Line shape variability in a sample of AGN with broad lines
D. Ilic, L. C. Popovic, A. I. Shapovalova, A. N. Burenkov, V. H. Chavushyan, A. Kovacevic
aa r X i v : . [ a s t r o - ph . GA ] O c t LINE SHAPE VARIABILITY IN A SAMPLE OF AGN WITH BROAD LINES
D.Ili´c a , L. ˇC. Popovi´c b,a , A. I. Shapovalova c , A. N. Burenkov c , V. H. Chavushyan d , A.Kovaˇcevi´c a a Department of Astronomy, Faculty of Mathematics, University of Belgrade, Studentski trg 16, 11000 Belgrade, Serbia b Astronomical Observatory, Volgina 7, 11060 Belgrade 74, Serbia c Special Astrophysical Observatory of the Russian Academy of Science, Nizhnij Arkhyz, Karachaevo-Cherkesia 369167, Russia d Instituto Nacional de Astrof´ısica, ´Optica y Electr´onica, Apartado Postal 51, CP 72000, Puebla, Pue. M´exico
Abstract
The spectral variability of active galactic nuclei (AGN) is one of their key features that enables us to study in moredetails the structure of AGN emitting regions. Especially, the broad line profiles, that vary both in flux and shape,give us invaluable information about the kinematics and geometry of the broad line region (BLR) where these lines areoriginating from.We give here a comparative review of the line shape variability in a sample of five type 1 AGN, those with broademission lines in their spectra, of the data obtained from the international long-term optical monitoring campaigncoordinated by the Special Astrophysical Observatory of the Russian Academy of Science. The main aim of this campaignis to study the physics and kinematics of the BLR on a uniform data set, focusing on the problems of the photoionizationheating of the BLR and its geometry, where in this paper we give for a first time a comparative analysis of the variabiltyof five type 1 AGN, discussing their complex BLR physics and geometry in the framework of the estimates of thesupermassive black hole mass in AGN.
Key words: galaxies:active-galaxies quasar:individual (Arp 102B, 3C 390.3, NGC 5548, NGC 4151, Ark 564)line:profiles
1. INTRODUCTION
In spite of decades of intensive investigations, the broadline region (BLR) of active galactic nuclei (AGN) is yet notfully understood. The direct detection of the BLR remainsa challenge for modern instruments, since the angular sizeof the BLR is less than 0 .
001 arcsec even for the closestAGNs. The only information from the BLR comes in theform of the broad emission lines (BEL), that are a veryprominent feature in the spectra of the so-called type 1AGN (Osterbrock & Ferland, 2006). Nevertheless, the in-vestigations of the BEL’s properties (flux and shape), andespecially of their variability, indicates that the BLR islinked with the accretion process onto the supermassiveblack hole in the center of an AGN. First of all, the BLRgas is photoionized by the continuum radiation from theaccretion disk, and secondly, at least part of the BLR fol-lows its geometry, i.e. even though the BLR geometry isnot ubiquitously described it is believed that one part ofthe BLR is a part of the accretion disk or has a disk-likegeometry (Popovi´c et al., 2004).An important method to indirectly map the geometryand kinematics of the the BLR is the “reverberation map-ping” (Blandford & McKee, 1982; Gaskell, 1988; Peterson,1993; Peterson et al., 2004), a method that is based onmultiple spectroscopic observation and the variability of
Email address: [email protected] (D.Ili´c) spectral properties. In particular, the BEL’s flux varieswith respect to the ionizing continuum flux with a cer-tain time delay (of the order of days to weeks), due to thelight-travel time from the source of the continuum photonto the source of the BEL photon, that is the BLR. There-fore, by applying the cross-correlation function (CCF) be-tween the continuum and BEL light curves an estimate ofthe time-lag τ between the two signals can be obtained,which is basically the photometric BLR radius R BLR = cτ ,where c is the speed of light. This is a powerful tool toobserve the unobservable, as one can estimate the size ofthe BLR and consequently the mass of the supermassiveblack hole, since the BLR gas is assumed to be virialized,and the method has been exploited in many papers (seee.g. Kaspi et al., 2000; Peterson et al., 2004; Bentz et al.,2009; Doroshenko et al., 2012; Kollatschny et al., 2014, etc.).The reverberation-mapped AGN are of particular impor-tance because they anchor the scaling relationships (BLRradius-luminosity relationship, see