Observations of a radio-bright, X-ray obscured GRS 1915+105
S. E. Motta, J. J. E. Kajava, M. Giustini, D. R. A. Williams, M. Del Santo, R. Fender, D. A. Green, I. Heywood, L. Rhodes, A. Segreto, G. Sivakoff, P. A. Woudt
MMNRAS , 1–10 (2020) Preprint 18th February 2021 Compiled using MNRAS L A TEX style file v3.0
Observations of a radio-bright, X-ray obscured GRS 1915+105
Motta, S. E. , , Kajava, J. J. E. , , Giustini, M. , Williams, D. R. A. , , Del Santo, M. , Fender, R. , ,Green, D. A. , Heywood, I. , , , Rhodes, L. , Segreto, A. , Sivakoff, G. , Woudt, P.A. Istituto Nazionale di Astrofisica, Osservatorio Astronomico di Brera, via E. Bianchi 46, 23807 Merate (LC), Italy University of Oxford, Department of Physics, Astrophysics, Denys Wilkinson Building, Keble Road, OX1 3RH, Oxford, United Kingdom Department of Physics and Astronomy, FI-20014 University of Turku, Finland Centro de Astrobiología (CSIC-INTA), Camino Bajo del Castillo s/n, Villanueva de la Cañada, E-28692 Madrid, Spain Jodrell Bank Centre for Astrophysics, School of Physics and Astronomy, The University of Manchester, Manchester, M13 9PL, UK Istituto Nazionale di Astrofisica, IASF Palermo, Via U. La Malfa 153, 90146, Palermo, Italy Department of Astronomy, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa Astrophysics Group, Cavendish Laboratory, 19 J. J. Thomson Avenue, Cambridge CB3 0HE, UK Department of Physics and Electronics, Rhodes University, PO Box 94, Makhanda 6140, South Africa South African Radio Astronomy Observatory, Cape Town, South Africa. Department of Physics, University of Alberta, CCIS 4-181, Edmonton, AB T6G 2E1, Canada
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
The Galactic black hole transient GRS 1915+105 is famous for its markedly variable X-ray and radio behaviour, and for being thearchetypal galactic source of relativistic jets. It entered an X-ray outburst in 1992 and has been active ever since. Since 2018GRS 1915+105 has declined into an extended low-flux X-ray plateau, occasionally interrupted by multi-wavelength flares. Herewe report the radio and X-ray properties of GRS 1915+105 collected in this new phase, and compare the recent data to historicobservations. We find that while the X-ray emission remained unprecedentedly low for most of the time following the decline in2018, the radio emission shows a clear mode change half way through the extended X-ray plateau in 2019 June: from low flux( ∼ Key words: accretion, accretion discs – black hole physics – X-rays: binaries – stars: jets
In black hole (BH) X-ray binaries a stellar mass black hole accretesvia an accretion disc formed by matter stripped from a low-masscompanion star. BH X-ray binaries are typically transient systems,i.e. they alternate between long states of quiescence, characterisedby a luminosity typically of the order of 𝐿 ∼ erg s − (seeWijnands et al. 2015), and relatively short outbursts, during whichtheir luminosity can reach ∼ erg s − . During outbursts thesesystems show clear repeating patterns of behaviour across variousaccretion states, each associated with mechanical feedback in theform of winds and relativistic jets (e.g., Fender et al. 2009, Ponti et al.2012).The hard states, characterised by highly variable X-ray emissiondominated by hard photons, are associated with steady radio jets(Fender et al. 2004). In a few occasions, cold (i.e., consistent withbeing not ionised) winds, which appear to co-exist with the radiojets, have been observed in the optical band during the hard state(Muñoz-Darias et al. 2016), casting doubts on the idea accordingto which jets and winds are associated to different accretion states, and therefore cannot co-exist. In the X-ray low-variability soft states,X-ray spectra are dominated by thermal emission from a geometricallythin, optically thick accretion disc, the radio emission is quenched ,and X-ray (ionised) winds are seen (Ponti et al. 2014; Tetarenko et al.2018). In between these two states lie the intermediate states, withproperties in between the hard and the soft state, and during whichshort-lived, powerful relativistic radio ejections are observed (Fenderet al. 2009). Quasi-simultaneous X-ray and radio observations of BHX-ray binaries have been fundamental to the study of the connectionbetween the accretion and the jet production mechanism, which ledto the establishment of a disc-jet coupling paradigm (Fender et al.2004). Such a coupling gives rise in the hard state to a well-knownnon-linear relation between the X-ray and the radio luminosity, knownas the radio–X-ray correlation (e.g., Gallo et al. 2003 and Corbelet al. 2003), which also encompasses AGN when a mass scaling term Any residual radio emission in the soft state has been so far associated withejecta launched before the transition to the soft state (see e.g. Bright et al.2020). © a r X i v : . [ a s t r o - ph . H E ] F e b S. Motta et al. is considered (Merloni et al. 2003; Falcke et al. 2004; Plotkin et al.2012; Gültekin et al. 2019).GRS 1915+105 is one of the most well studied Galactic BH X-raybinaries, which firsts appeared as a bright transient in August 1992and remained very bright in X-rays and in radio until recently (Negoroet al. 2018). GRS 1915+05 was the first Galactic source observed todisplay relativistic super-luminal radio ejections Mirabel & Rodríguez(1994), and it is still considered the archetypal galactic source ofrelativistic jets. This system, located at a radio parallax distance of8.6 + . − . kpc, hosts a stellar mass black hole (12.4 + . − . M (cid:12) Reid et al.2014), believed to accrete erratically close to the Eddington limit.Several characteristic X-ray variability patterns observed in the X-raylight curve of GRS 1915+105 (Belloni et al. 2000) are believed toreflect transitions from and to three accretion states: two soft states (Aand B), and a hard state, C, all slightly different from the canonical states seen in other BH binaries (Belloni & Motta 2016). State Aand B are characterised by limited X-ray variability, and a substantialcontribution from an accretion disc with a variable temperature thatcan reach 2 keV. State C shows high X-ray variability and no disccontribution to the X-ray spectrum, and is known to be associated withsteady radio jets (Rushton et al. 2010). Such jets appear as flat-topperiods in the radio light curve, characterised by relatively high radioflux densities ( ∼
100 mJy beam − ), an optically thick radio spectrum,and a flat low-flux X-ray light curve (Pooley & Fender 1997). In theradio–X-ray plane, state C corresponds to a high-luminosity extensionof the radio–X-ray correlation (Gallo et al. 2003). Radio Plateaus aregenerally preceded and followed by flaring periods due to the launchof relativistic ejections (Rodríguez & Mirabel 1999), which have beenrepeatedly resolved as extended radio jets on a range of scales from ∼ ∼ MJD 58300), after 26 years of extreme X-ray andradio activity, GRS 1915+105 entered an unusually long period oflow-flux in the X-rays and radio (Negoro et al. 2018; Motta et al. 2019),which made some to believe that quiescence was close. Around theend of 2019 March (MJD 58600), GRS 1915+105 entered a suddenfurther X-ray dimming (Homan et al. 2019; Rodriguez et al. 2019),which reinforced the hypothesis that the 26-year long outburst ofGRS 1915+105 was nearing an end. However, only days later, on2019 May 14th (MJD 58617), renewed flaring activity at differentwavelengths appeared to invalidate the quiescence hypothesis. Afterapproximately a month of marked multi-wavelength activity (Iwakiriet al. 2019; Miller et al. 2019b; Neilsen et al. 2019; Jithesh et al.2019; Vishal et al. 2019; Svinkin et al. 2019; Koljonen et al. 2019;Balakrishnan et al. 2019; Trushkin et al. 2019; Motta et al. 2019),GRS 1915+105 entered a new X-ray low-flux state. X-ray observationswith Swift, NuStar and Chandra taken during this second low phaseshowed hard spectra characterised by heavy and occasionally partialcovering absorption with equivalent column densities 𝑁 H >3 × cm − , i.e. over an order of magnitude larger then the usual equivalentcolumn density in the direction of GRS 1915+105 (Miller et al. 2019a; Instrument Energy/Frequency Time covered (MJD)Ryle Telescope 15.5 GHz (350 MHz) 49856–53898AMI-LA 15.5 GHz (5 GHz) 54615–56925MeerKAT 1.28 GHz (0.86 GHz) 57642–58926RXTE/ASM 2–12 keV 50088–55859MAXI/GSC 2–12 keV 55054–59169
Swift /BAT 15–50 keV 53347–59169
Table 1.
