Archival VLBA observations of the Cygnus A Nuclear Radio Transient (Cyg A-2) Strengthen the Tidal Disruption Event Interpretation
DDraft version September 10, 2020
Typeset using L A TEX default style in AASTeX62
Archival VLBA observations of the Cygnus A Nuclear Radio Transient (Cyg A-2) Strengthen the Tidal DisruptionEvent Interpretation
Steven J. Tingay, James C. A. Miller-Jones, and Emil Lenc International Centre for Radio Astronomy Research, Curtin University, Bentley, WA, Australia 6102 CSIRO Astronomy and Space Science, P.O. Box 76, Epping, NSW 1710, Australia (Received January 1, 2018; Revised January 7, 2018; Accepted September 10, 2020)
Submitted to ApJABSTRACTWe have analyzed archival VLBA data for Cygnus A between 2002 and 2013, to search for radioemission from the transient discovered in 2015 by Perley et al. (2017) approximately 0.4 (cid:48)(cid:48) from thenucleus of Cygnus A (Cyg A-2). Perley et al. (2017) use VLA and VLBA archival data (between1989 and 1997) to show that the transient rises in flux density by a factor of at least five in less thanapproximately 20 years. With the additional data presented here, we revise the rise time to betweenapproximately four years and six years, based on a new detection of the source at 15.4 GHz fromOctober 2011. Our results strengthen the interpretation of Cyg A-2 as the result of a Tidal DisruptionEvent (TDE), as we can identify the location of the compact object responsible for the TDE andcan estimate the angular expansion speed of the resulting radio emitting structures, equivalent to anapparent expansion speed of < . c . While our results are consistent with recent X-ray analyses, wecan rule out a previously suggested date of early 2013 for the timing of the TDE. We favour a timingbetween early 2009 and late 2011. Applying the model of Nakar & Piran (2011), we suggest a TDEcausing a mildly relativistic outflow with a (density-dependent) total energy > erg. Due to theimproved temporal coverage of our archival measurements, we find that it is unlikely that Cyg A-2 haspreviously been in a high luminosity radio state over the last 30 years. Keywords: galaxies: individual (Cygnus A) - radiation mechanisms: general - techniques: high angularresolution - radio continuum: general INTRODUCTIONUshered in by a new era of data processing capacity and high survey speed telescopes, significant effort has beenexpended over the past decade into large-scale surveys for radio transients at centimetre to metre wavelengths, andfor transient durations longer than a second, e.g. Kuiack et al. (2020); Hajela, Mooley, Intema et al. (2019); Bell etal. (2019); Fender et al. (2017); Murphy et al. (2013).These surveys have, in general, revealed remarkably few transient radio objects in a systematic manner, as elaboratedby Metzger, Williams & Berger (2015). It is likely that significantly higher sensitivity is required to uncover more ofthe slow transient and variable radio universe, for example using the Square Kilometre Array (Fender et al. 2015).However, some significant recent advances have been forthcoming from targeted follow-up observations of interestingevents, for example the multi-wavelength and multi-messenger observations of GW170817 (Abbott et al. 2017).It is still the case that many interesting radio transients are the result of serendipity, rather than surveys. One suchrecent discovery, by virtue of the fact that the transient resides within the host galaxy of one of the most powerful andnearby radio galaxies, Cygnus A, was reported by Perley et al. (2017).
