Fast infrared variability from the black-hole candidate MAXI J1535-571 and tight constraints on the modelling
F. M. Vincentelli, P. Casella, D. Russell, M.C. Baglio, A. Veledina, T. Maccarone, J. Malzac, R. Fender, K. O'Brien, P. Uttley
aa r X i v : . [ a s t r o - ph . H E ] F e b MNRAS , 1–11 (2020) Preprint 16 February 2021 Compiled using MNRAS L A TEX style file v3.0
Fast infrared variability from the black-hole candidateMAXI J1535 −
571 and tight constraints on the modelling
F. M. Vincentelli ★ , P. Casella , D. M. Russell , M. C. Baglio , A. Veledina , , ,T. Maccarone ,J. Malzac , R. Fender , K. O’Brien , P. Uttley Department of Physics and Astronomy, University of Southampton, SO17 1BJ, UK INAF, Osservatorio Astronomico di Roma Via Frascati 33, I-00078 Monteporzio Catone, Italy Center for Astro, Particle and Planetary Physics, New York University Abu Dhabi, PO Box 129188, Abu Dhabi, UAE Department of Physics and Astronomy, FI-20014 University of Turku, Finland Nordita, KTH Royal Institute of Technology and Stockholm University, Roslagstullsbacken 23, SE-10691 Stockholm, Sweden Space Research Institute of the Russian Academy of Sciences, Profsoyuznaya Str. 84/32, Moscow, 117997, Russia Texas Tech University, Physics & Astronomy Department, Box 41051, Lubbock, TX 79409-1051 IRAP, Universite’ de Toulouse, CNRS, UPS, CNES, Toulouse, France Department of Physics, Astrophysics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, UK Department of Physics, Durham University, South Road, Durham, DH1 3LE, UK Astronomical Institute, Anton Pannekoek, University of Amsterdam, Science Park 904, NL-1098 XH Amsterdam, Netherlands
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We present the results regarding the analysis of the fast X-ray/infrared (IR) variability of theblack-hole transient MAXI J1535 − XMM-Newton (X-rays: 0.7 −
10 keV), VLT/HAWK-I( 𝐾 s band, 2.2 𝜇 m) and VLT/VISIR ( 𝑀 and 𝑃𝐴𝐻 𝜇 m respectively).The cross-correlation function between the X-ray and near-IR light curves shows a strongasymmetric anti-correlation dip at positive lags. We detect a near-IR QPO (2.5 𝜎 ) at 2 . ± . 𝑓 = . ± . Key words: keyword1 – keyword2 – keyword3
Black hole low mass X-ray binaries (BH LMXB) have been his-torically studied mainly in X-rays, where, depending on the ac-cretion regime, great part of the dissipated gravitational energyis radiated away by either a geometrically-thin, optically-thick ac-cretion disc (Shakura & Sunyaev 1973) or a geometrically-thick,optically-thin inflow (Esin et al. 1997; Poutanen et al. 1997). How-ever, multiwavelength studies have shown that bright (non-thermal)emission is also present at lower (from optical-UV down to radio)frequencies (see e.g.: Fender et al. 2001; Corbel & Fender 2002;Hynes et al. 2003; Gandhi et al. 2011). This low-energy emissionis usually interpreted in terms of synchrotron radiation from ei-ther a hot magnetised geometrically-thick, optically-thin inflow, or ★ E-mail: [email protected] a compact collimated jet (Veledina et al. 2013a; Malzac et al. 2004;Malzac 2014).In the last 20 years the study of sub-second optical-infrared(O-IR) variable emission from BH LMXBs has improved signif-icantly our understanding of these systems. The first sub-secondobservations in the optical band revealed a complex phenomenol-ogy (Kanbach et al. 2001; Malzac et al. 2003, 2004; Hynes et al.2003). In particular, the first Cross-Correlation Function (CCF) ofXTE J118+40 showed an intriguing anti-correlation at negative lag(corresponding to X-ray delays), also known as the “precognitiondip”. This phenomenon was then confirmed by optical/X-ray obser-vations of Swift J1753.5 − © F. M. Vincentelli et al. (Soleri et al. 2010; Tomsick et al. 2015), the described behaviourwas associated to the hot inflow. In particular, a detailed modellingof this source showed that the observed CCF could be reproducedby assuming that the optical emission originates from both repro-cessed and synchrotron radiation coming from a hot, magnetisedinflow, while the X-rays are generated from the Comptonization ofthe synchrotron radiation (Veledina et al. 2011).Further fast O-IR photometry observations led to the discov-ery of a 0.1-second O-IR lag with respect to the X-ray variabil-ity (Gandhi et al. 2008; Casella et al. 2010). This behaviour wasattributed to the jet and was well reproduced by the so called“internal shock model" (Kobayashi et al. 1997; Spada et al. 2001;Jamil et al. 2010) when linking the shells’ velocities to the varia-tions observed in X-rays (Malzac 2013, 2014). This demonstratedthat fluctuations from the inflow can also be transferred into theoutflow and opened the possibility to put new constrains on the jetparameters (Casella et al. 2010; Kalamkar et al. 2016; Gandhi et al.2017; Vincentelli et al. 2019). Further multi-wavelength observa-tions permitted to deeply characterize the O-IR variability and tostudy the physical processes that take place inside these outflow.For example, there is now growing evidence that the O-IR jet emit-ting region is extended and probably stratified (Vincentelli et al.2018; Vincentelli & Casella 2019; Paice et al. 2019); moreover,Malzac et al. (2018) recently showed that Doppler boosting mod-ulation can also lead to an anti-correlation with X-rays on longtimescales.Another important feature which gave new insight to thegeometry of these systems are the quasi-periodic oscillations(QPO). These oscillations are very common in X-rays and havebeen attributed to Lense-Thirring precession of the hot inflow(Stella & Vietri 1998; Ingram & Done 2012; Motta et al. 2015),although this interpretation is still somewhat matter of debate(Ingram & Motta 2020; Marcel et al. 2020; Marcel & Neilsen 2020;Ma et al. 2020). At lower energies, similar and simultaneous QPOsin the optical band have been explained in terms of synchrotron ra-diation from a precessing magnetized inflow (Veledina et al. 2013b,2015). In GX 339 −