e.g. Kaspi et al., 2000;Bentz et al., 2006) that allow the estimate of the BLR ra-dius, and the supermassive black hole mass, from a single-epoch spectrum.However, this method is based on several assumptions,which makes it still pretty uncertain, especially for the su-permassive black hole mass estimate. It is assumed thatthe BLR gas is photoionized by the continuum comingfrom the central source, and that is gravitationally bound Preprint submitted to Journal of Astrophysics and Astronomy, Springer October 8, 2018 able 1: Spectral characteristics and the variability of the observed objects: object name and the monitoring period, redshift, AGN type,line profile characteristics and the FWHM of the mean spectrum, photometric BLR radius R BLR = cτ BLR , maximal and minimal flux ratio F max /F min , the variability parameter F var , mean continuum luminosity at 5100˚A, and the main references. If available values for both H β and H α line are listed. object z AGN Line Profile Shape c τ BLR [ld] F max /F min F var λ L λ (5100) Ref(period [years]) type FWHM [km/s] H β /H α H β /H α H β /H α [10 erg/s]NGC 5548 0.0172 Sy 1.0-1.8 strong shoulders 49 +19 − /27 +14 − ± ± ± +28 − /6 +27 − ± ± ± +28 − /120 +18 − ± ± ± +27 − /5 +6 − ± ± ± +20 − /21 +14 − ± ± ± to the supermassive black hole so that the virial theoremcan be used (Gaskell, 1988). Moreover, it strongly de-pends on the geometry of the BLR, and the kinematics andstructure of the BLR is not ubiquitously described (see e.g.Sulentic et al., 2000; Gaskell, 2009, and reference therein).The BLR is probably very complex, often with evidence formultiple components (e.g. Sulentic et al., 2000; Ili´c et al.,2006; Popovi´c et al., 2004; Bon et al., 2009), and can havesystematic motions such as infalls, outflows, circular mo-tions, which should be detected in the BEL profiles. If theline profile is varying, the change of the line shape shouldbe according to the change in the BLR geometry and kine-matics. In principle, the reverberation assumption can betested, e.g. one test is if the BEL fluxes correlate with thecontinuum flux. Therefore, detailed studies of differentbroad line profiles is required, and this can be done withlong-term (of the order of decade and longer) monitoringof type 1 AGN with different BEL properties.In this paper we give a comparative overview of ourinvestigation based on the international long-term opticalmonitoring campaign, coordinated by the Special Astro-physical Observatory of the Russian Academy of Science(SAO RAS). Within this campaign, several type 1 AGNhave been monitored for decades, out of which data werepublished for 5 objects (Shapovalova et al., 2001, 2004,2008, 2010a, 2012, 2013). The objects have been selectedso that representatives of type 1 AGN with different BELprofiles are in the sample: Arp 102B and 3C 390.3 haveBELs with double-peaked line profiles, Ark 564 is a Narrow-Line Seyfert 1 (NLSy1), and NGC 4151 and NGC 5548are well-studied Seyfert 1.5 with strongly variable BELprofiles, with bumps and asymmetries present in the lineprofiles (see Figure 1). All 5 AGN have been discussed in-dividually in referenced papers, but here we aim to give fora first time a comparative analysis of their variabilty prop-erties, that is to: i) summarize already obtained most im-portant results of their BEL variability, ii) give new com-parative analysis of the BEL properties, iii) discuss theircomplex BLR physics and geometry, all in the frameworkof the estimations of the mass of the supermassive blackhole in AGN, and suggestions for future exploitation of R e l a t i v e f l u x [ a r b i t r a r y un i t s ] Rest-frame wavelength [in A]H β region H α region3c390.3 (Feb 2007)Arp102B (Jul 1998)NGC 5548 (Mar 2003)NGC 4151 (Apr 2004)Ark 564 (Aug 2006) Figure 1: Comparison of optical spectra for all 5 objects (name andmonth of the observation denoted), given in arbitrary flux unites andshifted for better display. Rest-frame wavelengths are displayed onthe X-axis, and the spectral region of H β and H α line are marked. our large data sets. Our data sets are observed, reducedand analyzed using the same procedures, thus the totalanalyzed data form an uniform set of data.The paper is organized as follows, in Section 2 webriefly report on observation and data reduction, in Sec-tion 3 we present the main result, which are discussed inSection 4, and in Section 5 we outline our conclusions andfuture investigations.
