A Deep Chandra X-ray Study of Neutron Star Coalescence GW170817

Daryl Haggard, Melania Nynka, John J. Ruan, Vicky Kalogera, S. Bradley Cenko, Phil Evans, Jamie A. Kennea

I. Introduction

The detection of gravitational waves (GWs) by the Laser Interferometer Gravitational-wave Observatory (LIGO) is one of the most exciting advances in physics in decades. Abbott et al. 2016a reported the first LIGO detection of GWs, resulting from the merger of two black holes (BHs). The observed waveforms showed a near-perfect match to predictions from general relativity for the inspiral and merger of two black holes, ushering in the era of gravitational-wave astronomy. Extensive follow-up observations based on this GW event found no robust electromagnetic (EM) counterparts (Abbott et al. 2016b; Connaughton et al. 2016; Evans et al. 2016; Soares-Santos et al. 2016, e.g.,), consistent with theoretical predictions for stellar-mass BH mergers.

The next frontier is multi-messenger astronomy, where GW sources are associated with an EM emitter, connecting GW astronomy to our rich understanding of astrophysics. Core-collapse supernovae, mergers of two neutron stars (NSs), and mergers of NS-BH binaries are among the EM sources likely to have detectable GW signals. In particular, NS-NS mergers have long been predicted to be the progenitors of short $\gamma$ -ray bursts (Paczynski 1986; Narayan et al. 1992, GRBs;), and may produce kilonovae (Li & Paczyński 1998) that are responsible for the majority of $r$ -process nucleosynthesis in the Universe (Eichler et al. 1989).

On 17 August 2017 at 12:41:04 UTC, LIGO-Virgo detected event GW170817—its observed waveform traced the distinctive signal of a NS-NS inspiral and early analysis indicated a luminosity distance of $D_{L}=40\pm 8$ Mpc (LIGO Scientific Collaboration and Virgo Collaboration 2017). This discovery is the first in a new class of GW events stemming from neutron star binary coalescences, which are predicted to produce EM emission. Approximately 2 seconds after the GW trigger, the Gamma-ray Burst Monitor (GBM) instrument onboard the Fermi Gamma-ray Space Telescope was also triggered by the short-duration GRB 170817A (Connaughton et al. 2017; von Kienlin et al. 2017; Goldstein et al. 2017a; Goldstein et al. 2017b). Thanks to tight localization by LIGO-Virgo, follow-up ground-based optical imaging soon discovered the associated optical transient Swope Supernova Survey 17a (Coulter et al. 2017a; Coulter et al. 2017b, SSS17a, ), near the galaxy NGC 4993 at $z=0.0098$ (da Costa et al. 1998, $D_{L}=42.5\pm 0.3$ Mpc;).

Prior to the observations reported here, Chandra also observed the field of NGC 4993. The first observation occurred approximately 2 days post-trigger and reported a non-detection at the location of SSS17a (Margutti et al. 2017a; Margutti et al. 2017b). An observation 9 days post-trigger detected a source consistent with SSS17a, though no flux or luminosity values were reported (Troja et al. 2017a; Troja et al. 2017c).

In this Letter, we present two deep Chandra X-ray observations of the field of GW170817. In a 42′′ $\times$ 42′′ patch centered on NGC 4993 we detect four X-ray sources, including SSS17a and spatially-extended X-ray emission from the host galaxy. By constructing a Chandra X-ray light curve of SSS17a using these and earlier Chandra observations, we show that the X-ray emission from this NS-NS merger is consistent with the afterglow from an off-axis short GRB, with a jet axis angle of $\gtrsim$ 23∘. If confirmed, this makes GRB 170817A the first off-axis short GRB observed to date, in addition to being the first EM counterpart to a LIGO-Virgo GW detection.

