9.565 UT (δt ≈ 7.94 days) with the Advanced CCD Imaging Spectrometer (ACIS-S) on board the Chandra X-ray Observatory (ID: 16508492; PI: E. Troja), with a net exposure time of 14.9 ks. We retrieve the preprocessed Level 2 data from the Chandra archive and use the CIAO/wavdetect routine to perform a blind search of X-ray sources at or near the optical afterglow position. We find an X-ray source located at R.A. = 12^h 32^m 05099 and decl. = −10°56'0280, with a 1σ positional uncertainty of 038, coincident with the position of the optical afterglow (Section 2.1). Within a 12-radius aperture, we compute a net count rate of $9.5\times {10}^{-3}$ counts s⁻¹ (0.5–8 keV), which corresponds to a significance of ∼52σ.

To assess the variability of this X-ray source, we analyze a second set of Chandra/ACIS-S observations obtained at a mid-time of 2015 February 10.305 UT (δt ≈ 39.68 days; ID: 16708496, PI: A. Levan). Fixing the position and aperture to those derived from the first Chandra epoch, we compute a net count rate of $1.3\times {10}^{-3}$ counts s⁻¹ (0.5–8 keV), which corresponds to a significance of ∼6σ. This indicates that the source has faded by a factor of ≈7 between δt ≈ 7.94 and 39.68 days. Due to the fading of this source, as well as the spatial coincidence with the optical afterglow, we confirm this source as the X-ray afterglow of GRB 150101B. We note that the X-ray afterglow position and count rates are fully consistent with an independent analysis of the Chandra data by Xie et al. (2016).

To determine the flux calibration, we extract a spectrum from each of the Chandra epochs using CIAO/specextract. We fit each of the data sets with an absorbed power-law model with photon index, ${{\rm{\Gamma }}}_{{\rm{X}}}$, and intrinsic neutral hydrogen absorption column, ${N}_{{\rm{H}},\mathrm{int}}$, in excess of the Galactic column density in the direction of the burst, ${N}_{{\rm{H}},\mathrm{MW}}=3.48\times {10}^{20}\,{\mathrm{cm}}^{-2};$ (typical uncertainty of ∼10%; Kalberla et al. 2005; Wakker et al. 2011), using Cash statistics. For the first epoch, we find a best-fitting spectrum characterized by ${{\rm{\Gamma }}}_{{\rm{X}}}=1.67\pm 0.17$ and a 3σ upper limit of ${N}_{{\rm{H}},\mathrm{int}}\lesssim 3.0\times {10}^{21}$ cm⁻² (C-stat${}_{\nu }=0.85$ for 92 dof). For the second epoch, the best-fitting spectrum is characterized by ${{\rm{\Gamma }}}_{{\rm{X}}}=1.13\pm 0.49$ and ${N}_{{\rm{H}},\mathrm{int}}\lesssim 2.1\times {10}^{21}$ cm⁻² (C-stat${}_{\nu }=1.21$ for 13 dof). We use the spectral parameters to calculate the count-rate-to-flux conversion factors, and hence the unabsorbed fluxes, in the 0.3–10 keV energy band. The details of the Chandra observations, 0.3–10 keV fluxes, and best-fit spectral parameters are listed in Table 2. To enable comparison of the X-ray light curves to the optical and radio data, we also convert the X-ray fluxes to flux densities, ${F}_{\nu ,X}$, at a fiducial energy of 1 keV (${F}_{\nu ,X}\propto {\nu }^{{\beta }_{X}}$ where ${\beta }_{X}\equiv 1-{{\rm{\Gamma }}}_{{\rm{X}}}$); these values are listed in Table 2.

Table 2.  GRB 150101B Chandra Afterglow Observations

δt	Exposure Time	FX(0.3–10 keV)a	${F}_{\nu }$(1 keV)	${{\rm{\Gamma }}}_{X}$	${N}_{{\rm{H}},\mathrm{int}}$
(days)	(s)	(erg cm−2 s−1)	(μJy)		(cm−2)
7.94	14869	$(1.16\pm 0.12)\times {10}^{-13}$	$(1.08\pm 0.11)\times {10}^{-2}$	1.67 ± 0.17	$\lesssim 3.0\times {10}^{21}$
39.68	14862	$(2.08\pm 0.50)\times {10}^{-14}$	$(1.94\pm 0.47)\times {10}^{-3}$	1.13 ± 0.49	$\lesssim 2.1\times {10}^{21}$

Note. Uncertainties correspond to 1σ, and upper limits correspond to 3σ confidence.

^aUnabsorbed flux.

Download table as:  ASCIITypeset image

2.1.4. Radio Upper Limit

2.2.1. Optical Imaging and Identification

2.2.2. Near-IR

2.2.3. HST Observations

2.2.4. Spectroscopy

2.2.5. Archival Observations: A Low-luminosity AGN

To constrain the explosion properties and circumburst environment, we compare the radio through X-ray observations of GRB 150101B to the standard synchrotron model. We adopt a constant-density medium model (Sari et al. 1998; Granot & Sari 2002), a reasonable assumption for a nonmassive star progenitor. This model provides a mapping from the broadband afterglow observations to the burst physical properties: the isotropic-equivalent kinetic energy (${E}_{{\rm{K}},\mathrm{iso}}$), circumburst density (n), fractions of post-shock energy in radiating electrons (${\epsilon }_{e}$) and magnetic fields (${\epsilon }_{B}$), and the electron power-law distribution index (p), with $N(\gamma )\propto {\gamma }^{-p}$ for $\gamma \gtrsim {\gamma }_{\min },$ where ${\gamma }_{\min }$ is the minimum Lorentz factor of the electron distribution.

