Investigating the evolution of the dual AGN system ESO~509-IG066
P. Kosec, M. Brightman, D. Stern, F. Müller-Sánchez, M. Koss, K. Oh, R. J. Assef, P. Gandhi, F. A. Harrison, H. Jun, A. Masini, C. Ricci, D. J. Walton, E. Treister, J. Comerford, G. Privon
DDraft version October 11, 2017
Preprint typeset using L A TEX style emulateapj v. 12/16/11
INVESTIGATING THE EVOLUTION OF THE DUAL AGN SYSTEM ESO 509-IG066
P. Kosec , , M. Brightman , D. Stern , F. M¨uller-S´anchez , M. Koss , K. Oh , R. J. Assef , P. Gandhi , F. A.Harrison , H. Jun , , A. Masini , , C. Ricci , , , D. J. Walton , , E. Treister , J. Comerford and G. Privon Cahill Center for Astrophysics, California Institute of Technology, 1216 East California Boulevard, Pasadena, CA 91125, USA Institute of Astronomy, Madingley Road, CB3 0HA Cambridge, UK Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA Department of Astrophysical and Planetary Sciences, University of Colorado, Boulder, CO 80309, USA Institute for Astronomy, Department of Physics, ETH Zurich, Wolfgang-Pauli-Strasse 27, CH-8093 Zurich, Switzerland N´ucleo de Astronom´ıa de la Facultad de Ingenier´ıa, Universidad Diego Portales, Av. Ej´ercito Libertador 441, Santiago, Chile Department of Physics and Astronomy, University of Southampton, Highfield, Southampton SO17 1BJ, UK INAF Osservatorio Astronomico di Bologna, via Ranzani 1, I-40127 Bologna, Italy Dipartimento di Fisica e Astronomia (DIFA), Universit´a di Bologna, viale Berti Pichat 6/2, 40127 Bologna, Italy Instituto de Astrof´ısica, Facultad de F´ısica, Pontificia Universidad Cat´olica de Chile, Casilla 306, Santiago 22, Chile Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China Chinese Academy of Sciences South America Center for Astronomy and China-Chile Joint Center for Astronomy, Camino ElObservatorio 1515, Las Condes, Santiago, Chile
Draft version October 11, 2017
ABSTRACTWe analyze the evolution of the dual AGN in ESO 509-IG066, a galaxy pair located at z = 0 . XMM-Newton on this dual AGNfound evidence for two moderately obscured ( N H ∼ cm − ) X-ray luminous ( L X ∼ erg s − )nuclear sources. We present an analysis of subsequent Chandra , NuSTAR and
Swift /XRT observationsthat show one source has dropped in flux by a factor of 10 between 2004 and 2011, which could beexplained by either an increase in the absorbing column or an intrinsic fading of the central enginepossibly due to a decrease in mass accretion. Both of these scenarios are predicted by galaxy mergersimulations. The source which has dropped in flux is not detected by
NuSTAR , which argues againstabsorption, unless it is extreme. However, new Keck/LRIS optical spectroscopy reveals a previouslyunreported broad H α line which is highly unlikely to be visible under the extreme absorption scenario.We therefore conclude that the black hole in this nucleus has undergone a dramatic drop in accretionrate. From AO-assisted near-infrared integral-field spectroscopy of the other nucleus, we find evidencethat the galaxy merger is having a direct effect on the kinematics of the gas close to the nucleus ofthe galaxy, providing a direct observational link between the galaxy merger and the mass accretionrate on to the black hole. Keywords: galaxies: active — galaxies: individual (ESO 509-IG066) — galaxies: nuclei — galaxies:Seyfert — X-rays: galaxies INTRODUCTION
Interactions between galaxies are predicted to cause in-creased nuclear activity (e.g. Sanders et al. 1988; Hern-quist 1989). Massive gas flows triggered by gravitationalinteraction and resulting tidal forces can potentially fuelcentral supermassive black holes, creating luminous ac-tive galactic nuclei (AGN). This has been shown observa-tionally in large statistical samples of galaxy pairs, wherethe AGN fraction and AGN luminosity have both beenshown to increase as the separation between the galax-ies decreases, peaking at ∼
10 kpc (Alonso et al. 2007;Woods & Geller 2007; Ellison et al. 2011; Silverman et al.2011; Koss et al. 2012; Satyapal et al. 2014). In addition,it is naturally expected that such a gas build-up in thenucleus will not only fuel the growth of the super-massiveblack hole, but obscure it as well, at Compton-thick lev-els ( N H > cm − , Hopkins et al. 2005). Indeed, thishas been shown recently with a sample of interactinggalaxies at z ∼ z = 0 . D L ∼
150 Mpc). The galaxy pair was chosen forthis study as part of a
NuSTAR program to observe
Swift /BAT detected AGN (Harrison et al. 2013). Thenuclei of the two galaxies are separated by 16 (cid:48)(cid:48) on the sky,which at this redshift implies a physical projected sepa-ration of 10.9 kpc (assuming H = 67 . − Mpc − ,Ω m = 0 .
308 and Ω Λ = 0 . ◦ , − ◦ , henceforth known as the “West-ern source” and 203.6700 ◦ , − ◦ , henceforth known a r X i v : . [ a s t r o - ph . GA ] O c t Kosec et al.
Figure 1.