4, the first QPO observed in the IR band(Kalamkar et al. 2016) was found in harmonic relation with an X-ray QPO, similar to its optical counterpart found some years earlier(Motch et al. 1983). Both precessing hot inflow and the inflow plusjet can qualitatively account for this behaviour, but currently thequantitative agreement has been proved only for the latter alterna-tive, using the internal shocks model (Malzac et al. 2018).MAXI J1535 −
571 is an X-ray transient discovered by the Mon-itor of All-sky X-ray Image Gas Slit Camera on the InternationalSpace Station (Matsuoka et al. 2009) and by the Burst Alert Tele-scope on board of the
Neil Gehrels Swift Observatory in September2017 (Negoro et al. 2017; Kennea et al. 2017). From the first obser-vations it was noticed that the source displayed strong X-ray vari-ability, including low-frequency QPOs (Mereminskiy & Grebenev2017; Gendreau et al. 2017), and very bright radio emission with aflat radio spectrum (Russell et al. 2017). Strong emission in the IRwas also observed (Dinçer 2017). This led to identify the source asa BH LMXB. During the outburst, despite being heavily absorbed(neutral column density 𝑁 H ≈ cm − ), the source reachedextremely high flux levels, up to 5 Crab in the 2 −
20 keV band(Shidatsu et al. 2017).An intensive multiwavelength campaign was rapidly coordi-nated to track the evolution of the source (Sivakoff 2017). Due to itsposition in the sky, only few observations could have been made withthe Very Large Telescopes (VLTs) before the source became too lowon the horizon. In this paper we present the results from two strictly simultaneous near-IR and X-ray observations, aimed at studying thefast multi-wavelength variability of this source. We complement thestudy using also, for the first time, strictly simultaneous mid-IR ob-servations. We find that the properties of fast near-IR variability puttight constraints on the current jet and accretion flow models.
Strictly simultaneous multi-wavelength observations were takenduring the night of the 14th and the 15th of September 2017 (MJD58010–58011), while the source was in its hard-intermediate state(Baglio et al. 2018; Bhargava et al. 2019; Russell et al. 2019). Near-and mid-IR data were collected with HAWK-I and VLT Imagerand Spectrometer for the mid-infrared (VISIR) respectively, bothmounted at the ESO VLTs; X-rays data were collected with theEPIC-pn camera on board the ESA satellite
XMM-Newton (see Fig.1).The epochs of the observations are reported in Table 1.
XMM-Newton
X-ray data were collected with the EPIC-pn camera. Observationswere taken in
Burst mode (OBS ID: 0795712001 and 0795712101).The data reduction was carried out using the XMM Science AnalysisSystem (SAS v15). In particular, the source counts were extractedin the range RAWX: 28 −
48 (corresponding to an angular size of ≈
86 arcsec along RAWX). Due to the high absorption of the source,X-ray events were extracted only between 2 and 10 keV. The meancount rate of the two observations was found to be very similar:234 ± − for the first night, and 226 ± − for the second.Data were then barycentered with the barycen software and binnedin a light curve of 5.7 ms. Near-IR sub-second data were collected with HAWK-I (Pirard et al.2004) mounted at the VLT UT-4/Yepun under Program: 099.D-0068(A). 1-hour long observations were taken in the 𝐾 s band (ef-fective wavelength 2.2 𝜇 m) with the time resolution of 0.125 s.Images were stacked in “data-cubes” of 250 frames, separated by agap of ≈ − ◦ in order to have the target on the lower-leftquadrant (Q1) together with the reference star with a 𝐾 s magni-tude of 11.55 ± MNRAS , 1–11 (2020)
AXI J1535 −
571 multi- 𝜆 variability F l u x ( m Jy ) µ m4.85 µ m 150 200 250 3001000 1500 2000 2500 3000 3500 4000 4500 5000 T start = 00:15:30 UT2550 15 30 45 (s)(mJy) C oun t r a t e ( c t s - ) Time from 23:00:00 UT 14/09/2017 (s)2-10 keV 0 25 50 75 100 125 F l u x ( m Jy ) µ m11.88 µ m 150 200 250 3001000 1500 2000 2500 3000 3500 4000 4500 5000 T start = 23:59:30 UT2550 15 30 (s)(mJy) C oun t r a t e ( c t s - ) Time from 23:00:00 UT 15/09/2017 (s)2-10 keV Figure 1.
Lightcurves during X-ray / near- and mid- IR strictly simultaneous windows for the night of the 14th (top panel) and 15th (bottom panels) ofSeptember 2017 (N.B. X-ray observation for the 15th started around midnight, i.e. at ≈ observed fluxes (not de-reddened) as a function of time. Blue curve shows the 2 −
10 keV count rate. While for the near- and mid- IR lightcurves we used the time resolution of the original data, the X-ray light curve was rebinned with resolution 1s.MNRAS , 1–11 (2020)
F. M. Vincentelli et al.
Table 1.