Corresponding author: Steven [email protected] a r X i v : . [ a s t r o - ph . H E ] S e p Tingay, Miller-Jones & Lenc
Cygnus A has been observed with high angular resolution radio telescopes over decades, providing a high qualityand long timescale series of observations. So when VLA observations in 2015 revealed a never-before-seen compactradio source approximately 0.4 (cid:48)(cid:48) from the bright nucleus of Cygnus A, Perley et al. (2017) could examine historicaldata. They found that no such source had existed in the 1980s and 1990s, from VLA and VLBA data.Perley et al. (2017) made new observations in the frequency range 7.1 – 47 GHz with the VLA and VLBA to examinethe object, designated Cygnus A-2 (Cyg A-2), and from the sum total of the available data favoured a scenario inwhich Cyg A-2 resulted from accretion onto a secondary supermassive black hole, approximately 450 pc distant fromthe primary black hole forming the Cygnus A nucleus. The rapid rise in flux density, by a factor of at least five inapproximately 20 years, was cited as one of the distinctive characteristics of this object and evidence in favour of anexplanation in terms of accretion. Perley et al. (2017) note, in particular, the possibility that Cyg A-2 is the result ofa Tidal Disruption Event (TDE).Additional suggestions of a TDE origin have been presented by de Vries et al. (2019), from an analysis of new andhistorical X-ray observations of Cygnus A. They find an enhancement in the X-ray light curve in early 2013 and noteprevious results that indicate the presence of a short-lived, fast, ionised outflow near that epoch. de Vries et al. (2019)take this as evidence for a TDE at that time, with the radio emission seen by Perley et al. (2017) in 2015 due to anafterglow, although they do not rule out stochastic X-ray variation of the Cygnus A AGN, which is not resolved fromCyg A-2.We examine the TDE origin for Cyg A-2 in more detail, by the additional mining of archival data to constrainmore tightly the rise time of the radio emission. We have extracted archival VLBA data between 2002 and 2013,over frequencies ranging from 1.6 GHz to 15.4 GHz, and search for emission from the reported location of Cyg A-2.We detect Cyg A-2 at one epoch, at a frequency of 15.4 GHz, in late 2011. This detection has consequences for theinterpretation of the other available data for Cyg A-2 in the context of a TDE explanation. In § § DATA PROCESSING AND RESULTSAll data were extracted from the VLBA archive, following a search for data corresponding to the position of CygnusA, with the dates (column 1) and project codes (column 2) for the observations listed in Table 1. The data for projectsBL1788 and BP171 data were correlated with DiFX (Deller et al. 2011, 2007). All other data were correlated usingthe original VLBA hardware correlator. The central frequencies (column 3), bandwidths (column 4), channels persub-band (column 5), and observation durations (column 6) are also given in Table 1.The correlated data were imported into AIPS (van Moorsel, Kemball, & Greisen 1996) for standard processing viafringe-fitting and calibration of the visibility amplitudes. The partially calibrated data were then exported from AIPSand imported into DIFMAP (Shepherd, Pearson & Taylor 1994) for imaging and self-calibration. The imaging processinitially concentrated on the bright core and milliarcsecond-scale jet in Cygnus A, utilising this bright structure to self-calibrate the visibilities in amplitude and phase. For each observation, the expected core-jet structure was recovered,appropriately for the angular resolution (column 7 of Table 1) and sensitivity (column 9 of Table 1) of each observation.Once the data were well calibrated using the bright core-jet, a wider field of view image could be formed in orderto search for emission at the location of the transient: α =19:59:28.32345; δ = +40:44:01.9133 (J2000, ± σ (column 8 of Table 1). All other epochs processed resulted in non-detections. For the non-detections,the image RMS values vary with frequency, duration of observation, and overall quality of the data, as also listed inTable 1 (column 8).In general, the observations are not of long duration and do not contain the full set of VLBA antennas. However, alldatasets contain adequate information to produce good images. Other data exist in the VLBA archive, but we foundthem to be of low quality. The best limits and the detection are obtained from the longest-duration observations. We he Cyg A-2 Transient as a TDE RA (J2000) +40:44:01.90001.91001.920 D e c ( J ) Figure 1.
VLBA images of Cyg A-2. Color scale shows the 15.4 GHz image from 2011 October 3, with the beam size shown asthe blue ellipse at the bottom left. Contours show the stacked 8.4-GHz image from three observations taken in 2016 November(re-processing of the BP213 data presented by Perley et al. 2017). Contours are at levels of ± . √ n mJy beam − , where n = 1 , , ... . Table 1.