A log of the data used in this work. For MeerKAT, the Ryle telescopeand AMI-LA we give in parenthesis the bandwidth used.
Koljonen & Tomsick 2020; Miller et al. 2020; Balakrishnan et al.2020). The presence of intrinsic absorption in GRS 1915+105 hasnever previously been reported, and have been observed only rarelyin other X-ray binaries. Two notable exceptions are the BH V404 Cyg(see, e.g., Życki et al. 1999 and Motta et al. 2017a), which showedclear signatures of heavy and variable intrinsic absorption during boththe outbursts monitored in the X-rays, and SS 433, which is believedto be obscured by its own inflated accretion disc (Fabrika 2004). Inboth these systems obscuration was the consequence of erratic (V404Cyg) or sustained (SS 433) super-Eddington accretion, which in bothcases is associated with extreme activity of the radio jets (Spencer1979; Miller-Jones et al. 2019).In this paper we report on the behaviour of GRS 1915+105 basedon the long-term monitoring operated by a number of X-ray All-skymonitors and radio facilities. We compare the recent (as in 2020)evolution of the systems with its past behaviour, with the aim ofhighlighting the peculiarities of the current, highly unusual state.In Section 2 we describe our data reduction and analysis, in Sec. 3we present our results, and in Sec. 4 we will discuss our findings.Finally, in Sec. 5 we will summarise our main results and outline ourconclusions.
In this section we describe the reduction and analysis of the data fromthe radio and X-ray facilities used in this work. A log of the dataconsidered is given in Tab. 1.
From 1995 May to 2006 June (MJD 49850 to 53900) the Ryletelescope routinely observed GRS 1915+105 as part of an extensivemonitoring campaign on a number of bright radio transients. In2006 the Ryle telescope was partly converted into the ArcminuteMicrokelvin Imager Large Array (AMI-LA) and observations ofGRS 1915+105 continued until 2016 January (MJD 57400), whenthe array was switched off to allow the original analog correlator tobe upgraded with a digital one. Observations resumed in 2016 June(MJD 57640) and continued until March 2020, when AMI-LA had tobe shut down due to the Covid-19 outbreak (MJD 58926).Data from the Ryle telecope have been published by, e.g., Pooley& Fender (1997), Klein-Wolt et al. (2002), and Rushton et al. (2010),and we refer the reader to those works for details on the data reduction.The AMI-LA observations were conducted at a central frequencyof 15.5 GHz with a 5 GHz bandwidth, divided into 8 channels forimaging (for the digital correlator data there were originally 4096narrow channels). We used 3C286 as the flux/bandpass calibrator, andJ1922+1530 as the interleaved phase calibrator. We reduced the datawith a custom pipeline that uses the AMI reduce_dc software, which
MNRAS000
MNRAS000 , 1–10 (2020) radio-bright, X-ray obscured GRS 1915+105 automatically flags for radio frequency interference (RFI), antennashadowing, and hardware errors, performs Fourier transforms of thelag-delay data into frequency channels (for the analog correlatordata), and then applies phase and amplitude calibrations (e.g., Perrottet al. 2013). We carried out further flagging using the CommonAstronomical Software Applications (CASA) package (McMullinet al. 2007), which was also used for the interactive cleaning. Forimaging we use natural weighting with a clean gain of 0.1. To measurethe source flux density we use the CASA task imfit. The synthesisedbeam of the AMI-LA when observing at the declination of GRS1915+105 is 40 arcsec ×
30 arcsec. The target is unresolved in allepochs.The data have been binned differently based on the brightness ofthe target. When the target was relatively faint (flux density <
10 mJybeam − ) we report the average flux measured in each epoch, whichhave a variable total duration of 1 to 7 hr. When the target was brighter,we split each epoch into shorter segments (down to 6-minutes long),depending on the source flux. As part of the ThunderKAT large survey project (Fender et al. 2016)we observed GRS 1915+105 with the MeerKAT radio interferometer38 times. Data were obtained at a central frequency of 1.28 GHzacross a 0.86 GHz bandwidth consisting of either 4096 channels or32768 channels (in this second case data were binned for consistencyto 4096 channels before any further analysis). Observations coveredthe period between 2018 December and 2020 November (MJD 58460– 59168). We initially observed the target every several weeks. WhenAMI-LA stopped operations, we switched to a weekly monitoringof GRS 1915+105 with MeerKAT. The first MeerKAT observationconsisted of an observation with a total duration 90 min, of which 60min is on-source, 20 minutes is on the flux and band-pass calibrator,and 3 minutes on the phase calibrator. All other observations hada total on-source integration time of 15 minutes, and a flux andbandpass calibrator, and phase calibrator times of 10 and 4 minutes,respectively. We used J1939 − − , Heywood 2020). We used CASA to flag thefirst and final 100 channels from the observing band, autocorrelationsand zero amplitude visibilities. Then we further flagged the data toremove RFI in the time and frequency domain. Flux density scaling,bandpass calibration and complex gain calibration were all performedwithin CASA using standard procedures. A spectral model for thephase calibrator is derived starting from the flux and band-passcalibrator, by temporarily binning the data into 8 equal spectralwindows. We then averaged the data in time (8 s) and frequency (8channels) for imaging purposes, and we used WSClean (Offringaet al. 2012) to image the entire MeerKAT square degree field.We measured the fluxes averaging data in each epoch, so that eachMeerKAT point (blue diamonds in Fig. 2, panel (a) ) corresponds to a15 minutes of on-target time, except the first point (MJD 58460), whichcorresponds to a 60-min integration time. The target is unresolvedin all observations, and in this paper we will only consider the fluxdensities measured in the MeerKAT images using the imfit task https://github.com/IanHeywood/oxkat in CASA. More in-depth analysis of the MeerKAT maps will bepresented in a dedicated paper (Motta et al. in prep). We extracted long-term light curves for GRS 1915+105 using thepublic data available on the web pages of RXTE/ASM (ASM )and MAXI/GSC (MAXI ), and from the