HST /WFPC2 F606W image of the galaxy pairESO 509-IG066 from Malkan, Gorjian & Tam (1998). The Easterngalaxy is on the left and the Western galaxy is on the right. Theimage is 50 (cid:48)(cid:48) × (cid:48)(cid:48) . as the “Eastern source”. Guainazzi et al. (2005) (G05)analyzed the galaxy pair using XMM-Newton data from2004 and reported that both galaxies host luminous nu-clear X-ray sources with luminosities of ∼ erg s − .They found that the Western source is a moderately ob-scured AGN with a column density of ∼ cm − , whilethe Eastern source is almost unobscured with the columndensity less than 10 cm − . While the Western sourceis very weak in the very soft X-ray band (0.5 − − Swift /BAT(Cusumano et al. 2010; Baumgartner et al. 2013) witha 14 −
195 keV flux of 1.4 × − erg cm − s − , althoughwith a PSF of 10 (cid:48) , Swift /BAT cannot resolve these twonuclei. Furthermore, ESO 509-IG066 was detected byMAXI/GSC (Hiroi et al. 2011) with a 4 −
10 keV flux of1.7 × − erg cm − s − .In this paper we reanalyze the XMM-Newton datafrom 2004 and add new results from the
Chandra , NuS-TAR and
Swift observations. In addition to the X-raydata, we use data from the Catalina Sky Survey (Drakeet al. 2009) and the
Wide-field Infrared Survey Explorer ( WISE , Wright et al. 2010) to compare the X-ray varia-tions with the variability in the optical and infrared (IR).Furthermore, we present new Keck/LRIS optical spectro-scopic observations of the galaxies and a Keck/OSIRISnear-IR integral field spectroscopic observation of theWestern nucleus that yield insights into the system.The structure of this paper is as follows. In Section 2we describe the observational data used and the data re-duction, Section 3 briefly summarizes our X-ray spectralfitting methods and results, followed by results from op-tical and IR analysis listed in Section 4. We present newKeck/LRIS optical spectroscopy and Keck/OSIRIS near-IR integral field spectroscopy in Section 5. We discussour results in Section 6 and conclude in Section 7. OBSERVATIONS
Table 1
X-ray observations.Telescope ObsID Date Exposure (ks)(1) (2) (3) (4)
XMM-Newton
Chandra
NuSTAR
Swift /XRT 00080115002 2014-09-03 6.2
Note . — Column (1) gives the telescope name, column (2)lists the observation ID, column (3) gives the start date of theobservation and column (4) gives the exposure time in ks.
ESO 509-IG066 has been observed by
XMM-Newton , Chandra , NuSTAR and
Swift , where the
NuSTAR and
Swift observations were simultaneous. Figure 2 presentsthe X-ray images of the system from each of the obser-vatories and Table 1 summarizes the basic observationaldata. The following sections discuss the processing ofeach of these X-ray data sets, as well as ancillary datasets at optical and IR wavelengths.
XMM-NewtonXMM-Newton (Jansen et al. 2001) EPIC-pn (Str¨uderet al. 2001) data were reduced using sas v14.0, selectingevents from a circular region of radius 60 (cid:48)(cid:48) centered on thegalaxy pair corresponding to a ∼
90% encircled energyfraction. EPIC-MOS data were not considered due totheir lower hard X-ray sensitivity. A period of high back-ground at the beginning of the observation was filteredout, leaving 7.9 ks of science data. Background spec-tra were extracted from a nearby circular region of 75 (cid:48)(cid:48) radius on the same chip as the galaxies. Initially, bothof the nuclei were extracted in a single spectrum. Dur-ing subsequent analysis, we also extracted a spectrumfor each of the two objects. For this individual analy-sis we used circular regions of radius 8 (cid:48)(cid:48) for the Easternsource and 7 (cid:48)(cid:48) for the Western source respectively. Spec-tra were grouped with a minimum of 20 counts per bin.We carried out spectral fitting in the 0.2 − ChandraChandra (Weisskopf 1999) data of the Eastern andWestern sources were extracted using the ciao (v4.7,CALDB v4.6.5) tool specextract , from circular re-gions with a radius of 5 (cid:48)(cid:48) . A larger circular region on thesame chip as the galaxies was used to extract the back-ground spectrum. The spectrum of the Western source(the brighter one) was grouped to at least 10 counts perbin and the spectrum of the Eastern source was binnedto at least 5 counts per bin. Counts at energies below 0.5and above 7.5 keV were ignored as the efficiency of theinstrument drops quickly when out of this energy range.
NuSTAR
The
NuSTAR (Harrison et al. 2013) raw data were re-duced using the nustardas v 1.5.1 software. Initiallythe events were cleaned and filtered with the nupipeline script with standard parameters, then the nuproducts procedure was used to extract spectra and the corre-sponding response and auxiliary files. A single spectrumwas extracted for the galaxy pair because the size of PSF
SO 509-IG066 Figure 2.
XMM-Newton (0.2-10 keV),
Chandra (0.5-8 keV),
Swift /XRT (0.5-10 keV) and
NuSTAR (3-79 keV) images of ESO 509-IG066from 2004-2014 showing the progressive fading of the Eastern source (East is left in these images). The red circles mark the positions ofthe sources and have 5 (cid:48)(cid:48) radii. All images have the same scale which is marked on the
XMM-Newton image. The top panels show theunsmoothed images and the bottom images show images that have been smoothed with a Gaussian kernel with radius 5 (cid:48)(cid:48) for
XMM-Newton , Swift /XRT and
NuSTAR and 2 (cid:48)(cid:48) for
Chandra . of NuSTAR ( ∼ (cid:48)(cid:48) , Madsen et al. 2015) is larger than theseparation of the galaxies. The spectra were extractedfrom circular regions centered on the peak of emissionand with specific radii to maximize the signal-to-noiseratio. The background spectra were obtained from re-gions chosen to cover as much area as possible on thesame detector as the source while avoiding the sourceitself and its point-spread function. Data from both fo-cal plane modules (FPMA and FPMB) were extractedand used in simultaneous fitting without coadding. Both NuSTAR spectra were grouped by at least 20 counts perbin using the heasarc tool grppha . We ignore chan-nels below 3 keV as the calibration at lower energies isuncertain, and channels above the 79 keV cut off thatresults from absorption in the mirror coating.