Archival VLBA data for Cyg A-2. Columns indicate the observation date, VLBA project code, centralfrequency ( ν c ), number and bandwidth (in MHz) of each IF (∆ ν ), number of channels per IF ( N chan ), time on source(∆ T ), the major and minor axes and position angle of the synthesized beam ( θ max / θ min /PA) , the peak intensity ofdetections, the RMS noise level at the position of Cyg-A-2, and the expected signal loss due to bandwidth smearing( L ). Date Proj. ν c ∆ ν N chan ∆ T θ max / θ min /PA S peak RMS L(yyyy.mm.dd) (GHz) (MHz) (min) (mas/mas/ ◦ ) (mJy/beam) (mJy/beam) (%)2002.08.06 BC123 1.550 8 × −
20 - 2 2.02006.06.15 BL137 8.268 4 × × −
23 - 0.6 1.22011.10.03 BL178 15.357 8 × −
10 2.3 0.3 0.042013.07.17 BP171 4.360 8 ×
32 4096 7 5.9/1.8/ −
22 - 2 0.04
Tingay, Miller-Jones & Lenc have included data from BC123 at 1.6 GHz, as it produces a good image. However, the reader is cautioned regardingthis particularly low frequency as the scattering effects in this region of Cygnus A are significant (Perley et al. 2017).Finally, in order to verify our data processing methods, and provide confidence in our detection and non-detections,we downloaded the VLBA data presented by Perley et al. (2017), with project code BP213. We found that missingantennas on 2016 November 14 led to a higher noise level for that epoch, causing us to omit it from the stackedimage. Owing to the proximity of Cyg A-2 to the Cygnus A AGN radio source, we self-calibrated the data on CygnusA, and simultaneously imaged both fields within AIPS. Thus our complex gain solutions were derived for the targetdata themselves, rather than being interpolated to Cyg A-2 from neighbouring scans on Cygnus A, as was the case inPerley et al. (2017). Applying these methods, we accurately reproduced the results presented in Perley et al. (2017).Figure 1 shows our image from October 2011, with contours from our image of the Perley et al. (2017) data from 2016November.From Figure 1, we see that at 15.4 GHz in late 2011, we detect one of the possible two components detectedapproximately four years later by Perley et al. (2017). To align the two images, we had to correct for both thedifference in the positions of Cygnus A assumed at correlation (a shift of 0.53 mas along a position angle 309 ◦ E of N),as well as the known frequency-dependent core shift of Cygnus A (0.49 mas along a position angle of 284 ◦ ; Bach et al.2004; Nakahara et al. 2019). Having done this, we find that the 15.4-GHz detection from 2011 aligns with the westernlobe of the 8.4-GHz image from November 2016. Perley et al. (2017) note the possibility that the extension seen in2016 is due to phase referencing errors. However, we recovered the same structure by self-calibrating the data usingthe Cygnus A radio AGN in the same set of visibilities, negating the need for any phase transfer. We therefore suggestthat the extended structure is likely to be real, and take our 2011 detection to represent the early time, unresolvedradio emitting structure, presumably marking the origin of a jet-like feature that subsequently evolved and was thendetected by Perley et al. (2017). If the same jet-like structure seen in 2016 was present in late 2011, our image wouldhave the sensitivity to detect this extension.Under this interpretation, we can estimate the angular expansion speed for the jet-like feature. From Figure 1 wemeasure the source extension to be 1 . ± . ∼ − , corresponding to a rest-frame apparent expansion speed of 0 . c . However, recognisingthat the zero separation time likely occurred prior to the 15.4 GHz observation, we treat this value as an upper limit. DISCUSSION AND CONCLUSIONFigure 2 summarises our results in the context of the results published by Perley et al. (2017), showing that compa-rable upper limits at comparable frequencies now extend from 1989 to 2010, relative to the 2015 discovery observationsof Cyg A-2. Consistent with the upper limits presented in Perley et al. (2017), our upper limits (and errors on the15.4 GHz detection from 2011) are three times the measured RMS values from our images at the location of Cyg A-2 .We note that our upper limits span a range of 1.6 – 8.3 GHz and that the spectral behaviour of Cyg A-2 presented byPerley et al. (2017) could possibly contribute to non-detection at the lower frequencies, 1.6 – 4.8 GHz.The non-detections, plus our detection at 15.4 GHz in late 2011, immediately constrain the rise time for Cyg A-2to a period significantly shorter than the approximate 20-year period presented in Perley et al. (2017). We define therise time as the period between the TDE and the peak flux density observed by Perley et al. (2017) in mid 2015. OurOctober 2011 detection with the VLBA places a lower limit on the rise time of ∼ ∼ ∼ ∼ − Perley et al. (2017) do not list the significance of their upper limits, but we clarified them to be 3 σ via private communication withthe authors. he Cyg A-2 Transient as a TDE Time (yr) I n t en s i t y ( m Jy / bea m ) Figure 2.