survey data collected withthe BAT telescope on board the Neil Gehrels Swift Observatory( Swift ). We used the MAXI on-demand tool to extract the datacovering the 2–12 keV band in order to be able to directly compare thelight curves from MAXI and the ASM. We converted the ASM andMAXI count rates into fluxes using an approximate counts-to-fluxconversion fraction, based on the mean count rates of the Crab, whichcorresponds to 75 count s − for the ASM and 3.74 count s − for theMAXI, respectively. We note that such a count rate to flux conversionis not rigorous, but it is sufficient for our purposes. We account for anybias introduced by the conversion assuming a conservative uncertaintyon the flux of 20 per cent. Owing to the larger energy interval coveredby BAT (nominally 15–150 keV), a count rate to flux conversionwould not be accurate when using the Swift/BAT transient monitorresults (Krimm et al. 2013). Hence, we processed the BAT survey dataretrieved from the HEASARC public archive by using the BatImagercode developed by Segreto et al. (2010), which is dedicated to theprocessing of coded mask instrument data. BatImager performsimage reconstruction via cross-correlation and, for each detectedsource, generates light curves and spectra. We processed BAT surveydata from MJD 53347 to MJD 59169, and extracted one spectrumper day using the official BAT spectral redistribution matrix and alogarithmic binned energy grid. Then, we fit the spectra from 15 keVto 50 keV with a simple power-law and derived the observed flux.Using the fluxes obtained as described, we calculated a X-ray colour 𝐶 = 𝐹 hard / 𝐹 soft , where 𝐹 hard is the flux coming from BAT, and 𝐹 soft is the flux coming from either the ASM or MAXI. Higher colourmeans a harder spectrum.The ASM and MAXI data overlap with those from BAT data byseveral years, and overlap to each other for about two years (i.e. fromMJD 55054 to 55859). This allows us to confirm that ASM and MAXIprovides consistent data (see 2, panel (b) ). Variability on time-scalessignificantly shorter than a day is known to occur during variousaccretion states both in the X-rays (Belloni et al. 2000) and in radio(Pooley & Fender 1997) in the flux from GRS 1915+105. However,the aim of this work is to study the long term behaviour of this system.Thus we focused on the variability occurring on time-scales longerthan a few days, rather than the details of a particular flare. Therefore,we rebinned the ASM, MAXI and BAT light curves to the same 1-daylong time bins.We also extracted energy spectra in specific time intervals (seeSec. 3.2) using the on-demand MAXI tool with the default extractionparameters, and specifying the good time intervals to be used for theextraction. The spectra were then fitted within xspec (v 12.11.0). http://xte.mit.edu/ASM_lc.html http://maxi.riken.jp/top/lc.html http://maxi.riken.jp/mxondem/ MNRAS , 1–10 (2020)
S. Motta et al.
Figure 1.
AMI-LA and Ryle data (panel a ), RXTE/ASM data (panel b ) and BAT data (panel c ), covering over 16 years. Panel b , c and d are colour-coded basedon the spectral X-ray colour displayed panel d , calculated as ratio between the BAT and the ASM fluxes. Redder points correspond to softer spectra. The greypoints in panel b are from MAXI (the same as in Fig. 2, panel b ) and are plotted to allow a comparison with the ASM data. Similarly, the grey points in panel d correspond to the colours shown in Fig. 2 for comparison. Figure 1 displays, from the top: the radio light curve taken at a centralfrequency of 15.5 GHz (Ryle Telescope and AMI-LA data, light anddark cyan points), the soft X-ray light curve (ASM data covering the2–12 keV band), the hard X-ray light curve (BAT data covering the15-50 keV band), and the X-ray colour. Figure 2 shows, from thetop: the radio light-curve from data taken at a central frequency of15.5 GHz (AMI-LA, clear blue points) and 1.28 GHz (MeerKAT,blue diamonds), the soft X-ray light curve (MAXI, covering the 2–12keV band), and again the BAT light curve and the X-ray colour. Notethat the two figures are plotted using similar time scales, and overlapby several years, but were kept separated to allow for the inspectionof both the ASM and MAXI data. In both figures, all panels exceptthe top ones are colour-coded so that redder corresponds to a softerspectrum (displayed in panel ( d ) in both figures). Wherever the ASMor MAXI data did not overlap with the BAT data, we left the pointsblack. Part of the radio and X-ray data presented in Fig. 1 have beenalready published by, e.g., Pooley & Fender (1997), Fender et al.(1999), Pooley et al. (2010), Klein-Wolt et al. (2002) and Rushtonet al. (2010).All the light curves are characterised by periods of intense flaring,interleaved by relatively short and quiet phases, both in the X-rays andin radio. In Fig. 2 we can easily identify a time when both the radioand the X-ray behaviour of GRS 1915+105 changed, i.e. around MJD 58300 (2018 July), when the source entered a first low-flux phaseapproximately 11 months-long, to which we refer to as Plateau 1 .The average flux level observed during Plateau 1 is approximately 𝐹 ≈ . × − erg cm − s − and 𝐹 ≈ . × − erg cm − s − inthe MAXI and the BAT data, respectively, and is marked with a dottedline in panels (b) and (c) in both Fig. 1 and Fig. 2 for comparison.According to the X-ray colour plotted in panel ( d ) of Fig. 2, this decayled the source from a relatively soft state (around MJD 58000, withcolour of ≈ ≈ − , consistent with the lowerend of the radio flux densities observed over the 11 years of activitycovered by the Ryle telescope.On MJD 58613 GRS 1915+105 showed a fast decay to an evenlower X-ray flux – we refer to this phase as Pre-Flare Dip . The Pre-Flare Dip can be easily discerned both in the MAXI and in the BATdata, as shown in Fig. 3. Shortly afterwards, GRS 1915+105 entereda multi-band flaring period – which we refer to as the
Flaring Phase –that in the X-rays lasted approximately 1 month. The Flaring Phase,instead, appears as a few isolated points in both the MAXI and BATlight curves around MJD 58600. The inspection of the orbit-by-orbitBAT light curve (not displayed ) shows that the flares in this phase The orbit-by-orbit BAT light curve is available at http://swift.gsfc.nasa.gov/results/transients/GRS1915p105/
MNRAS000
MNRAS000 , 1–10 (2020) radio-bright, X-ray obscured GRS 1915+105 Figure 2.
