A Decline in the X-Ray through Radio Emission from GW170817 Continues to Support an Off-axis Structured Jet

The discovery of broadband synchrotron emission associated with the binary neutron star merger, GW170817 (Abbott et al. 2017b, 2017c), together with the prompt low-luminosity gamma-ray emission (Abbott et al. 2017a) provided the first direct evidence for the production of relativistic ejecta in such a system (Alexander et al. 2017; Granot et al. 2017; Haggard et al. 2017; Hallinan et al. 2017; Kasliwal et al. 2017; Margutti et al. 2017, 2018; Murguia-Berthier et al. 2017; Troja et al. 2017; Kathirgamaraju et al. 2018; Mooley et al. 2018b). Several lines of evidence suggest that short gamma-ray bursts (SGRBs) are produced by binary neutron star mergers viewed on-axis (Berger 2014), and observations of SGRB afterglows have provided a measure of the kinetic energies, collimation angles, and circumbinary densities (Fong et al. 2015). These observations provide a basis for comparison with GW170817, suggesting that viewing angle effects are the dominant cause of observed differences between GW170817 and these events (e.g., Fong et al. 2017).

In GW170817 the radio and X-ray emission were observed to rise gradually for the first ∼160 days, with a single spectral power law, f_ν ∝ ν^β with β ≈ −0.6, spanning ∼10⁹–10¹⁸ Hz (Alexander et al. 2017; Hallinan et al. 2017; Kim et al. 2017; D'Avanzo et al. 2018; Margutti et al. 2018; Mooley et al. 2018b; Pooley et al. 2018; Resmi et al. 2018; Ruan et al. 2018; Troja et al. 2018). This emission was accompanied by optical detections well after the radioactive "kilonova" component had faded away (Lyman et al. 2018; Margutti et al. 2018). These observed properties can be explained in the context of two primary models. First is a "successful structured jet" with properties similar to those inferred in SGRBs but viewed at an angle θ_obs ≈ 20°–30° off-axis (Lamb & Kobayashi 2017; Lazzati et al. 2018; Margutti et al. 2018; Xie et al. 2018). Successful structured jets consist of an initially highly relativistic collimated core surrounded by wings of mildly relativistic material at larger angles (e.g., Rossi et al. 2002; Kumar & Granot 2003); they are also sometimes called "successful jets with cocoons" in the literature (e.g., Nakar & Piran 2018). The alternative is that no successful jet was produced and the emission is quasi-isotropic. In this case, either a choked jet deposits all of its energy into a mildly relativistic cocoon that dominates the emission (referred to as a "pure cocoon" model in this Letter; Kasliwal et al. 2017; Gottlieb et al. 2018; Mooley et al. 2018b) or the emission arises from the fastest component of the dynamical ejecta expelled during the merger (Hotokezaka et al. 2018; Mooley et al. 2018b). We note that independent of the radio/X-ray data, a recent joint analysis of the Laser Interferometer Gravitational-Wave Observatory (LIGO)/Virgo gravitational wave data and the electromagnetic observations indicates that the inclination angle of the binary is ${32}_{-13}^{+10}$ deg (Finstad et al. 2018), in agreement with the viewing angle inferred from the successful structured jet model. Recently reported very long baseline interferometry (VLBI) observations of GW170817 also support the existence of a successful jet with a viewing angle of 20° ± 5° (Mooley et al. 2018a).

A possible way to distinguish the origin of the relativistic ejecta is to measure the long-term behavior of the post-peak emission (e.g., Gill & Granot 2018; Margutti et al. 2018; Nakar & Piran 2018; Troja et al. 2018). Recently, a turnover in the radio light curve at ≈200 days was reported (Dobie et al. 2018), while Chandra and XMM-Newton observations extending to ≈160 days have been suggestive of a flattening in the X-ray light curve (D'Avanzo et al. 2018; Margutti et al. 2018). Here, we report new radio, optical, and X-ray observations at ≈220–290 days that unambiguously show a decline at both X-ray and radio wavelengths, as well as a possible decline in the optical band. The declining behavior is fully consistent with the predicted behavior of our successful structured jet models (Margutti et al. 2018; Xie et al. 2018). Uncertainties are 1σ confidence intervals unless otherwise specified.

We present new radio, optical, and X-ray observations of GW170817, and analyze those jointly with all previous observations from our work and from the literature (Alexander et al. 2017; Haggard et al. 2017; Margutti et al. 2017, 2018; D'Avanzo et al. 2018; Dobie et al. 2018; Lyman et al. 2018; Mooley et al. 2018b).

2.1. VLA Observations

We observed GW170817 with the VLA on 2018 March 22 UT in the A configuration, with nearly continuous frequency coverage at 2–18 GHz (apart from small gaps introduced by radio frequency interference (RFI) and correlator setup), on May 1 UT at 2–4 GHz, and on May 17 UT and June 2 UT at 2–8 GHz. We analyzed the data with CASA (McMullin et al. 2007) using 3C286 as the bandpass calibrator and J1258–2219 as the phase calibrator. We imaged the data using standard CASA routines, using a bandwidth of 2, 4, or 6 GHz for each image, and determined the flux density by fitting a point source model using the imtool program within the pwkit package (Williams et al. 2017). Our March 22 3 GHz observation was impacted by unusually strong RFI, resulting in elevated noise in the image produced after our initial data reduction. We therefore reprocessed the data using a thorough independent manual flagging procedure. The final flux density values are provided in Table 1.