To determine the locations of the synchrotron break frequencies with respect to the X-ray, optical, and radio bands and thus constrain the burst explosion properties, we calculate the electron power-law index p, using a combination of temporal and spectral information. The temporal and spectral power-law indices are given by α and β, respectively, where ${F}_{\nu }\propto {t}^{\alpha }{\nu }^{\beta }$. To determine ${\alpha }_{X}$ and ${\alpha }_{\mathrm{opt}}$, we utilize ${\chi }^{2}$-minimization to fit a single power-law model to each light curve in the form ${F}_{\nu }\propto {t}^{\alpha }$. In this manner, we measure ${\alpha }_{X}=-1.07\pm 0.15$ and ${\alpha }_{\mathrm{opt}}=-{1.02}_{-0.55}^{+0.44}$ for the X-ray and optical light curves, respectively. Computing ${\alpha }_{\mathrm{opt}}$ and 1σ uncertanties using the detections alone marginally violates the upper limit at δt ≈ 10.70 days; therefore, we use the observation at ≈10.70 days to set the upper limit on the decline rate, resulting in asymmetric uncertainties. The afterglow light curves, along with the best-fit power-law models, are shown in Figures 3 and 6 below. From the X-ray spectral fit, we find ${\beta }_{X}\equiv 1-{{\rm{\Gamma }}}_{{\rm{X}}}=-0.67\pm 0.17$ (Section 2.1.3 and Table 2).

To determine the location of the cooling frequency, ${\nu }_{c}$, to the observing bands, we compare the temporal and spectral indices to the closure relations (Sari et al. 1999; Granot & Sari 2002). If ${\nu }_{c}\lt {\nu }_{X}$, then the independently derived values for p from the temporal and spectral indices are inconsistent: p = 2.1 ± 0.2 from ${\alpha }_{X}$, p = 1.3 ± 0.3 from ${\beta }_{X}$, and p = 2.0 ± 0.73 from ${\alpha }_{\mathrm{opt}}$. On the other hand, for ${\nu }_{c}\gt {\nu }_{X}$, the values are fully consistent: p = 2.4 ± 0.2 from ${\alpha }_{X}$, p = 2.3 ± 0.3 from ${\beta }_{X}$, and p = 2.4 ± 0.7 from ${\alpha }_{\mathrm{opt}}$. Furthermore, if ${\nu }_{c}\gt {\nu }_{X}$, then ${\beta }_{X}={\beta }_{\mathrm{OX}}$, where ${\beta }_{\mathrm{OX}}$ is the optical-to-X-ray spectral index. Indeed, extrapolating the Chandra data to the time of the optical afterglow using ${\alpha }_{X}$, we calculate ${\beta }_{\mathrm{OX}}\approx -0.61$, which is consistent with the value of ${\beta }_{X}=-0.67\pm 0.17$. Therefore, for the rest of our calculations, we utilize a value of p = 2.40 ± 0.17 as determined by the weighted average and assume ${\nu }_{c}\gt {\nu }_{X}$. We note that this is also equal to the average value for p inferred from the short GRB population (Fong et al. 2015).

Next, we determine a set of constraints on n and ${E}_{{\rm{K}},\mathrm{iso}}$ based on the X-ray and optical flux densities, the radio upper limit, and the condition that ${\nu }_{c}\gt {\nu }_{X}$. Using the X-ray flux density, z = 0.1343, luminosity distance ${d}_{L}=1.96\times {10}^{27}$ cm, ${\nu }_{x}=2.4\times {10}^{17}$ Hz (1 keV), ${\nu }_{\mathrm{opt}}=4.8\times {10}^{14}$ Hz (r-band), and p = 2.4, we obtain the relation (Granot & Sari 2002)

$$\begin{eqnarray}&&{n}^{0.5}{E}_{{\rm{K}},\mathrm{iso},52}^{1.35}{\epsilon }_{e,-1}^{1.4}{\epsilon }_{B,-1}^{0.85}\approx (1.6\pm 0.2)\times {10}^{-3}\end{eqnarray} \tag{ 1 }$$

where n is in units of cm⁻³, ${E}_{{\rm{K}},\mathrm{iso},52}$ is in units of 10⁵² erg, and ${\epsilon }_{e,-1}$ and ${\epsilon }_{B,-1}$ are in units of 0.1. The 1σ uncertainty is dominated by the uncertainty in the individual flux measurements. The constraint on ${E}_{{\rm{K}},\mathrm{iso}}-n$ parameter space from the X-ray afterglow is shown in Figure 2, where the width of the region represents the 1σ uncertainty. We calculate an independent constraint from the optical flux density, assuming that the optical frequency is between the peak frequency, ${\nu }_{m}$, and the cooling frequency, ${\nu }_{c}$, such that ${\nu }_{m}\lt {\nu }_{\mathrm{opt}}\lt {\nu }_{c}$. Thus, the optical and X-ray bands occupy the same spectral regime and have the same dependencies on the physical parameters. We calculate the constraint,

$$\begin{eqnarray}&&{n}^{0.5}{E}_{{\rm{K}},\mathrm{iso},52}^{1.35}{\epsilon }_{e,-1}^{1.4}{\epsilon }_{B,-1}^{0.85}\approx (8.3\pm 1.3)\times {10}^{-4},\end{eqnarray} \tag{ 2 }$$