Summary of the multi-wavelength campaign on the BH LMXB MAXI J1535 −
571 on the 14th and 15th of September 2017.Night of the Start Time Telescope / Instrument Energy Band Start Time (UT) End Time (UT)
XMM-Newton /Epic-PN 0.5 −
10 keV 16:47:04 01:08:4414/09 UT-4/HAWK-I 𝐾 s (2.2 𝜇 m) 23:26:21 00:21:08UT-3/VISIR 𝑀 (4.85 𝜇 m) 23:19:22 23:46:20 XMM-Newton /Epic-PN 0.5 −
10 keV 23:59:09 04:20:0215/09 UT-4/HAWK-I 𝐾 s (2.2 𝜇 m) 23:34:15 00:19:25UT-3/VISIR 𝑃 𝐴𝐻 𝜇 m) 23:29:54 23:53:26 𝐾 s magnitude of 11.52 ± ≈
16 mJy)for both nights. This in good agreement with the near-IR data inBaglio et al. (2018) taken on the same nights by REM. The time ofeach frame was then put in the Dynamical Barycentric Time system.
Mid-IR observations were obtained with VISIR (Lagage et al. 2004)mounted at the VLT-UT3 Melipal under Program 099.D-0884(A).The instrument was set in small-field imaging mode, with a pixelscale of 45 mas pixel − , and the 𝑀 and 𝑃 𝐴𝐻 . − . 𝜇 mwavelength range; see Table 1). Every observation consisted of1000s of time on source, characterized by 44 nodding cycles. Intotal, considering chopping and nodding between sky and source,the total exposure time was 1800–1900 s per observation.Reduction of data was performed using the VISIR pipeline,available in 𝑔𝑎𝑠𝑔𝑎𝑛𝑜 . Sensitivities were estimated thanks to theobservation of two standard stars (HD161096 and HD163376) onthe same night and telescope configuration. After recombination ofthe chop/nod cycle raw images, aperture photometry was performedby means of an aperture that was large enough to avoid that possi-ble seeing variations could affect the portion of the flux falling inthe aperture. We note that the target of the observations was brightenough to allow a detection in individual observations. The VISIRpipeline requires that observations are combined in groups of mul-tiples of two. In order to achieve a flux from each observation forthis work, we therefore had to combine each image with the firstobservation on each night, then take into account the flux of this firstobservation to obtain the light curve. Using this method we wereable to sample the data twice as fast with respect to Baglio et al.(2018). In particular, each time bin has 27.8 s exposure, with avarying gap of ≈
10 seconds between the bins due to the chop/nodcycle and read-out. Also in this case, the resulting lightcurve wasthen put in Dynamical Barycentric Time system.No clear variability of the background was detected during theentire VISIR observations (see also Baglio et al. 2018 for details).A variation of ( − ) % in the ADU/flux conversion factor over dif-ferent dates has instead been observed. This variability can howeverbe caused by the different weather and sky conditions for differentdates, as well as different air masses. Therefore, any possible mid-infrared variability of MAXI J1535 −
571 can be safely consideredas intrinsic to the source. The properties of the X-ray and near-IR variability were studiedthrough Fourier power spectral analysis. The X-ray and near-IRlight curves were divided in 16384-bin and 512-bin long segments,respectively. We computed the power spectral density (PSD) ofeach segment in squared root-mean-square (rms) normalization (seee.g. Belloni & Hasinger 1990; Miyamoto et al. 1991; Vaughan et al.2003) and then applied a logarithmic binning factor of 1.05 to theresulting average. The gaps between the cubes in the near-IR datawere filled with simulated data, following the procedure described inKalamkar et al. (2016). For the near-IR case, in order to reach lowerfrequencies without affecting the statistics of the higher frequencies,we also computed a PSD with the light curve rebinned to 10 s (muchlonger than the ≈ ≈ ≈ ≈ Astrosat
X-ray timing observations (Bhargava et al.2019; Sreehari et al. 2019). Results from the fit are reported inTable 2.The near-IR PSDs show clear differences with respect to theX-rays: the low-frequency variability appears to dominate the IRPSDs, with a clear break at around 1 Hz, while the X-ray PSDs aredominated by variability in the 0.1 −
10 Hz frequency range. Simi-larly to the X-rays, we modelled the near-IR PSDs with a numberof Lorentzian components. The broadband noise, which in this caseextends up to the Nyquist frequency, shows an excess at the highestfrequencies, suggesting the presence of aliasing. Therefore, the fitwas performed excluding frequencies above 3 Hz. During the firstnight the broadband noise was well described with 3 components,centred at 0 Hz (fixed), ≈ ≈ ≈ 𝜎 level) required by the fit. The centroid frequency of thisadditional component is consistent with that of the QPO detectedin the simultaneous X-ray light curve. In order to confirm this hy- MNRAS , 1–11 (2020)
AXI J1535 −
571 multi- 𝜆 variability f x PS D ( f r a c . r m s ) Frequency (Hz) 0.001 0.01 HAWK-I (K s band) f x PS D ( f r a c . r m s )
14 September 2017 0.001 0.01 0.01 0.1 1 10XMM-Newton (2 - 10 keV) f x PS D ( f r a c . r m s ) Frequency (Hz) 0.001 0.01 HAWK-I (K s band) f x PS D ( f r a c . r m s )
15 September 2017
Figure 2.
PSDs computed from the HAWK-I and
XMM-Newton light curves for the night of 2017 September 14th and 15th. The models fitted to the PSDsand their individual components are shown with black and dashed lines, respectively. Data above 3 Hz are not used for fitting purposes due to clear aliasingcontribution. Empty purple points represent the near-IR PSD computed with a 10 s light curve and 16 bins per segment.