Table 1.  New Observations of GW170817

Observatory	UT Date	Δt	Mean Freq.	Freq. Range/	Exp. Time	Flux Density	Image rms
	(UT)	(days)	(Hz)	Filter	(hr)	(μJy)	(μJy)
VLA	2018 Mar 22	216.91	3.0 × 109	2–4 GHz	0.6	69 ± 15	10
VLA	2018 Mar 22	216.88	6.0 × 109	4–8 GHz	0.7	39 ± 9	6
VLA	2018 Mar 22	216.85	10.0 × 109	8–12 GHz	0.7	28 ± 7	5
VLA	2018 Mar 22	216.80	15.0 × 109	12–18 GHz	1.8	21 ± 5	4
VLA	2018 May 1	256.76	3.0 × 109	2–4 GHz	0.7	55 ± 12	9
VLA	2018 May 17	272.67	3.0 × 109	2–4 GHz	1.3	44 ± 11	8
VLA	2018 May 17	272.61	6.0 × 109	4–8 GHz	1.3	36 ± 7	5
VLA	2018 Jun 2	288.61	3.0 × 109	2–4 GHz	1.3	46 ± 11	8
VLA	2018 Jun 2	288.55	6.0 × 109	4–8 GHz	1.3	35 ± 7	5
HST	2018 Mar 23	218.37	5.07 × 1014	F606W	0.58	<0.070a	0.023
Chandra	2018 May 3–5	259.99	2.4 × 1017	0.3–10 keV	27.2	${1.22}_{-0.15}^{+0.25}\times {10}^{-3}$	⋯

Note.

^aCorrected for Galactic extinction.

Download table as:  ASCIITypeset image

We also measure the flux density of the compact background source at (R.A., decl.) = 13^h9^m53911, −23°21'3449 (J2000) noted by Dobie et al. (2018) and find that it remains constant in comparison to our previous observations of this field. For the new data presented here, we obtain F_ν = 599 ± 5 μJy at 3 GHz, F_ν = 370 ± 5 μJy at 6 GHz, F_ν = 146 ± 8 μJy at 10 GHz, and F_ν = 57 ± 5 μJy at 15 GHz, where the first two measurements are average values and the second two come from our observations at 217 days.

2.2. HST Observations

2.3. Chandra Observations

The X-ray, optical, and 3 and 6 GHz radio light curves are shown in Figure 1, together with our successful structured jet models previously presented in Margutti et al. (2018) and Xie et al. (2018). Both radio light curves show clear evidence of a decline at ≳200 days; the 3 GHz flux density at 289 days is about a factor of 3 times fainter than its peak brightness at 163 days. To quantify the significance of this turnover, we scale all radio data from Table 1 and previous results (Alexander et al. 2017; Hallinan et al. 2017; Dobie et al. 2018; Margutti et al. 2018; Mooley et al. 2018b) to a common frequency of 5.5 GHz using a spectral index of β = −0.585 (Margutti et al. 2018) and fit the resulting light curve with a smoothed broken power law

$$\begin{eqnarray}&&{F}_{\nu }(t)={F}_{\nu ,b}{\left[\displaystyle \frac{1}{2}{\left(\displaystyle \frac{t}{{t}_{b}}\right)}^{-s{\alpha }_{1}}+\displaystyle \frac{1}{2}{\left(\displaystyle \frac{t}{{t}_{b}}\right)}^{-s{\alpha }_{2}}\right]}^{-1/s},\end{eqnarray} \tag{ 1 }$$

where α₁ and α₂ are the temporal indices before and after the break time, t_b, respectively, F_ν,b is the flux density at the time of the break, and s defines the sharpness of the transition. We additionally model possible calibration differences between observations taken at different facilities and reduced by different groups as an extra fractional uncertainty, f, on all data points.¹⁰ We use emcee (Foreman-Mackey et al. 2013) to perform a MCMC analysis to determine the posteriors of the model parameters. We use a logarithmic flat prior on s (−1.5 < log s < 1.5) because by definition s must be positive and the data do not provide any constraints once s becomes sufficiently large (corresponding to a sharp break that occurs over a time interval much smaller than our sampling time). We use flat linear priors on all other parameters: 10 < t_b < 1000 days, 1 < F_ν,b < 1000 μJy, −20 < α₁, α₂ < 20, and 0 < f < 1.