Swift
The
Swift /XRT (Gehrels et al. 2004; Burrows et al.2005) observation was taken simultaneously with the
NuSTAR observation. The data were preprocessed andthe spectrum extracted using automatic routines xrt-pipeline and xrtproducts before downloading. Be-cause of the low spatial resolution of XRT (HPD=18 (cid:48)(cid:48) at1.5 keV), only one spectrum was extracted for the AGNpair. We used default parameters (such as extractionradius) while generating the spectrum. The data werethen grouped by at least 3 counts per bin. We carry outspectral fitting in the 0.2 − Keck
We obtained observations of the ESO 509-IG066 sys-tem with the Keck telescope during 2016. Optical spec- troscopy of both nuclei was carried out using the Keck Itelescope and the dual-beam Low Resolution ImagingSpectrometer (LRIS, Oke et al. 1995). The 300 s spec-trum, obtained on UT 2016 June 9 in photometric con-ditions, used the 1 . (cid:48)(cid:48) (cid:96) mm − grism on the blue arm( λ blaze = 4000 ˚A), and the 600 (cid:96) mm − grating on the redarm ( λ blaze = 7500 ˚A). The 1 . (cid:48)(cid:48) ∼ ◦ in order to simultaneously observeboth galaxies in the system. We processed the data us-ing standard techniques within IRAF, and calibrated thespectrum using standard stars observed using the sameinstrument configuration on the same night.In addition to the optical spectroscopy, we acquirednear-IR integral field spectroscopy of the nucleus ofthe Western galaxy from the adaptive optics (AO)-assisted near-IR integral-field spectrograph (OSIRIS,Larkin et al. 2006; van Dam et al. 2006; Wizinowich et al.2006) on the Keck I Telescope taken on UT 2016 April22. The data were taken in the K -band using the Kbb filter and the 0 . (cid:48)(cid:48) . (cid:48)(cid:48) × . (cid:48)(cid:48)
4. The galaxy nucleus was usedas tip-tilt star for the Laser Guide Star AO system. Atotal of two sky and four on-source exposures of 600 seach at a position angle of 90 degrees were combined tomake the final data cube.The OSIRIS data were reduced using the OSIRIS datareduction pipeline (ODRP). This performs all the usualsteps needed to reduce near-IR spectra, but with the ad-ditional routines for reconstructing the data cube. Moredetails can be found in M¨uller-S´anchez et al. (2016). Fluxcalibration was performed using an A7V star HD 87035
Kosec et al. ( K = 7 . Other data
We used
WISE (Wright et al. 2010) and
NEOWISE-R (Mainzer et al. 2011) data to investigate the IR variabil-ity of the galaxy pair, which are spatially resolved bythe telescope. ESO 509-IG066 was observed three timesby
WISE in 2010 − − X-RAY SPECTRAL FITTING
We fit the X-ray data using xspec (Arnaud 1996)software version 12.9.0 and used the Cash (Cash 1979)statistic for fitting because of the low number of countsper bin in the
Swift data. Both AGN are modeledin the same way with an absorbed cut-off power-lawplus pexrav (Magdziarz & Zdziarski 1995) componentsimulating scattered radiation from the dusty torus plusa narrow iron line at 6.4 keV. We take into accountGalactic absorption with a wabs model component,the N H value of which was obtained from the Lei-den/Argentine/Bonn survey of Galactic HI (Kalberlaet al. 2005), and found to be 6 . × cm − . We alsoincluded cross-normalization constants between the X-ray instruments. A secondary power-law component wasadded, assuming that a fraction of the primary radiationescapes through a patchy absorber without reprocessing,or is scattered into the line of sight. In xspec , thismodel is written: constant*wabs*(constant*cutoffpl+ zwabs*cabs(cutoffpl+pexrav+zgauss)) . The zwabs*cabs component represents the reprocessing ofthe X-rays by photo-electric absorption and Comptonscattering local to the source. The constant*cutoffpl component represents the secondary power-law com-ponent. The pexrav component represents scatteredradiation from the torus. All the statistical errorscalculated by XSPEC are at 90 percent confidencelevel, unless explicitly stated otherwise.Initially, we extracted a single spectrum for both ofthe sources from the
XMM-Newton , Swift and
NuSTAR data and fitted the three data sets simultaneously. SinceG05 showed that the two sources have different spec-tral properties, specifically the level of absorption, weassumed that we could spectrally decompose the two nu-clei in the summed spectra. To do so we used the modeldescribed above multiplied by two in order to accountfor both AGN within the extraction region. At first wetied the parameters for each source across data sets un-der the assumption that they did not change betweenthe
XMM-Newton observation and the
Swift plus
NuS-TAR observations, however, the resulting fit was verypoor. A visual inspection of the spectrum revealed thatthe 2004
XMM-Newton spectrum was significantly dif-ferent from the 2014
NuSTAR and
Swift /XRT spectra,appearing harder (see Figure 3). For this reason, anotherconstant component was applied to the components ofone of the sources to account for possible variability ofone source with respect to the other. From this we ob-tained a very good fit with a C-stat of 556.21 from 589degrees of freedom (DOF). We found that the constantfor one of the sources drops to 0 for the
NuSTAR and
Swift /XRT data implying that it is negligible in the 2014
Figure 3. EF E spectra of the two sources from XMM-Newton (green),
NuSTAR (black and red) and
Swift /XRT (blue), whereboth sources have been included in the extraction regions and thespectra have been unfolded through the instrumental responses.The absorbed power-law (abs. pl), secondary power-law (pl2) andpexrav components used in the fit are marked for each epoch. data. The results of this fit are summarized in Table 2.We identify the source with the highest N H value as theWestern source and the other the Eastern source, sincethese match the parameters from G05 who carried outspatially resolved analysis on the galaxies.The first constant in the spectral model is used toaccount for differences between instruments and pos-sible variability effects. We fix it to 1 for NuSTAR
FPMA and let it float for all other instruments. Con-stants obtained by fitting the spectrum simultaneouslyare: 1.01 ± +0 . − . for XMM-Newton -pn and 1.01 +0 . − . for Swift /XRT. The cross-normalizationfor