AMI-LA and MeerKAT data (panel a ), MAXI data (panel b ), and BAT data (panel c ), covering approximately 14 years. The colour-coding is the sameas in Fig. 1, with the difference that the spectral X-ray colour is calculated as the ratio between the BAT and the MAXI fluxes. The grey points in panel b are fromthe ASM (the same as in Fig. 1, panel b ) and are plotted to allow a comparison with the MAXI data. Similarly, the grey points in panel d correspond to the coloursshown in Fig. 1 for comparison. have a variable duration of a few hours up to a few days. Given thecoarse 1 day binning that we employed to compare the MAXI andBAT data, the X-ray colour displayed in Fig. 2 is not sensitive tothe fast spectral changes occurring during this high-variability phase.Thus, the colour does not provide any specific information on thespectral properties of the flares, apart from the fact that they appearto be predominantly hard, hence more pronounced in the BAT curve.GRS 1915+105 subsequently entered a new X-ray plateau ( Plateau2 ) around MJD 58700, which was occasionally interrupted by shortflares lasting approximately 1 day or less (see also Neilsen et al.2020), again visible as isolated points in the BAT and MAXI lightcurves. This second plateau lasted over 13 months. Interestingly, theMeerKAT and AMI data show that the radio flaring did not ceasewith the X-ray and multi-band flaring, but continued for severalmonths, until at least MJD 59100. In contrast, the X-ray flux levelcontinued to slowly decline from 𝐹 ≈ . × − erg cm − s − to 𝐹 ≈ . × − erg cm − s − in the MAXI data (red dashed andred solid line in panel (b) in both Fig. 1 and Fig. 2), and from 𝐹 ≈ . × − erg cm − s − to 𝐹 ≈ . × − erg cm − s − inthe BAT data (red dashed and red solid line in panel (c) in the samefigures). The signal-to-noise ratio of both the data from MAXI andBAT was very limited in this phase, due to the low count rates fromthe source, but the X-ray colours measured in this phase suggest amarkedly hard spectral shape. The radio flaring sampled by AMI-LA in this second plateau does not qualitatively differ from that observed previously over almost three decades . The radio emissionis characterised by flares of variable amplitude and duration, spanninga flux range between 3 and 300 mJy beam − in the AMI-LA data,and up to 900 mJy beam − in the MeerKAT data, and variability ontime scales from minutes to several hours.More recently, around ∼ MJD 59050, the MAXI light curve and theevolution of the colour in Fig. 2 and Fig. 3 shows that GRS 1915+105returned to a flux comparable to that of Plateau 1, but with a muchsofter spectrum characterised by an X-ray colour of approximately0.05. This indicates that GRS 19105+105 has likely transitioned to asignificantly softer state, which, coherently with what was observedin the past (see, e.g., data around MJD 53600 in Fig. 1), features adiminished radio activity, with flux densities of approximately 10 mJybeam − and limited variability. This last Soft Phase lasted until ≈ MJD59140, when both the soft and hard X-ray flux dropped to valuescomparable to those observed during Plateau 2, and the radio flaringresumed, qualitatively similar to what observed prior to the softening.At no point in the past has GRS 1915+105 reached fluxes as low asthose measured during Plateau 1 and 2 (see also Negoro et al. 2018),despite the presence of several low-flux phases observed both in thesoft and in the hard X-rays, all in general associated with relatively lowradio flux and low colours, indicative of a relatively soft state (see alsoKlein-Wolt et al. 2002 and Fender & Belloni 2004). The soft plateau Noted that MeerKAT observes at a lower frequency than AMI-LA, and aoptically thin to optically thick flare would peak sooner and higher at 15.5 GHzthan at 1.28 GHz (van der Laan 1966). MNRAS , 1–10 (2020)
S. Motta et al.
Figure 3.
A zoom of Fig. 2, focusing on the most recent evolution ofGRS 1915+105. The vertical dashed line marks the time of the changein the radio behaviour occurred on ≈ MJD 58617. -2 -1 EF E ( k e V c m - s - ) Sp e ct r um A Sp e ct r um D Sp e ct r um B Sp e ct r um C . . Energy (keV)0.51.01.52.0 R a t i o Figure 4.
MAXI unfolded average spectra extracted during different phasesof the evolution of GRS 1915+105, and the ratios to the best fits. Spectraare taken around the peak of the soft flare preceding Plateau 1 (Spectrum A,red); during Plateau 1 (Spectrum B, blue), during Plateau 2 in a time intervalflare-free (Spectrum C, green), during the Soft Phase (Spectrum D, magenta).The best-fit parameters are listed in Tab. 2. occurred around MJD 54500 is, however, noteworthy. The lack ofradio observations during such a plateau unfortunately prevents usfrom drawing solid conclusions on the state GRS 1915+105 was in,but it is possible that the source entered a relatively soft state (asindicated by the steep photon index measured) similar to that sampledby the source around MJD 59000. X-ray Luminosity (erg/s)10 R ad i o Lu m i no s i t y ( e r g / s ) C state (Rushton+2010)Plateau 1Plateau 2Soft phase
Figure 5.
The radio–X-ray plane. We mark in grey the fluxes from all the BHtransients considered by Motta et al. (2017a), plus MAXI J1820+070, basedon Bright et al. (2020). The red dots mark the data from Rushton et al. (2010),who selected data corresponding to a a bright hard state, or state C. The threesymbols mark the three main phases described in Sec. 3: Plateau 1, Plateau 2,and soft.
