High-speed Ejecta from the Gamma-ray Binary PSR B1259-63/LS 2883

TL;DR

This study reports the detection of high-speed X-ray emitting clumps ejected from the gamma-ray binary PSR B1259-63/LS 2883, revealing episodic ejections possibly linked to periastron passages and characterized by synchrotron emission.

Contribution

First observation of relativistic X-ray clumps ejected from a gamma-ray binary, with analysis of their properties and implications for binary dynamics.

Findings

Clumps move at about 0.1c, possibly accelerating.

Ejections occur at least once per 3.4-year orbit, near periastron.

Clump spectra consistent with synchrotron emission.

Abstract

Observing the famous high-mass, eccentric X-ray and gamma-ray binary PSR B1259-63/LS 2883 with Chandra, we detected X-ray emitting clumps moving from the binary with speeds of about 0.1 of the speed of light, possibly with acceleration. The clumps are being ejected at least once per binary period, 3.4 years, presumably around binary periastrons. The power-law spectra of the clumps can be interpreted as synchrotron emission of relativistic electrons. Here we report the results of 8 observations of the clumps in 2011-2017 (two binary cycles) and discuss possible interpretations of this unique phenomenon.

Full text

Pulsars: individual (B1256–63) — X-rays: binaries

1 Introduction

High-mass $\gamma$-ray binaries (HMGBs) consist of a compact object (a black hole or a neutron star) and a massive B- or O-type companion (see Dubus 2013 for a review). If the compact object is a rotation-powered (non-accreting) pulsar, the collision of the relativistic pulsar wind with the companion’s wind produces a shock, and the shocked wind(s) generate synchrotron and inverse Compton (IC) emission from the radio to TeV $\gamma$-rays (Tavani & Arons, 1997). Observations of this emission allow one to study the properties of the winds and the binary companions.

Particularly interesting for such studies is the PSR B1259–63/LS 2883 binary (B1259 hereafter), one of only two HMGBs in which a rotation-powered pulsar has been detected222 The second such binary, PSR J2023+4127/MT91 213, was identified by Lyne et al. (2015). Its binary period of $\sim 50$ years is, however, too long to study the orbital dependence in our lifetime.. PSR B1259–63 has the following properties: period $P=48$ ms, characteristic age $\tau=330$ kyr, spin-down power $\dot{E}=8.3\times 10^{35}$ erg s*-1*, and distance $d\approx 2.7$ kpc. The pulsar’s companion, LS 2883, is a fast-rotating O9.5Ve star, with mass $M_{*}\sim 30M_{\odot}$, luminosity $L_{*}=6.3\times 10^{4}L_{\odot}$, and an equatorial disk inclined at $\approx 35^{\circ}$ to the orbital plane (see Negueruela et al. 2011). The results of the VLBI astrometry (Miller-Jones et al., 2018), combined with high-precision pulsar timing (Shannon et al., 2014), have provided a full orbital solution for the binary system, including orbital period $P_{\rm orb}=1236.7$ days, eccentricity $e=0.87$, semimajor axis $a\simeq 6$ AU, inclination angle $i\simeq 154^{\circ}$, and longitude of the ascending node $\Omega\simeq 189^{\circ}$.

Multi-epoch observations of B1259 with numerous X-ray observatories have revealed binary phase dependences of its flux, photon index, and absorption column density, with particularly large variations within a (–20,+30) days interval around periastron passages (Chernyakova et al., 2015). AGILE, Fermi and H.E.S.S. observations have shown that the passages of 3 consecutive periastrons (2010 Dec 14, 2014 May 5, 2017 Sep 22) were accompanied by strong $\gamma$-ray flares, which had no counterparts at lower energies (see Tavani et al. 2010, Abdo et al. 2011, H.E.S.S. Collaboration 2013, Chernyakova et al. 2015, Johnson et al. 2018). Unexpectedly, the flares occurred within 30–70 day intervals after peristron passages (10–50 days after the pulsar passed through the stellar disk). Based on observations of H$\alpha$ lines, whose equivalent widths allow one to estimate the mass of the equatorial disk of the O star, it was suggested in Chernyakova et al. (2014) that the GeV flares are associated with partial destruction of the disk by the passing pulsar. The flares could be produced by IC emission from interactions between the pulsar wind and the stellar photons (Khangulyan et al., 2012) or by Doppler boosting of synchrotron radiation from the termination shocks in the pulsar wind colliding with the fragments of the destroyed disk (Kong et al., 2012). The competing models are still rather uncertain and cannot explain all the observed features. However, since a fragment of the destroyed disk could be ejected from the binary, detection and study of such fragments could provide a clue for understanding the dramatic variations of electromagnetic radiation from B1259 around the periastron passages, particularly the $\gamma$-ray flares. In this work we describe the main properties of the X-ray emitting ejecta detected in a series of high-resolution Chandra observations during 2 binary cycles. Observations in the 2010-2014 binary cycle have been reported by Kargaltsev et al. (2014) and Pavlov et al. (2015), while details of the 2014-2017 cycle observations and their analysis will be presented in a separate publication.