shown in Figure 2. Next, we impose a constraint from the radio band under the assumption that the radio band is between the self-absorption frequency and the peak frequency such that ${\nu }_{a}\lt {\nu }_{\mathrm{radio}}\lt {\nu }_{m}$. Using the radio upper limit and ${\nu }_{\mathrm{radio}}=9.8\times {10}^{9}$ Hz, we obtain the constraint

$$\begin{eqnarray}&&{n}^{1/2}{E}_{{\rm{K}},\mathrm{iso},52}^{5/6}{\epsilon }_{e,-1}^{-2/3}{\epsilon }_{B,-1}^{1/3}\gtrsim 1.5\times {10}^{-2}.\end{eqnarray} \tag{ 3 }$$

where the constraint imposed on the ${E}_{{\rm{K}},\mathrm{iso}}-n$ is shown in Figure 2. Using the fact that ${\nu }_{c}\gt {\nu }_{X}$, we can use the relative location of the cooling frequency as a final constraint. Setting the cooling break to a minimum value, ${\nu }_{c,\min }=2.4\times {10}^{18}$ Hz (10 keV), at the upper end of the X-ray band, we calculate the relation

$$\begin{eqnarray}&&{n}^{-1}{E}_{{\rm{K}},\mathrm{iso},52}^{-1/2}{\epsilon }_{B,-1}^{-3/2}\gtrsim 3.0\times {10}^{4}.\end{eqnarray} \tag{ 4 }$$

We use Equations (1)–(4) and joint probability analysis (described in Fong et al. 2015) to solve for the isotropic-equivalent kinetic energy and circumburst density. For ${\epsilon }_{e}={\epsilon }_{B}=0.1$, we calculate $n={8.0}_{-6.1}^{+25.2}\times {10}^{-6}$ cm⁻³ and ${E}_{{\rm{K}},\mathrm{iso}}={6.1}_{-2.6}^{+4.5}\times {10}^{51}$ erg (Figure 2). For ${\epsilon }_{e}=0.1$ and ${\epsilon }_{B}=0.01$, we find $n={4.3}_{-3.9}^{+51.1}\times {10}^{-5}$ cm⁻³ and ${E}_{{\rm{K}},\mathrm{iso}}={1.4}_{-0.9}^{+2.3}\times {10}^{52}$ erg (1σ uncertainties). From a comparison of the optical and X-ray afterglow emission, we find no evidence for extinction instrinsic to the burst site or host galaxy (e.g., ${A}_{V}^{\mathrm{host}}=0$). We note that the values for the kinetic energy and density differ slightly from our previous analysis, which included GRB 150101B (Fong et al. 2015) due to a refinement of the optical and X-ray fluxes.

Zoom In Zoom Out Reset image size

Finally, we calculate the isotropic-equivalent γ-ray energy, by ${E}_{\gamma ,\mathrm{iso}}=4\pi {d}_{L}^{2}{(1+z)}^{-1}{f}_{\gamma }$ erg. Using the Swift/BAT 15–150 keV fluence and a bolometric correction factor of 5 to correspond to a wider γ-ray energy range of ≈10–1000 keV, we calculate ${E}_{\gamma ,\mathrm{iso}}=(1.3\pm 0.3)\times {10}^{49}$ erg.

We can use the temporal evolution of the afterglow to constrain the jet opening angle. We have no observations of the X-ray afterglow at δt ≲ 8 days and therefore no information on the early decline rate. However, the optical and X-ray bands are on the same spectral slope (Section 3), and their afterglows should exhibit the same temporal decline. Indeed, the optical afterglow at $\delta t\approx 1.7\mbox{--}2.7$ days has a decline rate of ${\alpha }_{\mathrm{opt}}\approx -1.02$, matching the X-ray decline rate at $\delta t\approx 7.9\mbox{--}39.7$ days of ${\alpha }_{X}\approx -1.07$ within the 1σ uncertainties. Thus, the combination of the early optical data and the late-time X-ray data implies that the afterglow is on a single power-law decline over $\delta t\approx 1.7\mbox{--}39.7$ days. Since jet collimation is predicted to produce a temporal "break" in the light curve ("jet break"; Sari et al. 1999), the single power-law decline can be used to place a lower limit on the opening angle assuming on-axis orientation. The late-time X-ray observations, in conjunction with the energy, density, and redshift, can be converted to a jet opening angle, ${\theta }_{j}$, using (Sari et al. 1999; Frail et al. 2001)

$$\begin{eqnarray}&&{\theta }_{j}=9.5\,{t}_{j,{\rm{d}}}^{3/8}{(1+z)}^{-3/8}{E}_{{\rm{K}},\mathrm{iso},52}^{-1/8}{n}_{0}^{1/8}\ \deg ,\end{eqnarray} \tag{ 5 }$$

where ${t}_{j,{\rm{d}}}$ is in days. Using ${t}_{j,{\rm{d}}}\gt 39.68$ days, which corresponds to the last X-ray observation (Table 2), and using the values for the isotropic-equivalent kinetic energy and circumburst density as determined in Section 3, we calculate a lower limit on the opening angle of ${\theta }_{j}\gtrsim 9^\circ$.

