Mon.Not.R.Astron.Soc.000,1–10(0000) Printed6January2015 (MNLATEXstylefilev2.2) Relationship between X-ray spectral index and X-ray Eddington ratio for Mrk 335 and Ark 564 5 1 R. Sarma1⋆, S. Tripathi2, R. Misra2, G. Dewangan2, A. Pathak3 and J. K. Sarma3 0 1Department of Physics, Hojai College, Hojai, 782435, India 2 2Inter-UniversityCentre For Astronomy and Astrophysics, Post Bag 4, Ganeshkind, Pune-411007, India n 3Department of Physics, Tezpur University,784028, India a J 5 ] ABSTRACT A We present a comprehensive flux resolved spectral analysis of the bright Narrow line G Seyfert I AGNs, Mrk 335 and Ark 564 using observations by XMM-Newton satellite. . The meanandthe flux resolvedspectra are fitted by an empiricalmodel consistingof h two Comptonizationcomponents,one for the low energy soft excess andthe other for p - the high energy power-law. A broad Iron line and a couple of low energies edges are o required to explain the spectra. For Mrk 335, the 0.3 - 10 keV luminosity relative to tr the Eddingtonvalue, LX/LEdd,varied from0.002to 0.06.The index variationcanbe s empiricallydescribedasΓ=0.6log10LX/LEdd+3.0for0.005<LX/LEdd <0.04.At [a LX/LEdd ∼0.04thespectralindexchangesandthencontinuestofollowΓ=0.6log10 LX/LEdd+2.7,i.e.onaparalleltrack.Weconfirmthattheresultisindependentofthe 1 specificspectralmodelusedbyfittingthedatainthe3-10keVbandbyonlyapower- v lawandanIronline.For Ark 564,the index variationcanbe empirically describedas 8 Γ=0.2log10LX/LEdd+2.7withasignificantlylargescatterascomparedtoMrk335. 0 Our results indicate that for Mrk 335, there may be accretion disk geometry changes 9 which lead to different parallel tracks. These changes could be related to structural 0 changes in the corona or enhanced reflection at high flux levels. There does not seem 0 to be any homogeneous or universal relationship for the X-ray index and luminosity . 1 for different AGNs or even for the same AGN. 0 5 Key words: Key Words: galaxies: Seyfert, X-rays: galaxies 1 : v i X 1 INTRODUCTION AGN,Γvariesasafunctionoftime.Understandingthebe- r haviourofΓwithothersourcepropertiesisexpectedtogive a ActiveGalactic Nuclei (AGN)are known toemit X-raysas importantinsightintothenatureoftheradiativemechanism aresultofphysicalprocessesactiveintheinnermostregions and thegeometry of theinner regions. near the central super-massive black hole. X-ray spectra of a typical AGN in the 2-10 keV range show primarily the Previous studies have shown the existence of a corre- signature of a power-law continuum and an iron line. At lation between the X-ray photon index Γ and the source energies<2keVthereisoftenasoftexcessoverthepower- flux e.g., (Perola et al. 1986; Singh, Rao & Vahia 1991; lawemission.Thepower-lawemissioniswidelyconsideredas Mushotzky et al. 1993; Done, Madejski, & Z˙ycki 2000; theoutcomeofinverseComptonscatteringofthermallypro- Nandra& Papadakis 2001). For many Seyfert 1 AGNs, the duced accretion disk seed optical/UV photons by a corona power-law index shows significant variation and generally of hot electrons close to thedisc (Haardt & Maraschi 1991, follows the trend of steeper Γ with increasing source inten- 1993; Zdziarski, Poutanen, & Johnson 2000). However, the sity. During 1997, MCG 6-30-15 was observed with RXTE geometryandsizeofthecoronaaswellasthephysicalmech- for a duration of 8 days ∼ 910 ks. The range of Γ ob- anismgoverningtheenergytransferbetweenthetwophases served was 1.8-2.2 with flux variation of ∼ 3.3 in the en- are not well understood. ergy band 2-10 keV.This observation showed a tight corre- While, the average value of photon index of the pri- lationbetweenthephotonindexΓandthesourcefluxinthe mary power-law for AGN has been found to be Γ ∼ sensethatthepower-lawsteepensasthesourcegetsbrighter 1.9, there is a large variation with Γ ranging from 1.5- (Vaughan & Edelson2001).ThesteepeningofΓ(range1.89- 2.5 (Nandra& Pounds 1994; Page et al. 2005). For a given 2.34) with fluxhasalso been observed inIRAS13224-3809, wherea10daylongASCAobservationrevealedachangein thepower-lawfluxbyafactorof∼3.2inthe2-10keVrange ⋆ e-mail:[email protected] (Dewangan et al. 2002).RXTE observedNGC5548 on five 2 R. Sarma et al. occasions in 1998 during which a clear positive correlation galaxiesasaclassfollowtheMBH−σ∗ relationifthewidths betweenthephotonindex(range1.75-1.93)andthe2-10keV of emission lines not strongly affected by outflow compo- flux was seen with flux variation of ∼ 1.84.(Chiang et al. nentsare used as asurrogate for σ∗ (Komossa & Xu2007). 2000).Forthesoft energy band(0.3-2.0 keV),similar trend Duetoalltheseextremeandambiguousproperties,NLS1as betweenphotonindex(range1.7-2.0)andfluxwasobserved aspecial class of AGN seem to challenge theUnified model fora2daylongBEPPOSAX observationof3C120withflux and need a more careful investigation. Recent observations variation of ∼ 1.5 (Zdziarski & Grandi 2001). The Seyfert and surveys have pointed towards the presence of smaller galaxyNGC4151wasobservedonsevenoccasionsbyGinga black hole masses in NLS1 galaxies (Barth, Greene, & Ho duringMay1987-January1989(∼21months).Inthisobser- 2005; Botte et al. 2004) and which could possibly represent vation significant flux variation (∼ 3) has been seen along an important connection with the less explored intermedi- with the correlation between the photon index and flux ate mass black holes. It is believed that the extreme X- (Yaqoob & Warwick 1991). The 2009 XMM-Newton