The structure of AGNs from X-ray absorption variability
aa r X i v : . [ a s t r o - ph . C O ] D ec Co-Evolution of Central Black Holes and GalaxiesProceedings IAU Symposium No. 267, 2009B.M. Peterson, R.S. Somerville, & T. Storchi-Bergmann, eds. c (cid:13) The structure of AGNs from X-rayabsorption variability
Guido Risaliti , INAF - Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, Firenze, Italyemail: [email protected] Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA
Abstract.
We present new evidence of X-ray absorption variability on time scales from a fewhours to a few days for several nearby bright AGNs. The observed N H variations imply thatthe X-ray absorber is made of clouds eclipsing the X-ray source with velocities in excess of10 km s − , and densities, sizes and distances from the central black hole typical of BLR clouds.We conclude that the variable X-ray absorption is due to the same clouds emitting the broademission lines in the optical/UV. We then concentrate on the two highest signal-to-noise spectraof eclipses, discovered in two long observations of NGC 1365 and Mrk 766, and we show thatthe obscuring clouds have a cometary shape, with a high density head followed by a tail withdecreasing N H . Our results show that X-ray time resolved spectroscopy can be a powerful wayto directly measure the physical and geometrical properties of BLR clouds. Keywords. galaxies: active; galaxies: Seyfert; X-rays: galaxies; X-ray: invividual (NGC 1365,Mrk 766)
1. Introduction
X-ray absorption variability is a common feature in Active Galactic Nuclei (AGN).An analysis of a sample of nearby X-ray obscured AGN with multiple X-ray observa-tions, performed a few years ago (Risaliti et al. 2002) revealed that column density (N H )variations are almost ubiquitous in local Seyfert galaxies. More recent observations per-formed with XMM-Newton , Chandra and
Suzaku further confirmed this finding. Thephysical implications of these measurements are that the circumnuclear X-ray absorber(or, at least, one component of it) must be clumpy, and located at sub-parsec distancesfrom the central source. The comparison between different observations, typically per-formed at time distances of months-years, only provides upper limits to the intrinsictime scales of N H variations. An improvement of these estimates can only be obtainedthrough observational campaigns within a few weeks/days, and/or through the search forN H variations within single long observations. Such short time-scale studies have beenalready performed for a handful sources: NGC 1365 (Risaliti et al. 2005, 2007, 2009),NGC 4388 (Elvis et al. 2004), NGC 4151 (Puccetti et al. 2007). In particular, in the caseof the AGN in NGC 1365 we revealed extreme spectral changes, from Compton-thin (N H in the range 10 cm − ) to reflection-dominated (N H > cm − ) in time scales from acouple of days to ∼
10 hours. Such rapid events imply that the absorption is due to cloudswith velocity v > km s − , at distances of the order of 10 gravitational radii (assum-ing that they are moving with Keplerian velocity around the central black hole). Thephysical size and density of the clouds are of the order of 10 cm and 10 -10 cm − ,respectively. † All these physical parameters are typical for Broad Line Region (BLR) † For a detailed derivation and discussion of these parameters, we refer to Risaliti et al. 2009.A rough estimate is also mentioned here in Section 2.
Figure 1.
Four examples of new N H variations on time scales of a few days discovered withcampaigns of snapshot observations. clouds, strongly suggesting that the X-ray absorber and the clouds responsible for broademission lines in the optical/UV are one and the same.Here we present more evidence, based on time resolved X-ray spectroscopy, of variableX-ray absorption due to BLR clouds, and we show how in the best cases we can useX-ray observations to directly probe the nature and physical state of BLR clouds.
2. X-ray absorption from BLR clouds in AGNs
X-ray absorption due to BLR clouds crossing the line of sight to the observer impliesvariability time scales from hours to weeks, depending of the physical size of the X-raysource. We can obtain a rough estimate of the occultation times assuming a cloud movingwith Keplerian velocity around the central black hole, and a size of the X-ray source of10 R G : T ∼ × M v − , where M =M/(10 M ⊙ ) and v =v/(10 km s − ). Two obser-vational methods have been used to investigate N H variations at these time scales:1) Snapshot observations of the same source, repeated every few days. After the first suc-cessful campaign of six short (10 ks) Chandra observations of NGC 1365, we repeated thisapproach for other bright Compton-thin sources with column densities between 10 and10 cm − , using both archival data and dedicated observations. In this way we discov-ered clear cases of N H variations in NGC 4151 and UGC 4203 (one of the few ”changinglook” sources, known for having been observed in the past in both Compton-thin andreflection-dominated states). A further case of such variations has been discovered byBianchi et al. (2009) in NGC 7582 (Fig. 1).2) Study of the spectral variations of bright sources during long observations. This methodconsists of a two-phase analysis: we first use the hardness-ratio light curve to select the H variability in AGNs Figure 2.