We utilize our observations at δt ≳ 10 days to place limits on the presence of SN or kilonova emission associated with GRB 150101B. First, we compare our optical r-band observations in the range of $\delta t\approx 10\mbox{--}50$ days to the light curves of known SNe associated with long-duration GRBs to place limits on SN emission associated with GRB 150101B. We utilize optical light curves of three SNe that represent the range of luminosities observed for GRB-SNe: GRB-SN 1998bw (Galama et al. 1998; Clocchiatti et al. 2011), GRB-SN 2006aj (Mirabal et al. 2006), and GRB-SN 2010bh (Olivares et al. 2012), redshifted to z = 0.1343 (Figure 3). At δt ≈ 10.7 days, the limit on the optical flux density from GRB 150101B is ${F}_{\nu ,\mathrm{opt}}\lesssim 0.76\,\mu \mathrm{Jy}$, a factor of ≈6–60 below the expected brightness of GRB-SNe (Figure 3). Similarly, the limits at δt ≈ 40.7 and 48.4 days are a factor of ≈30–65 below the expected luminosity of GRB-SN 1998bw (Figure 3). Thus, we can rule out the presence of an SN associated with GRB 150101B to deep limits.

Zoom In Zoom Out Reset image size

To place limits on emission from a kilonova associated with GRB 150101B, we compare our deep optical and near-IR observations in the range of $\delta t\approx 10.7\mbox{--}28.9$ days to kilonova models from the literature. We note that the available models truncate at $\delta t\gt 30$ days. The deepest optical limits during this time range are ${r}_{\mathrm{AB}}\gtrsim 24.2$ mag at δt ≈ 10.7 days and ${Y}_{\mathrm{AB}}\gtrsim 23.6$ at δt ≈ 14.7 days. Similarly, the deepest near-IR limits in this time range are ${J}_{\mathrm{AB}}\gtrsim 23.0$ mag, ${H}_{\mathrm{AB}}\gtrsim 23.4$ mag, and ${K}_{\mathrm{AB}}\gtrsim 22.4$ mag (Table 1 and Figure 4).

Zoom In Zoom Out Reset image size

We consider emission from four sets of kilonova models: (i) spherically symmetric ejecta from NS–NS mergers for a range of ejecta velocities (0.1c–0.3c) and ejecta masses (${10}^{-3}\mbox{--}{10}^{-1}\,{M}_{\odot };$ Barnes & Kasen 2013), (ii) dynamical ejecta from NS–NS mergers considering "soft" and "stiff" equations of state for the merger components (Hotokezaka et al. 2013; Tanaka & Hotokezaka 2013), (iii) NS–BH mergers considering "soft" and "stiff" equations of state for the neutron star and varying spin for the black hole (Tanaka et al. 2014), and (iv) winds from an accretion disk surrounding a long-lived central compact remnant for remnant lifetimes of 0 ms to $\infty$ (Kasen et al. 2015). We utilize models that have employed r-process element (as opposed to iron-like) opacities, where model (i) assumes opacities calculated from a few representative lanthanide elements, while models (ii)–(iii) incorporate all of the r-process element opacities. For models (ii)–(iv), we select the pole-on orientation light curves. For each observing band, we use models in the corresponding rest-frame band (e.g., ${\lambda }_{\mathrm{rest}}={\lambda }_{\mathrm{obs}}{(1+z)}^{-1}$), convert the provided light curves to apparent magnitude, and shift the light curves to the observer frame. The models, along with the late-time optical and near-IR limits, are shown in Figure 4. Although these observations place among the deepest limits on kilonova emission following a short GRB to date (see Section 5.3), they do not constrain the brightest models. This demonstrates the difficulty of placing constraints on kilonovae associated with short GRBs at cosmological redshifts based on the current era of kilonova models.

Since the optical afterglow is faint and rapidly fading, we rely on the host galaxy to determine the redshift of GRB 150101B. Thus, we fit the IMACS spectrum of the host galaxy using simple stellar population (SSP) models at fixed ages of $\tau =0.64\mbox{--}11\,\mathrm{Gyr}$ (Bruzual & Charlot 2003). We fit models over the wavelength range 4400–8500 Å as the signal-to-noise ratio is too low to contribute significantly to the fit at wavelengths outside this range. Using ${\chi }^{2}$-minimization, the resulting best-fit model has a redshift of z = 0.1343 ± 0.030, determined primarily by the location of the main absorption features of Hβ, Mg b λ5175 and NaDλ5892 (Figure 5), and a stellar population age of 2.5 Gyr (${\chi }_{\nu }^{2}\approx 5.0$ for 1471 degrees of freedom) assuming solar metallicity. If we allow models to deviate from solar metallicity, adequate fits are also found for stellar population ages of 1.4 and 5 Gyr, while poorer fits are found for SSPs with younger or older ages. Due to the deep absorption features, lack of emission lines, and old inferred stellar population age, we conclude that this host is an early-type galaxy. The spectrum does not require any intrinsic extinction, consistent with the afterglow. We note that this redshift is fully consistent with the reported redshift from the VLT (Levan et al. 2015).

Zoom In Zoom Out Reset image size

While the SSP templates provide adequate matches to several of the main absorption features, a strong Hα absorption feature is notably absent (Figure 5). Two possible explanations are emission from the underlying AGN (see Section 2.2.5; Xie et al. 2016) or star formation from a younger stellar population, both of which would cause the Hα absorption to be "filled in."