XMM-Newton is higher, possibly due to variabilitybetween 2004 and 2014. However it is still consistentwith unity within the 90 percent uncertainties. Noticingthis, we investigated variability of the Western source.We tried freeing the photon index and column densitiesof this source in
XMM-Newton data, though the result-ing values stayed constant within the uncertainties. Theother cross-normalization constants are consistent withcalibrated values from Madsen et al. (2015).To investigate the variability of the two sources in-dividually, we analyzed the
Chandra observation from2011. For
XMM-Newton , while the 16 (cid:48)(cid:48) separationof the sources is similar to the FWHM of the tele-scope’s PSF, we extracted the spectra of each sourceusing small 7 − (cid:48)(cid:48) radius apertures following G05, bear-ing in mind that each spectrum will be contaminatedby the other’s PSF wings. First we extracted XMM-Newton and
Chandra spectra of the Western source only.We simultaneously fit the spectra using a simple spec-tral model in the form constant*wabs(constant*po +zwabs*cabs(po + zgauss)) . The power-law did not re-quire a cut-off and the pexrav component was not re-quired since these features are not significant in the softX-ray range (below 10 keV). As the
Chandra observationis only 5 ks, we were not able to constrain all the parame-
SO 509-IG066 Table 2
Results of simultaneous fitting of both sources using
NuSTAR , Swift and
XMM-Newton data.Source N H Γ E C Power-law norm. f pl2 Pexrav norm. Iron line norm.(1) (2) (3) (4) (5) (6) (7) (8)West 7.3 +1 . − . +0 . − . +150 − +0 . − . × − +25 − l × − +2 . − . × − +0 . − . × − East 0.60 +0 . − . +0 . − .
500 5.9 +1 . − . × − +1 . − l ———– ———— Note . — Column (1) gives source name, column (2) gives the N H value in units of 10 cm − , column (3) gives the photon index ofthe cut-off power-law, column (4) shows the exponential cutoff energy of the cut-off power-law in keV. Here ‘ − l ’ signifies that the lowerlimit on this parameter is unconstrained, in this case consistent with zero. Column (5) lists the normalization of the power-law in unitsof photons keV − cm − s − at 1 keV, column (6) gives the fraction of the leaked power-law model to the primary one (’-l’ indicates thatthis fraction is unconstrained at the lower end), column (7) shows the normalization of the pexrav component in units of photons keV − cm − s − at 1 keV, and column (8) lists the normalization of Gaussian component representing the iron line at 6.4 keV in units of totalphotons cm − s − in the line. The equivalent width of this line is 150 eV. ters of the spectral model. We achieved a very good fit of69.49/92 C-stat/DOF when freeing f pl2 (the fraction ofthe secondary power-law to the primary one), N H and thecross-normalization constant, and fixing all other param-eters. The results of the fit are shown in Table 3 and inFigure 4. We notice an increase in the N H of the absorberfrom 6 ± × cm − to 1.2 ± . × cm − , thoughthe uncertainties increased as well. The equivalent widthof the Fe K α line is 82 +95 − eV during the XMM-Newton observation and 68 +162 − eV during the Chandra obser-vation. The cross-normalization constant between themeasurements is consistent with unity at the 90 percentconfidence level.We extracted the spectrum of the Eastern source from
XMM-Newton and
Chandra observations as done withthe Western source. The source is detected in both ob-servations, but the flux from the 2011
Chandra observa-tion is much lower than from the 2004
XMM-Newton observation. We fit the spectrum simultaneously us-ing the same model as for the Western source. In thiscase, the zgauss component is not significant anymoreand again the pexrav component and high-energy cut-offwere not required. We obtained a good fit with 85.18/115C-stat/DOF by tying all the parameters for
Chandra to the
XMM-Newton parameters except for the cross-normalization constant, which now accounts mostly forthe drop in flux of the object. Upon freeing the other
Chandra spectral parameters, the fit becomes uncon-strained. The results are summarized in Table 3 andFigure 5. We find that the normalization of the East-ern source decreased by a factor of ∼
10 between 2004,when
XMM-Newton observed it and 2011, the date ofthe
Chandra observation.In the final part of X-ray analysis, we focus on fluxesfrom both of the sources and their change over time.We calculate the flux using cflux , which provides theobserved flux of the combined spectral models. We cal-culate the flux in two different bands: soft (0.5 − − Chandra and
XMM-Newton data we obtain fluxes of the sources separately,while for
Swift plus
NuSTAR data, the calculated fluxis the sum of both sources. However, we list it as theflux of Western source only since the Eastern source isundetected. We include the upper limit at 90 percentconfidence level for the Eastern source flux calculated
Figure 4. EF E spectra of the Western source from XMM-Newton (black) and
Chandra (red) where the spectra have been unfoldedthrough the instrumental responses. The absorbed power-law (abs.pl) and secondary power-law (pl2) components used in the fit aremarked for each epoch. using spectral modeling. We plot the results from thisanalysis in Figure 6.The system was reported by
Swift /BAT as having14 −
195 keV flux of 1.4 × − erg cm − s − (Cusumanoet al. 2010; Baumgartner et al. 2013), which is an averageover the period 2005 − −
195 keVflux of ESO 509-IG066 from the 2014
NuSTAR observa-tion by extrapolating our spectral model up to 195 keV.We find that the flux during the
NuSTAR observationis 1.5 × − erg cm − s − , which is consistent with theSwift/BAT flux reported, implying there is no evidencefor a drop in X-ray flux in this band. However, this is notsurprising since the Western source, which has remainedrelatively constant, dominates at high energies. LONG-TERM OPTICAL AND IR LIGHT CURVES
To investigate the cause of the X-ray variability ofthe Eastern source, we examined the optical and IRlightcurves. We used the Catalina Real-Time TransientSurvey (CRTS) to determine the V -band optical bright- Kosec et al.