2. OBSERVATIONS AND DATA SETS
With the international long-term optical monitoringcampaign, coordinated by SAO RAS several type 1 AGNhave been monitored for decades, out of which data werepublished for 5 objects in 2001–2013 (see Table 1 for de-tails). The high quality spectra were taken with 6 differenttelescopes: (i) 6-m and 1-m telescopes of SAO RAS; (ii)2.1-m telescope of Guillermo Haro Astrophysical Obser-vatory, Mexico; (iii) 2.1-m telescope of the ObservatorioAstronomico Nacional at San Pedro Martir, Mexico, and2 F l u x ( H β ) Julian Date [2400000+]Ark 564 0.2 0.4 0.6 0.8 1 F l u x ( H β ) NGC 4151 0.2 0.4 0.6 0.8 1 F l u x ( H β ) NGC 5548 0.2 0.4 0.6 0.8 1 F l u x ( H β ) Arp 102B 0.2 0.4 0.6 0.8 1 1988 1992 1996 2000 2004 2008 2012 F l u x ( H β )
3C 390.3
Figure 2: H β line light curves for all 5 objects (name denoted in bot-tom left). The line flux and corresponding error-bars are normalizedto the maximal flux for better comparison and X-axis shows timein units of modified Julian Date. (iv) the 3.5-m and 2.2-m telescopes of Calar Alto Observa-tory, Spain. The data acquisition, reduction and differentcalibration (e.g. corrections for different position angle,seeing and aperture effects), and flux measurements weredescribed in details in Shapovalova et al. (2001, 2004,2008, 2010a, 2012, 2013). Table 1 gives a summary of thedata (object name, monitored period, redshift and type,line properties, and references).For all objects we measured different line parameters,such as total line fluxes, broad line fluxes , line widths, andline-flux time-lags, that will be investigated here (see refer-ences listed in Table 1 for details on other line parameters).Some of the measured H β and H α line parameters aregiven in Table 1: the full-width half-maximum (FWHM)of the mean spectrum, the photometric radius of the BLR R BLR = cτ BLR , where τ BLR is the line-flux time-lag ob-tained from the CCF analysis, the maximal-to-minimalflux ratio F max /F min , the variability parameter F var , andthe continuum luminosity at 5100 ˚A. The variability pa-rameter F var is used to estimate the amount of the line-variability and is calculated according to O’Brien et al.(1998). For the estimates of time-lags from the CCF analy-sis a Z-transformed Discrete Correlation Function (ZDCF)was used (Alexander, 1997). Luminosities are calculatedusing the mean continuum flux at 5100 ˚A and the on-line calculator for luminosity distance (Wright, 2006), forwhich we adopted the cosmological parameters Ω = 0 . Λ = 0 . k = 0, and a Hubble constant, H =69 . − . All flux measurements from this cam-paign are publicly available in publications listed in Table1 and in the corresponding VizieR Online Data Catalog.The same procedures were performed for the observa-tions, data reduction, measurements and analysis, so thatthe resulting data sets of these 5 objects basically are uni-form. This makes them valid for further analysis and com-parison.