obser- ray and other observed properties of NLS1 galaxies may vation of Mrk 335 has also shown ‘softer when brighter’ be due to an extreme value of a fundamental physical pa- property (Grupeet al. 2012). In all the above examples of rameter related to the accretion process. This fundamental AGNs, the flux variation during the observations does not parameter is most likely to be the accretion rate relative exceed a factor of ∼4. to the Eddington rate. This is well supported by more re- Studies of a sample of AGNs have also provided ev- cent studies, theoretical considerations and by black hole idence that there is significant positive correlation be- mass estimates from optical emission-line and continuum tween the X-ray photon index and the bolometric Ed- measurements that NLS1 are accreting close to their Ed- dington ratio, λ = L /L , where L is the dington rates (Boller, Brandt, & Fink 1996; Boroson 2002; Bol Edd Edd Eddington luminosity (Laor et al. 1997; Lu & Yu 1999; Xu et al. 2003; Grupe 2004; Warner, Hamann, & Dietrich Wang, Watarai, & Mineshige 2004; Shemmeret al. 2006). 2004;Collin et al.2006)andhenceshouldbeconsideredim- With larger samples involving sources with higher redshifts portanttestbedsofaccretionmodels.RecentX-raystudyof andgreaterluminosities,itwasobservedthatthehardX-ray NLS1 using an optically selected SDSS sample have shown photonindexcorrelateswiththebolometricEddingtonratio that some NLS1 show steep X-ray spectra and strong Fe when λ & 0.01 (Porquet et al. 2004; Shemmeret al. 2006; II emission while some do not. In this study, a strong cor- Saez et al. 2008; Sobolewska & Papadakis 2009; Cao 2009; relation was also found between Γ and the luminosity at Zhou, Xin,& Zhao 2010; Liu et al. 2012; Brightman et al. 1 keV, L1keV suggesting differences in Lbol/LEdd among 2013) but when λ < 0.01, especially for low-luminosity theNLS1sinthesample(Williams, Pogge, & Mathur2002; AGNs,anti-correlation isseen betweenΓ andλ(Gu & Cao Williams, Mathur,& Pogge 2004). 2009; Constantin et al. 2009; Youneset al. 2011; Xu 2011). Mrk 335, also known as PG003+199 is a nearby NLS1 In general, there have been several studies to investi- galaxy at a redshift z=0.026 (Longinotti et al. 2007a) and gate the relation between the photon index and λ for has a well measured black hole mass of 1.4 ×107 M⊙ from different samples of AGN (e.g. Risaliti, Young,& Elvis reverberation mapping (Peterson et al. 2004; Grier et al. 2009; Jin, Ward,& Done 2012). Studies incorporating 2012). It has been the target of most X-ray observatories. ROSAT and ASCA observations showed the correlation XMM-Newton observed Mrk 335 for the first time in 2000. between Γ and full width at half maximum (FWHM) AnalysingtheRGSdata,anabsorptionedgeat0.54keVdue of the Hβ emission line (Boller, Brandt, & Fink 1996; toGalacticoxygenwasreportedandthesoftexcesswasde- Brandt, Mathur, & Elvis 1997; Dewangan et al. 2002). scribed as a combination of bremsstrahlung emission and There is a need to study the relation between spec- ionised reflection from the accretion disk (Gondoin et al. tral index and Eddington ratio spanning a larger range in 2002). The XMM-Newton observation was later reanalysed L/LEdd. This is provided by XMM-Newton data for the and a narrow absorption feature at 5.9 keV was found bright and highly variable Narrow Line Seyfert 1 (NLS1) (Longinotti et al. 2007a). In 2006 January, XMM-Newton galaxy Mrk 335. Forcomparison we also analyse theexten- re-observed Mrk 335 for 133 ks which revealed a double- sive data available for another NLS1,Ark 564. peaked Fe emission feature with peaks at 6.4 and 7.0 keV Narrow line Seyfert 1 galaxies (NLS1) form a sub- (O’Neill et al. 2007). (Larsson et al. 2008) studied Mrk 335 set of AGN which exhibit exceptional features in terms of usinga151 ksSuzaku observation performed in2006. They emission-line and continuum properties. Unlike Seyfert 2 modelled the data using a power law and two reflectors in galaxies which show narrow optical emission lines, NLS1 which an ionised, heavily blurred, inner reflector produces show the broad emission-line optical spectra of Seyfert 1 mostofthesoftexcess,whileanalmostneutralouterreflec- galaxies, but with the narrowest Balmer lines from the tor (outside ∼ 40rg) produces most of theFe line emission. broad line region (full width at half maximum (FWHM) . Theyalsoverifiedtheirmodelusingthe2006XMM-Newton 2000 kms−1withrelativelyweak[OIII]λ5007emission)and data and did not see any correlation between photon in- prominentopticalFeIIemission(Osterbrock & Pogge1985; dexandpowerlawflux.Butsubsequentlyamarginaltrend, V´eron-Cetty,V´eron,& Gonc¸alves 2001). NLS1 also show a i.e., the source becomes softer with increasing count rate number of other extreme properties in X-rays e.g., strong has been reported for the same data (Grupeet al. 2012). soft excess emission below 1 keV, steep 2-10 keV power- When observed by XMM-Newton in July 2007 for 22 ks, law continuum and very rapid and large X-ray variability Mrk 335 was found to be in extremely low X-ray flux state (Leighly 1999; Gallo et al. 2004). Their X-rayspectra often (Grupeet al. 2008). The spectrum of this low flux state present complex behaviour with the presence of cold and of Mrk 335 was explained by partial covering and blurred ionised absorption, partial covering and reflection compo- reflection models. Mrk 335 was again observed by XMM- nents ((Komossa 2008) and references therein). Also NLS1 Newton in 2009 for 200 ks spread over two consecutive or- Variation of X-ray spectral index with X-ray Eddington ratio 3 bits.TheX-raycontinuumandtimingpropertieshavebeen Table 1. Observation log of Mrk 335 and Ark564 by XMM- described by using a blurred reflection model (Gallo et al. Newton 2013).The2009XMM-Newtondatahavealsobeenanalysed byGrupe et al.