To further investigate the properties of the emission fromGRS 1915+105 over the phases we described, we extracted fourtime-averaged MAXI energy spectra in specific phases of the evolu-tion of the source, with variable exposure times chosen to avoid timesof variable emission, and to guarantee similar signal-to-noise ratios.Figure 4 shows the spectra we obtained and their best fit: SpectrumA (in red, extracted in the time interval between from MJD 58034and 58035), extracted at the peak of a soft flare preceding Plateau 1;Spectrum B, extracted during Plateau 1 (in blue, MJD 58450–58490);Spectrum C, extracted during Plateau 2 in a time interval flare-free(in green, MJD 58830–58850); Spectrum D, extracted during the SoftPhase (in magenta, MJD 59095–59116). The the best fit parametersare reported in Tab. 2.We fitted the four spectra with phenomenological models consti-tuted by either a powerlaw, or a disc blackbody continuum, eachmodified by interstellar absorption (tbfeo in xspec), and an ad-ditional partially covering absorber (tbpcf in xspec, see Wilmset al. 2000), so that the models used in xspec have the form: tbfeo × tbpcf × (powerlaw) or tbfeo × tbpcf × (discbb) . We fixed theinterstellar absorption parameter to 𝑁 H = . × cm − (Milleret al. 2016; Zoghbi et al. 2016). In order to compare the two softspectra (A and D) and the two hard spectra (B and C) directly, we fittedspectra A and D, and B and C with the same underlying model, andattempted to reproduce the different spectral shapes by applying addi-tional absorption. Spectra A and D are well-fitted by a hot disc, withcharacteristic temperature of ≈ 𝐾 bb ≈ 𝐾 ≈ ( . ± . ) × cm − . The observed fluxes in the 2-10keV band from spectrum A and D are ∼ . × − erg cm − s − and2 × − erg cm − s − , respectively. Fitted individually, both spectrareturn best fit parameters consistent with those reported above, and MNRAS000
To further investigate the properties of the emission fromGRS 1915+105 over the phases we described, we extracted fourtime-averaged MAXI energy spectra in specific phases of the evolu-tion of the source, with variable exposure times chosen to avoid timesof variable emission, and to guarantee similar signal-to-noise ratios.Figure 4 shows the spectra we obtained and their best fit: SpectrumA (in red, extracted in the time interval between from MJD 58034and 58035), extracted at the peak of a soft flare preceding Plateau 1;Spectrum B, extracted during Plateau 1 (in blue, MJD 58450–58490);Spectrum C, extracted during Plateau 2 in a time interval flare-free(in green, MJD 58830–58850); Spectrum D, extracted during the SoftPhase (in magenta, MJD 59095–59116). The the best fit parametersare reported in Tab. 2.We fitted the four spectra with phenomenological models consti-tuted by either a powerlaw, or a disc blackbody continuum, eachmodified by interstellar absorption (tbfeo in xspec), and an ad-ditional partially covering absorber (tbpcf in xspec, see Wilmset al. 2000), so that the models used in xspec have the form: tbfeo × tbpcf × (powerlaw) or tbfeo × tbpcf × (discbb) . We fixed theinterstellar absorption parameter to 𝑁 H = . × cm − (Milleret al. 2016; Zoghbi et al. 2016). In order to compare the two softspectra (A and D) and the two hard spectra (B and C) directly, we fittedspectra A and D, and B and C with the same underlying model, andattempted to reproduce the different spectral shapes by applying addi-tional absorption. Spectra A and D are well-fitted by a hot disc, withcharacteristic temperature of ≈ 𝐾 bb ≈ 𝐾 ≈ ( . ± . ) × cm − . The observed fluxes in the 2-10keV band from spectrum A and D are ∼ . × − erg cm − s − and2 × − erg cm − s − , respectively. Fitted individually, both spectrareturn best fit parameters consistent with those reported above, and MNRAS000 , 1–10 (2020) radio-bright, X-ray obscured GRS 1915+105 neither is better described by a power law continuum. We fitted spec-trum B and C using a power law continuum, and we linked the powerlaw photon index and normalisation across the two spectra, obtaininga photon index of 𝛤 = . ± .
04. Spectrum C required additionalabsorption of ( + − ) × cm − , with a partial covering factor of0 . ± .
02. The observed fluxes in the 2–10 keV band from spectraB and C are ∼ . × − erg cm − s − and 3 × − erg cm − s − ,respectively.We note that the above analysis is based on simple phenomenolo-gical models with the aim to provide clues regarding the nature of theaccretion state(s) sampled by GRS 1915+105 after July 2018. To better compare the current behaviour of GRS 1915+105 with itspast behaviour, as well as with other BH transients, we placed iton the radio–X-ray plane. We measured the radio and X-ray fluxescorresponding to Plateau 1, Plateau 2, and the Soft Phase takingthe mean in the three phases. Figure 5 shows the radio–X-ray points(in grey) for a number of BH transients considered in Motta et al.(2017a), plus MAXI J1820+070 from Bright et al. (2020). The errorbars account for both the scatter in the values, and the uncertainty onthe counts-to-flux conversion described above. GRS 1915+105 wastraditionally considered an outlier on the radio–X-ray correlation thatthe vast majority of BH transients follow, as shown by the red dotsin the figure, which mark the position that GRS 1915+105 occupiedduring its C state phases of its 27-years long outburst (points takenfrom Rushton et al. 2010). While during Plateau 2 and during the SoftPhase GRS 1915+105 still clearly lies away from the other sourceson the plane, during Plateau 1 GRS 1015+105 falls on the correlation,incidentally approximately in the same position occupied by Cyg X–1.