3. RESULTS
Our sample consists of 5 type 1 AGN with differentBEL profiles (Figure 1): Arp 102B and 3C 390.3 haveBELs with double-peaked line profiles, Ark 564 is a NLSy1,and NGC 4151 and NGC 5548 are strongly variable Seyfert1.5. The results of our long-term monitoring campaign aresummarized in Table 1, and Figure 1 and 2, that show thecomparison of optical spectra for all 5 objects, given inarbitrary flux unites and shifted for better display (Figure1), and the H β line flux light curves for all 5 objects (Fig-ure 2), where the line flux and corresponding errors arenormalized to the maximal flux for better comparison.Similar as in Kollatschny et al. (2006), in Figure 3 weplot the ratio of the maximal-to-minimal line flux against To study the broad components showing the main BLR charac-teristics, the narrow lines and the forbidden lines were subtractedeither by constructing observational templates or by multi-Gaussianfittings. F m a x / F m i n FWHM [km s -1 ]H β H α c τ B L R [ l d ] NLSy1 Seyfert 1 DPL AGNArk 564 NGC 4151 3C 390.3NGC 5548 Arp 102b
Figure 3: Photometric radius of the BLR c τ BLR (upper) and the vari-ability indicator, the ratio of maximal-to-minimal flux F max /F min (bottom) versus the FWHM of H β (circles) and H α (diamonds) linesfor all 5 objects. The vertical dashed lines correspond to the lineFWHM of 2000 and 10000 km / s.Table 2: Line and continuum flux correlations for all 5 objects. Thecorrelation coefficient and the corresponding P null value (in brackets)are given for each pair of data. object H β vs F cnt H α vs F cnt NGC 5548 0.90 (0.1E-28) 0.85 (0.2E-13)NGC 4151 0.94 (0.0) 0.88 (0.0)3C390.3 0.91 (1.1E-19) 0.82 (1.1E-12)Ark 564 0.59 (0.7E-9) 0.71 (0.8E-8)Arp 102B 0.37 (0.5E-4) 0.25 (0.02)the FWHM of H β (circles) and H α line (diamonds). Wedivided the F max /F min vs. FWHM plane according to theline FWHM in three groups of objects: 1) NLSy1 havingFWHM < < FWHM < > β line andcontinuum flux for all objects (name indicated on eachplot). The correlation coefficient and the correspondingP-value are listed in Table 2. The two objects, Arp 102Band Ark 564 have very low correlation coefficients (0.37and 0.59, respectively), indicating a weak response of theline to the continuum variability.In principle, the geometry of the BLR should be seenin the line shape, e.g. in the most obvious case if the rota-tional motion is present in the BLR, then the line profile should have two peaks, the red corresponding to the gasmoving away, and the blue one moving towards the ob-server. Moreover if there is a change in the BLR geometryand kinematics, the line shape should vary accordingly. Inorder to show how the line profiles vary in two DPL ob-jects, we plot the mean and rms profile for the H β line of3C 390.3 (Figure 5, upper panel) and Arp 102B (Figure 5,bottom panel). From the shape and intensity of the rmsprofiles, it is obvious that in case of 3C 390.3 the line pro-files vary significantly, while in Arp 102B the line profilesremain almost unchanged.Finally, in Figure 6, the radius of the BLR for H β lineversus the continuum luminosity at 5100 ˚A, is given. Thesolid line represents the scaling relation from Bentz et al.(2006), log R BLR = K + α log[ λL λ (5100) / ], where α isthe slope of the BLR radius-luminosity relationship and K is the scaling factor, and in Figure 6 α = 0 .
533 and K =1 .
527 from Bentz et al. (2013) are plotted. The dashed lineis a simple linear fit of the above equation (not consideringthe error-bars) through all objects except from Ark 564,obtaining the fitting parameters of α = 0 .
887 and K =2 .