To assess these contributions, we subtract the 2.5 Gyr template from the host spectrum (Figure 5). We find a clear emission feature at the location of Hα with an integrated flux of ${F}_{{\rm{H}}\alpha }\approx 1.3\times {10}^{-15}$ erg s⁻¹ cm⁻², or ${L}_{{\rm{H}}\alpha }\approx 5.5\,\times {10}^{40}$ erg s⁻¹ at z = 0.1343. To help distinguish between an AGN and star formation, we use archival 1.4 GHz observations as part of the NRAO VLA Sky Survey (Condon et al. 1998), where the host galaxy is detected with ${F}_{\nu ,1.4\mathrm{GHz}}\approx 10.2$ mJy. Using standard relations between radio luminosity, Hα emission, and star formation (Kennicutt et al. 1994; Yun & Carilli 2002; Murphy et al. 2011), we find that if the radio emission is due to star formation, the expected Hα luminosity is ≈500 times above the observed value. A large SFR is also in contradiction with the early-type galaxy spectrum and lack of emission lines; thus, star formation is not playing a large role. Instead, the Hα emission can naturally be explained by the AGN, as narrow Hα emission is common in the optical spectra of a wide variety of AGNs (e.g., Véron-Cetty & Véron 2000). Since Hα is contaminated by the AGN, we use the observed Hα flux to set an upper limit on the SFR. Using standard relations from Kennicutt et al. (1994), we calculate SFR ≲ 0.4 ${M}_{\odot }$ yr⁻¹ for the host galaxy of GRB 150101B. We note that we do not detect any continuum emission or spectral features at the position of the afterglow. In particular, the lack of emission lines indicates that there is no ongoing star formation at the location of GRB 150101B.

To determine the stellar mass and stellar population age, we use the Fitting and Assessment of Synthetic Templates (FAST; Kriek et al. 2009) code to fit the host galaxy optical and near-IR photometry to a stellar population synthesis template. We fix the redshift to z = 0.1343 and assume solar metallicity, allowing the extinction, stellar population age, and stellar mass to vary. We include all available host photometry in the fit: grizJK bands from ground-based telescopes and F606W/F160W from HST (Table 1). We find a best-fit template characterized by a stellar population age of $\tau ={2.0}_{-0.7}^{+0.8}$ Gyr and stellar mass of ${M}_{* }={7.1}_{-2.3}^{+1.2}\times {10}^{10}\,{M}_{\odot }$ (1σ uncertainties; ${\chi }_{\nu }^{2}=1.5$). Our results are independent of the choice of template library or initial mass function within the 1σ uncertainties. We find poorer fits when assuming subsolar or supersolar metallicities. We note that the age derived from the host galaxy photometry is in good agreement with the results from spectroscopic fitting.

We utilize the host spectrum to calculate the expected B-band magnitude as none of the observing bands correspond to rest-frame B band. We calculate the average spectral flux over 100 Å centered at ${\lambda }_{\mathrm{rest}}=4450$ Å and find ${F}_{\lambda }\approx 7.6\times {10}^{-16}$ erg cm⁻² s⁻¹ Å⁻¹; this value does not significantly change for assumed widths of 10–500 Å. The apparent rest-frame B-band magnitude is therefore ≈17.2 AB mag, corresponding to an absolute magnitude of ${M}_{B}=-21.7$, or ${L}_{B}\approx 6.9\times {10}^{10}\,{L}_{\odot }$. From a comparison to the local B-band luminosity function, which gives ${M}_{B}^{* }=-20.2$ (Ilbert et al. 2005), we infer ${L}_{B}\approx 4.3\,{L}^{* }$ for the host of GRB 150101B.

To determine the shape and size of the host galaxy, we use the IRAF/ellipse routine to generate elliptical intensity isophotes and construct one-dimensional radial surface brightness profiles. We utilize the late-time HST observations as these provide the highest angular resolution. For each observation, we allow the center, ellipticity, and position angle of each isophote to vary.

Using a ${\chi }^{2}$-minimization grid search, we fit each profile with a Sérsic model with index n (n = 1 is equivalent to an exponential disk profile, and n = 4 is the de Vaucouleurs profile typical of elliptical galaxies), effective radius r_e, and surface brightness ${\mu }_{e}$. In our grid search, n, r_e, and ${\mu }_{e}$ are the three free parameters. A single Sérsic component provides an adequate fit (${\chi }_{\nu }^{2}\approx 1.0$) in both filters. The best-fit models are characterized by n = 5.0 ± 0.1 and ${r}_{e}=9.5\pm 0.3\,\mathrm{kpc}$ in the optical F606W filter and n = 4.1 ± 0.1 and ${r}_{e}=7.2\,\pm 0.2\,\mathrm{kpc}$ in the F160W near-IR filter. Both fits demonstrate an elliptical morphology.

We determine the location of GRB 150101B with respect to its host galaxy center and light distribution. The angular offset is δR = 3.07 ± 003 (Section 2.2.1), which translates to a projected physical offset of 7.35 ± 0.07 kpc at z = 0.1343. Using the effective radius of the host galaxy measured in the F606W and F160W filters, the host-normalized offsets are $0.8{r}_{e}$ and $1.0{r}_{e}$ in the optical and near-IR bands, respectively.

To determine the brightness of the galaxy at the burst location with respect to the host light distribution, we calculate the fraction of total light in pixels fainter than the afterglow position ("fractional flux"; Fruchter et al. 2006). Since this relies on high angular resolution, we use the HST observations, where the 1σ uncertainties on the afterglow positions are ${\sigma }_{{\rm{F}}160{\rm{W}}}=0\buildrel{\prime\prime}\over{.} 07$ and ${\sigma }_{{\rm{F}}606{\rm{W}}}=0\buildrel{\prime\prime}\over{.} 04$ as determined by relative astrometry (Section 2.2.3). We calculate the average flux among the pixels spanned by the afterglow and create an intensity histogram of a region centered on the host galaxy. We consider pixels with a 1σ brightness level above the Gaussian sky brightness distribution to be part of the host galaxy light (e.g., corresponding to a signal-to-noise ratio of 1). We then plot the pixel flux distribution and determine the fraction of light in pixels fainter than the afterglow pixel. In this manner, we calculate fractional flux values of 0.21 (F606W) and 0.34 (F160W).