(2012)alongwiththeSwift data.Partialab- Mrk335 sorptionandblurredreflectionmodelsgivesequallygoodfit OBS.ID Duration(s) Date Pile-up tothespectrum.ThenumberofobservationsofMrk335by XMM-Newton over time provides an opportunity to study 0101040101 36910 2000-12-25 yes thevariationofthehighenergyphotonindexwithluminos- 0306870101 133251 2006-01-03 no ity in a systematic manner. 0510010701 22580 2007-07-10 no At a redshift z=0.02469 (Huchraet al. 1999), Ark 564 0600540601 132315 2009-06-11 no 0600540501 82615 2009-06-13 no is the brightest NLS1 galaxy in the 2.0-10.0 keV range, L(2−10)keV = 2.4 × 1043 erg s−1 (Turneret al. Ark564 2001). Ark 564 has been studied across all wavebands 0006810101 34466 2000-06-17 no (Shemmeret al.2001;Romano et al.2004).Inthe2000and 0006810301 16211 2001-06-09 no 2001 XMM-Newton observations of Ark 564, Vignali et al. 0206400101 101774 2005-01-05 no (2004) reported an edge-like feature in the EPIC data at 0670130201 59500 2011-05-24 yes ∼ 0.73 keV and interpreted it as the OVII K absorp- 0670130301 55900 2011-05-30 no tion edge. The ∼100 ks 2005 XMM-Newton observation of 0670130401 63620 2011-06-05 no Ark564 havebeen described as eitherapowerlaw and two 0670130501 67300 2011-06-11 yes black-bodies or a relativistically blurred photo ionised disk 0670130601 60900 2011-06-17 no reflection model (Papadakis et al. 2007). Dewangan et al. 0670130701 64420 2011-06-25 no (2007)analysedthe2005dataandfoundtwowarmabsorber 0670130801 58200 2011-06-29 yes phases. During 2011, Ark 564 has been observed by XMM- 0670130901 55900 2011-07-0 yes Newton for eight occasions. Legg et al. (2012) confirmed a significantsoft laginthe0.3-1.0 keVand4.0-7.5 keVbands andsuggestedadistantreflectionorigin.Thus,Ark564has wereperformed withXSPECversion 12.8.0 (Arnaud1996). been also observed several times by XMM-Newton making Thebackgroundsubtractedlightcurveshavebeenproduced it another good candidate. with the tool EPICLCCORR which are binned with 400 s Hereweinvestigatethedependenceofpower-lawindex bins. withtheX-rayEddingtonratioforthecombinedEPICspec- traofMrk335andArk564.Usingfluxresolvedspectroscopy we study the power law index variation against X-ray Ed- 3 PHENOMENOLOGICAL MODEL dington ratio. The paper is organised as follows. Section 2 describestheobservationsandthedatareductionprocedure WefittheEPIC-pnspectraldatainthe0.3-10.0keVrange, andSection3explainsthespectralmodelusedfortheanal- byasimplephenomenological modelconsistingoftwother- ysis. The details of theflux resolved spectroscopy are given malComptonizationmodels.Inparticular,thesoftexcessis insection4.Finally,weconcludethisworkwithasummary describedbytheXSPECmodel“nthComp”andforthehard of the results and discussion in section 5. X-ray emission, we have used XSPEC convolution model “Simpl” (Steiner et al. 2008). Galactic absorption is taken tobeNH=3.99×1020 cm−2forMrk335andNH=5.34×1020 2 OBSERVATIONS AND DATA REDUCTION cm−2 forArk564 (Kalberla et al.2005).ForMrk335there areclearresidualsataround∼6keVsignifyingthepresence For the analysis, we use all the available archival data of ofabroad Ironline whichwemodelled usingthe“diskline” Mrk 335 and Ark 564 from the XMM-Newton observa- (Fabian et al. 1989). The inner radius of the line emitting tory(Jansen et al.2001).Thelistofobservationsareshown region Rin is fixed at 6rg and the outer radius rout is fixed in Table 1. The XMM-Newton data have been processed at1000rg.Theinclinationofthediskisfixedat40◦.Forthe in the standard way using the SAS version 12.0. For all “nthcomp” model we find that the results are not sensitive cases we have considered data from the EPIC-pn camera totheinputseedphotonswhichwefixtobeablackbodyat (Struderet al. 2001) only. 0.05keV.Forboththesources,additionof2or3lowenergy The data were cleaned for high background flares and absorptionedgesimprovesthespectralfitting.Foroneofthe were selected using the conditions PATTERN ≦ 4 and Mrk 335 observations there is also a hint of a narrow emis- FLAG==0. Wehavecheckedfor pile-up in all cases using sionlineat7keV.Figure1presentstheunfoldedspectrafor SAStaskepatplot andfoundthatforboththesourcessome the observations 0101040101, 0306870101, 0600540501 and observations are affected by pile-up (Table 1). We reduced 0600540601ofMrk335.Thebestfitspectralparametersand thepile-upbyexcludingtheinnermostsourceemissionusing the reduced χ2 for all the observations are listed in Tables an annular region as considered by (Legg et al. 2012). Ex- 2 and 3. ceptforthepile-upaffectedobservations,thesourcespectra The reduced χ2 of the fits to the observations range have been extracted from circular regions of radius 35 arc from 1 to 1.6, which suggests that the underlying spectra scentred on themaximum source emission. Theredistribu- may be more complex than the phenomenological model tion matrices and auxiliary response files were created by used here, especially at low energies. Howeverfor themoti- theSAStaskespecget. Spectraweregroupedsuchthateach vation of this work, the phenomenological model used here bin contained at least 30 counts. Spectral fits to the data is adequate since even for a reduced χ2 ∼ 1.6, the model 4 R. Sarma et al. 0.02 keV (Photons cm s keV)2−2−1−1 5×01.00−13 keV (Photons cm s keV)2−2−1−1 00..0012 2 4 2 χ 0 χ 0 −2 −2 0.5 1 2 5 0.5 1 2 5 Energy (keV) Energy (keV) rmisra 30−Oct−2014 18:09 rmisra 30−Oct−2014 18:20 5×10−3 keV (Photons cm s keV)2−2−1−1 25××1100−−33 keV (Photons cm s keV)2−2−1−1 2×10−3 10−3 4 2 2 χ 0 χ 0 −2 −2 −4 −4 0.5 1 2 5 0.5 1 2 5 Energy (keV) Energy (keV) rmisra 30−Oct−2014 18:03 rmisra 30−Oct−2014 18:23 Figure 1. EPIC PN unfolded spectra and residuals (∆χ) of Mrk 335 observations i.e. 0101040101 (top right), 0306870101 (top left), 0600540501 (bottom right)and0600540601 (bottom left)forthephenomenological modelwithbestfitparameterslistedinTable2. 