Table 3
Simultaneous fitting results of both sources using
XMM-Newton and
Chandra data.Source N H Γ Power-law norm. f pl2 Iron line norm. Cross-normalization constant(1) (2) (3) (4) (5) (6) (7)Western source
XMM-Newton +0 . − . ± .
20 1.49 +0 . − . × − +0 . − . +8 . − . × − fixed to 1 Chandra +1 . − . tied tied 0.011 +0 . − . tied 0.88 +0 . − . Eastern source
XMM-Newton +0 . − . ± .
12 5.57 +1 . − . × − +0 . − . ———– fixed to 1 Chandra tied tied tied tied ———– 0.107 +0 . − . Note . — Column (1) lists the instrument that measured the data, column (2) gives the N H value in units of 10 cm − , column (3)gives the photon index of the power-law, column (4) lists the normalization of the power-law in units of photons keV − cm − s − at 1 keV,column (5) gives the fraction of the secondary power-law to the primary one, column (6) gives the normalization of the zgauss componentin units of total photons cm − s − in the line, and column (7) shows the cross-normalization constant between the two measurements. Figure 5. EF E spectra of the Eastern source from XMM-Newton (black) and
Chandra (red) where the spectra have been unfoldedthrough the instrumental responses. The absorbed power-law (abs.pl) and secondary power-law (pl2) components used in the fit aremarked for each epoch. ness of the Eastern source over time. The measurementswere taken from August 2005 until July 2013 so the sur-vey covers a large time period between our X-ray obser-vations.We then analyzed the mid-IR brightness of bothsources using
WISE and
NEOWISE data as describedin Section 2. The sources were observed during sevenepochs, three times in 2010 − − χ of the standard profile fit-ting technique for WISE point sources listed in the All-WISE catalog indicates that the sources are extended.For this reason, we use small-aperture photometry, ex-tracted from regions of 5.5 (cid:48)(cid:48) also listed in the AllWISEcatalog. Lastly, since the W3 and W4 passbands are notavailable for
NEOWISE , we do not include them in ouranalysis. We present this photometry in Table 4.Firstly we note that the mid-IR colors of the Easterngalaxy are relatively blue, with W1-W2 (cid:39) .
3, which in-dicates that the bands are dominated by stellar emission.For the Western galaxy W1-W2 (cid:39) .
7, which is more con-sistent with being dominated by the AGN (Stern et al. 2012). Therefore any drop in mid-IR flux from the AGNin the Eastern galaxy will probably be washed out bythe host galaxy. While the flux in W1 from the Easterngalaxy does shows a ∼
20% drop during the 2010 − WISE observations and the AllWISE catalog gives it themaximum probability that the flux was not constant withtime, the drop in flux is not sustained, as seen in the2014 − NEOWISE data which show a recovery ofthe initial flux. Finally, the expected optical to K-bandtime lag using our assumed cosmology, is 25 days, a muchshorter timescale than the cadence of the
WISE data(although the W1, W2 emission regions may be slightlylarger than the K-band emission region).We convert the V -band magnitude from CRTS and the WISE
W1 and W2 magnitudes to νF ν fluxes in orderto compare to the X-ray data. We plot this multibandlightcurve in Figure 6. While the X-ray flux from theEastern galaxy has dropped by a factor of 10 in the X-ray bands, the flux at optical and mid-IR wavelengthshas remained relatively constant over the long base line.No major variations are seen in the flux from the Westerngalaxy at X-ray, optical or mid-IR wavelengths. OPTICAL AND IR SPECTROSCOPY
In order to gain further insight into the nature of the in-teraction between the galaxies and the drop in X-ray fluxfrom the Eastern nucleus, we analyzed the optical spectraof the two nuclei. The processed Keck/LRIS spectra areshown in Figure 7. We use the penalised PiXel Fittingsoftware ( pPXF , Cappellari & Emsellem 2004) to measurestellar kinematics and the central stellar velocity disper-sion with the Indo-U.S. CaT, and MILES empirical stel-lar library (3465 − PYSPECKIT software followingBerney et al. (2015) and correct the narrow line ratios(H α /H β ) assuming an intrinsic ratio of R = 3 . β non-detection, we assume the 3 σ upperlimits for the extinction correction.The optical spectrum of the Western galaxy exhibitsstrong forbidden transition lines from [O iii] and [N ii] and BPT diagnostics confirm that the galaxy hosts aSeyfert 2 nucleus (Figure 8). The Balmer decrementcorrected [O iii] flux is 1.91 × − erg cm − s − . Wemeasure a velocity dispersion of 118 ±
37 km s − in theCaH+K and Mgb region and 124 ±
27 km s − in the Cal- SO 509-IG066 Table 4
WISE and NEOWISE photometry of both galaxies.Eastern source Western sourceW1 W2 W1 W2(1) (2) (3) (4) (5)AllWISE 1 12.67 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note . — Column (1) shows the observational epoch and columns (2)-(5) list the
WISE and
NEOWISE magnitudes (Vega) of bothgalaxies in the W1 and W2 bands.
Figure 6.