We compare the inferred afterglow and burst explosion properties of GRB 150101B to previous events. With an observed duration of ${T}_{90}\approx 0.012\,{\rm{s}}$, GRB 150101B has one of the shortest durations for an event detected by Swift or Fermi to date (Lien et al. 2016). This holds true when comparing its intrinsic, rest-frame duration of ≈0.011 s to the population of Swift-detected short GRBs with determined redshifts (Lien et al. 2016). The temporal evolution of the X-ray and optical afterglows is characterized by power-law decline rates that are consistent with the median values for the short GRB population of $\langle {\alpha }_{X}\rangle \approx -1.07$ and $\langle {\alpha }_{\mathrm{opt}}\rangle \approx -1.07$ (Fong et al. 2015). The detection of the X-ray afterglow of GRB 150101B extends to δt ≈ 40 days, corresponding to a rest-frame time of $\delta {t}_{\mathrm{rest}}\approx 35$ days.

To put GRB 150101B in the context of other events, we collect X-ray light curves for all Swift short GRBs that have multiple detections beyond $\delta {t}_{\mathrm{rest}}\gtrsim 1$ day and determined redshifts. These selection criteria result in four events: GRB 050724A (Grupe et al. 2006), GRB 051221A (Burrows et al. 2006; Soderberg et al. 2006), GRB 120804A (Berger et al. 2013b), and GRB 130603B (Fong et al. 2014). For GRB 150101B and each of the previous events, we convert the X-ray flux (0.3–10 keV) to luminosity and shift the light curve to the rest frame of the burst. The resulting X-ray light curves are shown in Figure 6. GRB 150101B has an X-ray luminosity of ${L}_{X}\approx 5.6\times {10}^{42}$ erg s⁻¹ at $\delta {t}_{\mathrm{rest}}\approx 7.0$ days, similar to the X-ray luminosities of previous short GRBs at the same epoch (Figure 6). Previously, short GRBs have only been detected to $\delta {t}_{\mathrm{rest}}\approx 20$ days. The low redshift of GRB 150101B, coupled with the moderate decline rate of ${L}_{X}\propto {t}^{-1.07}$ at late times, allows for the detection of the X-ray afterglow to an unprecedented time of $\delta {t}_{\mathrm{rest}}\approx 35$ days.

Zoom In Zoom Out Reset image size

We compare our inferred values of the explosion properties for GRB 150101B to those for the population of Swift short GRBs. We calculate median values for the isotropic-equivalent energies of ${E}_{\gamma ,\mathrm{iso}}\approx 1.3\times {10}^{49}$ erg and ${E}_{{\rm{K}},\mathrm{iso}}\approx (6\mbox{--}14)\,\times {10}^{51}$ erg, where the range is due to uncertainty in the microphysical parameters. From the combination of the X-ray and optical afterglows, we infer a lower limit on the jet opening angle of ${\theta }_{j}\gtrsim 9^\circ$ for GRB 150101B. Using 9^◦ as the minimum opening angle, we find minimum beaming-corrected energies of ${E}_{\gamma }\gtrsim 1.6\times {10}^{47}$ erg and ${E}_{{\rm{K}}}\gtrsim (7.4\mbox{--}17)\times {10}^{49}$ erg. Compared to Swift short GRBs, the isotropic-equivalent γ-ray energy of GRB 150101B is one of the lowest values for short GRBs, while the inferred isotropic-equivalent kinetic energy is at the 60%–70% level of the population (Fong et al. 2015). Assuming a constant-density medium, we also infer an extremely low median circumburst density of $\approx (0.8\mbox{--}4.3)\times {10}^{-5}$ cm⁻³ (for ${\epsilon }_{B}=0.01\mbox{--}0.1$), which is in the bottom ≈1/4 of all events, and at the median level when compared to short GRBs that originated in elliptical host galaxies (Fong et al. 2015). These low densities are commensurate with predictions for NS–NS mergers (Montes et al. 2016), as population synthesis for mergers in large, 10¹¹ M_⊙ galaxies predicts that a significant fraction of systems should occur in environments with gas densities of $\lesssim {10}^{-3}$ cm⁻³ (Belczynski et al. 2006).

At z = 0.1343, GRB 150101B is the closest short GRB with an early-type host galaxy to date. The host association is extremely robust, with a probability of chance coincidence of ${P}_{\mathrm{cc}}\approx 5\times {10}^{-4}$. A single more nearby event,²² GRB 080905A, likely occurred in a star-forming host galaxy at z = 0.1218, with ${P}_{\mathrm{cc}}\approx 0.01$ (Rowlinson et al. 2010). Overall, short GRBs in early-type galaxies comprise ≈30% of the population and are detected to z ∼ 1 (Fong et al. 2013), similar to the host galaxy demographics of SNe Ia (e.g., Sullivan et al. 2006). Both populations are indicative of older stellar progenitors. In contrast, long GRBs exclusively occur in star-forming galaxies (Wainwright et al. 2007; Savaglio et al. 2009), consistent with their massive star progenitors. Furthermore, the lack of associated SNe demonstrates that GRB 150101B did not form from a massive star progenitor.