Flux (top) and hardness ratio (bottom) light curves from the
XMM-Newton longobservation of Mrk 766. The observation is made in five consecutive
XMM-Newton orbits. time intervals where strong spectral variations occurred; we then perform a completeanalysis of the spectra obtained from these intervals, in order to measure possible N H varations (and to check if the spectral changes are due to other effects, such as variationsof the slope of the continuum emission).This approach is illustrated in Fig. 2 for the long XMM-Newton observation of Mrk 766.A complete analysis of these data is presented in Turner et al. 2007 and Miller et al. 2007.With respect to these works, our analysis is in many aspects less detailed (though theresults are in full agreement), but is more effective in isolating the effects of possibleN H variations. We note that Mrk 766 is a Narrow Line Seyfert 1, so on average we donot expect to observe complete X-ray absorption of the X-ray source. However, as weshow below, isolated clouds occasionally cross the line of sight, producing measurableabsorption in the X-ray spectrum. The upper panel of Fig. 2 shows the standard 2-10 keV flux light curve for this observation, with the well known strong variability ontime scales of thousands of seconds, or even shorter. The lower panel shows the light curveof the (6-10 keV)/(2-5 keV) flux ratio. In general, this light curve shows much smallervariations, indicating that the continuum shape remains the same during most of theluminosity variations. However, clear exceptions are observed in at least three intervals,highlighted in Fig. 2. During these intervals it is possible that a cloud with N H of theorder of 10 cm − has covered the central source, strongly decreasing the observed fluxin the soft band, without affecting the hard band, and therefore increasing the observedhardness ratio.Following this interpretation, in the first interval we should be observing a cloud un-covering the X-ray source (with the covering phase occurred before the beginning of theobservation); in the second interval another cloud shoud be covering the X-ray source,with the uncovering phase not observed due to the ”dead time” between two consecutive22 G. Risaliti Figure 3.
Results from the spectral analysis of the eclipses observed in Mrk 766. Top two panels:spectra, best fit model and residuals from the third and fourh orbit (Fig. 2), where no spectralchanges are observed. Bottom panels: difference between the spectra in the three intervals withspectral variations (Fig. 2) and the best fit model for the third and fouth orbit.
XMM-Newton orbits. Finally, in the third interval we should be observing the wholeeclipse.In order to check this scenario, we performed a complete analysis of the spectra ob-tained from the three highlighted intervals, and of those obtained from the third andfourth orbit, representing the standard spectral state of the source. In this analysis weallowed all the main spectral parameters of the model to vary among the different inter-vals. The results of the this study, illustrated in Fig. 3 are the following: 1) the 2-10 keVspectrum obtained from the third and fourth orbit (the ”standard” state) is well re-produced by a typical model for type 1 AGNs, consisting of a power law, a reflectioncomponent and an iron emission line; 2) the spectral variations observed in the three in-tervals discussed above are completely reproduced by three absorption components withcolumn densities in the range 1-3 10 cm − .The analysis briefly summarized here has been applied to several sources with longarchival observations by XMM-Newton and
Suzaku . Up to now, we obtained unambiguousevidence for N H variations in NGC 4388, MCG-6-30-15, Mrk 766, NGC 1365. A furthercase of similar variations has been found in a Suzaku observation of NGC 3227 (Liu etal. 2010, subm.). H variability in AGNs NGC 1365 MRK 766
Figure 4.
Detailed analysis of a
Suzaku observation of NGC 1365 (left panel) and of the firsttwo intervals of the Mrk 766 observation highlighted in Fig. 2 (right panel). In each panel, weshow, from top to bottom, the light curves for: flux, hardness ratio (in the case of Mrk 766 thesetwo plots are just a zoom of the light curves shon in Fig. 2), N H , and covering factor. A summary of all the positive detections obtained so far is shown in Table 1. In allcases the inferred physical parameters for the obscuring clouds clearly indicate that theX-ray absorption is due to BLR clouds.