Table 2.
Parameters of the fit to the PSD with multiple Lorentzian components, 𝐿 ( 𝑓 ) = 𝐴 /[( 𝑓 − 𝑓 ) + ( Δ / ) ] . We additionally define Q= 𝑓 / Δ . Thefractional rms of each Lorentzian was obtained from the squared of the integration over the whole frequency range .Date Band Comp. 𝑓 Δ A rms Q 𝜒 / d.o.f.(Hz) (Hz) (10 − Hz)14/09 X-rays 1 0.12 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
14 5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
24 3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± pothesis we fixed the frequency and width of this component to theone measured in the X-rays, leaving only the normalization as afree parameter. The amplitude of the QPO in this case was foundsignificant at a 3 𝜎 level, strengthening the evidence for a QPO inthe infrared band. The correlated variability between the X-ray and near-IR lightcurves was investigated using both time- and frequency-domaintechniques. In order not to distort the results we used only theoriginal light curves (i.e. without filling the near-IR gaps) and ex-cluding cubes affected by frame-losses. We computed the CCF(Edelson & Krolik 1988) for all the simultaneous observations us-ing the procedure described in Gandhi et al. (2010). The computed
MNRAS000
MNRAS000 , 1–11 (2020)
F. M. Vincentelli et al. -0.1-0.08-0.06-0.04-0.02 0 0.02 0.04 0.06-10 -5 0 5 10 CC F IR Lag (s) 14 Sep15 Sep
Figure 3.
Near-IR/X-ray CCF computed using the
XMM-Newton /Epic-PN(2 −
10 keV) and HAWK-I 𝐾 s light curves for the two nights of observations(September 14 and 15, red and blue, respectively). The 1 𝜎 confidence levelis shown in grey. An anti-correlation dip at positive lags is clearly detectedin both nights. C oun t s Figure 4.
Histogram of the number of consecutive data-cubes with no framelosses. For both nights, the majority of the segments of the cleaned lightcurve consisted of a single data-cube. This means that the distance betweenthe segments was (almost always) too large to be filled with standard methods(see also: Kalamkar et al. 2016).
CCFs in the two nights present a strong asymmetric dip at posi-tive lags (positive lags correspond to near-IR lagging behind theX-rays), with a . ∼ 𝑁 = couples of uncorrelated light curveswith the same power-spectral properties of our dataset. The resultingdistribution of of the 𝑁 uncorrelated CCFs has a standard deviationof 0.016 (grey-shaded area in Fig. 3). The anti-correlation dip ap-pears statistically significant, while the peaks observed at lags largerthan ± - π - π /2 0 π /2 π P ha s e Lag (r ad . ) Frequency (Hz) 0 0.1 0.2 0.3 0.4 0.5 C ohe r en c e Figure 5.
Intrinsic coherence (top panel) and phase lags (bottom panel)computed between near-IR and X-ray light curves, combining data fromboth nights. 64-bin segments were used. Vertical dashed line shows theposition of the QPO. card a significant fraction of the data-cubes in the cross-correlationanalysis. The remaining “cleaned” light curve presented many in-terrupted segments, with gaps much longer than those seen betweentwo consecutive data-cubes (see Fig. 4). In order to compute thecross spectral densities, we used 64-bin long segments, which al-lowed us to explore frequencies down to 0.125 Hz. In order to probelonger timescales, one would have to either deal with much lowerstatistics (because of the lower number of sufficiently long intervals)or fill the gaps between intervals. In the latter case, even though theresulting statistics can be formally high, the resulting cross spectrawould appear significantly distorted, as confirmed by simulations.However, given the width of the dip in the CCF, frequencies in the ≈ . − .
25 Hz range are still expected to contribute to the observedanti-correlation. Therefore our choice is suitable for our purpose.Due to the low statistics, and given the similar timing propertiesduring the two observations, cross-spectral analysis was performedcombining the data from both nights together. We also checked thedata for the individual nights but no significant variation from thedescribed results was found. The results are shown in Fig. 5.The intrinsic coherence of the two datasets is very low for al-most all the probed frequencies. Nevertheless the phase lags, thoughscattered, seem to show a clear trend. At low frequencies there isevidence for a (nearly) constant phase lag at ∼ − 𝜋 /
2, while at fre-quencies higher than the QPO the phase lags are unconstrained,thus oscillate between − 𝜋 and 𝜋 . A somewhat higher coherence of0 . ± . 𝜎 error) is measured over the frequency range wherethe QPO is detected in the X-ray light curve. In this range we also MNRAS , 1–11 (2020)
AXI J1535 −
