VLBA Discovery of a Resolved Source in the Candidate Black Hole X-Ray Binary AT2019wey

Black Hole X-ray binaries are comprised of stellar mass black holes accreting from companion stars. Most of the electromagnetic emission from these systems arises from accretion disks and relativistic jets, and strong coupling is observed between properties of the disks and jets. The presence of a jet is dependent on which X-ray spectral state the black hole binary is observed to be in. This phenomenon was first observed in the correlated X-ray and radio intensities of Cyg X-1 (Tananbaum et al. 1972), where the X-ray emission was interpreted as originating from the disk, and the radio emission from the jet. The two main X-ray spectral states of interest are the thermal state (formerly known as the high/soft state) and the hard state (formerly known as the low/hard state). In the thermal state, >75% of the observed flux is contributed by the accretion disk. The predominantly thermal spectrum is accompanied by a steep power law extending to energies higher than ∼10 keV. In the hard state the disk appears cooler and contributes very weakly, and >80% of the flux is contributed by a nonthermal power-law spectrum with a photon index of 1.4 < Γ < 2.1. For detailed reviews, see Fender (2003), McClintock & Remillard (2009), and Remillard & Mcclintock (2006).

The first observation of a resolved jet in an X-ray binary was conducted with the Very Large Array (VLA) on SS 433 (Hjellming & Johnston 1981). Subsequent observations of the X-ray binary systems GRS 1915+105 (Mirabel & Rodríguez 1994) and GRO 1655 − 40 (Tingay et al. 1995), with the VLA and the Southern Hemisphere VLBI Experiment (SHEVE), respectively, revealed the first Galactic examples of superluminal motion of relativistic jet components. The ejection of superluminal components was linked to X-ray outbursts. In its hard "plateau" X-ray state, however, GRS 1915+105 was observed to host a compact steady radio jet with mildly relativistic component velocities (Dhawan et al. 2000). Compact steady jets have also been observed in two other sources, Cyg X-1 (Stirling et al. 2001) and MAXI J1836−194 (Russell et al. 2015), in canonical hard states. The link between the ejection of highly relativistic jet components and X-ray outbursts in the thermal state, first observed in GRS 1915+105 (Dhawan et al. 2000), established a causal link between the accretion rate and jet properties (Vadawale et al. 2003; Fender et al. 2004).

Radio observations of the hard-state compact steady jets of GRS 1915+105 and Cyg X-1 that comfortably resolve the emission region were used to derive the jet powers and speeds (Dhawan et al. 2000; Stirling et al. 2001). These measurements are consistent with the model of conical synchrotron jets to describe radio emission from black hole accretors (Blandford & Konigl 1979; Hjellming & Johnston 1988), and motivated the derivation of scaling relations between jet power and accretion-disk luminosity (Falcke & Biermann 1995, 1999). Adding to the handful of spatially resolved black hole X-ray binary jets in the hard state is critical toward refining physical models for disk/jet coupling.

1.1. The Candidate Black Hole X-Ray Binary AT2019wey

Figure 1 shows X-ray and radio light curves for AT2019wey. The X-ray data are taken by the Monitor of All-sky X-ray Image (MAXI) telescope in the 2–10 keV band (Matsuoka et al. 2009). The radio data point depicted on May 27 was taken with the VLA at 6.0 GHz while the observations between August 2 and 28 were taken with the VLA at 3.5 GHz. These measurements were scaled to 4.8 GHz using the spectral indices published by Yao et al. (2020c). The radio data point shown on October 17 was taken by Giroletti et al. (2020) with the European VLBI Network (EVN) at 6.7 GHz; this point was scaled to 4.8 GHz on the light curve as well using the spectral index from Yao et al. (2020c) on August 28.

Zoom In Zoom Out Reset image size

The two epochs of observations of AT2019wey at 4.8 GHz were obtained with the VLBA on September 6 and 12 and were processed with the DiFX correlator (Deller et al. 2011). Both 3 hour epochs were phase referenced, with alternating scans of 3.5 minutes on AT2019wey and 40 s on the phase reference (J0418+5457). The phase reference is at an assumed location of R.A. = 04^h18^m193401920 and decl. = 54°57'15334490, an angular separation of 247 from the target, and was chosen because of its inclusion in the third realization of the International Celestial Reference Frame (ICRF3; Charlot et al. 2020). Each epoch also contained 4 minute observations of a check source (J0427+5618), and 6 minute observations of a bandpass calibrator (J0555+3948).

