A DEEP RADIO SURVEY OF HARD STATE AND QUIESCENT BLACK HOLE X-RAY BINARIES

Black hole X-ray binaries spend the majority of their duty cycles in a low-luminosity (<10^33.5 erg s⁻¹), quiescent state. Theoretical models to explain the low luminosities and the observed spectra either consider radiatively inefficient accretion flows (e.g., Narayan & Yi 1995), reduce the inner accretion rate to zero (Narayan et al. 2000; Quataert & Gruzinov 2000), invoke winds or outflows (Blandford & Begelman 1999), or some combination of these effects. More recently, jets have been demonstrated to represent a significant component of the power output of accreting black holes at low luminosities (Fender et al. 2003; Gallo et al. 2005a), making the coupling between inflow and outflow processes crucial to understanding the nature of low-luminosity accretion.

Simultaneous multi-wavelength observations (Gallo et al. 2003; Corbel et al. 2003) have shown that the radio and X-ray luminosities (L_R and L_X, respectively) of hard state black hole X-ray binaries are correlated via the nonlinear relation L_R∝L^0.6_X, implying a fundamental coupling between the accretion and ejection processes in hard state systems, which was later demonstrated to extend unbroken into the quiescent state (Gallo et al. 2006). With the inclusion of a mass term, the correlation was further extended to include supermassive black holes (Merloni et al. 2003; Falcke et al. 2004; Gültekin et al. 2009). However, the universality of this correlation has been called into question. From an observational perspective, a population of radio-faint X-ray binaries lying well below the correlation was discovered in the relatively bright hard X-ray state (e.g., Gallo 2007), although Coriat et al. (2011) found evidence that this population may rejoin the correlation at low luminosities (≲ 5 × 10⁻⁵L_Edd). More theoretical work also suggested that the correlation should break down below a fraction 10⁻⁵ of the Eddington luminosity L_Edd, with the power-law index steepening as the synchrotron X-ray emission from the jet begins to dominate the Compton-upscattered emission from the accretion flow (Yuan & Cui 2005). Although two of the most sub-Eddington systems detected in the radio band, Sgr A* and A0620-00, both appear to rule out this model, it is possible that the proposed threshold luminosity below which the correlation steepens is either significantly lower than proposed by Yuan & Cui (2005), or that it is not the same for all sources. Thus, the true nature of the coupling between inflow and outflow at the lowest luminosities has been difficult to confirm.

The extremely low luminosity of the quiescent state of black hole X-ray binaries has rendered it inaccessible to radio observations for all but the closest and brightest systems, A0620-00 and V404 Cyg, respectively. A0620-00, at a distance of only d = 1.06 ± 0.12 kpc (Cantrell et al. 2010), is currently the faintest radio-emitting black hole X-ray binary known, with a measured 8.5 GHz radio flux density of 51 ± 7 μJy and a simultaneously determined 2–10 keV X-ray luminosity of only 7.1 × 10³⁰(d/1.2 kpc)² erg s⁻¹ (Gallo et al. 2006). V404 Cyg, at the recently measured parallax distance of 2.39 ± 0.14 kpc (Miller-Jones et al. 2009), has a relatively high quiescent luminosity of 4 × 10³² erg s⁻¹ (Corbel et al. 2008). It is a persistent radio source with a flat radio spectrum (Gallo et al. 2005b) and a flux density of ∼0.3 mJy, although occasional small flares have been detected in the quiescent state, in which the flux density increases by a factor of three on timescales of ∼30 minutes (Miller-Jones et al. 2008). The persistent radio emission is believed to arise from a steady, partially self-absorbed, compact jet (e.g., Blandford & Konigl 1979). The persistent, core-dominated nature of this emission makes black hole X-ray binaries ideal astrometric targets when they are in their hard or quiescent states. High-precision astrometric observations with very long baseline interferometry (VLBI) would allow us to measure source proper motions, which can provide information about the formation mechanisms of the black holes (e.g., Mirabel et al. 2001; Mirabel & Rodrigues 2003), and in some cases, model-independent distances (e.g., Miller-Jones et al. 2009). Prior to embarking on an astrometric campaign, we need to determine which hard state and quiescent systems are bright enough to be detected with the High Sensitivity Array (HSA).

While several black hole X-ray binaries have been monitored during their decay to quiescence following an outburst (e.g., Jonker et al. 2004, 2010), in every case, the sources have faded below the detection threshold before reaching their stable quiescent luminosity. Using the enhanced sensitivity of the new wideband receiver (CABB; the Compact Array Broadband Backend) on the Australia Telescope Compact Array (ATCA), Calvelo et al. (2010) attempted to detect the two quiescent systems GRO J1655−40 and XTE J1550−564, finding mildly constraining 3σ upper limits on their quiescent radio emission of 26–27 μJy at 5.5 GHz and 47 μJy at 9 GHz. Deeper radio observations are required to identify astrometric targets and to accurately determine the nature of the correlation between radio and X-ray emission in the quiescent state.