Multiband lightcurves of both Eastern (top) and West-ern (bottom) galaxies covering the period 2004 − cium triplet absorption lines. We show the fit to theCalcium triplet lines in Figure 9.In the LRIS spectrum of the Eastern galaxy a broad H α line is detected with a width of 4226 km s − characteristicof a Seyfert 1 nucleus (Figure 10), however the H β line is very weak, and so would be classified as a Seyfert 1.9(Osterbrock 1981). Using an upper limit to the flux ofthe narrow H β line, we find that the BPT diagnostics alsoconfirm the presence of a Seyfert nucleus in this source(Figure 8). The Eastern galaxy has a velocity dispersionthat is consistent with the instrumental resolution ( < − ) in the CaH+K and Mgb region. However, due tothe broad H α line, the Calcium triplet absorption linesare likely contaminated by AGN emission.Sekiguchi & Wolstencroft (1992) presented opticalspectroscopic observations of the two galaxies, taken withthe 1.9-m South African Astronomical Observatory, alsofinding the Western nucleus to be a Seyfert 2. They,however, classify the Eastern nucleus as LINER or H ii galaxy. It is unclear if the broad H α line was undetectedin their observations or not present at that time whenthey were made, between 1987 and 1990. Our new de-tection of broad H α strongly suggests that our view of theEastern nucleus is largely unobscured. Some reddeningmay be present in order to explain the non-detection of abroad H β line. We can estimate the amount of reddeningfrom the flux ratio of the H α and H β lines, known as theBalmer decrement. Given the upper limit on the flux ofthe H β line, the lower limit on the Balmer decrement is4.7. Assuming an intrinsic value of 3.1, this correspondsto a lower limit on the reddening of E ( B − V ) = 0 . N H ∼ cm − . In order to suppress theX-ray flux from the Eastern nucleus such that it is notdetected by NuSTAR , the obscuration must be at least4 orders of magnitude higher, around 10 cm − . If thiswere the case, the implied reddening in the optical meansthat the broad H α line would not be detectable. A pos-sibility remains, however, that the broad H α line resultsfrom scattered light from the nucleus which would notbe subjected to the heavy line of sight absorption thatthe X-rays may be subjected to.In addition to the optical spectroscopy of the two nu-clei, we also acquired near-IR K -band integral field spec-troscopy of the inner ∼ kpc of both nuclei. The integratedspectrum of the Western nucleus (Figure 11) reveals sev-eral transitions of molecular hydrogen, Br γ and Br δ fromatomic hydrogen, as well as transitions from ionized gas([Si vi ] and [He i ]). The molecular hydrogen transitionsindicate the presence of large amounts of molecular gaswhile the emission from highly ionized gas confirms thepresence of a powerful AGN. The integrated spectrum Kosec et al.
Figure 7.
Keck/LRIS optical spectra of the nucleus of the West-ern galaxy (top) showing strong forbidden lines typical of a Seyfert2 and the Eastern galaxy (bottom) which reveals a broad H α be-neath narrow H α +[N ii] lines. This is the first reported detection ofbroad optical lines from this galaxy and indicates that the nucleusis not strongly absorbed, arguing against heavy absorption behindthe drop in X-ray flux. of the Eastern nucleus is however featureless, with noemission lines detected.We use the strong H − µ mto map the velocity of the gas within the inner ∼ kpc ofthe Western galaxy, fitting it with a single Gaussian foreach pixel in the field of view. Figure 12 presents the flux,velocity and the velocity dispersion inferred from thesemeasurements. We find that within the inner 200 −
300 pcof the galaxy, the gas rotates in an ordered fashion, with asystematic velocity towards us to the west of the nucleus,and a systematic velocity away from us to the east of thenucleus. The velocity dispersion is also low ( <
100 kms − ). However, at ∼ − (typical error 10–20 km s − ). Ahigh dispersion is an indication of shocks and perturbedkinematics. These usually correspond to outflows, butthey are also associated with inflows, particularly frommerger processes (Medling et al. 2015; M¨uller-S´anchezet al. 2016). We have mapped the gas outflow from theWestern AGN using the high-ionisation [Si vi ] line whichshows a different morphology from the molecular gas,orientated in the north-south direction. We thereforefind it unlikely that the perturbed molecular gas in theeast is caused by an AGN outflow. Since this galaxyappears to be interacting with its neighbor to the east, weinterpret these observations as signatures of an inflow ofgas caused by a physical interaction between the galaxies.While a region of high velocity dispersion is also seen tothe north west of the nucleus, the signal to noise is lowand has significantly lower velocity dispersion (300 kms − ) than the region to the east. Furthermore, there isno evidence that the velocity is systematically differentto the ordered rotation seen in the rest of the nucleus.Therefore we conclude that while the nuclei are sepa-rated by ∼
11 kpc, the effect of the interaction is seen onthe gas within the inner ∼ DISCUSSION
One of the notable features of the X-ray observationsspanning 12 years is the drop, by a factor of 10, of the fluxof the Eastern source. There are two possible straightfor-ward explanations for this observed drop in X-ray flux.The decrease in flux could be caused by an increase inthe column density of the absorber. A cloud of gas anddust might be passing in front of the nucleus along theline of sight obscuring the source (e.g. NGC 1365, Risal-iti et al. 2009; Rivers et al. 2015b). Provided the columndensity is extremely high (above 10 cm − ), the onlyreceived X-ray radiation would be that escaping throughgaps between clouds, assuming that the covering frac-tion is not 100%, or light that has been scattered intoour line of sight. The spectrum would then resembleresults obtained from the 2011 Chandra observation, be-ing lower in flux with approximately the same spectralshape. An extreme N H would be required so that evenemission above 10 keV is suppressed by Compton scat-tering since the Eastern source is not visible in the NuS-TAR image (Figure 2). With an angular separation of16 (cid:48)(cid:48) , the nuclei are far enough apart to be distinguishablewith
NuSTAR , whose PSF has a 18 (cid:48)(cid:48)
FWHM (Harrisonet al. 2013; Madsen et al. 2015). The X-ray emission inthe
NuSTAR image peaks strongly at the position of theWestern source, with no indication of the Eastern one.Furthermore, there is no evidence for Fe-K α emission inthe Chandra spectrum of the source. In AGN, signifi-cant obscuration is usually, but not always, associatedwith Fe-K α fluorescence emission. The absence of an Fe-K line suggests absorption is not responsible for the fluxdecrease.The second possibility is that the intrinsic X-ray lumi-nosity of the AGN itself decreased by a factor of at least10 over the past 10 years, due to a decrease in coronalactivity which could have been caused by a drop in mass SO 509-IG066 Western
Eastern
Figure 8.