Radio light curve compiled from published results of Perley et al. (2017) (black upper limits and measured points)and archival data processed and presented here (red upper limits and detection). Rather than plot all the measurements fromPerley et al. (2017) from 2015 - 2017, we plot representative points in 2015 and 2017 at the common frequency of 10.5 GHz.The vertical blue line marks the February/March 2013 epoch of the 2 - 10 keV peak noted by de Vries et al. (2019) as theirproposed timing of the TDE. Upper limits and error bars are all 3 σ . but does not rule them out. However, the challenges for supernova models based on energetics described in Perley etal. (2017) remain.In terms of scenarios involving accretion onto a secondary black hole, radio emission from a relatively steady stateaccretion is well known to be associated with AGN and stellar mass black holes in our Galaxy. Powerful, relativisticallybeamed jets in AGN regularly display flares with rise times consistent with the rise time limits and flux density increasefactors for Cyg A-2 (Park & Trippe 2016). Generally many such flares (often overlapping in time) are seen in multi-decade periods. The apparent luminosity of Cyg A-2 being many orders of magnitude below the apparent luminositiesof AGN jets and the singular appearance of Cygnus A-2 in a ∼
30 year period do not argue in favour of this scenario,although a lower limit of four years on the rise time may admit AGN flare timescales; significantly shorter rise timesthan this would be more difficult for AGN models to achieve.Galactic microquasars such as GRS 1915+105 (Mirabel & Rodr´ıguez 1994) and GRO J1655 −
40 (Tingay et al. 1995)have also been show to have relativistic jets and, in contrast to AGN, often remain dormant for long periods of time(decades). However, if such an object were placed at the distance of Cygnus A and directed at us, unrealistically largeDoppler boosting factors are required to produce a radio source as bright as the few mJy seen for Cyg A-2.Thus, the current evidence points to a luminosity consistent with accretion onto a massive black hole, but that isnot steady state or episodic and frequent. This favours a discrete accretion event, such as a TDE, an option noted byPerley et al. (2017) based on the presence of an optical/NIR counterpart assumed to be persistent. Our results provideadditional weight to this scenario by showing that the near-infrared counterpart observed in 2002 May (Canalizo etal. 2003) existed prior to the radio transient and so was not itself a transient. As noted by Perley et al. (2017), anexplanation of the near infrared counterpart as either steady emission from a secondary active galactic nucleus (Perleyet al. 2017) or the stripped core of a merging galaxy (Canalizo et al. 2003) would be consistent with a TDE scenariofor the radio transient. Ongoing monitoring of the infrared counterpart to Cyg A-2 could assist in confirming thissuggestion.A handful of TDEs and candidate TDEs have been shown to produce radio emission, consistently showing radio lightcurves with rise times of months following high energy flares, as comprehensively reviewed by De Colle & Lu (2019)and more recently by Alexander et al. (2020). An analysis of approximately 31 years of X-ray data for Cygnus A byde Vries et al. (2019) shows a mild enhancement in the 2 - 10 keV luminosity in February/March 2013; the authorsalso note previously published evidence for a short-lived, fast, ionised outflow in the NuSTAR spectrum near the sameepoch. de Vries et al. (2019) suggest that Cyg A-2 is the afterglow of a TDE marked by this enhanced X-ray emission.
Tingay, Miller-Jones & Lenc t dec (days) − − − − F p e a k ( m Jy ) ∝ t . . . . . v/c E ( e r g ) Figure 3.