In this Letter, we describe a deep radio survey of quiescent black hole X-ray binaries carried out by the Very Large Array (VLA) in order to identify possible targets for an astrometric observing campaign, assuming the existence of more radio-loud quiescent sources similar to A0620-00, V404 Cyg, and Sgr A*. We also report on higher-sensitivity follow-up observations of three targets using the wide bandwidths now available with the Expanded Very Large Array (EVLA; Perley et al. 2011).

2.1. VLA

2.2. EVLA

In our original VLA survey, only Swift J1753.5−0127 was detected, with a flux density of 0.4–0.5 mJy in each of the observations (the exact flux densities have already been reported by Soleri et al. 2010). The other 10 sources were not detected, and are therefore too faint for an astrometric HSA observing campaign. Of the three systems that we observed with the EVLA (GRO J0422+32, XTE J1118+480, and GRO J1655−40), none were detected at a statistically significant level. The observations of GRO J1655−40 were hampered by the array configuration (C configuration; given the low declination of the source, a hybrid array with an extended northern arm would have been more appropriate) and the presence of significant extended emission in the field. To filter this out, we selected only data with a uv-spacing of >5 kλ. Our best radio upper limits on all 11 sources in our sample are shown in Table 2.

While not formally significant, the measured radio brightnesses at the known source positions of XTE J1118+480 and GRO J1655−40 were 6.4 ± 2.6 and 9.0 ± 4.7 μJy beam⁻¹, respectively. This suggests that these are good candidates for deeper observations, using either longer integration times or further increased bandwidth following the completion of the EVLA upgrade. Considering the entire sample, a Kolmogorov–Smirnov test on all 11 non-detections gave a probability of 0.126 that the measured pixel values at the source positions (as a fraction of the image noise) are consistent with the expected Gaussian distribution of zero mean and unit standard deviation. Thus, we cannot formally exclude the slightly positive pixel values at the positions of XTE J1118+480 and GRO J1655−40 being due to chance.

For all the systems in our sample with constraints on the quiescent X-ray luminosity, we have plotted our radio upper limits on the radio/X-ray correlation in Figure 1, together with a sample of other sources from the literature. We used WebPimms⁹ to convert all X-ray luminosities to a 3–9 keV range, for compatibility with the observations of V404 Cyg and GX 339−4 that form the bulk of the sample (Corbel et al. 2003, 2008). We calculated the 8.4 GHz radio luminosities by multiplying the measured monochromatic luminosities by a nominal observing frequency of 8.4 GHz, assuming a flat radio spectrum in all cases. We performed a fit to the correlation using data from A0620-00, GX 339−4, and V404 Cyg, finding L_R∝L^0.67_X. All three of our EVLA upper limits fall on or above our best fitting correlation, so we cannot constrain whether the correlation continues to hold at the lowest luminosities.

Zoom In Zoom Out Reset image size

The only quiescent sources with radio detections are A0620-00 (Gallo et al. 2006) and V404 Cyg (Gallo et al. 2005b). The fitted correlation in Figure 1 shows that the three sources observed with the EVLA could just have been detected if they were as bright as A0620-00 at a similar X-ray luminosity. However, when taking into account the known variability in quiescent luminosity and the uncertainties in source distances and the fitted correlation index, we cannot definitively state that the sources are more radio faint in quiescence than expected.

4.1. Uncertainties

4.2. Stellar Radio Emission

We have made deep VLA observations of the fields surrounding 11 confirmed or candidate black hole X-ray binaries. The only source detected was Swift J1753.5−0127, at a level of 0.4–0.5 mJy beam⁻¹. The typical 3σ upper limits on the radio brightnesses of the remaining sources were in the range 30– 50 μJy beam⁻¹. Three targets were selected for follow-up with the EVLA, using 2 GHz of bandwidth to enhance the sensitivity. We found 3σ upper limits of 7.8, 8.3, and 14.2 μJy beam⁻¹ for XTE J1118+480, GRO J0422+322, and GRO J1655−40, respectively. We find that most quiescent systems are beyond the reach of current high-sensitivity VLBI arrays, and cannot therefore be used as astrometric targets. Uncertainties in the source distances (Jonker & Nelemans 2004), in the empirical correlation between radio and X-ray emission, and in the X-ray luminosities at the times of the observations imply that we cannot rule out these sources falling on the radio/X-ray correlation of Gallo et al. (2003). However, while deeper constraints on the quiescent jet emission will be possible on completion of the EVLA, the existing upper limits suggest that it may be difficult to disentangle jet emission from stellar radio emission at such low radio luminosities.

J.C.A.M.-J. thanks Stephane Corbel for making available his original data on GX339−4. 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.

Facilities: EVLA - Expanded Very Large Array, VLA - Very Large Array