BPT narrow emission-line diagnostic diagrams for the Western (black dot) and Eastern (black bar) nuclei. The solid blackcurve shows the separation between star forming galaxies, which lie below the curve, and AGN, which lie above the curve, from Kewleyet al. (2001). The dashed curved line shows the same separation, but from Kauffmann et al. (2003). The solid straight line shows theseparation between Seyferts, which fall left of the line, and LINERs, which fall to the right of the line, from Kewley et al. (2006). TheWestern nucleus is in the Seyfert section in all three diagrams. The Eastern nucleus only has an upper limit on the H β flux, but thecorresponding lower limit of the ratio is in the Seyfert region, thus both galaxies are classified as Seyferts from our data. Figure 9.
Zoom in of the Keck/LRIS optical spectrum of theWestern galaxy in the region where the Calcium triplet absorptionoccurs, which we use to measure the velocity dispersion of the stars.
Figure 10.
Zoom in of the Keck/LRIS optical spectrum of theEastern galaxy in the region where the broad H α line was detectedwhich shows the spectral decomposition. Figure 11.
Integrated Keck/OSIRIS spectrum of the Westernnucleus in the K -band (rest frame). The individual spectra wereadded over an aperture of 0.6 (cid:48)(cid:48) diameter centered at the peak ofcontinuum emission in the near-IR. Several transitions of molecularhydrogen can be seen, where the 2.12 µ m H − vi ], Br δ , Br γ and [He i ]), confirming the presence of a powerfulAGN. accretion rate. This would explain the spectral shapeseen in the 2011 Chandra observation, which is fittedwell by a model with similar physical parameters suchas column density and the power-law slope. Assuming asimilar luminosity of the nucleus in 2014 and 2011, thisscenario also agrees with the non-detection of the AGNby
NuSTAR and its very faint detection by
Swift /XRT.The Eastern nucleus was weak in the hard X-ray bandin 2004 with respect to the Western source so would beundetectable by
NuSTAR above 3 keV after a decreaseby a factor of 10. Additional evidence in favor of thedrop in accretion rate comes from the optical spectrum,which reveals a weak, but significantly detected broad H α Kosec et al.
Figure 12.
Maps of the H − − and the angu-lar scale is 700 pc/ (cid:48)(cid:48) at the redshift of the system and our assumedcosmology. The central 200 −
300 pc shows ordered rotation (PAof the kinematic major axis ∼ ◦ ), whereas the gas at ∼ − ) ve-locity dispersion, pointing towards a physical interaction betweenthe galaxies (bottom). The contours delineate the molecular gasmorphology and are normalized to the peak of emission. Eachcontour represents a change in flux of 10%. line from the nucleus of the Eastern galaxy, which mustcome from close to the central engine. If the dimmingwere due to obscuration, it would require an extremelylow dust-to-gas ratio for the X-ray flux to have under-gone such suppression, while the H α line remains visible,although the possibility still exists that the H α line maybe scattered light.However, no emission lines were detected in the NIRfrom the Eastern nucleus. This implies that there arenot enough ionizing photons to produce emission linesof ionized gas in the near-IR (like Br- γ , [He i ] or [Si vi ]seen from the Western nucleus). Also, the lack of molec-ular hydrogen indicates that there is not sufficient gasto maintain the active nucleus. This is consistent withour interpretation that the accretion rate of the easternnucleus has dropped.We discuss the scenario that the merging of the twogalaxies is directly linked to the change in accretion rateof the Eastern AGN. Firstly, galaxy merger simulationspredict large fluctuations in black hole accretion rateduring the final stages of a merger (e.g. Van Wassen-hove et al. 2012; Gabor et al. 2015). Although the timeresolution of most simulations ( ∼ years) is muchlonger than our observational time scale, results on muchshorter time scales ( ∼
10 years) also reveal similar accre-tion rate fluctuations (J. Gabor, private communication).It should be noted, however, that fluctuations are alsopredicted from simulations of isolated AGN (e.g. Novak,Ostriker & Ciotti 2011) and have been observed as well(LaMassa et al. 2015), although this could be related toa tidal disruption event (Merloni et al. 2015). Secondly,the motion of the gas in the central ∼ kpc of the Westerngalaxy as revealed by integral field spectroscopy is highlysuggestive that the galaxy merger is directly affecting thekinematics of the gas within the nuclear region, provid-ing a direct observational link between the galaxy mergerand the change in mass accretion rate on to the blackhole.To better place this AGN pair in the context ofgalaxy simulations, we estimate the masses of the centralSMBHs. For the Eastern galaxy, a broad H α line was de-tected, which we use for the M BH estimation. Greene &Ho (2005) presented a method for estimating the blackhole mass from the width and luminosity of the H α line.From their equation 6, given that we measure a widthof 4226 km s − and a luminosity of 2.1 × erg s − , weobtain M BH ≈ . × M (cid:12) . Since no broad line wasdetected in the Western galaxy, we use the velocity dis-persion of the stars in the center of the galaxy to estimatethe black hole mass. Using the M BH − σ ∗ relation fromKormendy & Ho (2013) and the Calcium triplet measure-ment implies a black hole mass of M BH = 3 . +5 . − . × M (cid:12) for the Western galaxy. This then implies that theblack hole mass ratios of the two galaxies is 10:1, whichis rather larger than the 4:1 or 2:1 M BH ratios consideredin recent simulations by Gabor et al. (2015).It is interesting that the AGN in this system withthe smallest black hole mass has exhibited the greatestX-ray variability since it is well known that the vari-ability timescale correlates with black hole mass. I.e.The variability timescale increases with M BH (e.g. Pa-padakis 2004). However, these timescales are muchshorter ( ∼ s) than the timescale of the drop in X- SO 509-IG066 λ Edd in theX-ray spectrum of the Eastern source is expected sincethere is a known correlation between λ Edd and Γ (e.g.Shemmer et al. 2006; Risaliti et al. 2009; Brightman et al.2013, 2016). During the 2004
XMM-Newton observation,the Eastern source had an absorption-corrected L X of7.5 × erg s − . Applying a bolometric correction of10 (Lusso et al. 2012) implies L Bol =7.5 × erg s − ,which in turn yields λ Edd =0.12 for our M BH estimate.At the time of the Chandra observation, L X had reducedby a factor of 10, meaning a decline in λ Edd by the samefactor. For the observations in 2014, the X-ray emis-sion from the Eastern nucleus was undetectable, thus λ Edd (cid:46) .