Calibration and imaging for the observations were carried out in AIPS (Greisen 2003) using standard procedures. To derive phase solutions, we performed global fringe fitting on the phase calibrator followed by two rounds of phase-only self-calibration and one round of amplitude+phase self-calibration. The self-calibration was done with 2 minute solution intervals and assumed a point-source model with the catalog flux density for the phase calibrator. Once phase variations of less than ±5° were reached for all stations, the phase solutions were applied and no further self-calibration was performed on either the check source or AT2019wey. All images were created with natural weighting to maximize sensitivity. Images of the phase calibrator and check source are shown in Figure 2 and images of AT2019wey are shown in Figure 3. The FWHM of the Gaussian model used to approximate the synthesized beam (shown in the corners of the images) is approximately a 4 mas × 1.5 mas ellipse in both epochs. The estimated deconvolved component for AT2019wey for each epoch is represented by the red dotted ellipse in Figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The images and uv-amplitudes of AT2019wey both show that the source is resolved. The images (Figure 3) show a source that is clearly wider and oriented at a different position angle than the elliptical model of the synthesized beam. Plots of the uv-amplitudes, coherently averaged for 20 minute intervals, are shown in Figure 4 along with uncertainties. The amplitudes are clearly not constant as a function of uv-distance, as would be expected for a point source.

Zoom In Zoom Out Reset image size

We attempted to fit a 2D Gaussian component to the visibilities using the uvfit task in AIPS; however, low signal-to-noise in individual visibility measurements yielded unreliable results for a 2D Gaussian model fit to visibilities. Instead, we performed an image plane fit using the AIPS task jmfit, which uses a least squares approach to fit a 2D Gaussian component to the image and then deconvolves the CLEAN beam from the fitted component to estimate a 2D Gaussian model for the true source geometry. The jmfit task also provides uncertainties for all of these parameters; however, the task will report the lower limits of the deconvolved major and minor axes as 0 mas if a reliable uncertainty cannot be derived. As we are confident that the source is resolved, in these cases we assume symmetric uncertainties based on the estimate of the upper limit for use in any calculations. To achieve convergence with the jmfit task, we fix the position of the 2D Gaussian to the position of the phase center of the image. The errors on the R.A. and decl. of the source position are then dominated by the cataloged position errors of the phase reference.

To investigate time variability in the phase calibrator and AT2019wey, we image and perform model fitting with jmfit for three hour-long blocks in each epoch. The fitted flux density, position angle reported east of north, the major axis, and the minor axis for each hour of both epoch are shown along with uncertainties in Figure 5. The mean best-fit values and uncertainties are summarized in Table 1. We see that AT2019wey has a flux density of 1.35 ± 0.02 mJy and an approximated 2D Gaussian geometry with major and minor axes of 2.13 ± 0.10 mas and 0.80 ± 0.18 mas, respectively, at a position angle of 122° ± 4°. These values closely agree with those derived from performing a model fit on data combining the two epochs. Though the formal uncertainties on these parameters are small, this does not imply the true source geometry is well modeled by a Gaussian; however, subtracting this model from the visibilities and imaging the residuals reveals only noise, an indication that it would be difficult to extract more complex structure from this data. The model does illustrate relatively stable source properties and facilitates a rough estimate of source power.

Zoom In Zoom Out Reset image size

Table 1. Properties for AT2019wey, Approximating the Source Structure as a 2D Gaussian

R.A. (J2000)	Decl. (J2000)	Flux Density	Major Axis	Minor Axis	Position Angle
04h35m2327345 ± 000028	55° 22' 34''28715 ± 000017	1.35 ± 0.02 mJy	2.13 ± 0.10 mas	0.80 ± 0.18 mas	122° ± 4°

Note. Though this geometry does not reflect the true source structure, more complex features cannot be extracted from these observations.

Download table as:  ASCIITypeset image

Although the source geometries of AT2019wey remain relatively consistent within the uncertainties across both epochs, there are some notable deviations. The best-fit parameters are anomalous for the first hour of the first epoch. Inspection of the phase calibrator showed no significant phase or gain fluctuations during this time, indicating that the anomalous measurement is likely not caused by calibration error. As a light-crossing time of 1 hour is equivalent to a distance of several au, a scale similar to the synthesized beam-width for low distance estimates, it cannot be ruled out that this measurement is due to true variability in the X-ray binary. The fitted flux density is also around 15% higher in the second epoch than the first. As the source is seen to be fading in the radio light curve presented in Yao et al. (2020c), we do not expect the second epoch to show significant increase in brightness. A similar level of variation is seen in the flux density of the check source, which was fitted to 45.6 mJy in epoch 1 compared to 54.9 mJy in epoch 2. This may indicate residual phase error in the first epoch leading to a lower flux density measurement, but intrinsic source variability on these small angular scales cannot be ruled out. The use of dynamical imaging and attempts to super-resolve the source, left to a later work, would provide a way to verify potential short timescale evolution.