We have presented the results of a comparative radio and X-ray studyof GRS 1915+105, primarily focusing on the evolution of the sourcefrom MJD 58300 (July 2018) to MJD 59200 (November 2020). Weare motivated by the fact that in 2018 the source underwent a fluxdecline in the X-rays that might have been interpreted as a transition toa canonical hard state, possibly preceding a long-expected quiescence(Truss & Done 2006). Our data show that, despite the low X-rayfluxes displayed lately, GRS 1915+105 is still very active in radio,and during the last several months has been showing marked activitythat is qualitatively very similar to that observed in the past. Thisprovides evidence for two important facts: first, that the correlationbetween radio and X-ray emission that for many years characterisedGRS 1915+105 (Fender & Belloni 2004) ceased to exist sometime inJune 2019; and second, that the X-ray behaviour alone–in the absenceof such radio–X-ray correlation–is very misleading, as it offers only apartial view of the current state of the source. GRS 1915+105 has notentered quiescence, and might not be approaching it either (see alsoNeilsen et al. 2020 and Koljonen & Hovatta 2021 for a discussionof this topic). Rather, the accretion processes that must be at workto feed the jets responsible for the observed radio behaviour are forsome reason not directly visible to the observer.The low X-ray fluxes observed during Plateau 1, i.e. after theexponential decline occurred in 2018 (see also Negoro et al. 2018)appear as highly unusual for GRS 1915+105. Furthermore, the almosttotal lack of variability in the X-ray emission, as opposed to markedradio flaring observed after June 2019 that characterises Plateau 2(referred to as the obscured phase by Miller et al. 2020) is certainly unprecedented. Based on the data reported here, and on older datapublished by other authors (e.g., Klein-Wolt et al. 2002) Plateau 2is the longest, lowest flux, and hardest X-ray plateau ever observedin GRS 1915+105, and the only one associated with marked radioflaring. Any previous plateau was a plateau both in radio and in theX-rays, and any Flaring Phase has occurred in both bands.Since the beginning of Plateau 2, absorption is a constant charac-teristic of the energy spectra of GRS 1915+105. Our spectral analysisresults are fully consistent with those reported by previous works(Koljonen & Tomsick 2020; Miller et al. 2020; Neilsen et al. 2020;Balakrishnan et al. 2020; Koljonen & Hovatta 2021): an obscuringin-homogeneous medium is required to explain the average spectrumwe extracted from Plateau 2. This agrees with the results from theChandra spectra, which were taken when GRS 1915+105 was ob-served in a deep obscuration state (Miller et al. 2020). Absorptionalso impacted the occasional flares captured by
NICER during theobscured state , which still reached a high luminosity, indicatingthat the intrinsic X-ray luminosity of GRS 1915+105 is likely not farfrom the Eddington limit (Neilsen et al. 2020). The evolution of theX-ray emission during one particular flare reported by Neilsen et al.(2020), shows that the flares are characterised by harder-when-brighterspectra, and require high density and variable in-homogeneous localabsorption, properties very reminiscent of the behaviour of V404Cyg during a vast majority of the flares observed in 2015 (Motta et al.2017b). Also reminiscent of V404 Cyg are the spectral propertiesof the obscured state around the occasional flares observed. A highequivalent column density absorber, which was almost completelycovering the emission from the central portion of the accretion flow,was responsible for the spectrally hard and low flux emission observedduring a number of plateaus during the 2015 outbursts of V404 Cyg(Motta et al. 2017a; Kajava et al. 2018).The behaviour and properties of GRS 1915+105 during the ob-scured state are consistent with those observed in all the systemsdisplaying phases of strong, variable local absorption: all tend toshow high-amplitude flares. Some noteworthy examples are V4641Sgr (Revnivtsev et al. 2002), Swift J1858–0814 (Hare et al. 2020;Muñoz-Darias et al. 2020), and even some Seyfert II Galaxies (Moranet al. 2001), but perhaps the best example remains that of V404 Cyg.In the case of V404 Cyg, the behaviour observed both during thelow-flux phases and during flares has been ascribed to the presenceof an inflated accretion disc. Such a thick disc was likely sustained byradiation pressure induced by high accretion rate episodes, and wasfragmenting out in the clumpy outflow responsible for the local andvariable Compton-thick absorption which affected the spectra. Some-thing similar might be happening in GRS 1915+105 in the obscuredstate, even though the obscured state in this case is significantly moreextended than for V404 Cyg.Owing to the better energy resolution and signal-to-noise ratioachieved with Chandra and NICER , respectively, observations ofGRS 1915+105 in the obscured state offered a deeper insight intothe properties of this state. Both Neilsen et al. (2020) and Milleret al. (2020) showed that the absorber in GRS 1915+105, besidesbeing in-homogeneous, requires a multi-temperature profile, and it islikely radially stratified. On the one hand, Miller et al. (2020) showedthat at least two types of media form this absorber: an inner bound(failed) magnetic wind, and an outer, cooler component which couldbe a thermally-driven outflow. In this scenario, the failed wind wouldbe responsible for the obscuration. On the other hand, Neilsen et al. Some of these flares were also detected by BAT, when sufficiently long tobe seen in the 1-day averaged light curve. MNRAS , 1–10 (2020)