571 multi- 𝜆 variability find the phase lag consistent with zero ( ≈ 𝜋 /6 rad. 3 𝜎 upper limit).We tested the significance of this variation by integrating the phaselags and coherence within the QPO frequencies (2.05-2.3 Hz) andin an equally large bin, centered just before the QPO (1.7-2.05 Hz).The phase lags pass from -0.8 ± ± 𝜎 ); the coherence, instead, increases from 0.08 ± ± 𝜎 ). This clearly shows the presence of an addi-tional correlated component associated with the QPOs observed inboth X-rays and IR at ≈ The variability properties of the mid-IR light curve have alreadyanalysed by Baglio et al. (2018). During the first night (4.85 𝜇 m)the source showed a flux of 62.4 mJy and a fractional rms of 17.2 ± 𝜇 m) a flux of 90.2 mJy anda 14.9 ± 𝜎 confidence levels. For thefirst night, we did not find any significant correlation between X-ray and mid-IR nor between near-IR and mid-IR light curves. Avisual inspection of the light curves and the PSDs suggests that thereason behind this non-detection is the lack of variability for Fourierfrequencies lower than ≈ ≈ 𝜎 -level at 0 s and 100 s lags (mid-IR lags behind near-IR). We note, however, that during the second night, the mid-IR lightcurve shows a slowly increasing long-term trend, which could affectthe CCF. We therefore computed the CCF after removing a lineartrend to the light curve (see Fig. 6). The correlation in the secondcase is below 2 𝜎 (simulations showed that the confidence levelsdid not change significantly after the de-trending). This shows thatthe observed correlation is due to the long-term trend, and not orig-inating from the fast timescales variations. In order to visualize thisbetter, we also plotted the mid-IR vs near-IR correlation diagram(Fig. 7), averaging the HAWK-I values within the 27.8s VISIR ex-posures. The Pearson correlation coefficient between the two seriesis 𝜌 =0.56, which according to a simple t-distribution is significantat a ≈ 𝜎 level. We also quantified the relation between the near andthe mid-IR variations by fitting as a power law 𝐹 mid − IR ∝ 𝐹 𝛽 near − IR ,finding a slope of 𝛽 =0.7 ± We observe strongly variable near-IR emission from the BH LMXBMAXI J1535 −
571 in its intermediate state. The broadband vari-ability extends down to sub-second timescales, albeit with a clearbreak in the PSD at ≈ ∼ . ±
1% vs 6 ± − ≈ − -10 − Hz; Veledina et al.2017), however, as long as the oscillation is also detected in X-rays, there are no principle restrictions on the QPO frequency. If weassume that the observed near-IR frequency 𝜈 = . × Hzis the frequency at which the synchrotron spectrum cuts off(towards longer wavelengths, the so-called turnover frequency,Wardziński & Zdziarski 2001, see also eq. 3 in Veledina et al.2013a), it is possible to set a limit to the size of the partially-absorbedsynchrotron-emitting region. To comply with the required near-IRfrequency, we obtain constraints on the electron Thomson opticaldepth ( 𝜏 ) of the power-law electrons in the hot flow ∼ − − − ,and the magnetic field in the medium ∼ − G, which can beachieved at 𝑅 / 𝑅 g ∼ −
50 (for the black hole mass of 10 solarmasses). This estimate is to be compared to the outer radius of theprecessing accretion flow, as inferred from the Fourier frequencyof the QPO, 𝑓 = . 𝑎 & .
8, radial power-law dependence of the disc surfacedensity Σ ∝ 𝑅 − . (typical for advection-dominated accretion flows,Narayan & Yi 1994) and 𝐻 / 𝑅 ∼
1. Hence, in order to explain thesimultaneous near-IR and X-ray QPO within the hot flow model,we require high spin and high disc aspect ratio. The obtained zerophase lag between the X-ray and near-IR QPO can be explained bythe solid-body precession of the hot flow, if the orbital inclination is & ◦ (Veledina et al. 2013b). The high spin and orbital inclination MNRAS000
1. Hence, in order to explain thesimultaneous near-IR and X-ray QPO within the hot flow model,we require high spin and high disc aspect ratio. The obtained zerophase lag between the X-ray and near-IR QPO can be explained bythe solid-body precession of the hot flow, if the orbital inclination is & ◦ (Veledina et al. 2013b). The high spin and orbital inclination MNRAS000 , 1–11 (2020)
F. M. Vincentelli et al. -0.3-0.2-0.1 0 0.1 0.2 0.3 0.4 -400 -300 -200 -100 0 100 200 300 400 CC F Mid-IR (vs near-IR) Lag (s)14 Sep, 4.5 µ m15 Sep, 11.88 µ m15 Sep (de-trended) -0.1-0.05 0 0.05 0.1 -400 -300 -200 -100 0 100 200 300 400 CC F Mid-IR (vs X-ray) Lag (s)14 Sep,4.85 µ m Figure 6.
Mid-IR/near-IR (left) and mid-IR/X-ray (right) CCFs. 1-, 2- and 3 𝜎 confidence levels are plotted with different grey shades. Dotted-dashed line inthe left panel shows the CCF for the second night after the linear de-tremding was applied.
40 50 60 70 80 90 12 14 16 18 20 22F mid-IR α F V I S I R F l u x ( m Jy ) HAWK-I Flux (mJy)
Figure 7.