35 c e 4 s / 30 s t n u 3 o C 25 e t a 2 R 20 Mrk335(0306870101) 1 0 5x104 1x105 2x105 Time 7 7 Mrk335(0600540601) Mrk335(0600540501) ec 6 ec 6 4 s s s/ 5 4 s/ t t 5 n n 3 u 4 u o 3 o C C 4 3 2 e 2 e t t a a 3 R 2 R 1 1 2 1 0 5x104 1x105 2x105 0 2x104 4x104 6x104 8x104 Time Time Figure 2.EPICPNlightcurve(in400sbins)ofMrk335. Thelightcurves wereextracted intheenergyrangeof(0.3-10.0)keV. The timeranges forflux resolvedspectroscopy have been selected corresponding to differentcounts s −1 as shown bythe horizontal dotted lines. Variation of X-ray spectral index with X-ray Eddington ratio 5 50 70 40 70 Ark564(0670130701) Ark564(0206400101) Ark564(0006810301) Ark564(0006810101) 60 40 3 4 3 3 nts/sec nts/sec 50 3 nts/sec 35 nts/sec 60 Cou 30 2 Cou 40 2 Cou 2 Cou 2 ate ate 30 ate 30 ate 50 R 20 R R R 1 1 20 1 1 10 10 25 40 0 2x104 4x104 6x104 0 2x1044x1046x1048x1041x105 0 5x103 1x104 2x104 0 5x103 1x104 2x104 Time Time Time Time 70 60 100 90 Ark564(0670130301) Ark564(0670130901) Ark564(0670130201) 60 3 50 3 80 3 80 3 ec ec ec ec ounts/s 50 ounts/s 40 2 ounts/s 60 2 ounts/s 70 2 e C 40 2 e C e C e C 60 at at at at R R 30 R 40 1 R 30 1 50 1 1 20 20 Ark564(0670130801) 20 40 0 1x1042x1043x1044x1045x1046x104 0 2x104 4x104 6x104 0 2x104 4x104 6x104 0 2x104 4x104 6x104 Time Time Time Time 80 70 70 Ark564(0670130501) Ark564(0670130401) ate Counts/sec 45670000 32 ate Counts/sec 5600 32 ate Counts/sec 456000 23 R R 40 R 301 30 Ark564(0670130601) 1 1 20 30 20 0 2x104 4x104 6x104 0 2x104 4x104 6x104 0 2x104 4x104 6x104 Time Time Time Figure 3.EPIC PNlightcurve (in400s bins)ofArk564. The lightcurves wereextracted inthe energyrange of(0.3-10.0) keV. The timeranges forflux resolvedspectroscopy have been selected corresponding to differentcounts s −1 as shown bythe horizontal dotted lines. Table 2.Spectral parametersforMrk335derivedfromdifferentXMM-Newton observationsinthe(0.3-10.0)keV range. Model parameter 0101040101 0306870101 0510010701 0600540601 0600540501 zedge1 Ec (keV) 0.63(froze) 0.361+−00..000034 1.047+−00..003325 0.730+−00..000043 0.730+−00..000044 τ 0.097+0.028 0.228+0.027 0.369+0.091 0.776+0.027 0.553+0.029 −0.031 −0.006 −0.093 −0.042 −0.017 zedge2 Ec (keV) 1.127+−00..004426 1.120+−00..023340 1.509+−00..006603 0.943+−00..001112 0.922+−00..001141 τ 0.063+0.031 0.026+0.006 0.377+0.101 0.313+0.030 0.238+0.024 −0.031 −0.006 −0.095 −0.041 −0.040 zedge3 Ec (keV) - - 0.686(Froze) - - τ - - 0.524+0.104 - - −0.100 Diskline† Ec (keV) 6.313+−00..316824 6.275+−00..003421 6.009+−00..005668 6.022+−00..005601 5.952+−00..007604 β −2.609+0.749 −1.326+0.643 −9.668+3.607 −7.416+2.297 −5.061+1.497 −5.900 −0.414 −8.006 −4.753 −4.475 Norm(×10−4) 0.449+0.254 0.217+0.044 0.549+0.074 0.276+0.034 0.292+0.042 −0.168 −0.038 −0.076 −0.033 −0.022 zgaussian Ec - 7.004+−00..003301 - - - Norm(×10−3) - 0.007+0.002 - - - −0.002 Simpl Γ 2.094+0.024 1.885+0.009 1.064+0.073 1.676+0.018 1.787+0.018 −0.049 −0.005 −0.046 −0.007 −0.008 FracScat 0.136+0.011 0.087+0.002 0.310+0.324 0.233+0.012 0.193+0.007 −0.009 −0.001 −0.104 −0.007 −0.005 nthComp Γ 2.862+0.052 3.277+0.008 1.980+0.094 1.922+0.087 2.043+0.069 −0.057 −0.023 −0.092 −0.213 −0.219 kT(keV) 0.266+0.051 0.823+0.019 0.210+0.019 0.153+0.008 0.159+0.008 −0.031 −0.020 −0.019 −0.018 −0.020 Norm(×10−2) 0.428+0.029 0.705+0.007 0.093+0.084 0.060+0.003 0.091+0.006 −0.032 −0.037 −0.039 −0.008 −0.009 reducedχ2/d.o.f. 1.04/144 1.48/163 1.70/130 1.40/161 1.52/159 †Thevaluesofrin androut havebeenfixedat6rg and1000rg respectivelywhiletheinclinationhasbeenfixedat40◦. 6 R. Sarma et al. Table 3.Spectral parametersforArk564derivedfromdifferentXMM-Newton observationsinthe(0.3-10.0)keV range. Model Parameter 0006810101 0670130301 0670130801 0670130901 0670130201 0670130601 zedge1 Ec (keV) 0.508+−00..001135 0.708+−00..000088 0.345+−00..001135 0.723+−00..001100 0.330+−00..001118 0.347+−00..000056 τ 0.111+0.022 0.105+0.012 0.244+0.050 0.093+0.014 0.359+0.274 0.269+0.044 −0.022 −0.010 −0.051 −0.013 −0.044 −0.021 zedge2 Ec (keV) 0.712+−00..001113 1.054+−00..004415 0.713+−00..001026 1.071+−00..005632 0.751+−00..002203 0.714+−00..001112 τ 0.147+0.030 0.033+0.013 0.072+0.013 0.029+0.019 0.045+0.013 0.063+0.012 −0.022 −0.014 −0.010 −0.011 −0.007 −0.012 zedge3 Ec (keV) 1.122+−00..102964 1.231+−00..003389 1.160+−00..001198 1.231+−00..105897 1.217+−00..003331 1.127+−00..002119 τ 0.022+0.023 0.047+0.015 0.086+0.013 0.023+0.020 0.057+0.014 0.072+0.011 −0.022 −0.014 −0.007 −0.018 −0.014 −0.011 Simpl Γ 2.474+0.024 2.513+0.014 2.418+0.014 2.495+0.018 2.448+0.019 2.428+0.015 −0.023 −0.013 −0.008 −0.017 −0.009 −0.014 FracScat 0.193+0.008 0.192+0.004 0.153+0.004 0.194+0.006 0.152+0.005 