From fitting of the optical to near-IR SED, we infer stellar population properties for the host: a stellar mass of $\approx 7\times {10}^{10}\,{M}_{\odot }$ (log $({M}_{* }/{M}_{\odot })\approx 10.9$), stellar population age of ≈2–2.5 Gyr (at solar metallicity), SFR of $\lesssim 0.4\,{M}_{\odot }$ yr⁻¹, and B-band luminosity of $\approx 6.9\times {10}^{10}\,{L}_{\odot }$ ($\approx 4.3{L}^{* };$ Table 3). These properties are consistent within 2σ with the values inferred from an independent analysis of the radio to X-ray SED (Xie et al. 2016). The stellar mass and stellar population age are well matched to the eight previous short GRBs with early-type host galaxies, which have stellar masses of log(${M}_{* }/{M}_{\odot })\approx 9.7\mbox{--}11.8$ and stellar population ages of ≈0.8–4.4 Gyr (Table 3). The host of GRB 150101B is one of the most luminous early-type host galaxies, surpassed only by GRB 050509B with $5.0{L}^{* }$ (Bloom et al. 2006). This is commensurate with its large relative size of ≈8 kpc. Overall, the global host galaxy properties indicate an evolved stellar population and are consistent with those expected for an older stellar progenitor. Notably, this is the first evidence for an AGN in a GRB host galaxy; the paucity of GRB-AGN systems is perhaps not surprising given the rates of low-luminosity AGNs and well-studied GRB hosts (Xie et al. 2016).

Table 3.  Short GRB Early-type Host Galaxy Properties

GRB	z	Pcc	log(M*)	Age	LB	SFR	References
			(M⊙)	(Gyr)	(L*)	(M⊙ yr−1)
050724A	0.257	$2\times {10}^{-5}$	10.8	0.94	1.0	≲0.05	1–3
050509B	0.225	$5\times {10}^{-3}$	11.6	3.18	5.0	≲0.15	3, 4–5
060502B	0.287	0.03	11.8	1.3	1.6	0.8	3, 6
070809	0.473	0.03	11.4	3.07	1.9	≲0.1	3, 7
070729	0.8	0.05	10.6	0.98	1.0	≲1.5	3
090515	0.403	0.15	11.2	4.35	1.0	0.1	3, 7–8
100117A	0.915	$7\times {10}^{-5}$	10.3	0.79	0.5	≲0.2	3, 9
100625A	0.452	0.04	9.7	0.8	0.2	≲0.3	10
150101B	0.1343	$5\times {10}^{-4}$	10.9	2–2.5	4.3	≲0.4	This work

Note. The columns are as follows: (1) GRB name, (2) host galaxy redshift, (3) probability of chance coincidence, (4) stellar mass, (5) stellar population age as determined from fitting to single stellar population models, (6) rest-frame B-band luminosity, (7) star formation rate.

References. (1) Berger et al. 2005; (2) Berger 2009; (3) Leibler & Berger 2010; (4) Gehrels et al. 2005; (5) Bloom et al. 2006; (6) Bloom et al. 2007; (7) Berger 2010; (8) Berger 2014; (9) Fong et al. 2011; (10) Fong et al. 2013.

Download table as:  ASCIITypeset image

Another important diagnostic to evaluate the progenitor is the location of GRB 150101B with respect to its host galaxy. Indeed, a hallmark signature of NS–NS/NS–BH mergers are natal kicks imparted to the compact objects during their formation (e.g., Fryer et al. 1998; Fryer 2004). Coupled with the ∼1 Gyr delay times, this results in a population of events with a range of separations from their hosts (Perna & Belczynski 2002; Belczynski et al. 2006; Behroozi et al. 2014). We calculate a kick velocity for GRB 150101B by assuming that the progenitor system formed with the most recent stellar population of ≈2–2.5 Gyr. Using the projected physical offset of ≈7.4 kpc, this implies a minimum kick velocity of ${v}_{\min }\approx 2.9$ km s⁻¹ for the progenitor of GRB 150101B. However, a more reasonable value for the kick velocity takes into account the host velocity dispersion (${v}_{\mathrm{disp}}$), such that ${v}_{\mathrm{kick}}\approx \sqrt{{v}_{\min }{v}_{\mathrm{disp}}}$. Assuming ${v}_{\mathrm{disp}}\approx 250$ km s⁻¹ as determined from $\approx {10}^{11}\,{M}_{\odot }$ elliptical galaxies (Forbes & Ponman 1999), we find ${v}_{\mathrm{kick}}\approx 27\mbox{--}30$ km s⁻¹ for GRB 150101B. This is on the low end of the inferred natal kick velocities for the eight known Galactic NS–NS binaries, which range from ≈5 to 500 km s⁻¹ (Fryer & Kalogera 1997; Willems et al. 2004; Wong et al. 2010). We note that Xie et al. (2016) inferred an older stellar population age of $\approx 5.7\pm 1.0\,\mathrm{Gyr}$, which would translate to a lower kick velocity of ${v}_{\mathrm{kick}}\approx 16\mbox{--}20$ km s⁻¹. The fractional flux value of ≈20%–35% demonstrates that GRB 150101B occurred on a faint region of its host rest-frame optical light and is thus weakly correlated with local stellar mass. Furthermore, there is no evidence for ongoing star formation at the position of GRB 150101B. These findings are also commensurate with NS/BH kicks.