2 Observations and analysis of fast-moving ejecta

A southward extension in the B1259 X-ray image was first noticed by Pavlov et al. (2011) in a 28 ks Chandra ACIS-I observation taken on 2009 May 14, when the binary was close to its apastron ($\Delta t=667$ days after periastron passage). ACIS-I observations in the 2010-2014 cycle were taken at $\Delta t=370$, 886 and 1151 days after periastron passage. Each of them showed an extended source with a characteristic size of $3^{\prime\prime}$ ($\approx 10^{17}$ cm at $d=2.7$ kpc) at progressively larger distances from the binary (see the upper panels of Fig. 1). Assuming that it was the same source moving south-southwest from the binary (i.e., along the major axis of the binary orbit in the periastron-apastron direction) with a constant velocity, its projected velocity and launch time were $v_{\perp}=(0.07\pm 0.01)\,c$ and $t_{\rm launch}=-20^{+120}_{-160}$ days, where $t_{\rm launch}=0$ corresponds to the time of periastron passage (see Fig. 2, left panel). The 0.5–8 keV luminosity of the source was decreasing from $8\times 10^{31}$ to $2\times 10^{31}$ erg s*-1* (at $d=2.7$ kpc), but the hard spectrum of the fading source (photon index $\Gamma\approx 1.2$) did not show any softening. The source was interpreted as a clump of stellar disk matter ejected from the binary through the interaction of the pulsar with the O-star’s disk around the 2010 periastron (possibly at the time of the $\gamma$-ray flare); its X-ray emission could be due to synchrotron radiation of the shocked pulsar wind (Pavlov et al., 2015). Hydrodynamical simulations of clump ejection were presented in Barkov & Bosch-Ramon (2016).

To check repeatability of the discovered phenomenon and better understand the clump’s properties, we again observed B1259 in the 2014-2017 binary cycle with the same observational setup and exposure times of about 70 ks. Five observations at $\Delta t=352$, 620, 978, 1086, and 1175 days from the 2014 periastron again showed a clump apparently moving in the same direction (see the lower panels in Fig. 1, where the images from 3 later observations are shown). However, the clump became discernible only at the second observation. Moreover, fitting the clump separations from the binary with a linear function of time, we obtained $v_{\perp}=(0.12\pm 0.02)\,c$ and $t_{\rm launch}=420^{+78}_{-103}$ days. This places the launch time far after the periastron passage and after passage through the stellar disk (see Fig. 2, left panel). To reconcile the time dependence of the clump separation from the binary with the plausible assumption that it was launched near periastron, we have to include an acceleration term in the separation versus time fit, which gives an acceleration $a=49\pm 2$ cm s${}^{-2}=15,400\pm 600$ (km/s) yr*-1* (Fig. 2, left panel).

These observations also showed puzzling variations of the clump’s shape and flux at larger $\Delta t$ (see the lower panels in Fig. 1), such as the narrow structure stretched perpendicular to the apparent velocity direction (which we call “whiskers”) in the third observation and, in particular, the unexpected brightening by a factor of 1.8 in the fourth observation. Another unexpected feature was evident in the fourth observation: an additional clump adjacent to the binary, which disappeared in the fifth observation.

The evolution of the 0.5–8 keV clump flux values in the two binary cycles is shown in Figure 2 (right panel). The corresponding clump luminosities vary in the range of (2–$8)\times 10^{31}$ erg s*-1*, or $\sim 1\%$–3% of the (variable) binary luminosity in the same energy range. The clump spectra could be described by a power-law model in each of the cycles, with photon indices in the range $\Gamma\approx 1.0$–1.6, with a mean value of about 1.4.

3 Summary and discussion

3.1 Summary of observational results

Our main findings from the 8 high-resolution Chandra observations of B1259 can be briefly summarized as follows.

• •

B1259 ejects clumps of X-ray emitting matter along the periastron-apastron directions, at least one clump per binary period.

• •

A typical observed size of the resolved clumps is about $3^{\prime\prime}$ ($\sim 10^{17}$ cm). The clumps were detected up to at least $8^{\prime\prime}$ from the binary. They showed significant shape (and size) variations, particularly in the 2014-2017 binary cycle.