A comparison of these VLBA observations of AT2019wey and observations taken of similar systems, specifically GRS 1915+105 and MAXI J1836−194, provides compelling evidence that the radio source we observe is likely a steady compact jet. Compact radio sources were observed while both of these systems were in hard X-ray spectral states. Dhawan et al. (2000) report a well-resolved, elongated radio source in GRS 1915+105 that exhibits a shallow radio spectral index, α ≲ 0.5, and a steady position angle over two years of observations. Although Russell et al. (2015) report only a marginally resolved radio source in MAXI J1836−194 with a steeper spectral index of α ≲ 0.8, the source also shows a stable position angle over two months of observations. The fact that AT2019wey was observed in the hard X-ray state with a nearly flat radio spectral index of α ∼ 0.2 and the position angle of the VLBA source remains stable between the two epochs of observation indicate the presence of a compact steady jet.

The measurement of an angular size enables us to estimate the total energy in the source following standard synchrotron theory (following Pacholczyk 1970). We assume an ellipsoidal structure for the source, with a projected shape corresponding to the 2D Gaussian described above. At a fiducial distance of D = 3 kpc, we adopt a source volume of V = 10⁴² cm⁻³. We assume a flat radio spectrum between 1 and 12 GHz only (spanning the VLA observations of AT2019wey), with a flux density of 1.35 mJy. The minimum energy required to power the synchrotron source (relativistic particles and magnetic fields) is then

$$\begin{eqnarray}&&{E}_{\min }\approx 5\times {10}^{38}{\left(\displaystyle \frac{D}{3\,\mathrm{kpc}}\right)}^{\tfrac{17}{7}}{\left(\displaystyle \frac{V}{{10}^{42}\,{\mathrm{cm}}^{3}}\right)}^{\tfrac{3}{7}}\,\mathrm{erg}.\end{eqnarray} \tag{ 1 }$$

The corresponding mean magnetic field strength is ∼0.07 G, implying a relativistic-lepton Lorentz factor of γ ≈ 250 for 12 GHz emission. Assuming a characteristic particle acceleration timescale corresponding to the light-crossing time of the source of ∼3 × 10³ s at 3 kpc, the power dissipation in the source is approximately

$$\begin{eqnarray}&&P\gtrsim 2\times {10}^{35}{\left(\displaystyle \frac{D}{3\,\mathrm{kpc}}\right)}^{\tfrac{10}{7}}{\left(\displaystyle \frac{V}{{10}^{42}\,{\mathrm{cm}}^{3}}\right)}^{\tfrac{3}{7}}\,\mathrm{erg}\,{{\rm{s}}}^{-1}.\end{eqnarray} \tag{ 2 }$$

We emphasize that this is a lower limit given the limited band used to calculate the total radio luminosity, and the minimum-energy assumption.

The inferred power is remarkably close to the ∼10³⁶ erg s⁻¹ X-ray luminosity of AT2019wey found by Yao et al. (2020d) for a 3 kpc distance. The luminosity of the thermal emission from the disk is likely a few tens of percent of this total. This correspondance has been observed previously in the hard and plateau states of the GRS 1915+105 (e.g., Dhawan et al. 2000), and is a critical assumption of models for symbiotic disk-jet systems (Falcke & Biermann 1999). Although we resolve the radio source in AT2019wey, we have no compelling morphological evidence for a jet. Nonetheless, the panchromatic properties of AT2019wey are closely similar to low-mass black hole X-ray binaries in which relativistic jets have been observed (Yao et al. 2020c). We therefore interpret the resolved source as a compact steady jet, with a power that is comparable to the accretion-disk X-ray luminosity.

We present here two epochs of 4.8 GHz VLBA observations of candidate black hole low-mass X-ray binary system AT2019wey following a period of X-ray and radio brightening. The observations revealed a resolved source with deconvolved source geometries that are relatively constant across both epochs. Together with the observed X-ray spectrum, we interpret these results to indicate the presence of a compact, steady jet. Using the angular scale derived from image plane fits of a 2D Gaussian component to the source, we show that the power dissipation from the jet is comparable to the X-ray luminosity, consistent with a standard assumption of models for disk/jet coupling.

Thus far, spatially resolved compact jets in X-ray binaries in the hard spectral state have only been observed for five systems, including AT2019wey. The next-generation Very Large Array (ngVLA) and the Square Kilometre Array (SKA) are ideal instruments to expand on this limited sample (Fender et al. 2015; Maccarone et al. 2018). With their combination of long baselines and extreme sensitivity, the ngVLA and SKA will enable high-cadence monitoring of the flux densities and multiscale morphologies of an extended sample of X-ray binaries. The ngVLA may in fact prove to be a discovery engine for accretion Galactic black holes through astrometric surveys (Maccarone et al. 2019). Observations of these systems during a variety of states and state transitions are required to broaden our understanding of the connections between accretion states and jets.

We thank Tim Pearson, Gregg Hallinan, and Katie Bouman for useful discussions on interpreting these results. These observations were conducted with the Very Long Baseline Array. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This research has made use of NASA's Astrophysics Data System.