3. Structure of the BLR clouds from X-ray spectroscopy
The analysis presented above can be further refined in the few cases where the statis-tics is high enough to perform a more detailed study, or where multiple occultation eventoccur. In particular, we briefly summarize two results from the analysis of the two so-farbest studied sources, Mrk 766 and NGC 1365.
1) Velocity distribution of BLR clouds.
The analysis of the spectral variations inMrk 766 (Fig. 2-4) directly provides contraints on the distribution of BLR cloud veloci-ties, from the comparison of the durations of the observed occultations. The actual valuesof the cloud velocity in each case cannot be precisely determined, being dependent on theexact size of the X-ray source. However, the ratio between the durations of the differenteclipses is an estimate of the velocity ratios. In particular, the duration of the two firstoccultation events (Fig. 2), is about 4-5 times longer than that of the third eclipse. Thissuggests a similar spread in the velocity of BLR clouds.
2) Structure of BLR clouds.
The model adopted to reproduce the observed occul-tation events consists of an absorption component with a constant (during the eclipse)N H , and a variable covering fraction, C F . This is an extreme simplification of the cloudstructure, which is assumed to have constant column density, and very sharp edges. Themodel can be improved by releasing the assumption of constant N H . Doing so, in mostof the cases analyzed so far we found that it is impossible to significantly constrain the24 G. Risaliti − ∆ χ Energy (keV)5 − ∆ χ Energy (keV) 5 − − ∆ χ Energy (keV)5 − ∆ χ Energy (keV)
EAB D
Figure 5.
Left panel: same as in Fig. 3, for the second orbit of the
XMM-Newton observationof Mrk 766. Right panel: residuals for the intervals indicated in the left panel. The presence ofiron absorption lines suggest the presence of a ionized tail. values of N H and C F during the different phases of an eclipse, due to their strong de-generacy. In other words, it is only possible to study the evolution of one parameter,freezing the other one. The only exceptions found so far, i.e. spectra with enough signal-to-noise to analyze the evolution of the two parameters at the same time, are Mrk 766and NGC 1365.In Fig. 4A we show the results of a recent analysis of a Suzaku observation of NGC 1365(Maiolino et al. 2010, subm), where the hardness ratio light curve clearly shows at leasttwo possible occultation events. We performed a complete spectral analysis of severalshort intervals before, during and after the eclipses, and we obtained the results shownin the two lower panels of Fig. 4, showing the light curves of the covering factor and thecolumn density. The most interesting results come from the second eclipse: the coveringfactor C F increases during the occultation phase, with a constant N H ∼ × cm − .Then, in the subsequent phase, C F remains compatible with 1 (i.e. complete cover-ing) while the column density decreases. This behaviour can only be explained with a”cometary shape” of the obscuring cloud, with a high column density head, and a tailwith decreasing N H .An analogous result has been obtained (though with smaller statistical significance) forthe first two eclipses in Mrk 766 (Fig. 4B). The third eclipse of Mrk 766 is instead too fastto allow a detailed spectral analysis (the spectral counts in such short time intervals arenot enough to break the degeneracy between C F and N H , as shown in Fig. 5). However,a strong indication of a cometary tail comes from the residuals with respect to the bestfit model during and after the eclipse, showing strong absorption lines due to Fe XXVin outflow with a velocity of 10-15000 km s − . These features completely disappear after ∼
40 ks. A complete analysis of these spectra, where we demonstrate the high statisticalsignificance of the lines detections, is presented in a dedicated paper (Risaliti et al. 2010,subm.) The highly ionized component revealed by the absorption lines strongly suggestthe presence of a ionized outflowing tail associated to the obscuring cloud.We note that this highly ionized component is not present in the first two occultation H variability in AGNs Table 1.