Zoom In Zoom Out Reset image size

We find ${\alpha }_{1}={0.86}_{-0.04}^{+0.05}$ and ${\alpha }_{2}=-{1.6}_{-0.3}^{+0.2}$, with ${t}_{b}={150}_{-6}^{+7}$ days, indicative of a clear transition from a rising to a declining light curve; see Figure 2. We can reject a non-declining light curve (i.e., α₂ ≥ 0) at 7σ confidence level. These results are consistent¹¹ with the analysis of Dobie et al. (2018), but provide a much stronger indication of a break due to the longer time baseline of our observations. Both successful structured jet models and pure cocoon models are consistent with the observed α₁. Unlike the simplest analytic models, which predict α₂ ≈ −2.2 (Sari et al. 1999; Nakar & Piran 2018), the Xie et al. (2018) successful structured jet simulations predict a shallower initial post-peak decline (Figure 1). Therefore, our derived α₂ does not strongly discriminate between successful structured jet and pure cocoon or dynamical ejecta models, although the latter are mildly disfavored (Lamb et al. 2018). Of the two Xie et al. (2018) models shown in Figure 1, our most recent data at 273 days and 289 days mildly favor the dashed model, which in comparison to the solid model implies a higher circumbinary density of 10⁻⁴ cm⁻³, a lower fraction of the shock energy imparted to electrons (_e = 0.02), and an observer viewing angle of 20°. The model light curves diverge more at later times, so future observations will allow us to place tighter constraints on these parameters. We note that the data require f = 0.09 ± 0.03, which is broadly consistent with the combination of the calibration uncertainties reported by Dobie et al. (2018) and the known typical flux density calibration accuracy of radio observations at the VLA (5%).

Zoom In Zoom Out Reset image size

The new X-ray data independently support the evidence for temporally decaying emission from GW170817. Comparing to the previous epoch of Chandra observations at ≈160 days, and applying a simple binomial test, we find a probability of P ∼ 2 × 10⁻⁶ that the observed decrease in count-rate arises from a random statistical fluctuation. We can thus reject the hypothesis of a random fluctuation with ∼4.8σ confidence, and conclude that these Chandra observations provide the first statistically significant evidence for fading X-ray emission from GW170817.¹²

Notably, the X-ray spectral index ${\beta }_{x}\equiv 1-{\rm{\Gamma }}=-{0.51}_{-0.27}^{+0.26}$ is consistent with previously reported values (e.g., β_x =−0.61 ± 0.17 at ∼160 days; Haggard et al. 2017; Margutti et al. 2017, 2018; Troja et al. 2017, 2018; Ruan et al. 2018). We observe no evidence for a steepening of the X-ray spectral index from β_x to β_x + 0.5, predicted when the synchrotron cooling frequency passes below the X-ray band (Granot & Sari 2002). This behavior is consistent with our successful structured jet models, which predict that the cooling frequency will remain above the X-ray band until ≳10⁴ days (Margutti et al. 2018), but rules out some dynamical ejecta models, which predict an earlier passage of the cooling break (e.g., Hotokezaka et al. 2018; Nakar & Piran 2018). The radio-to-X-ray spectral index is β_XR = −0.583 ± 0.013 at t ∼ 260 days, confirming that the radio and X-ray bands still lie on the same spectral segment. This is fully consistent with the radio-to-X-ray spectral index of β_XR = −0.584 ± 0.006 reported at t ∼ 160 days (Margutti et al. 2018), indicating no spectral evolution across the peak. The radio-only spectral index at 217 days is β_R = −0.74 ± 0.20, consistent with this value. Finally, our HST observations are also consistent with the observed X-ray and radio decline.

We present new X-ray, optical, and radio observations of GW170817 at ≈220–290 days. Our new broadband measurements show that the synchrotron emission from GW170817 has passed its peak brightness and has begun to decline. We find that the data continue to be well described by a single power law extending over ≈10⁹–10¹⁸ Hz, with no sign of the spectral index change in the X-rays that would signify the passage of the synchrotron cooling break. We find clear evidence (≳5σ) for a turnover in the radio and X-ray light curves, and full agreement with the predicted evolution of our previously published successful structured jet models (Margutti et al. 2018; Xie et al. 2018). The steep decline rate seen in our new radio and X-ray data is characteristic of all published successful structured jet models (Lazzati et al. 2018; Xie et al. 2018), and mildly disfavors some dynamical ejecta and pure cocoon models (e.g., Hotokezaka et al. 2018; Nakar & Piran 2018; Troja et al. 2018). Continued observations will allow us to better constrain the post-peak decline rate, providing further insights into the ejecta structure. If the emission continues to decay as predicted by the successful structured jet models, GW170817 should remain detectable with current radio and X-ray facilities until ∼1000 days post-merger.

We would like to thank the referee for helpful comments that improved this manuscript. The Berger Time-Domain Group at Harvard is supported in part by the NSF through grant AST-1714498, and by NASA through grants NNX15AE50G and NNX16AC22G. This work is partially based on observations acquired by the Chandra X-ray Observatory. The Chandra X-ray Observatory Center is operated by the Smithsonian Astrophysical Observatory for and on behalf of the National Aeronautics Space Administration under contract NAS8-03060. W.F. acknowledges support from Program number HST-HF2-51390.001-A, provided by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Incorporated, under NASA contract NAS5-26555. B.D.M. is supported in part by NASA ATP grant NNX16AB30G. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

Facilities: VLA - Expanded Very Large Array, HST - , CXO. -

Software: CASA, emcee, Numpy, pwkit.