Mid-IR vs near-IR correlation diagram for the second night. Asfor Fig. 1 the fluxes are not de-reddened. A correlation of 3 𝜎 is detected. seems to be consistent with recent X-ray spectral measurements(Miller et al. 2018; Xu et al. 2018; Sreehari et al. 2019).The alternative scenario which is usually invoked for O-IRQPOs is the LT precession of a relativistic jet together with theX-ray emitting hot inflow. Recent 3D GRMHD simulations haveshown that the jet can precess along with the hot inflow (Liska et al.2018). Moreover, Malzac et al. (2018) showed that the amplitudeand the width of the IR QPO observed in GX 339 − Γ 𝛽 ≈ . − . Γ is the jetLorentz factor and 𝛽 = p − Γ − is the dimensionless jet velocity)and a precession angle of ≈ ◦ would produce the required 3% rmsobserved in the IR.The observed zero phase lag then imposes constraints on thedistance between the X-ray and near-IR emission sites. This canbe used to constrain the maximum inclination of the jet, and to seta lower limit to its speed ( 𝛽 ). By taking into account projectioneffects, the distance between the two emitting regions will be 𝑅 = 𝛽𝑐 Δ 𝑡 /( − 𝛽 cos 𝑖 ) , where cos 𝑖 is the cosine of jet inclination anglewith respect to the line of sight, and Δ 𝑡 is the near-IR/X-ray time lag. By integrating the cross-spectrum on the frequencies where thephase lag is consistent with 0, we obtained a 3 𝜎 upper limit to thetime lag of 0.04 s.Simulations from the internal shocks model have shown thatthe peak of the near-IR emission region (R) can span, dependingon the jet physical conditions, a range of values from ≈ × 𝑅 g to ≈ × 𝑅 g (Malzac 2014; Malzac et al. 2018). Given also thelimits to the jet first acceleration zone imposed by Russell et al.(2020), we set a conservative value for R of 5 × 𝑅 g , which leadsto a required jet Lorentz factor of Γ & 𝑖 . ◦ . It is interesting to notice that this is in good agreementwith radio observations of MAXI J1535 −
571 performed close toour campaign, which revealed a relativistic ejection with a reportedinclination ≤ ◦ (Russell et al. 2019). Given that X-ray spectralmeasurements seem to indicate a highly inclined disk (Miller et al.2018; Xu et al. 2018; Sreehari et al. 2019), this suggests the pres-ence of a misalignment between the orbital spin and jet axis, whichmay cause the jet precession. Rapid changes of jet orientation onthe sky, recently seen in V404 Cyg and Cir X-1 (see e.g Coriat et al.2019; Miller-Jones et al. 2019), support the possibility for such pre-cession.We notice that a combination of the two scenarios is also pos-sible. For instance, the stronger IR variability observed at lowerfrequencies may indicate the presence of a significant contributionfrom the jet (Veledina et al. 2011; Malzac et al. 2018). Therefore,the weak QPO signal measured in this dataset, could then be sim-ply be reproduced by assuming a higher 𝜏 and B (i.e. smaller radiiwith higher density and magnetic energy density), as the IR bandwould be falling in the self-absorption regime. A detailed simula-tion is however necessary in order to disentangle the contributionfrom these two components and to better understand the observeddifferences with respect to GX 339-4. Our analysis reveals a clear connection between the X-ray and the IRbands. The CCF presents a significant asymmetric anti-correlationat positive lags, without any evidence of positive correlation. Anti-correlations between X-ray and O-IR variability have already beenseen in other BH LMXBs. One of the the most notable cases is prob-ably the optical/X-ray CCF of Swift J1753.5 − MNRAS , 1–11 (2020)
AXI J1535 −
571 multi- 𝜆 variability tion at positive lags (Durant et al. 2008; Veledina et al. 2017). Suchbehavior was successfully reproduced by invoking a combinationsynchrotron self-Compton radiation of the hot flow and reprocessingfrom the outer disc (Veledina et al. 2011, 2013a). The CCF observedin MAXI J1535 − ≈ ≈ − − − 𝜋 / ∼ − . An impulse response function defined as thederivative of a function is known to give a phase lag of 𝜋 / ≈ − − 𝜋 / We present for the first time the analysis of the correlated variabilitybetween simultaneous X-rays, near and mid-IR data. No significantcorrelation was detected during the first night, while during thesecond one correlation was found at a ≈ 𝜎 level, dropping to below2 𝜎 once the long-term trend of the light curves is removed. Sucha behaviour can be explained by the fact that both X-ray and near-IR PDS show a low-frequency break at ≈ In the internal shocks model (Malzac 2014), the shell velocity is propor-tional to the X-ray fluctuations. The shocks occur because of the differencein velocity between the shells, which leads to a differential response. both the near-IR and the X-ray light curves show a small amount ofvariability, hampering the possibility to detect any correlation.A detailed discussion of the mid-IR observations ofMAXI J1535 −
571 carried out during our campaign has been pre-sented in Baglio et al. (2018). The authors reported significant mid-IR variability on timescales of minutes which, given also the strongexcess shown in the spectral energy distribution (SED), was in-terpreted in terms of synchrotron emission from a collimated jet.Contribution from the hot-inflow at these wavelengths seems alsobe unlikely when considering the preliminary parameters obtainedin the previous section. The same authors, instead, conclude thatin the near-IR band there may be potential contribution from bothjet and accretion disc (Baglio et al. 2018; Russell et al. 2020). Wetherefore discuss our results regarding the mid/near-IR connectionaccording to these two possible scenarios:
Both mid-IR and near-IR from jet : Being part of the samephysical component, a tight correlation should be present betweenthe two bands, with possible delays of the order of few seconds(Malzac et al. 2018). We notice, however, that although we do notfind correlation on timescales of tens of seconds, this does notnecessarily point against this scenario. As mentioned above, neitherX-ray nor near-IR emission display strong variability on timescalescomparable with VISIR’s time resolution: thus, a low connectionis somehow expected in this case. New detailed simulations whichinclude the low-frequency end of the X-ray fluctuation PSD mayhelp to quantitatively test this scenario with this dataset.
Mid-IR jet, near-IR hot inflow : On short timescales, the ex-pected correlation strongly depends on the physical parameters ofthe system, which can affect the responses of the two components.The shape of the CCF may have a complex shape difficult to predictwithout proper modelling. Nevertheless, if the same input mass-accretion rate fluctuations travel through the hot-inflow (where itemits in the near-IR) and then in the jet (emitting in the mid-IR),a lag of the order of seconds is expected to appear. On longertimescales, instead, the two bands should be correlated, accordingto the well known strong inflow-outflow connection of these sources(see e.g. Gallo et al. 2018, and references there in).Given that both scenarios can reproduce the observed correla-tion on longer timescales, a self consistent modelling with both thecomponents is necessary in order to actually quantify the contribu-tion of hot-inflow and jet in the near-IR. More importantly, the twoscenarios have strongly different predictions when probing (sub-)second timescales, which could not be reached with the currentdataset. Thus, new higher time resolution observations will providecrucial physical information.