We note that magnetars have also been suggested as viable progenitors of short GRBs. In particular, the accretion-induced collapse (AIC) of a WD or WD–WD merger product may lead to the delayed formation of a magnetar, subsequently producing a flare that would be seen as a short GRB (Levan et al. 2006; Metzger et al. 2008; Bucciantini et al. 2012). Such events are predicted to trace stellar mass and thus occur primarily in early-type galaxies (Levan et al. 2006). However, unlike NS–NS/NS–BH mergers, these systems would not experience significant kicks. GRB 150101B is weakly correlated with its local stellar mass, making NS/BH kicks a more viable option for this event. Furthermore, the low predicted rates of such AIC events make them unlikely to contribute significantly to the short GRB rate (Dessart et al. 2007).

A predicted signal of NS–NS/NS–BH mergers is transient emission from the radioactive decay of heavy elements produced in the merger ejecta (r-process "kilonova"; Li & Paczyński 1998; Kulkarni 2005; Metzger et al. 2010). The signal is predicted to be dominant in the near-IR bands due to the heavy-element opacities (Barnes & Kasen 2013; Rosswog et al. 2013; Tanaka & Hotokezaka 2013), although models incorporating a long-lived NS remnant predict bluer colors (Metzger & Fernández 2014; Kasen et al. 2015). A near-IR excess detected with HST observations following the short GRB 130603B was interpreted as kilonova emission and the first direct evidence that short GRBs originate from NS–NS/NS–BH mergers (Berger et al. 2013a; Tanvir et al. 2013).

For GRB 150101B, we place limits of $\approx \,(2\mbox{--}4)\,\times {10}^{41}$ erg s⁻¹ on kilonova emission (Figure 7). To compare these limits to searches for late-time emission following previous short GRBs, we collect all available data from the short GRB afterglow catalog (Fong et al. 2015), constraining the sample to bursts with upper limits at $\delta {t}_{\mathrm{rest}}\gtrsim 0.1$ days to match the timescale of kilonova light curves. We only include events that have either rest-frame r- or J-band observations. For 15 bursts with no determined redshift, we assume the median of the short GRB population, z = 0.5, to convert to luminosity. In addition to GRB 150101B, 25 short GRBs have rest-frame optical limits, and seven events have rest-frame near-IR limits. These limits, along with the GRB 130603B kilonova detection (Berger et al. 2013a; Tanvir et al. 2013) and four sets of kilonova models (described in Section 3.3), are displayed in Figure 7. We note that since GRB 061201 has a relatively uncertain association with a galaxy at z = 0.111, we also display the limit if this burst originated at the median redshift of z = 0.5 (Figure 7).

Zoom In Zoom Out Reset image size

In the rest-frame optical band, GRB 150101B has one of the deepest limits on optical kilonova emission to date with $\approx 2\times {10}^{41}$ erg s⁻¹, and the most stringent for a short GRB with a secure redshift. For GRB 061201, if the true redshift is z = 0.111, this event has the deepest limit of $\approx 6\times {10}^{40}$ erg s⁻¹. This limit can rule out the optically brightest models that invoke an indefinitely stable NS remnant (Kasen et al. 2015), while an assumption of a higher-redshift origin at z = 0.5 is not stringent enough to place any meaningful constraints (Figure 7).

The sample of short GRBs with rest-frame near-IR follow-up is significantly smaller, spanning a range of $\approx {10}^{42}\mbox{--}{10}^{44}$ erg s⁻¹ with most limits clustered at $\approx (0.8\mbox{--}3)\times {10}^{42}$ erg s⁻¹. Thus, with constraints of $\approx (2\mbox{--}4)\times {10}^{41}$ erg s⁻¹, GRB 150101B has the deepest limit on the luminosity of a near-IR kilonova (Figure 7). For comparison, the detection of the near-IR kilonova following GRB 130603B had a luminosity of $\approx 1.5\times {10}^{41}$ erg s⁻¹, which mapped to an ejecta mass and velocity of $\approx 0.03\mbox{--}0.08\,{M}_{\odot }$ and ≈0.1c–0.3c (Berger et al. 2013a; Tanvir et al. 2013). In the case of GRB 150101B, optical observations of comparable depth at earlier epochs of $\delta {t}_{\mathrm{rest}}\approx 2\mbox{--}5$ days or deeper near-IR observations at $\delta {t}_{\mathrm{rest}}\lesssim 10$ days would have helped to confirm or rule out the brightest kilonova models (Figure 7). This demonstrates the difficulty of performing effective kilonova searches following short GRBs based on the current era of kilonova models and the necessity of more sensitive instruments (e.g., space-based facilities or ∼30 m ground-based telescopes) in this effort.

A more promising route to kilonova detection is searches following NS–NS/NS–BH mergers detected by the Advanced LIGO-VIRGO (ALV) network, which is expected to detect such systems to 200 Mpc at design sensitivity (LIGO Scientific Collaboration et al. 2013). Due to their isotropic emission and distinct red color, kilonovae are expected to be premier counterparts in the optical and near-IR bands (Metzger & Berger 2012). Given current models, shallow searches that reach depths of ${r}_{\mathrm{AB}}=21$ mag and ${J}_{\mathrm{AB}}=20$ mag, corresponding to ≈few $\times {10}^{41}$ erg s⁻¹ at 200 Mpc, will only be marginally effective in detecting additional kilonovae (Figure 7). However, deep searches reaching depths of ${r}_{\mathrm{AB}}=24$ mag and ${J}_{\mathrm{AB}}=23$ mag, corresponding to $\approx 2\times {10}^{40}$ erg s⁻¹, will be sensitive to a more meaningful range of existing kilonova models and thus improve the chance of detection (Figure 7).