4. DISCUSSION
We have analyzed the variability of the optical spec-tra for five type 1 AGN with different properties of thebroad emission lines. The data sets were processed in auniform way, which makes them valid for further analysisand comparison.During the long-term monitoring, that is by rule longerthan a decade, the line and continuum flux for all objectsare varying. Most of the objects are strongly varying, i.e.NGC 5548, NGC 4151, and 3C 390.3, that is best seen fromthe normalized light curves (Figure 2), but also from theratio of the maximal-to-minimal flux (Figure 3, Table 1)and the variability parameter F var that is ∼
40% (Table 1).This is supported with the correlation analysis, as thesestrongly varying objects are showing significant correlationbetween the line and continuum fluxes (Figure 4, Table 2).On the other hand, the line fluxes of the Ark 564, thatis a NLSy1, stay almost constant during the monitoredperiod (only changing by ∼ ∼ ∼ β of Arp 102B it is r = 0 . P null =0.5E-04), and even worse for H α of the same object( r = 0 . P null =0.22). The lack of variability can be dueto poor sampling of the data, e.g. in case of Arp 102B themean sampling is 40 days (Kovaˇcevi´c et al, 2014). How-ever, the lack of correlation between the line and contin-uum flux also points to additional sources of ionization inthe BLR apart from the central AGN continuum source.4 F li ne [ - e r g s - c m (cid:226)(cid:136)(cid:146) ] F cnt [10 -16 erg s -1 cm (cid:226)(cid:136)(cid:146)2 A -1 ] Arp 102B F li ne [ - e r g s - c m (cid:226)(cid:136)(cid:146) ] F cnt [10 -15 erg s -1 cm (cid:226)(cid:136)(cid:146)2 A -1 ] Ark 564 F li ne [ - e r g s - c m (cid:226)(cid:136)(cid:146) ] F cnt [10 -15 erg s -1 cm (cid:226)(cid:136)(cid:146)2 A -1 ]
3C 390.3 F li ne [ - e r g s - c m (cid:226)(cid:136)(cid:146) ] F cnt [10 -14 erg s -1 cm (cid:226)(cid:136)(cid:146)2 A -1 ] NGC 4151 F li ne [ - e r g s - c m (cid:226)(cid:136)(cid:146) ] F cnt [10 -15 erg s -1 cm (cid:226)(cid:136)(cid:146)2 A -1 ] NGC 5548
Figure 4: Correlation between the line and continuum flux for theH β line for all 5 objects (name indicated on each plot). The dashedline represents the linear best fit through all data. N o r m a li z ed f l u x Velocity [km/s]meanrms 0 0.2 0.4 0.6 0.8 1 N o r m a li z ed f l u x meanrms Figure 5: Mean (solid line) and rms (dashed line) line profiles for theH β line of 3C 390.3 (upper panel) and Arp 102B (bottom panel). c τ B L R [ l d ] λ L λ (5100 A) [erg/s] Ark 564NGC 41513C 390.3NGC 5548Arp 102b H β this workBentz et al. 2013 Figure 6: The photometric radius of the BLR versus the contin-uum luminosity at 5100 ˚A, for H β line. The solid line representsthe BLR radius-luminosity scaling relation from Bentz et al. (2013)where R BLR ∼ L . , while the dashed line is our fit through allobjects except Ark 564, where R BLR ∼ L . .