Left panel: The peak flux density, F peak , as a function of time to reach the peak, t dec , for a range of values of outflowenergy, E , and outflow speed, β , as described in the text, for n = 1 cm − . The dark blue lines indicate constraints from theobserved flux density of 4 mJy and the limit on the observed rise time, , indicating the family of models that producepeak flux densities of plausible magnitude and timescale. The red diagonal line indicates the pre-peak flux density evolutionproportional to t , such that values of E and β above that line produce a pre-peak flux density greater than 4 mJy, within of the TDE. Right panel: all values of E and β that produce a peak flux density above 4 mJy within of the TDElie above the line. In this case, the blue line represents the nominal model we have adopted, with n = 1 cm − . The orange linerepresents the model with n = 10 cm − , and the green line represents the model with n = 100 cm − , as described in the text. The timing of the X-ray enhancement is shown in Figure 2, in relation to the radio measurements of Perley et al.(2017) and our new VLBA archival measurements. Our new detection at 15.4 GHz prior to this suggested 2013 timingof the TDE clearly rules out this suggestion. de Vries et al. (2019) do not rule out that the 2013 X-ray enhancementcan be explained as the stochastic variation of the Cygnus A AGN. Our result is consistent with this interpretation.However, jetted TDEs with X-ray emission such as SWIFT J1644+57 show a rapid rise in X-rays, variation in ahigh state over 100s of days, and sharp/gradual declines over 1000s of days (e.g. Levan et al. 2016). Other similarexamples are known (Komossa 2015). Also, the ionised outflow noted by de Vries et al. (2019) could have appearedat any point between 2005 and 2013. The X-ray lightcurve of de Vries et al. (2019) is sparse, with measurements fromboth
NuSTAR and
Swift contributing to the identification of the X-ray enhancement in early 2013, but the closestmeasurement previous to that was in late 2008. de Vries et al. (2019) also note the uncertainties on the errors in theX-ray luminosities when noting the statistically significant X-ray enhancement in 2013. We suggest that, while a 2013TDE date is ruled out by our new VLBA detection, the X-ray enhancement seen in 2013 may be the result of a TDEapproximately two to four years earlier, but representing a decaying X-ray emission rather than marking the timing ofthe TDE itself. de Vries et al. (2019) place the observed X-ray enhancement at approximately 2 × erg s − abovethe long term value for Cygnus A, which appears plausible for an X-ray afterglow of a jetted TDE on a timescale ofyears post-disruption; the jetted TDEs Swift J1644+57 and Swift J2058+05 maintained an X-ray luminosity above10 erg s − for more than a year post-disruption (Levan et al. 2016; Auchettl et al. 2017).Our estimate of the apparent expansion speed of the radio structure, originating from an apparently fixed point wetake to be the location of the black hole and TDE, implies a mildly relativistic intrinsic expansion. The interpretationas a jetted TDE is consistent with our identification of a location of origin and a jet-like evolution of the structure.The radio structure is one-sided and has an apparent expansion speed of < . c , implying a relatively small angle tothe line of sight for the jet, although the true apparent speed is unlikely to be as high as c .Based on this scenario, and as typically adopted for TDEs (van Velzen et al. 2016; Stone & Metzger 2016), we usea canonical blast wave model from a mildly-relativistic outflow (Nakar & Piran 2011) to explore the energetics andtimescales of Cyg A-2. While the jet kinetic power is expected to evolve on timescales of days (Metzger, Giannios &Mimica 2012), at late times the blast wave (producing the radio emission) is likely to become mildly relativstic andspherical (e.g. Stone & Metzger 2016). In this model, the late-time synchrotron emission from a spherically expandingoutflow at frequency ν obs peaks at a time t dec (following the X-ray flare) at a peak flux density of F peak , as a functionof the total energy injected into the outflow, E , and the outflow speed, β = v/c .We adopt the canonical values of Nakar & Piran (2011) for the density of the environment into which the outflowexpands ( n = 1 cm − ) and the fraction of the total energy carried by electrons and magnetic fields ( (cid:15) B = (cid:15) e = 0 . he Cyg A-2 Transient as a TDE p , in the range 2 −