• •

The apparent velocities $v_{\perp}$ of the two clumps detected in the 2010-2014 and 2014-2017 cycles were around 10% of the speed of light.

• •

Assuming that the clump travels at constant velocity in a given binary cycle, its launch time was close to the 2010 periastron in the 2010-2014 cycle, but it was $\sim 400$ days after the 2014 periastron in the 2014-2017 cycle, in which $v_{\perp}$ was a factor of 1.7 higher.

• •

The assumption that the clumps are launched around the nearest preceding periastrons requires a significant acceleration, $\sim 50$ cm s*-2*, in the 2014-2017 cycle.

• •

The clump detected in the 2010-2014 cycle faded with increasing separation from the binary, while the 2014-2017 cycle clump showed a factor of 1.8 brightening in one of the observations.

• •

The clump spectra did not show significant slope variations with time, despite the varying flux.

• •

An additional clump was detected in the binary vicinity in one of the 2014-2017 observations, but it disappeared by the next observation 89 days later.

3.2 Inferences from the clump obervations

A detailed analysis of the observational results will be presented elsewhere. Here we only briefly mention the most important inferences.

Repeatability.

In each of the two binary cycles covered by our observations we saw a relatively bright clump moving in the same direction. However, the different temporal dependences of the clump-binary separation (Fig. 2, left) suggest that either the clumps were launched at different binary phases and with different velocities or, if they leave the binary at the same phase (e.g., at periastron), they move with different accelerations. The evolution of the clump shape is noticeably different in the two cycles (Fig. 1). The measured X-ray flux values were similar in the two cycles, but their temporal behavior was different (Fig. 2, right). The lack of full repeatability could be due to the turbulent nature of disk fragmentation and the interaction of the pulsar wind with the stellar wind (Barkov & Bosch-Ramon, 2016).

The nature of the clump X-ray emission.

The power-law spectra of the clumps suggest that their emission was produced by relativistic electrons, likely supplied by the pulsar. Possible emission mechanisms are synchrotron radiation in the clump’s magnetic field and IC scattering of UV photons from the luminous high-mass companion off the relativistic electrons. For the IC up-scattering of $\sim 10$ eV photons to X-ray energies, the Lorentz factors of the relativistic electrons should be $\gamma\sim 10$–30. To provide the observed X-ray flux with so low $\gamma$, about $10^{50}$ simultaneously emitting electrons is required, which is too large a number to be supplied by the pulsar or stellar wind. Therefore, we conclude that the clump X-rays are emitted by the synchrotron mechanism. For a reasonable magnetic field $B\sim 10$–100 $\mu$G, the Lorentz factors of the X-ray emitting electrons should be in the range $\gamma\sim(0.3$–$3)\times 10^{8}$. Such parameters are consistent with the lack of synchrotron cooling on a 1000 days timescale.

Clump motion and evolution.

In both binary cycles the clumps were moving in the periastron-apastron direction. A lack of any deceleration of the clumps implies an exremely low density of the ambient medium (unless the clump is unrealistically heavy) (Kargaltsev et al., 2014; Pavlov et al., 2015). Such a medium could be created by the pulsar wind, which sweeps out the matter of the stellar wind bubble in a cone confined by the shock that separates the colliding stellar and pulsar winds (Tavani & Arons, 1997; Bogovalov et al., 2008). Although this cone rotates around the massive star together with the pulsar, the cone’s axis is close to the periastron-apastron direction during most part of the very eccentric orbit, creating a low-density channel, which explains the ejection direction (Barkov & Bosch-Ramon, 2016). Moreover, the pulsar wind pressure can accelerate the clump, even when it is at large distances from the binary (Pavlov et al., 2015). The apparent acceleration in the 2014-2017 cycle can be reached for clump masses m_{\rm cl}\lower 4.0pt\hbox{\buildrel<\over{\sim}}10^{23} g.

Conclusion.

The clumps are likely composed of a mixture of stellar matter with the shocked pulsar wind. They are formed by the complex interaction of the stellar and pulsar winds. Formation and initial acceleration of the clumps can be closely connected with the post-periastron $\gamma$-ray flares, but this connection is not certain yet. The X-ray emitting electrons could be accelerated to $\gamma\sim 10^{8}$ in shocks or by magnetic field reconnection. The reconnection in the turbulent clump plasma could be responsible for the observed brightening.

Support for this work was provided by NASA through Chandra Awards GO5-16065, GO7-18056 and DD7-18088 issued by the Chandra X-ray Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of the National Aeronautics Space Administration under contract NAS8-03060.