5e should note here that our result for observed low vari-ability in NLS1 Ark 564 is in agreement with a comparativestudy of NLS1 and broad line Seyfert 1 galaxies (BLS1)variability given by Ai (2013). They found that the ma-jority of NLS1-type AGN show variability on timescalesfrom several days to a few years, but with the variabiltyamplitudes smaller than in the case of BLS1-type AGN.Moreover, there are some results in particular study casesthat indicate peculiar BLR in NLS1 AGN, as e.g. in NLS1Mrk 493 (see Popovi´c et al., 2009) where the broad lineshave not changed in a period of couple of years (betweentwo observations), and it is suspected that the weak broadH α and H β may come partly from the unresolved centralBLR, but also may be partly produced by the violent star-burst in the circumnuclear ring (clearly resolved on HSTimage). It seems that NLS1-type AGN have some specificphysical properties that have to be considered in monitor-ing campaigns.Another peculiarity to note is the fact that the line fluxappears to flatten at high continuum flux, that is especiallyseen in the case of NGC 4151 for which the line flux satu-rates above the continuum flux of 7 × erg cm − s − ˚A − (see thorough analysis and discussion in Shapovalova et al.,2008). Here this effect can be also noticed in the case ofNGC 5548 and only weakly in 3C 390.3. This implys thateither there is more ionizing to optical flux than expectedfor a typical AGN spectrum, or that, again, lines are notproduced purely by the photoionization from the centralcontinuum source.The photometric radius of the BLR c τ BLR increaseswith the width of the BEL’s profiles (Figure 3), exceptfor the case of Arp 102B. This is in accordance with thegeneral picture of the central engine, as it appears thatin objects with lower accretion rate (
L/L edd ), those withlarger line widths (Marziani et al., 2014) the BLR is larger.Arp 102B though remains a mystery, as here some othergeometry, apart from the obvious accretion disk geometrythat is usually used to explain the double-peaked line pro-files, may be considered (see Popovi´c et al., 2014). In favorto this is the fact that during more than two decades, theline profiles of this object have not changed significantly(Figure 5, bottom panel). This is not expected in caseof the relatively compact accretion disk ( ∼ ∼
7% for H β line) and poor significanceof the obtained BLR radius. Therefore, a simple fit wasdone through all objects except Ark 564, and we obtained α = 0 .
887 (Figure 6). This result is different from Bentzet al. (2013), who obtained α = 0 . λ L λ (5100) ∼ − erg/s) of the empirical radius-luminosity scaling re-lation is very sensitive to the measured continuum lumi-nosity and BLR radius. One reason could be the contribu-tion of the host-galaxy continuum flux, that was not con-sidered here, apart from subtracting the extended sourcecorrection factor G ( g ), that is an aperture-dependent cor-rection factor standardly used in our campaign to accountfor the host-galaxy contribution. Moreover, the sample isfar from being statistically significant. However, this stillillustrates that new results of the monitoring campaignsmay help in obtaining more accurate results and all AGNproperties should be taken into account (e.g. level of linevariability or line production mechanism), as well as ob-taining direct distances need for calculations of the lumi-nosity as discussed in Bentz et al. (2013).It is obvious that for some objects, the BLR geometry ischanging during decades of observation, e.g. in case of 3C390.3 the rms profile of H β shows strong variability espe-cially in the blue peak (Popovi´c et al., 2011). One possiblescenario to model this variability is with the orbiting brightspots in accretion disk (Jovanovi´c et al., 2010). Other pro-posed scenario, e.g. for NGC 4151 that also has stronglyvariable BEL’s profiles (Shapovalova et al., 2010b) is thatthis well-studied object is hosting a binary supermassiveblack hole with two BLR orbiting around them (Bon et al.,2012). Bon et al. (2012) found by analyzing the 20-yearlong NGC 4151 H α light curve and radial velocity curvesof the line components the evidence for a sub-parsec scalesupermassive binary black hole system with an orbital pe-riod of ∼
5. CONCLUSIONS
The comparative study of the main results of SAORAS long-term monitoring campaign of five type 1 AGNwith BELs having different properties are given. The vari-ations of the BELs parameters of the H β and H α line areinvestigated with the aim to constrain the properties of theBLR and test the reverberation mapping as a method toestimate the size of the BLR, and consequently the massof the supermassive black hole in the center of an AGN.The main conclusions are:(i) the five selected objects are varying during the moni-tored period, where some are strongly variable ( ∼ Acknowledgments
This work was supported by the Ministry of Education,Science and Technological Development of Republic of Ser-bia through the project Astrophysical Spectroscopy of Ex-tragalactic Objects (176001), INTAS (grant N96-0328),RFBR (grants N97-02-17625 N00-02-16272, N03-02-17123,06-02-16843, N09-02-01136, 12-02-00857a, 12-02-01237a,N15-02-02101), CONACYT research grants 39560, 54480,and 151494. D.I. has been awarded L’Or´eal-UNESCO”For Women in Science” National Fellowship for 2014.We would like to thank the anonymous referees for veryuseful comments.
References