II. Observations

We report analysis of two 46.69 ks Chandra X-ray observations, ObsID 20899 and ObsID 18988, which cover a patch of the LIGO-Virgo high confidence localization for GW170817 (LIGO Scientific Collaboration and Virgo Collaboration GNCs 21505 & 21509). ObsID 20899 (PI Troja) began 2017 September 01 at 15:22:22 ( $\sim$ 15 days post-trigger) and ObsID 18988 (PI Haggard) began approximately 13 hours later on 2017 September 02 at 04:53:25 ( $\sim$ 16 days post-trigger). Both observations were acquired using Chandra’s ACIS-S3 chip in VFAINT mode. Data reduction and analysis were performed with CIAO v.4.8 tools (Fruscione & Burke 2016, CALDB v4.7.2;). We reprocessed the level 2 events files, applied the latest calibrations via CIAO’s repro script, and extracted the 0.5–7 keV images and X-ray spectra described in §III. Our small field of view (Figure 1) includes the optical transient SSS17a (Coulter et al. 2017a; Coulter et al. 2017b), the Swift X-ray detection (Evans et al. 2017a; Evans et al. 2017b), and several other X-ray sources of interest.

Continued monitoring observations of this field with Chandra (as well as Swift) were prohibited by Sun constraints beginning in mid-September 2017 and continuing until early December 2017 (grey region in the left panel of Figure 4).

III. X-ray Analysis and Source Properties

In a small, on-axis patch $\sim 0.5^{\prime}$ on a side (Figure 1), we detected X-ray emission from three point sources and from one extended source: (1) point-source X-ray emission at the location of the optical transient SSS17a (Coulter et al. 2017a; Troja et al. 2017a; Troja et al. 2017b; Fong et al. 2017; Haggard et al. 2017), (2) another point source, CXOU J130948, near the location of the Swift X-ray emission (Evans et al. 2017b; Fong et al. 2017; Haggard et al. 2017), (3) emission from another previously unidentified X-ray point source, CXOU 130946, and (4) extended emission from the host galaxy NGC 4993 (Evans et al. 2017b; Margutti et al. 2017a).

We used CIAO’s wavdetect to obtain the centroid position for each source in the broadband ( $0.5-7$ keV) images. We selected a 1 $\mathrel{\mathop{\kern 0.0pt.}\limits^{\prime\prime}}$ 97 extraction region for the three point sources, corresponding to a $\sim$ 90% encircled energy fraction near Chandra’s aim point. A $2<math xmlns="http://www.w3.org/1998/Math/MathML" display="block"><semantics><mrow><mo><mover><mo><mspace width="0em"/><mi mathvariant="normal">.</mi></mo><mrow><mo mathvariant="normal">′</mo><mo mathvariant="normal">′</mo></mrow></mover></mo></mrow><annotation encoding="application/x-tex">\mathrel{\mathop{\kern 0.0pt.}\limits^{\prime\prime}}</annotation></semantics></math>.′′95$ extraction radius for NGC 4993 was chosen to enclose as much of its emission as possible while minimizing contamination from nearby CXOU J130948. Care was also taken to extract background photons from a large region that did not enclose other sources. Source IDs, positions, and extraction regions are reported in Table 1 and visualized in Figure 1.

For each of the two ObsIDs, we extracted spectra and response files for the X-ray sources using CIAO’s specextract tool. We then fit the spectra using XSPEC v12.9.0 (Arnaud 1996), with atomic cross sections and abundances from Verner et al. 1996 and Wilms et al. 2000, respectively. The data from the two observations were first jointly-fit with an absorbed power law ( $S\propto E^{-\Gamma}$ ). The absorption column in all cases was fixed to NH $=7.5\times 10^{20}$ cm-2, derived by converting the Galactic optical extinction A ${}_{\rm V}=0.338$ (Schlafly & Finkbeiner 2011) to a hydrogen column density via the relation NH(cm ${}^{-2})\approx 2.21\times 10^{21}$ AV (Güver & Özel 2009). The photon index was tied between the data sets while the normalization was left free. To obtain better constraints on both the photon indices and fluxes, the spectra (and response files) from the two observations were co-added and fit again with an absorbed power law. The resultant best-fit parameters are shown in Table 1 and Figures 2 and 4.