0.151+0.004 −0.007 −0.005 −0.004 −0.006 −0.005 −0.004 nthComp Γ 2.295+0.052 2.292+0.010 2.676+0.059 2.353+0.016 2.795+0.063 2.862+0.051 −0.047 −0.019 −0.050 −0.022 −0.055 −0.044 kT(keV) 0.184+0.020 0.195+0.009 0.220+0.027 0.192+0.011 0.240+0.032 0.254+0.031 −0.020 −0.008 −0.020 −0.010 −0.023 −0.022 Norm 0.014+0.0008 0.012+0.0003 0.011+0.0004 0.013+0.0004 0.015+0.0005 0.011+0.0004 −0.0009 −0.0003 −0.0003 −0.0003 −0.0004 −0.0003 redχ2/d.o.f. 1.03/149 1.58/161 1.62/160 1.17/158 1.03/159 1.30/164 Model Parameter 0670130501 0006810301 0206400101 0670130701 0670130401 zedge1 Ec (keV) 0.346+−00..000067 0.338+−00..000067 0.513+−00..000066 0.587+−00..004357 0.341+−00..000188 τ 0.406+0.039 0.595+0.226 0.093+0.008 0.035+0.019 0.303+0.041 −0.044 −0.226 −0.009 −0.020 −0.041 zedge2 Ec (keV) 0.718+−00..001029 0.489+−00..001122 0.706+−00..000057 0.716+−00..001131 0.710+−00..000095 τ 0.069+0.011 0.188+0.024 0.116+0.011 0.126+0.017 0.075+0.012 −0.011 −0.017 −0.012 −0.024 −0.012 zedge3 Ec (keV) 1.197+−00..002230 0.710+−00..001157 1.111+−00..001189 1.143+−00..002237 1.164+−00..003302 τ 0.068+0.012 0.102+0.017 0.054+0.009 0.061+0.014 0.046+0.011 −0.012 −0.016 −0.009 −0.014 −0.011 Simpl Γ 2.437+0.011 2.376+0.009 2.472+0.009 2.436+0.016 2.416+0.012 −0.014 −0.012 −0.004 −0.016 −0.015 FracScat 0.144+0.004 0.085+0.015 0.183+0.003 0.183+0.004 0.159+0.005 −0.004 −0.015 −0.003 −0.005 −0.004 nthComp Γ 2.892+0.029 3.127+0.024 2.381+0.019 2.175+0.022 2.748+0.049 −0.029 −0.021 −0.022 −0.027 −0.051 kT(keV) 0.250+0.032 0.240+0.006 0.188+0.009 0.191+0.009 0.233+0.032 −0.023 −0.004 −0.008 −0.008 −0.022 Norm 0.012+0.0004 0.009+0.00003 0.010+0.0001 0.009+0.0002 0.011+0.0003 −0.0002 −0.0001 −0.0003 −0.0003 −0.0003 redχ2/d.o.f. 1.35/163 1.16/148 1.62/165 1.51/163 1.49/164 described the data well at a few percentage level and most 0600540601 and 0600540501 were 12.7, 22.0, 3.1, 9.0 and of the discrepancies are at low energies. The high energy 12.0 respectively. More importantly, the best fit high en- photon index is not too sensitive to the actual model used ergyspectralindexesobtainedwere2.05±0.05,1.78±0.06, asweshowlaterwhenwefitonlythehighenergypart(3-10 1.75±0.05 and 1.81±0.02. When compared with the high keV)of thespectraand obtain qualitatively similar results. energyspectralindexesobtainedfromthephenomenological Nevertheless, we caution against over-interpretation of the model,onecanseethattheerrorsarelargerandanychange best fitvaluesobtained inthesefits,especially for thecom- in the index is less than 0.1. In contrast, for the deep low ponentsaffecting thelow energy part of thespectra. flux state, ID 0510010701, the ∆χ2 = 45 and the spectral indexobtainedwas1.55±0.15whichissignificantlydifferent from thephenomenological model. 4 THE EFFECT OF COMPLEX ABSORPTION Next we consider the possible effect of a complex rel- AND REFLECTION COMPONENT ativistically blurred reflection component in the spectra. Since,theionised reflection component producesacomplex ThespectraofMrk335,maybeaffectedbycomplexab- soft excess which needs tobe modelled with absorption, we sorptionandtheremaybearelativisticallyblurredreflection limit our analysis to energies > 2 keV. This is adequate component which may dominate at low and high energies. since our interest here is to study the effect of the compo- Given the quality of the data many of these complex mod- nentonthehighenergyindex.Sinceforthephenomenolog- els may be degenerate, however in the present context it is ical model, the spectra shows the presence of a narrow and usefultoquantifytheeffectof thesemodelson thehighen- broad Iron lines, we consider a power-law and two reflec- ergy spectral index. The motivation here is not to obtain tion components represented by the table model “reflionx” a physically self consistent model but rather to understand (Ross & Fabian 2005). For one of the the reflection com- their effect of the high energy spectral index and hence in ponent we convolve it using the relativistic blurring model theprimary result of this work. “kdblur” where we fix the inner and outer radii, but al- To take into account the possibility of complex ab- low for the emissivity index to be free. The best fit power- sorption,weincludeinthephenomenologicalmodel,partial law index for thefour high fluxstate observations with IDs ionisedabsorberrepresentedbytheXSPECroutine“zxipcf” 0101040101,0306870101, 0600540601 and 0600540501 were (Miller et al.2007).Theabsorptionischaracterisedbythree 2.09±0.05,2.16±0.02,1.77±0.03and1.64±0.05.Wenote parametersnamelythecolumndensity,thecoveringfraction that for theobservation with ID.0306870101 thechange in and the ionisation parameter of the absorber. For the ad- spectral index as compared to the phenomenological model dition of three parameters the changes in χ2 for the four is ∼0.28 while for the others it is <0.15. For the deep low high flux observations with IDs 0101040101,0306870101, flux state, ID 0510010701, the spectral index obtained was Variation of X-ray spectral index with X-ray Eddington ratio 7 2.6 ID. 0306870101, which incidentally is also the highest flux OBS-ID0306870101 OBS-ID0600540601 state in the sample. OBS-ID0600540501 2.4 OBS-ID0101040101 5 FLUX-RESOLVED SPECTROSCOPY 2.2 The light curves for each of the observations of Mrk 335 and Ark 564 are shown in Figure 2 and Figure 3 which re- 2 Γw vealsthatfornearlyallobservationsthereissignificantflux wer-la 1.8 variation in timescales of ∼104 s. To investigate the varia- Po tion of the photon index, each observation was split into 3 1.6 or 4 flux levels (marked by horizontal dotted lines in Fig- 1.4 ures 2 and 3) and the corresponding spectrum was gener- ated. We do not consider the 2007 observation of Mrk 335 1.2 (ID0510010701) for the flux resolved spectroscopy since as shownintheprevioussectionthespectraisclearlycomplex 1 0.01 0.1 andthephenomenological modelusedhereisnotadequate. Lx/Ledd Each of the flux level spectra was fitted with the same Figure4. Thehighenergyphotonindex,ΓversustheX-rayEd- phenomenologicalmodelusedfortheaveragespectra.How- dingtonratioforMrk335.TheX-rayEddingtonratioisLX/LEdd ever, due to the lower statistics of the flux resolved spec- whereLX istheunabsorbedluminosityinthe0.3-10keVrange. tra, there were several parameters which were either not The solid line is a fit to three of the data sets and has a slope constrained or were consistent within error bars to be not m = 0.64±0.04 and intercept c = 3.08±0.08 with a reduced varying during the observations. Thus, several spectral pa- χ2 = 0.57. The bottom dotted straight line is a fit only to the rameters were fixed to their best fit values obtained from data set (ID0306870101) and has a slope m = 0.65±0.04 and interceptc=2.76±0.07withareducedχ2=0.76. the fitting of the average spectra. These were, the energies of the different edges and emission lines, the temperature and normalisation of the soft photon Comptonization (i.e. kT and normalisation of the “nthcomp” component) and bb the normalisation and emissivity index of the Iron line (i.e. β and normalisation of the“diskline” component). 2.3 OBS-ID0306870101 OBS-ID0600540601 For each flux resolved spectra, the unabsorbed flux in 2.2 OOBBSS--IIDD00610001504400510011 the0.3-10keVbandwascomputedusingtheXSPECmodel 2.1 “cflux”. The fluxes were converted into luminosities, L by using theluminosity distances of 103 Mpc for Mrk 335 and 2 98.5 Mpc for Ark 564, which are the average values quoted Γw 1.9 in the NED website. For Mrk 335, the black hole mass was Power-la 1.8 2ta0k1e2n)wtohbileef1o.4rA×r1k0756M4⊙we(Paedtoeprstoanveatluael.o2f020.46;1G×rie1r06etMa⊙l. 1.7 (Botte et al. 2004) to obtain their respective Eddington lu- minositiesL .Thebestfitparametersforthefluxresolved 1.6 Edd spectra are listed in Table 4. 1.5 Inthiswork weconsidertheX-rayEddingtonratio i.e. 1.4 LX/LEdd insteadoftheBolometric oneanddiscusstheim- 0.001 0.01 plication of this in the last section. In Figure 4, the high LHX/Ledd energy photon index is plotted against the X-ray Edding- Figure 5. Same as Figure4 except that only the energy range ton ratio for the different flux resolved spectra. Three of 3-10 keV is taken into account. The spectra are fitted with a the observations cover an order of magnitude range in X- power-lawandIronline.TheluminosityLHX correspondstothe ray Eddington ratio from 0.004 to 0.04 and show a tight energy range 3-10 keV. Note the similaritywith Figure 4 which correlation between thephoton indexand X-rayEddington impliesthatthequalitativeresultsarenotsensitivetothespectral ratio.Astraightlinefittothesethreeobservationsetsgives modeladopted. a slope m = 0.64±0.04, intercept c = 3.08±0.08 and re- duced χ2 = 0.57. The observation with the highest X-ray Eddington ratio >0.04, does not follow the linear relation. Insteadfittingthefluxresolvedspectraforthatobservation only, one gets a straight line with slope m = 0.65±0.04, 3.00±.07whichisradicallydifferentthanthevalueobtained intercept c = 2.76±0.07 and reduced χ2 = 0.76. In other earlier. words,thedatafollowsaparalleltrack.Notethatthelowest Thus, the effect of complex absorption or reflection is luminosityobservation(LX/LEdd ∼0.002)wasnotincluded dramatic for the deep low state. This is expected based on in thestraight line fit. Nevertheless, the fit to thehigh flux theanalysisofGrupeet al.(2008).Fortheotherhigherflux data, passes through thelowest flux one. states, complex absorption may change the index by < 0.1 To verify that the results do not depend sensitively on from those of the phenomenological model. The effect of the phenomenological spectral model used, we repeat the Complex reflection on the index is also modest < 0.15 but flux resolved spectroscopy analysis using only the energy wenotethatitmaybelarge∼0.28fortheobservationwith band3-10keV.Wefitthisenergybandwithapower-lawand 8 R. Sarma et al. 6 DISCUSSION 2.6 ID0006810301 ID0670130801 ID0006810101 ID0670130201 Figure7summarises theresultsoftheworkbyshowing the 2.55 ID0670130901 ID0206400101 variation of the high energy photon index versus the X-ray ID0670130701 IIDD00667700113300530011 EddingtonratioforbothMrk335andArk564.ForMrk335, 2.5 there are two parallel tracks with the possibility that the Γw sources shifts to the lower track for LX/LEdd > 0.04. For wer-la 2.45 Ark 564 the points are clustered at a larger Γ and while Po thereisapositivecorrelation,theslopeisflatterandpoints 2.4 show much more scatter than thepoints for Mrk 335. Itshouldbenotedthattheluminosityusedinthiswork 2.35 is in the energy range 0.3-10 keV and is not the bolomet- riconeandhencetheseresultscannotbecompareddirectly with those where the bolometric luminosity or X-ray lumi- 2.3 0.01 0.1 nosity in some other energy range has been used. The con- Lx/Ledd version to Bolometric luminosity often involves several un- Figure6.Thehighenergyphotonindex,ΓversustheX-rayEd- certain