In this paper we report the discovery of correlated X-ray/near-IRvariability from the BH LMXB MAXI J1535 − • Power spectral analysis reveals the presence of complex broad-band noise down to sub-second timescales in both X-ray and near-IRbands. No significant differences were found in the PSDs during thetwo consecutive nights of our campaign. A QPO at ≈ . 𝜎 , but the fact that the measured lagsat the QPO frequencies have relatively small uncertainties demon-strates the QPO is present in the near-IR light curve. This is thefirst unambiguous detection of a near-IR and X-ray QPO at the MNRAS , 1–11 (2020) F. M. Vincentelli et al. same frequency. The high QPO frequency and nearly zero phaselag (with 𝜋 /6 radians 3 𝜎 upper limit) between the bands puts tightconstraints on the models. We discuss the origin of the near-IRQPOs in terms of two specific scenarios: Lense-Thirring precessionof the hot accretion flow (Veledina et al. 2013b) or the simultaneousprecession of the jet and the inflow (Malzac et al. 2018). Both sce-narios provide constraints that seem to confirm previous indicationsof a misalignment between the disk and the jet. • The cross-correlation function shows a puzzling asymmetricanti-correlation at positive lags. Such feature corresponds to nearlyconstant phase lags ≈ − 𝜋 / − − • We performed the first correlated analysis of the variable mid-IR (4.85–11.88 𝜇 m) and near-IR (2.2 𝜇 m) emission from a BHLMXB. No significant correlation is found during the first night, be-cause of the lack of variability of the near-IR lightcuve on timescalesprobed by the VISIR observation. No significant correlation is foundon the timescales up to 400 s. The flux-flux correlation diagramfor the second night revealed instead a 3 𝜎 level correlation, likelyassociated with a clearly visible long-term trend in the mid-IR lightcurve. The trends can be associated with oscillations on timescalesmuch longer than the corresponding segment of the light curve. Inour case, from the trend visible in the lower panel of Fig. 1, we canonly say that the oscillation has a characteristic timescale ≫ JWST ; Gardner et al. 2006), will help toaddress many remaining open questions on these objects.
DATA AVAILABILITY STATEMENT
X-ray data is accessible from the XMM-Newton online archive .Data from HAWK-I and VISIR instead are publicly available on theESO online archive ACKNOWLEDGEMENTS
The authors are extremely grateful to the VLT and XMM-Newtonastronomers, who allowed us to successfully collect strictly si-multaneous data from 3 different telescopes. The authors wouldlike to thank the referee for the useful comments which signifi-cantly improved the paper.The discussion of this paper benefitedfrom the meeting ’Looking at the disc-jet coupling from differentangles’ held at the International Space Science Institute in Bern,Switzerland. FMV acknowledges support from STFC under grant http://nxsa.esac.esa.int/nxsa-web/search http://archive.eso.org/eso/eso_archive_main.html ST/R000638/1. This work was supported by the Programme Na-tional des Hautes Energies of CNRS/INSU with INP and IN2P3,co-funded by CEA and CNES. AV acknowledges the Academy ofFinland grants 309308 and 321722.
REFERENCES
Baglio M. C., et al., 2018, ApJ, 867, 114Belloni T., Hasinger G., 1990, A&A, 227, L33Belloni T., Psaltis D., van der Klis M., 2002, ApJ, 572, 392Bhargava Y., Belloni T., Bhattacharya D., Misra R., 2019, MNRAS, 488, 720Casella P., Belloni T., Stella L., 2005, ApJ, 629, 403Casella P., et al., 2010, MNRAS, 404, L21Corbel S., Fender R. P., 2002, ApJ, 573, L35Coriat M., Fender R. P., Tasse C., Smirnov O., Tzioumis A. K., BroderickJ. W., 2019, MNRAS, 484, 1672Dhillon V. S., et al., 2007, MNRAS, 378, 825Dinçer T., 2017, The Astronomer’s Telegram, 10716Durant M., Gandhi P., Shahbaz T., Fabian A. P., Miller J., Dhillon V. S.,Marsh T. R., 2008, ApJ, 682, L45Edelson R. A., Krolik J. H., 1988, ApJ, 333, 646Esin A. A., McClintock J. E., Narayan R., 1997, ApJ, 489, 865Fender R. P., Hjellming R. M., Tilanus R. P. J., Pooley G. G., Deane J. R.,Ogley R. N., Spencer R. E., 2001, MNRAS, 322, L23Fragile P. C., Blaes O. M., Anninos P., Salmonson J. D., 2007, ApJ, 668, 417Gallo E., Degenaar N., van den Eijnden J., 2018, MNRAS, 478, L132Gandhi P., et al., 2008, MNRAS, 390, L29Gandhi P., et al., 2010, MNRAS, 407, 2166Gandhi P., et al., 2011, ApJ, 