3. Here we adopt p = 2 . F peak at t dec , the emission rises proportionally to t . In recognition of the uncertainty inthe value of n that we adopt (the comprehensive study of the infrared counterpart by Canalizo et al. 2003 is silent onthe density of its environment), we also note results for n = 10 cm − and n = 100 cm − (encompassing the range ofdensities inferred for the environment of the TDE XMMSL1 0740 −
85; Alexander et al. 2017), to demonstrate theirsensitivity to n . We also explore models that assume rise times of 4, 5, and 6 years, in order to cover our observationalestimates.The left panel of Figure 3 shows how F peak varies as a function of t dec , for n = 1 cm − and a range of values of E and β , relative to a rise time for the TDE of 5 years (1825 days), representing the mid-point of our 4 − ∼ ν obs = 10 GHz. The figure showsthat a range in E and β values can produce an appropriate peak flux density on the correct timescales for Cyg A-2.The diagonal line on the figure labeling the t rise for the pre-peak flux density denotes values of E and β above whichthe flux density reaches 4 mJy prior to peak flux density, but within of the TDE. All of these solutions arevalid in the framework of Nakar & Piran (2011), as ν obs = 10 GHz is greater than both the characteristic synchrotronfrequency for the typical electron Lorentz factor ( ≤ ≤ E and β that produce a flux density greater than 4 mJy within of the TDE event are reflectedas the regions above the lines in the right panel of Figure 3, for our full range of assumed densities ( n = 1, 10 and100 cm − ). The minimum β required to accommodate the observational constraints drops from 0.36 to 0.10 as thedensity increases from 1 to 100 cm − over the rise time range. Lower values of β and lower densities both require higherenergies; as β decreases, the minimum required energy increases from ∼ × to ∼ × erg for n = 1 cm − and arise time of 4 years, whereas for n = 100 cm − , the minimum required energy goes from ∼ × to ∼ × erg fora rise time of 6 years. From an outflow energetics point of view, these numbers appear reasonable, as typical outflowkinetic energies for TDEs range from 10 erg for thermal TDEs to several times 10 erg for the non-thermal TDE SwJ1644+57 (Alexander et al. 2020), and can easily be accommodated by the expected energy release for a TDE thatinvolves half a stellar mass (10 − erg; Lu & Parwan 2018). This is also consistent with the TDE bolometricenergies and efficiencies of rest mass conversion to energy compiled by Mockler & Ramirez-Ruiz (2020), which spanranges of approximately 10 − erg and 0 . − .
1, respectively.In summary, our new archival VLBA measurements of Cyg A-2, along with the radio data from Perley et al. (2017),the analysis of de Vries et al. (2019), and consideration of a simple model for the radio emission in temporal andenergetic terms therefore strengthen the general interpretation of a TDE origin for Cyg A-2.We thank the anonymous referee for providing comments that led to significant improvements to the paper, byprompting us down a route of analysis that yielded important new results (the detection at 15.4 GHz). This researchwas partially supported by the Australian Government through the Australian Research Council’s Discovery Projectsfunding scheme (project DP200102471). This work made use of the Swinburne University of Technology softwarecorrelator, developed as part of the Australian Major National Research Facilities Programme and operated underlicence. This research has made use of NASA’s Astrophysics Data System. AIPS is produced and maintained bythe National Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperativeagreement by Associated Universities, Inc.
Facility:
VLBA REFERENCES
Abbott, B.P., Abbott, R., Abbott, T.D., Acernese, F.,Ackley, K., Adams, C. et al. 2017, ApJ, 848, L12Alexander, K., Wieringa, M.H., Berger, E., Saxton, R.D., &Komossa, S. 2017, ApJ, 837, 153Alexander, K., van Velzen, S., Horesh, A., & Zauderer,B.A. 2020, arXiv:2006.01159 Bach, U. 2004, PhD thesis, RheinischenFriedrich-Wilhelms-Universit¨at BonnAuchettl, K., Guillochon, J., & Ramirez-Ruiz, E. 2017,ApJ, 838, 149Bell, M.E., Murphy, T., Hancock, P.J., Callingham, J.R.,Johnston, S., Kaplan, D.L. et al. 2019, MNRAS, 482,2484