S. Motta et al.
Table 2.
Best fit parameters for the spectra shown in Fig. 4. The interstellar hydrogen column was fixed to 𝑁 ISMH = × cm . From top to bottom in column 1:multiplicative constant 𝐾 ; local absorption 𝑁 locH ; local partial covering fraction 𝑃𝐶𝐹 ; disc-blackbody temperature 𝑇 bb ; photon index 𝛤 ; observed flux 𝐹 in the2–10 keV band. The parameters marked with ∗ ( 𝛤 in Spectra B and C, and 𝑇 bb in Spectra A and D, respectively), are linked, so that the same continuum is used tofit each pairs of spectra. The xspec models used have the form: tbfeo × tb_pcf × ( powerlaw ) or tbfeo × tb_pcf × ( discbb ) . Parameter A B (Plateau 1) C (Plateau 2) D (soft phase) 𝐾 . ± . 𝑁 locH [ × 𝑐𝑚 ] - - 220 + − . ± . 𝑃𝐶𝐹 - - 0 . ± .
02 1 (fixed) 𝑇 bb [keV] 1 . ± . ∗ - - 1 . ± . ∗ 𝛤 - 1 . ± . ∗ . ± . ∗ - 𝐹 [ × − erg cm s − ] 3.6 0.16 0.03 0.2 𝜒 /d.o.f. 134 . /
119 158 . /
159 12 . /
12 165 . / (2020), who analysed data taken at a different time, showed that theirresults are consistent with the presence of radially stratified puffed-upouter disc, which would hide the regions closer to the BH. It seemstherefore plausible that the absorber in GRS 1915+105 covers a largeportion of the accretion disc, from a few hundreds gravitational radii,to the outer disc, at radii larger than tens of thousands 𝑅 g . In thisrespect GRS 1915+105 differs from V404 Cyg. In the latter bothneutral and ionised winds were being launched from the outer disc,but a cold, optically thick and partially covering absorbing materialwas launched from within a few hundreds of gravitational radii fromthe black hole, and was clearly outflowing with a velocity of the order0.05 c (Motta et al. 2017b).Considering GRS 1915+105 in the context of the radio–X-rayplane, we see that for the first time since its discovery, during Plateau1 the system was consistently located on the correlation traced bythe majority of known BH transients. During both Plateau 2 and theSoft Phase the position of GRS 1915+105 in the radio–X-ray planeconfirms that a large fraction of the X-ray flux is lost. If, for the sakeof the argument, we assume that the correlation holds during Plateau2, we estimate that the obscuration of the inner accretion flow mightbe causing a loss of approximately a factor ∼
200 in X-ray luminosity,consistent with the obscuration scenario. However, note that the radio–X-ray correlation strictly holds only during the canonical hard state(Fender & Belloni 2004), when a compact steady jet is detected, andthe radio and X-ray emission can be directly compared. The aboveestimate is intended only as an indication of the overall behaviour ofthe source, the emission being so heavily modified by absorption, wehave no means to confirm the X-ray state of the source.Based on our results, it seems that Plateau 1 was a truly dim state,very similar to the canonical low-flux hard state typical of morewell-behaved BH transients, characterised by low X-ray and radioflux, as well as low long-term variability in both bands. Such a state iscertainly unusual for GRS 1915+105, which has never been observedbefore in a low-luminosity canonical hard state (but see Gallo et al.2003). Instead, despite the fact that Plateau 2 has been dubbed ‘theobscured state’, it is not really a state, but rather a condition directlydependent on the presence of local absorption along our the line ofsight. The accretion processes that must be feeding the markedlyvariable jet observed in radio must be happening beneath a complexlayer of material local to the source, which shields the inner part of theaccretion flow, and thus blocks a large portion of the X-ray emission.Behind this Compton-thick curtain of material, GRS 1915+105 ismost likely evolving through various states and transitions, consistentwith what it had been doing for 25 years until June 2018. Thismeans that perhaps, as already proposed by Miller et al. (2020),GRS 1915+105 did not really enter the outburst phase in 1992, whenit was discovered, but simply emerged from an obscured phase similar to the one we are witnessing at the time of the observations presentedhere.GRS 1915+105 is an important system for a number of reasons,including being in many ways a small-scale AGN. As in the caseof many other Galactic BHs, its different states may correspond toa number of AGN classes, but the specific multi-band behaviour ofGRS 1915+105 has clear counterparts in AGN (Miller et al. 2020). So,how does this new observed state of GRS 1915+105–heavily absorbedin X-rays and active in radio–compare to AGN? In AGN, differentabsorbers on different scales affect the X-ray spectrum: from dustlanes in the host galaxy, to the parsec-scale torus, to accretion-discscale clouds/winds (see Ramos Almeida & Ricci 2017, for a recentreview). Large column densities 𝑁 H ∼ − × cm − , comparableto those measured in the X-ray spectra of GRS 1915+105 in Plateau 2,are inferred from hard X-ray observations of local radio galaxies (seeUrsini et al. 2018, and references therein) and young radio sources(e.g., Mrk 668, Guainazzi et al. 2004). The radio jet activity of youngradio sources is also intermittent, possibly due to the interaction withthe X-ray absorbing dense circumnuclear medium. However, most ofthe X-ray absorbers found in these sources are likely found on parsecscales and beyond. Compton-thick absorbers on accretion disc–scalesare observed in some Changing-Look AGN (e.g., Risaliti et al. 2005),but their time scales for variability, once scaled down to stellar massblack holes, are much shorter than what is observed in GRS 1915+105.Thus, a phenomenon similar to the one observed in GRS 1915+105has not been identified in AGN, yet.Finally, while accreting super-massive BHs and stellar-mass blackholes are connected by the same fundamental physics (Merloni et al.2003; Falcke et al. 2004), which governs their inner accretion flow,their larger-scale structure differ quite appreciably. In particular,accreting stellar mass BH have a companion star, the behaviour ofwhich can potentially greatly affect the long-term behaviour of theaccretion disc. Neilsen et al. (2020) speculated that if a verticallyextended outer accretion disc is responsible for the obscuration inGRS 1915+105, its formation might have been triggered by changesin the companion star, e.g. an increase in the mass supply intothe accretion disc, which necessarily would trigger a change in theaccretion flow from the outside-in. The fact that an equivalent of theobscured state seen in GRS 1915+105 has not been seen in AGNmight support this hypothesis. In 2018, the black hole binary GRS 1915+105, after 25 years ofhigh-luminosity X-ray activity, decayed to a prolonged low-flux X-raystate. Due to this relatively sudden change in the X-ray behaviour ofGRS 1915+105, some were led to believe that its outburst, the longest
MNRAS000
MNRAS000 , 1–10 (2020) radio-bright, X-ray obscured GRS 1915+105 ever observed from a black hole X-ray binary, was approaching itsend.We analysed the simultaneous X-ray and radio data collected overalmost 3 decades with various facilities, focusing on the most recentevolution of the system. Our data show that at the beginning of thedim X-ray state GRS 1915+105 was also relatively faint in radio.Since June 2019 the system has been showing marked radio activity,characterised by the signatures of relativistic jets, and X-ray spectraaffected by high and variable in-homogeneous absorption. Our resultsshow that more recently GRS 1915+105, while still affected by heavyabsorption, transitioned to a softer state, which was accompanied bya decrease in the radio flaring that resumed when GRS 1915+105moved back to a hard(er) state.We argue that GRS 1915+105 first transitioned to a low-luminosityhard state, similar to the canonical hard state shown by many otherblack hole X-ray binaries, and then entered a prolonged obscuredphase. In this phase the highly variable radio jets we have beenobserving for months must be fed by the same sort of accretionprocesses that have been seen often in the past in GRS 1915+105,and are now happening behind a complex layer of absorbing material.We therefore conclude that GRS 1915+105 is far from being inquiescence, even though a substantial change in the accretion flow–perhaps the launch of a powerful outflow and/or the thickening ofthe outer disc–must have occurred at some point around June 2019.The behaviour of GRS 1915+105 in the obscured state appears tohave no counterpart in its super-massive relatives, the AGN, wherethe time-scales typical of similar radio-bright obscured phases (oncethey are scaled up in mass) are either much longer, or much shorterthan in GRS 1915+105. ACKNOWLEDGEMENTS
SEM acknowledges the Violette and Samuel Glasstone ResearchFellowship programme, and the UK Science and Technology FacilitiesCouncil (STFC) for financial support. SEM and DW acknowledge theOxford Centre for Astrophysical Surveys, which is funded throughgenerous support from the Hintze Family Charitable Foundation.JJEK acknowledges support from the Academy of Finland grant333112 and the Spanish MINECO grant ESP2017-86582-C4-1-R.MG is supported by the “Programa de Atracción de Talento” of theComunidad de Madrid, grant number 2018-T1/TIC-11733. MDSand AS acknowledge financial contribution from the agreement ASI-INAF n.2017-14-H.0 and from the INAF mainstream grant. PAWacknowledges financial support from the University of Cape Town andthe National Research Foundation. We also acknowledge support fromthe European Research Council under grant ERC-2012-StG-307215LODESTONE.We thank the staff of the Mullard Radio Astronomy Observatory,University of Cambridge, for their support in the maintenance andoperation of AMI. We thank the staff at the South African RadioAstronomy Observatory (SARAO) for scheduling the MeerKATobservations. The MeerKAT telescope is operated by the SouthAfrican Radio Astronomy Observatory, which is a facility of theNational Research Foundation, an agency of the Department ofScience and Innovation. This research has made use of MAXI dataprovided by RIKEN, JAXA and the MAXI team (Matsuoka et al.2009).All the authors wish to heartily thank Guy Pooley, who sadly recentlypassed away. His work was instrumental for the understanding of theradio properties of GRS 1915+105.
DATA AVAILABILITY
The un-calibrated MeerKAT visibility data presented in this paperare publicly available in the archive of the South African RadioAstronomy Observatory at https://archive.sarao.ac.za, subject to astandard proprietary period of one year. Data from the
Swift /BAT andthe RXTE/ASM are publicly available in the NASA’s HEASARCData archive. The MAXI/GSC data are made available by RIKEN,JAXA and the MAXI team. Data that are not available thoroughpublic archives, and all source code, will be shared on reasonablerequest to the corresponding author.