factors and hence is avoided in this work. Since for dingtonratioforArk564.TheX-rayEddingtonratioisLX/LEdd Mrk335,wefindthetheindexcorrelateswellwiththeX-ray whereLX istheunabsorbedluminosityinthe0.3-10keVrange. luminosityfornearlyanorderofmagnitudechangeinlumi- nosity,thismayindicatethatoverthisrangethebolometric correctionisnearlyconstant.Ontheotherhand,itmaywell 2.6 Mrk335 bethatforthissourcetheX-rayindexcorrelatesbetterwith Ark564 2.4 theX-rayluminosity and not with thebolometric one. The uncertaintiesandmodeldependencyofestimatingthebolo- 2.2 metric luminosity (including the general non-availability of simultaneousmulti-wavelengthdata)doesnotallowforcon- 2 Γw cretestatements.ForMrk335,thereisonlyoneobservation wer-la 1.8 forwhichtheX-rayEddingtonratioislarger than0.04 and Po the source follows a lower parallel track. It is necessary to 1.6 confirm this behaviour with future observations when the source is equally luminous. It might be that there is a real 1.4 transition at the X-ray Eddington ratio ∼ 0.04 or that the 1.2 sourcecanactuallyfollowanyofthetwotracksatthesame luminosity. 1 0.01 0.1 IntheComptonization context,thehighenergypower- Lx/Ledd law index is inversely proportional to the Compton Ampli- Figure 7. Summary figure for the high energy photon in- fication factor A, which is the ratio of the luminosity of dex, Γ versus the X-ray Eddington ratio for both Mrk 335 and Comptonizingcloud,Lc totheinputsoftphotonluminosity Ark 564. The X-ray Eddington ratio is LX/LEdd where LX is Linp i.e. A = Lc/Linp. The input luminosity Linp depends theunabsorbedluminosityinthe0.3-10keVrange.ForMrk335 ontheluminosityofthesoftphotonsourceandthefraction two parallel tracks are observed with a possible switching at of photons which enter the Comptonizing region. The lat- LX/LEdd ∼ 0.04, while for Ark 564 the correlation is flatter terdependson theaccretion geometryof thesystem.Thus, andwithsignificantlymorescatter. the correlation between the photon index and the observed luminosity could occur if A decreases with luminosity or in abroadIronline.Figure5showstheresultsofthisanalysis other words Lc varies less rapidly with the observed lumi- wherethephotonindexisplottedagainstLHX/LEdd,where nosity than Linp. The shift to the lower parallel track can LHX iscomputedusingtheunabsorbedfluxinthe3-10keV be explained if there is a decrease in Linp at the X-ray Ed- band. dington ratio of ∼0.04, perhaps caused by a change in the Figure 6 shows the results of the flux resolved analy- fraction of photons entering the Comptonizing region. This sis (for the complete 0.3-10 keV band) for Ark 564. Here, wouldmeanthattheaccretiongeometryforthetwoparallel despite the larger number of observations, the flux varia- tracksare different. tionismodestwiththeX-rayEddingtonratiorangingfrom Thehighenergyphotonindexcouldalsobeaffectedby 0.02-0.07. Although there is more scatter than the case of the presence of a strong reflection component, especially if Mrk 335, there is a correlation between the spectral index the reflection is from partially ionised matter and/or is rel- and X-ray Eddington ratio. A straight line fit gives a slope ativistically blurred such that it has a significant contribu- m=0.22±0.08,interceptc=2.75±0.1andalargereduced tiontothespectrabelow10keV.Thereflectioncomponent χ2 = 2.75. From the Figure, it seems that for the flux re- would tend to flatten the high energy spectrum and hence solvedspectraofeachindividualobservationthecorrelation the correlation may be caused by the reflection component may be better and perhaps the different observations are decreasing as the source becomes more luminous. Indeed, parallelly shifted with respect to each other. However, the theanomalouslowluminosityobservationofMrk335ofJuly statistics is not good enough to make any concrete state- 2007, can be modelled as being dominated by a blurred re- ments. flectioncomponent(Grupeet al.2008).Itmayalsobethat Variation of X-ray spectral index with X-ray Eddington ratio 9 the primary correlation between the index and luminosity Collin S., Kawaguchi T., Peterson B. M., Vestergaard M., is due to variation of the Compton Amplification factor, 2006, A&A,456, 75 whiletheshift toadifferentparallel trackisduetoappear- ConstantinA.,GreenP.,AldcroftT.etal.,2009,ApJ,705, ance of astrong reflection component.The appearance of a 1336 stronger reflection component would also mean that there Done C., Madejski G. M. & Z˙ycki P. T. 2000, ApJ, 536, was a change in the accretion disk geometry. Thus, a qual- 213 itative change in the accretion disk geometry seems to be Dewangan G. C., Boller th., Singh K.P. & Leighly K.M., required to explain the parallel tracks seen for Mrk 335. 2002, A&A,390, 65 Changes in the accretion disk geometry at the same X-ray Dewangan G. C., Griffiths R. E., Dasgupta S. & Rao, A. luminosity seems to indicate that the accretion flow is not R.,2007, ApJ,671, 1284 determinedonlybythelocalaccretionratebutratheritmay Fabian A.C., Rees M . J., Stella L. & White N.E., 1989, also depend on the previous history of the accretion rate MNRAS,238, 729 variation. Our brief analysis of including a blurred reflec- Gu, M. & Cao, X., 2009, MNRAS,399, 349 tion component suggests that this indeed may be the case. Gallo L. C., Boller T., Tanaka Y., Fabian A. C., Brandt Itisparticularlyinterestingtonotethatforthehighestflux W.N.,WelshW.F.,AnabukiN.,HabaY.,2004,MNRAS, observation (ID 101040101) a reflection component may in- 347, 269 crease the index by 0.28 while a more modest variation is Gallo L.C. et al.,2013, MNRAS,428, 1191 expected for the others (Section 4). If indeed this highest Grier et al., 2012, ApJ,744, 4 fluxobservationhasastrongerreflectioncomponentandits Grupe D., 2004, AJ, 127, 1799 spectral indexis more close to ∼2 then it would seem that Grupe D., Komossa S., Gallo L.C., Fabian A., Larsson J., the index has a positive correlation with flux but saturates PradhanA.K.,XuD.