Dynamic shocks powered by a wide, relativistic, super-Eddington outflow launched by an accreting neutron star in the mid-20th century

TL;DR

Using two decades of radio data from ATCA and MeerKAT, this study resolves parsec-scale shocks around Circinus X-1 as synchrotron-emitting caps formed by interaction with the ambient medium. The shocks are powered by a hidden wide-angle, mildly relativistic outflow launched in the mid-1970s, with deprojected speeds around $\sim0.1$–$0.3c$ and an opening angle $>17^{\circ}$, implying a kinetic power near $\sim40\,L_{\rm Edd}$ when accounting for ambient density. Broadband modeling shows electrons can reach $\sim0.7$ PeV and protons up to $\sim20$ PeV, with a hadronic gamma-ray component potentially detectable by next-generation facilities, positioning Cir X-1 as a rare Galactic PeVatron site. The results demonstrate extreme feedback and particle acceleration in a young X-ray binary and suggest ultra-fast outflows or transient energetic events can dominate energy transfer to the ISM in such systems.

Abstract

Accreting systems can launch powerful outflows which interact with the surrounding medium. We combine new radio observations of the accreting neutron star X-ray binary (XRB) Circinus X-1 (Cir X-1) with archival radio observations going back 24 years. The $\sim3$ pc scale wide-angle radio and X-ray emitting caps found around Cir X-1 are identified as synchrotron emitting shocks with significant proper motion and morphological evolution on decade timescales. Proper motion measurements of the shocks reveal they are mildly relativistic and decelerating, with apparent velocity of $0.14c\pm0.03c$ at a propagation distance of 2 pc. We demonstrate that these shocks are likely powered by a hidden relativistic ($\gtrsim0.3c$) wide-angle conical outflow launched in $1972\pm3$, in stark contrast to known structures around other XRBs formed by collimated jets over 1000s of years. The minimum time-averaged power of the outflow required to produce the observed synchrotron emission is $\sim0.1L_\text{Edd}$, while the time-averaged power required for the kinetic energy of the shocks is $\sim40 \left(\frac{n}{10^{-2} \text{cm}^{-3}}\right)L_\text{Edd}$, where $n$ is the average ambient medium number density. This reveals the outflow powering the shocks is likely significantly super-Eddington. We measure significant linear polarisation up to $52\pm6\%$ in the shocks demonstrating the presence of an ordered magnetic field of strength $\sim200~μ\text{G}$. We show that the shocks are potential PeVatrons, capable of accelerating electrons to $\sim0.7~\text{PeV}$ and protons to $\sim20~\text{PeV}$, and we estimate the injection and energetic efficiencies of electron acceleration in the shocks. Finally, we predict that next generation gamma-ray facilities may be able to detect hadronic signatures from the shocks.

Figures (10)

• Figure 1: 2.6 GHz MeerKAT image in 2023 of Cir X-1 and its natal SNR, the Africa nebula. Structures of radio emission associated with the SNR and areas where the outflows from Cir X-1 interact with the surrounding region are visible and labelled. Cir X-1 and its precessing jets on parsec scales are seen as radio sources in the centre of the nebula. To the north-west and south-east asymmetric jet inflated bubbles are visible at the edge of the SNR. Roughly half way between Cir X-1 and the edge of the SNR the radio cap structures are seen to the north-west and south-east. An unlabelled version of this plot is available in the online supplementary material.
• Figure 2: Three images showing Cir X-1 and the radio caps in 2001 at 1.4 GHz, 2011 at 2.1 GHz, and 2023 at 2.6 GHz. Evolution in the morphology and position of the radio caps over the different epochs is evident. Despite this, the straight dashed lines represent an opening angle of $29\degree$, illustrating the constant opening angle over the three epochs. Edges found by the gradient based edge detection algorithm described in text are shown, and the beam size is illustrated by an ellipse in the lower left. Despite the fact that different interferometer sampling can cause apparent differences in images of diffuse structure we are confident in the structural and positional changes observed (see text for more). A .gif showing the evolution of the caps over all 5 epochs is provided in the online supplementary material.
• Figure 3: Image of Cir X-1 and the radio caps in 2023 with 6 and 12 $\sigma$ contours from the 2001, 2011, and 2023 images over-plotted, clearly showing the motion of the radio caps away from Cir X-1. The position angle of the brightest region in the caps, the hotspot, also varies over the 3 epochs. The beams of all 3 observations are shown in the bottom left.
• Figure 4: Surface brightness profiles of the NW cap along a position angle of $-48\degree$ E of N, perpendicular to the long axis of the caps. In one case the surface brightness profile crosses the hotspot and in the other case it does not. The lines along which the profiles are measured are shown in an inset in the top left in addition to an example of the ellipse used for flux density measurements discussed in text. The surface brightness profiles are aligned to their maximum for ease of comparison. An asymmetry in both profiles is apparent, with a sharp drop in surface brightness on the side of the cap furthest from Cir X-1, expected in the case of a shock. The shaded region represents the uncertainties on surface brightness measurements from the image noise. The size of the beam, the expected size scale of correlated image noise, is shown in the bottom left by the double headed arrow.
• Figure 5: Separation of the caps from Cir X-1 as a function of time for the NW edge and hotspot (left), and the SE edge and hotspot (right). For the edge positions the best fit deceleration model is shown, with parameters given in Table \ref{['tab:position_fits']}. An asymmetry in the separation from Cir X-1 between the NW and SE caps and evidence for deceleration in both cases is evident.