References
Balakrishnan M., Tetarenko B., Corrales L., Reynolds M., Miller J. M., 2019,The Astronomer’s Telegram, 12848Balakrishnan M., Miller J. M., Reynolds M. T., Kammoun E., Zoghbi A.,Tetarenko B. E., 2020, arXiv e-prints, p. arXiv:2012.15033Belloni T. M., Motta S. E., 2016, Transient Black Hole Binaries. Springer-Verlag Berlin Heidelberg, p. 61, doi:10.1007/978-3-662-52859-4_2Belloni T., Klein-Wolt M., Méndez M., van der Klis M., van Paradijs J., 2000,A&A, 355, 271Bright J. S., et al., 2020, Nature Astronomy, 4, 697Chatterjee R., et al., 2009, ApJ, 704, 1689Corbel S., Nowak M. A., Fender R. P., Tzioumis A. K., Markoff S., 2003,A&A, 400, 1007Dhawan V., Mirabel I. F., Rodríguez L. F., 2000, ApJ, 543, 373Done C., Wardziński G., Gierliński M., 2004, MNRAS, 349, 393Fabrika S., 2004, Astrophysics and Space Physics Reviews, 12, 1Falcke H., Körding E., Markoff S., 2004, A&A, 414, 895Fender R., Belloni T., 2004, ARA&A, 42, 317Fender R. P., Garrington S. T., McKay D. J., Muxlow T. W. B., Pooley G. G.,Spencer R. E., Stirling A. M., Waltman E. B., 1999, MNRAS, 304, 865Fender R. P., Belloni T. M., Gallo E., 2004, MNRAS, 355, 1105Fender R. P., Homan J., Belloni T. M., 2009, MNRAS, 396, 1370Fender R., et al., 2016, in MeerKAT Science: On the Pathway to the SKA.p. 13 ( arXiv:1711.04132 )Gallo E., Fender R. P., Pooley G. G., 2003, MNRAS, 344, 60Guainazzi M., Siemiginowska A., Rodriguez-Pascual P., Stanghellini C., 2004,A&A, 421, 461Gültekin K., King A. L., Cackett E. M., Nyland K., Miller J. M., Di Matteo T.,Markoff S., Rupen M. P., 2019, ApJ, 871, 80Hare J., et al., 2020, ApJ, 890, 57Heywood I., 2020, oxkat: Semi-automated imaging of MeerKAT observations(ascl:2009.003)Homan J., Neilsen J., Steiner J., Remillard R., Altamirano D., Gendreau K.,Arzoumanian Z., 2019, The Astronomer’s Telegram, 12742Iwakiri W., et al., 2019, The Astronomer’s Telegram, 12787Jithesh V., Maqbool B., Dewangan G. C., Misra R., 2019, The Astronomer’sTelegram, 12805Kajava J. J. E., Motta S. E., Sánchez-Fernández C., Kuulkers E., 2018, A&A,616, A129Klein-Wolt M., Fender R. P., Pooley G. G., Belloni T., Migliari S., MorganE. H., van der Klis M., 2002, MNRAS, 331, 745Koljonen K. I. I., Hovatta T., 2021, arXiv e-prints, p. arXiv:2102.00693Koljonen K. I. I., Tomsick J. A., 2020, A&A, 639, A13Koljonen K., Vera R., Lahteenmaki A., Tornikoski M., 2019, The Astronomer’sTelegram, 12839Krimm H. A., et al., 2013, ApJS, 209, 14Marscher A. P., Jorstad S. G., Gómez J.-L., Aller M. F., Teräsranta H., ListerM. L., Stirling A. M., 2002, Nature, 417, 625Matsuoka M., et al., 2009, PASJ, 61, 999McMullin J. P., Waters B., Schiebel D., Young W., Golap K., 2007, inShaw R. A., Hill F., Bell D. J., eds, Astronomical Society of the PacificConference Series Vol. 376, Astronomical Data Analysis Software andSystems XVI. p. 127 MNRAS , 1–10 (2020) S. Motta et al.
Merloni A., Heinz S., di Matteo T., 2003, MNRAS, 345, 1057Miller-Jones J. C. A., McCormick D. G., Fender R. P., Spencer R. E., MuxlowT. W. B., Pooley G. G., 2005, MNRAS, 363, 867Miller-Jones J. C. A., Rupen M. P., Fender R. P., Rushton A., Pooley G. G.,Spencer R. E., 2007, MNRAS, 375, 1087Miller-Jones J. C. A., et al., 2019, Nature, 569, 374Miller J. M., et al., 2016, ApJ, 821, L9Miller J. M., Balakrishnan M., Reynolds M., Fabian A. C., Kaastra J., KallmanT., 2019a, The Astronomer’s Telegram, 12771Miller J. M., Reynolds M. T., Zoghbi A., Fabian A. C., Kaastra J. S., KallmanT., Proga D., Raymond J., 2019b, The Astronomer’s Telegram, 12788Miller J. M., et al., 2020, ApJ, 904, 30Mirabel I. F., Rodríguez L. F., 1994, Nature, 371, 46Moran E. C., Kay L. E., Davis M., Filippenko A. V., Barth A. J., 2001, ApJ,556, L75Motta S. E., Kajava J. J. E., Sánchez-Fernández C., Giustini M., Kuulkers E.,2017a, MNRAS, 468, 981Motta S. E., et al., 2017b, MNRAS, 471, 1797Motta S., Williams D., Fender R., Titterington D., Green D., Perrott Y., 2019,The Astronomer’s Telegram, 12773Muñoz-Darias T., et al., 2016, Nature, 534, 75Muñoz-Darias T., et al., 2020, ApJ, 893, L19Negoro H., et al., 2018, The Astronomer’s Telegram, 11828Neilsen J., et al., 2019, The Astronomer’s Telegram, 12793Neilsen J., Homan J., Steiner J. F., Marcel G., Cackett E., Remillard R. A.,Gendreau K., 2020, ApJ, 902, 152Offringa A. R., van de Gronde J. J., Roerdink J. B. T. M., 2012, A&A, 539,A95Perrott Y. C., et al., 2013, MNRAS, 429, 3330Plotkin R. M., Markoff S., Kelly B. C., Körding E., Anderson S. F., 2012,MNRAS, 419, 267Ponti G., Fender R. P., Begelman M. C., Dunn R. J. H., Neilsen J., Coriat M.,2012, MNRAS, 422, L11Ponti G., Muñoz-Darias T., Fender R. P., 2014, MNRAS, 444, 1829Pooley G. G., Fender R. P., 1997, MNRAS, 292, 925Pooley D., Homan J., Heinke C., 2010, The Astronomer’s Telegram, 2974Ramos Almeida C., Ricci C., 2017, Nature Astronomy, 1, 679Reid M. J., McClintock J. E., Steiner J. F., Steeghs D., Remillard R. A.,Dhawan V., Narayan R., 2014, ApJ, 796, 2Revnivtsev M., Gilfanov M., Churazov E., Sunyaev R., 2002, A&A, 391, 1013Risaliti G., Elvis M., Fabbiano G., Baldi A., Zezas A., 2005, ApJ, 623, L93Rodríguez L. F., Mirabel I. F., 1999, ApJ, 511, 398Rodriguez J., Chenevez J., Cangemi F., Corbel S., 2019, The Astronomer’sTelegram, 12755Rushton A., et al., 2007, MNRAS, 374, L47Rushton A., Spencer R., Fender R., Pooley G., 2010, A&A, 524, A29Segreto A., Cusumano G., Ferrigno C., La Parola V., Mangano V., Mineo T.,Romano P., 2010, A&A, 510, A47Soleri P., Belloni T., Casella P., 2008, MNRAS, 383, 1089Spencer R. E., 1979, Nature, 282, 483Svinkin D., et al., 2019, The Astronomer’s Telegram, 12818Tetarenko B. E., Lasota J. P., Heinke C. O., Dubus G., Sivakoff G. R., 2018,Nature, 554, 69Trushkin S. A., Nizhelskij N. A., Tsybulev P. G., Bursov N. N., ShevchenkoA. V., 2019, The Astronomer’s Telegram, 12855Truss M., Done C., 2006, MNRAS, 368, L25Ursini F., Bassani L., Panessa F., Bazzano A., Bird A. J., Malizia A., UbertiniP., 2018, MNRAS, 474, 5684Vishal J., Banerjee Dipankar K. P., 2019, The Astronomer’s Telegram, 12806Wijnands R., Degenaar N., Armas Padilla M., Altamirano D., Cavecchi Y.,Linares M., Bahramian A., Heinke C. O., 2015, MNRAS, 454, 1371Wilms J., Allen A., McCray R., 2000, ApJ, 542, 914Zoghbi A., et al., 2016, ApJ, 833, 165Życki P. T., Done C., Smith D. A., 1999, MNRAS, 309, 561van der Laan H., 1966, Nature, 211, 1131This paper has been typeset from a TEX/L A TEX file prepared by the author.MNRAS000