List of sources with N H variations on short time scales Name ∆(N H ) a ∆(T) b Method c Ref.NGC 1365 > < ×
10 hours Continuous 3,4NGC 4388 2 ×
15 hours Continuous 1NGC 4151 2 ×
20 hours Continuous 5NGC 4151 10 <
20 hours Snapshot 6Mrk 766 3 ×
10 to 20 hours Continuous 1,7MCG-6-30-15 10
10 hours Continuous 1UGC 4203 3 × <
15 days Snapshot 1NGC 3227 7 × a : N H variations in cm − ; b duration of the observed eclipse; c : Observational method: repeated snapshotobservations, or analysis of long continuous observations. References: 1: this work; 2: Maiolino et al. 2010, subm.;3: Risaliti et al. 2007; 4: Risaliti et al. 2009; 5: Puccetti et al. 2007; 6: Bianchi et al. 2009; 7: Risaliti et al. 2010,subm.; 8: Liu et al. 2010, in prep. events in Mrk 766, but only in the third, much faster event. Even if one case is clearlynot enough to build a complete model, the data suggest a scenario where the obscuringclouds are distributed in a large range of distances from the center (if we assume Keplerianvelocity, the factor ∼ ∼
20 of spreadin distances), and where the clouds closer to the center are also the more ionized ones.Thsi simple scheme is illustrated in Fig. 6.
4. Conclusions, and future work
We have presented new time-resolved spectral studies of several bright AGNs, showingcolumn density variability on time scales for a few hours to a few days. The main resultsof our analysis are:1) Fast (hours-days) column density variability is common among AGNs. This impliesthat the observed variable X-ray absorption is due to clouds with velocities, densities,sizes and distances from the central black hole of the same order of those of BLR clouds.2) In the highest signal-to-noise studies, it is possible to investigate the structure andshape of the single obscuring clouds in detail. This reveals a ”cometary” profile, with ahigh column density head and a tail with decreasing N H .These results show that the X-ray absorption is at least in part due to BLR clouds, andthat X-ray spectroscopy can be a powerful tool to diretly measure the physical propertiesof the broad line region in AGNs.The work presented here is only the first part of an on-going comprehensive analysis ofall the bright AGNs with long high quality X-ray observations. At present, the evidenceof ”common” short time scale N H variations presented here is based on a sparse sam-ple of sources, with no homogeneous selection criteria. The next major step is thereforethe selection and the homogeneous analysis of a representative sample of the local AGNpopulation. This work will provide a quantitative estimate of the occurrence of N H vari-ability on short time scales, and hopefully, will lead to the discovery of more high signalto noise spectra of occultations, such as the ones found in NGC 1365 and Mrk 766, inorder to perform other studied of the physical propoerties of BLR clouds. Acknowledgements. . This work has been partially supported by NASA grants NNX08AX78Gand G08-9107X, and by grant ASII-INAF I/088/06/026 G. Risaliti
Figure 6.
A schematic view of the X-ray absorbing clouds, as estimated from the observationsdescrived here. The eclipsing cloud in NGC 1365 is the one with the best-measured properties,with a column density going from ∼ × cm − in the head, to a few 10 cm − in the tail(Maiolino et al. 2010). In Mrk 766, we suggest that the fast cloud with the ionized tail is muchcloser to the center than the two slower, neautral clouds. The ionized tail in the fast cloud isprobably in outflow (not shown here; a complete analysis is presented in Risaliti et al. 2010). References
Bianchi, S., Piconcelli, E., Chiaberge, M., Bail´on, E. J., Matt, G., & Fiore, F. 2009,
ApJ , 695,781Elvis, M., Risaliti, G., Nicastro, F., Miller, J. M., Fiore, F., & Puccetti, S. 2004,
ApJ , 615, L25Miller, L., Turner, T. J., Reeves, J. N., George, I. M., Kraemer, S. B., & Wingert, B. 2007,
A&A , 463, 131Puccetti, S., Fiore, F., Risaliti, G., Capalbi, M., Elvis, M., & Nicastro, F. 2007,
MNRAS , 377,607Risaliti, G., Elvis, M., & Nicastro, F. 2002,
ApJ , 571, 234Risaliti, G., Elvis, M., Fabbiano, G., Baldi, A., & Zezas, A. 2005,
ApJ , 623, L93Risaliti, G., Elvis, M., Fabbiano, G., Baldi, A., Zezas, A., & Salvati, M. 2007,
ApJ , 659, L111Risaliti, G., et al. 2009, apj , 696, 160Turner, T. J., Miller, L., Reeves, J. N., & Kraemer, S. B. 2007,