740, L13Gandhi P., et al., 2017, Nature Astronomy, 1, 859Gardner J. P., et al., 2006, Space Sci. Rev., 123, 485Gendreau K., et al., 2017, The Astronomer’s Telegram, 10768Hynes R. I., et al., 2003, MNRAS, 345, 292Hynes R. I., Brien K. O., Mullally F., Ashcraft T., 2009, MNRAS, 399, 281Ingram A., 2019, MNRAS, 489, 3927Ingram A., Done C., 2012, MNRAS, 419, 2369Ingram A., Motta S., 2020, arXiv e-prints, p. arXiv:2001.08758Ingram A., Done C., Fragile P. C., 2009, MNRAS, 397, L101Jamil O., Fender R. P., Kaiser C. R., 2010, MNRAS, 401, 394Jenkins G. M., Watts D. G., 1969, Spectral analysis and its applications.Holden DayKalamkar M., Casella P., Uttley P., O’Brien K., Russell D., Maccarone T.,van der Klis M., Vincentelli F., 2016, MNRAS, 460, 3284Kanbach G., Straubmeier C., Spruit H. C., Belloni T., 2001, Nature, 414, 180Kennea J. A., Evans P. A., Beardmore A. P., Krimm H. A., Romano P.,Yamaoka K., Serino M., Negoro H., 2017, The Astronomer’s Telegram,10700Kobayashi S., Piran T., Sari R., 1997, ApJ, 490, 92Lagage P. O., et al., 2004, The Messenger, 117, 12Liska M., Hesp C., Tchekhovskoy A., Ingram A., van der Klis M., MarkoffS., 2018, MNRAS, 474, L81Lubow S. H., Ogilvie G. I., Pringle J. E., 2002, MNRAS, 337, 706Ma X., et al., 2020, Nature Astronomy,Malzac J., 2013, MNRAS, 429, L20Malzac J., 2014, MNRAS, 443, 299Malzac J., Belloni T., Spruit H. C., Kanbach G., 2003, A&A, 407, 335Malzac J., Merloni A., Fabian A. C., 2004, MNRAS, 351, 253Malzac J., et al., 2018, MNRAS, 480, 2054Marcel G., Neilsen J., 2020, arXiv e-prints, p. arXiv:2011.09032Marcel G., et al., 2020, A&A, 640, A18Matsuoka M., et al., 2009, PASJ, 61, 999Mereminskiy I. A., Grebenev S. A., 2017, The Astronomer’s Telegram,10734Miller-Jones J. C. A., et al., 2019, Nature, 569, 374Miller J. M., et al., 2018, ApJ, 860, L28Miyamoto S., Kimura K., Kitamoto S., Dotani T., Ebisawa K., 1991, ApJ,383, 784 MNRAS , 1–11 (2020)
AXI J1535 −
571 multi- 𝜆 variability Motch C., Ricketts M. J., Page C. G., Ilovaisky S. A., Chevalier C., 1983,A&A, 119, 171Motta S. E., Casella P., Henze M., Muñoz-Darias T., Sanna A., Fender R.,Belloni T., 2015, MNRAS, 447, 2059Narayan R., Yi I., 1994, ApJ, 428, L13Negoro H., et al., 2017, The Astronomer’s Telegram, 10699Paice J. A., et al., 2019, MNRAS, 490, L62Pirard J.-F., et al., 2004, in Ground-based Instrumentation for Astronomy.pp 1763–1772, doi:10.1117/12.578293Poutanen J., Krolik J. H., Ryde F., 1997, MNRAS, 292, L21Poutanen J., Veledina A., Revnivtsev M. G., 2014, MNRAS, 445, 3987Rao F., Belloni T., Stella L., Zhang S. N., Li T., 2010, ApJ, 714, 1065Ratti E. M., Belloni T. M., Motta S. E., 2012, MNRAS, 423, 694Russell T. D., Miller-Jones J. C. A., Sivakoff G. R., Tetarenko A. J., JacpotXrb Collaboration 2017, The Astronomer’s Telegram, 10711Russell T. D., et al., 2019, ApJ, 883, 198Russell T. D., et al., 2020, MNRAS,Shakura N. I., Sunyaev R. A., 1973, A&A, 24, 337Shidatsu M., et al., 2017, The Astronomer’s Telegram, 10761, 1Sivakoff G. R., 2017, The Astronomer’s Telegram, 10714Soleri P., et al., 2010, MNRAS, 406, 1471Spada M., Ghisellini G., Lazzati D., Celotti A., 2001, MNRAS, 325, 1559Sreehari H., Ravishankar B. T., Iyer N., Agrawal V. K., Katoch T. B., MandalS., Nand i A., 2019, MNRAS, 487, 928Stella L., Vietri M., 1998, ApJ, 492, L59Tomsick J. A., Fuerst F., Belloni T., 2015, The Astronomer’s Telegram, 7962Uttley P., Casella P., 2014, Space Sci. Rev., 183, 453Uttley P., Cackett E. M., Fabian A. C., Kara E., Wilkins D. R., 2014,Astronomy and Astrophysics Review, 22, 72Vaughan B. A., Nowak M. A., 1997, ApJ, 474, L43Vaughan S., Edelson R., Warwick R. S., Uttley P., 2003, MNRAS, 345, 1271Veledina A., Poutanen J., Vurm I., 2011, ApJ, 737, L17Veledina A., Poutanen J., Vurm I., 2013a, MNRAS, 430, 3196Veledina A., Poutanen J., Ingram A., 2013b, ApJ, 778, 165Veledina A., Revnivtsev M. G., Durant M., Gandhi P., Poutanen J., 2015,MNRAS, 454, 2855Veledina A., Gandhi P., Hynes R., Kajava J. J. E., Tsygankov S. S., RevnivtsevM. G., Durant M., Poutanen J., 2017, MNRAS, 470, 48Vincentelli F. M., Casella P., 2019, Astronomische Nachrichten, 340, 319Vincentelli F. M., et al., 2018, MNRAS, 477, 4524Vincentelli F. M., et al., 2019, ApJ, 887, L19Wardziński G., Zdziarski A. A., 2001, MNRAS, 325, 963Xu Y., et al., 2018, ApJ, 852, L34This paper has been typeset from a TEX/L A TEX file prepared by the author.MNRAS000