&MiniuttiG.,2008a,ApJ,681,982 to ∼2 at high Eddington ratios. Interestingly a similar be- havior is observed in Fig. 5, where the results for a simple GrupeD.,KomossaS.,GalloL.C.,LonginottiA.L.,Fabian power law plus Iron line fit to the 3-10 keV energy range A.,PradhanA.K.,GruberbauerM.&XuD.APJS,2012, areshown. However,thiscan beconfirmed only with broad 199, 17 band data covering energies > 10 keV where the reflection Gondoin P., Orr A., Lumb D. & Santos-Lleo M., 2002, component will bebetter constrained. A&A,388, 74 The results for Ark 564 are significantly different than Haardt F. & Maraschi L. 1991, ApJ,380, 51 that for Mrk 335, with a flatter correlation and more scat- Haardt F. & Maraschi L. 1993, ApJ,413, 507 ter.ThissuggeststhatthebehaviourofAGNsisnothomo- Huchra J. P., Vogeley M. S. & Geller M. J. 1999, ApJS, geneous and perhaps cannot be understood by studying a 121, 287 large sample of them. Long term sensitive and broad band Jansen F. et al. 2001, A&A,365, 1 monitoringof individualAGNsmaybeexpectedtoprovide Jin C., Ward M. & DoneC., 2012, MNRAS,425, 907 betterinsights. KalberlaP.M.W.,BurtonW.B.,HartmannDap,ArnalE. M.,Bajaja E.,MorrasR.W.&PppelW.G.,2005, A&A, 440, 775 ACKNOWLEDGMENTS Komossa S., 2008, RMxAC,32, 86 Komossa S., Xu D., 2007, ApJ, 667, L33 RSacknowledgesIUCAAvisitorprogram andAPacknowl- Larsson J., Miniutti G., Fabian A. C., Miller J. M., edges seed money grant from Tezpur University and visit- Reynolds,C.S. & Ponti G., 2008, MNRAS,384, 1316 ing associateship IUCAA, Pune. This research made use of Laor A., Fiore F., Elvis M., Wilkes B. J. & McDowell J. data obtained from the High Energy Astrophysics Science C., 1997, ApJ,477, 93 ArchiveResearchCenter(HEASARC),providedbyNASA’s Legg E., Miller L., TurnerT. J., GiustiniM., ReevesJ.N. Goddard Space Flight Center. The authors would like to & Kraemer S. B., 2012, ApJ,760, 73 thank the anonymous referee for useful suggestions which Leighly K. M., 1999, ApJS,125, 317 has enhanced thequality of thepaper. LiuJ.Y.,LiuB.F.,QiaoE.L.&MineshigeS.,2012,ApJ, 754, 81 Longinotti A. L., Sim S. A., Nandra K. & Cappi M.,2007, MNRAS,374, 237 REFERENCES Lu Y. & Yu Q., 1999, ApJL, 526, 5 ArnaudK.,ASPConf.Ser.101:Astronomical DataAnalysis Mushotzky R.M., DoneC., PoundsK. A., 1993, ARA&A Software and Systems, 101, 1996, 17 31, 717 Barth A. J., Greene J. E., Ho L. C., 2005, ApJ, 619, L151 MillerL.,TurnerT.J.,ReevesJ.N.,GeorgeI.M.,Kraemer Boller T., Brandt W. N.& Fink H., 1996, A&A,305, 53 S.B., Wingert B., 2007, A&A,463, 131 Boroson T. A., 2002, ApJ, 565, 78 NandraK. & Pounds K.A., 1994, MNRAS,268, 405 Brandt W.N.,MathurS.&ElvisM.,1997, MNRAS,285, NandraK.,LeT.,GeorgeI.M.,EdelsonR.A.,Mushotzky 15 R.F.,PetersonB.M.&TurnerT.J.,2000,ApJ,544,734 Brightman M. et al. 2013, MNRAS,433, 2485 NandraK., & Papadakis I. E., 2001, ApJ,554, 710 Botte V., Ciroi S., Rafanelli P. & Mille Dr. F., 2004, AJ, Osterbrock D.E., Pogge R.W., 1985, ApJ, 297, 166 127, 3168 O’Neill P. M., Nandra K., Cappi M., Longinotti A. L. & Cao X.-W. 2009, MNRAS,394, 207 Sim S.A., 2007, MNRAS 381, 94 Chiang J. et al., 2000, ApJ, 528, 292 PageK.L.,ReevesJ.N.,O’BrienP.T.&TurnerM.J.L., 10 R. Sarma et al. 2005, MNRAS,364, 195 Papadakis I. E., Brinkmann W., Page M. J., McHardy I. & Uttley,P. 2007, A&A,461, 931 Perola G. C. et al., 1986, ApJ, 306, 508 Peterson B.M., et al., 2004, ApJ, 613, 682 PorquetD.,ReevesJ.N.,OBrienP.&BrinkmannW.2004, A&A,422, 85 Risaliti G., YoungM. & Elvis M., 2009, ApJ, 700, 6 RomanoP.,MathurS.,TurnerT.J.etal. 2004, ApJ,602, 635 Ross R. R., Fabian A. C., 2005, MNRAS,358, 211 Saez C., Chartas G., Brandt W. N., Lehmer B. D., Bauer F. E., Dai X. & Garmire G. P. 2008, AJ, 135, 1505 Singh K. P., Rao A. R. & Vahia M. N., 1991, A&A, 248, 37 Struder, L., Briel U., Dennerl K., et al., 2001, A&A, 365, 18 SteinerJ.F.,NarayanR.,McClintockJ.E.,&EbisawaK., 2008, PASP,121, 1279 Shemmer O., Romano P., Bertram R., et al. 2001, ApJ, 561, 162 ShemmerO.,BrandtW.N.,NetzerH.,MaiolinoR.,Kaspi S., 2006, ApJL, 646, 29 Sobolewska M. A. & Papadakis I. E. 2009, MNRAS, 399, 1597 Tananbaum H et al., 1978, ApJ, 223, 74 Turner T. J., Romano P., George I. M., et al. 2001, ApJ, 561, 131 VignaliC.,BrandtW.N.,BollerT.,FabianA.C.,Vaughan S., 2004, MNRAS,347, 854 Vaughan S.& Edelson R., 2001, ApJ, 548, 694 V´eron-CettyM.-P.,V´eronP.,Gonc¸alvesA.C.,2001,A&A, 372, 730 Warner C., Hamann F., Dietrich M., 2004, ApJ, 608, 136 Wang J. M., Watarai K. Y. & Mineshige S., 2004, ApJ, 607, 107 Williams R. J., Pogge R. W., Mathur S., 2002, AJ, 124, 3042 Williams R. J., Mathur S., Pogge R. W., 2004, ApJ, 610, 737 Xu Ya-Di,2011, ApJ,729, 10 Xu D. W., Komossa S., Wei J. Y., Qian Y., Zheng X. Z., 2003, ApJ, 590, 73 Yaqoob T. & Warwick R. S., 1991, MNRAS,248, 773 Younes G., Porquet D., Sabra B. & Reeves J. N., 2011, A&A,530, 149 ZdziarskiA.A.,PoutanenJ.&JohnsonW.N.,2000,ApJ, 542, 703 Zdziarski A.A & Grandi P., 2001, ApJ,551, 186 Zhou Xin-Lin& Zhao Yong-Heng,2010, ApJ, 720, 206