The X-ray counterpart of the optical source SSS17a was well-fit with a power law index of $\Gamma\sim 2.4$ . Analysis of the spectra did not reveal a statistical difference in either the flux values or the count rates of SSS17a between the two observations. When we co-added the spectra and response files, the improved statistics yielded an absorbed flux of $\sim 3.6\pm 0.1\times 10^{-15}$ erg s-1 cm-2. This is consistent with the upper limits observed by Swift (Evans et al. 2017c, see also Figure 2).

IV. Discussion

It is challenging to explain our Chandra X-ray detection of the afterglow of GRB 170817A at 16 days post-trigger, in combination with the Chandra non-detection at 2 days post-trigger, and the detection at 9 days post-trigger, as described in Section II. For short GRBs, standard afterglow models predict that after the prompt emission fades on timescales of $<$ 2s post-burst, the relativistic jet will be decelerated by the ambient medium. This leads to X-ray emission that decays as $t^{-2}$ on the timescales of $10^{5-6}$ s, i.e., exactly the timescales covered by our X-ray observations. Figure 4 (left panel) displays our X-ray light curve of the afterglow of GRB 170817A, including the detection at 16 days post-trigger, an upper limit for the non-detection at 2 days post-trigger, and a lower limit for the detection at 9 days post-trigger. Figure 4 (left panel) also displays theoretical 1.5 keV X-ray afterglow light curves for short GRBs for a range of jet axis angles, scaled to the observed flux of our Chandra X-ray observations at 16 days. These light curve models are from the relativistic hydrodynamic simulations of van Eerten & MacFadyen 2011, which includes radiative transfer for synchrotron emission, and assumes that the beaming-corrected total energy in both jets is 1048 erg, the number density of the ambient medium is 10-3 cm-3, and the jet half opening angle is 11∘. The model light curve in Figure 4 for a jet observed at $0^{\circ}$ off-axis (i.e., directly along the line of sight) predicts afterglow X-ray emission that is a factor of $>$ 103 higher than the upper limit from the non-detection at 2 days post-trigger. Thus, the standard on-axis GRB scenario is disfavored by these X-ray observations.

GRB jets observed off-axis can produce afterglow emission that is faint at early times, but becomes luminous and observable as the jet beaming becomes less severe and the jet opening angle spreads into the line of sight (Granot et al. 2002). Figure 4 (left panel) also displays model light curves at a range of jet axis angles from the line of sight. Our observed X-ray light curve is consistent with afterglow models for an off-axis short GRB, with jet axis angle of $\gtrsim$ 23∘. If confirmed, this makes GRB 170817A the first observed off-axis short GRB, in addition to the first electromagnetic counterpart to a GW event.

Further X-ray monitoring of GRB 170817A can tightly constrain the jet axis angle. If the jet axis angle is at $\gtrsim$ 23∘, the X-ray afterglow has already reached its peak and will continue to fade. However, at larger jet axis angles, the X-ray afterglow can still be brightening (e.g., see the model light curve for a 46∘ axis angle in Figure 4), until reaching a peak at up to 430 days post-burst for a jet at a 90∘ axis angle, using our assumed model parameters. Deep Chandra observations after sunblock ( $\sim$ December 2017; Figure 4) could easily distinguish between these possibilities.

The off-axis GRB scenario also predicts other multi-wavelength properties, including late-time radio emission from the afterglow that peaks on timescales of order 10 days after the X-ray peak. Indeed, a previously-undetected radio source associated with SSS17a was reported approximately 15 days post-burst (Mooley et al. 2017; Corsi et al. 2017). Thus, continued multi-wavelength monitoring of GRB 170817A will be key to unveiling its nature and understanding its properties.

A Deep Chandra X-ray Study of Neutron Star Coalescence GW170817

I. Introduction

II. Observations

III. X-ray Analysis and Source Properties

IV. Discussion

IV.2. Gamma-ray Evidence for an Off-Axis Short GRB

References