01 at that time. From Brightman et al. (2013),Γ = (0 . ± . λ Edd +(2 . ± . ≈ − . λ Edd ≈ −
1. For our anal-ysis presented in Section 3, we tie the Γ values to eachother for both flux levels. If we perform the same anal-ysis, but with the Γ parameter not linked between theobservations, we obtain Γ = 1 . ± .
12 for the
XMM-Newton observation and Γ = 1 . +0 . − . for the Chandra observation. The uncertainties are therefore too large toconstrain ∆Γ at the requisite level. SUMMARY AND CONCLUSIONS
We conducted a multi-wavelength analysis of thegalaxy pair ESO 509-IG066 using X-ray, optical, near-IR and mid-IR data taken between 2004 and 2016. Thepair of galaxies, located at a distance of 150 Mpc with aprojected separation of 10.9 kpc, were both reported tohost an AGN of L X ∼ erg s − by G05 using XMM-Newton data. In an analysis of all available data, wefound that since the
XMM-Newton observation in 2004,the Eastern nucleus has shown a strong decrease in X-ray flux revealed first by a
Chandra observation in 2011.The galaxy remained at this level or lower during a joint
NuSTAR and
Swift /XRT observation in 2014. The X-ray emission from the Western source remained relativelyconstant during this period. Although the 16 (cid:48)(cid:48) angularseparation of the galaxy pair causes significant overlapgiven the
NuSTAR
PSF, there is no evidence for theEastern source in the
NuSTAR image from 2014. Thisargues against a rise in obscuration behind the drop inX-ray flux, unless it is extreme. New Keck/LRIS opticalspectroscopy taken after the drop in X-ray flux revealsa broad component to the H α line from the Eastern nu-cleus, which also strongly argues against heavy obscura-tion. We therefore conclude that the AGN has droppedintrinsically in luminosity, most likely due to a decreasein mass accretion rate. From AO-assisted near-infraredintegral-field spectroscopy, we find that the kinematics ofthe gas close to the Western nucleus show evidence thatthe galaxy merger is having a direct effect close in tothe black hole, providing an observational link betweenthe galaxy merger and the mass accretion rate on to theblack hole.We thank the referee for their constructive input onour manuscript. We also thank Mislav Balokovi´c forintroducing us to these interesting galaxies and JaredGabor for useful discussion. RJA was supported byFONDECYT grant number 1151408. AM acknowledges support from the ASI/INAF grant I/037/12/0-011/13.CR acknowledges financial support from the CONICYT-Chile grants “EMBIGGEN” Anillo ACT1101, FONDE-CYT 1141218, Basal-CATA PFB–06/2007 and from theChina-CONICYT fund. This work was supported un-der NASA Contract No. NNG08FD60C, and made useof data from the NuSTAR mission, a project led bythe California Institute of Technology, managed by theJet Propulsion Laboratory, and funded by the NationalAeronautics and Space Administration. We thank the
NuSTAR
Operations, Software and Calibration teamsfor support with the execution and analysis of theseobservations. This research has made use of the
NuS-TAR
Data Analysis Software (NuSTARDAS) jointly de-veloped by the ASI Science Data Center (ASDC, Italy)and the California Institute of Technology (USA). Thework presented here was also based on observations ob-tained with
XMM-Newton , an ESA science mission withinstruments and contributions directly funded by ESAMember States and NASA. This publication makes useof data products from the Wide-field Infrared SurveyExplorer, which is a joint project of the University ofCalifornia, Los Angeles, and the Jet Propulsion Labo-ratory/California Institute of Technology, funded by theNational Aeronautics and Space Administration. Thedata presented herein were obtained at the W.M. KeckObservatory, which is operated as a scientific partner-ship among the California Institute of Technology, theUniversity of California and the National Aeronauticsand Space Administration. The Observatory was madepossible by the generous financial support of the W.M.Keck Foundation. The authors wish to recognize and ac-knowledge the very significant cultural role and reverencethat the summit of Mauna Kea has always had withinthe indigenous Hawaiian community. We are most for-tunate to have the opportunity to conduct observationsfrom this mountain. This research has also made use ofdata and software provided by the High Energy Astro-physics Science Archive Research Center (HEASARC),which is a service of the Astrophysics Science Division atNASA/GSFC and the High Energy Astrophysics Divi-sion of the Smithsonian Astrophysical Observatory. Fur-thermore, this research has made use of the NASA/IPACExtragalactic Database (NED) which is operated bythe Jet Propulsion Laboratory, California Institute ofTechnology, under contract with the National Aero-nautics and Space Administration. AM acknowledgessupport from the ASI/INAF grant I/037/12/0-011/13,CR acknowledges financial support from the CONICYT-Chile grants “EMBIGGEN” Anillo ACT1101, FONDE-CYT 1141218, Basal-CATA PFB–06/2007 and from theChina-CONICYT fund.
Facilities: NuSTAR , XMM-Newton pn,
Chandra , Swift (XRT)
WISE , NEOWISE , Keck/LRIS, Keck/OSIRISREFERENCES