Eclipsing the X-ray emitting region in the active galaxy NGC 6814
DD RAFT VERSION J ANUARY
15, 2021Typeset using L A TEX twocolumn style in AASTeX63
Eclipsing the X-ray emitting region in the active galaxy NGC 6814 L UIGI
C. G
ALLO , A DAM
G. G
ONZALEZ , AND J ON M. M
ILLER Department of Astronomy & Physics, Saint Mary’s University, 923 Robie Street, Halifax, Nova Scotia, B3H 3C3, Canada Department of Astronomy, The University of Michigan, 1085 South University Avenue, Ann Arbor, Michigan, 48103, USA (Received xxx yyy zzz; Revised xxx yyy zzz; Accepted xxx yyy zzz)
Submitted to ApJLABSTRACTWe report the detection of a rapid occultation event in the nearby Seyfert galaxy NGC 6814, simultaneouslycaptured in a transient light curve and spectral variability. The intensity and hardness ratio curves capture distinctingress and egress periods that are symmetric in duration. Independent of the selected continuum model, thechanges can be simply described by varying the fraction of the central engine that is covered by transitingobscuring gas. Together, the spectral and timing analyses self-consistently reveal the properties of the obscuringgas, its location to be in the broad line region (BLR), and the size of the X-ray source to be ∼ r g . Our resultsdemonstrate that obscuration close to massive black holes can shape their appearance, and can be harnessed tomeasure the active region that surrounds the event horizon. Keywords: galaxies: active – galaxies: nuclei – galaxies: individual: NGC 6814 – X-rays: galaxies INTRODUCTIONThe innermost region around an accreting supermassiveblack hole produces the bulk of the radiation that defines ac-tive galactic nuclei (AGN). This central engine is unresolvedwith current detectors, but simple calculations predict a lighttravel time of minutes-to-hours across the compact region.The very hot gas in this region produces X-rays, whichcan be used to measure the extreme relativistic effects andsize scales close to the black hole (Fabian et al. 2009; Al-ston et al. 2020; Wilkins et al. 2017; Chartas et al. 2017;Chainakun & Young 2017; Wilkins & Gallo 2015). X-rayeclipses of this region are then particularly incisive, becausethey can deliver clear constraints on size scales in the ab-sence of imaging (Risaliti et al. 2007, 2009, 2011; Kaastraet al. 2018; Turner et al. 2018; Zoghbi et al. 2019). Whensize scales are known, the radiation processes that power theAGN and the nature of the inner flow are determined.NGC 6814 ( z = 0 . ) is a Seyfert 1.5 active galaxycharacterized by moderate absorption in its optical and X-rayspectrum (Leighly et al. 1994; Rosenblatt et al. 1994; Waltonet al. 2013). The X-rays exhibit rapid variability on short(hours, e.g. Walton et al. (2013)) and long timescales (years, Corresponding author: Luigi C. [email protected] e.g. Mukai et al. (2003)). The X-ray spectrum (Walton et al.(2013); Tortosa et al. (2018); Waddell & Gallo (2020)) is typ-ical of Seyfert 1.5 AGN (e.g. Waddell & Gallo (2020)) pos-sessing weak excesses at low ( (cid:46) ) and high ( (cid:38)
10 keV )energies, and narrow emission in the Fe K α band.Here we report the detection of a rapid occultation event,simultaneously captured in a transient light curve and spec-tral variability. Together, the spectral and timing analysesself-consistently reveal the size of the X-ray source and theproperties and location of the obscuring gas (Mizumoto et al.2018; Miller et al. 2010; Turner et al. 2017; Gallo et al. 2004). OBSERVATIONS AND DATA REDUCTIONNGC 6814 was observed for ∼
131 ks with
XMM-Newton (Jansen et al. 2001) starting 08 April, 2016. During the obser-vations the EPIC detectors (Str¨uder et al. 2001; Turner et al.2001) were operated in large-window mode and with themedium filter in place. The
XMM-Newton
Observation DataFiles (ODFs) were processed to produce calibrated eventlists using the
XMM-Newton
Science Analysis System (SAS V ∼
15 ks of the observation. These pe-riods were neglected during analysis rendering a good-timeexposure of ∼
112 ks .Spectra were extracted from a circular region with a radiusof arcsec centred on the source. The background photons a r X i v : . [ a s t r o - ph . H E ] J a n G ALLO , G
ONZALEZ & M
ILLER were extracted from an off-source circular region on the sameCCD with a radius of arcsec. Pile-up was negligible dur-ing the observations. Single and double events were selectedfor the pn detector, and single-quadruple events were selectedfor the Metal Oxide Semi-conductor (MOS) detectors. EPICresponse matrices were generated using the SAS tasks ARF-GEN and RMFGEN. The MOS and pn data were comparedfor consistency and determined to be in agreement withinknown uncertainties. The RGS (den Herder et al. 2001) spec-tra were extracted using the SAS task RGSPROC and re-sponse matrices were generated using RGSRMFGEN. Thecombined spectra are displayed for presentation purposes.The spectra were optimally binned (Kaastra & Bleeker2016) and the backgrounds were modelled. Spectral fit-ting was performed using XSPEC V C -statistic (Cash 1979),since optimal binning allows bins to have a small numberof counts (i.e. < ) and a Gaussian distribution cannot beassumed. All parameters are reported in the rest frame of thesource unless specified otherwise, but figures remain in theobserved frame. The quoted errors on the model parameterscorrespond to a 90% confidence level for one interestingparameter. A value for the Galactic column density towardNGC 6814 of . × cm − (Willingale et al. 2013) isadopted in all of the spectral fits with appropriate abundances(Wilms et al. 2000).The AGN was also observed with Swift during this timeand the XRT light curve was created with the
Swift -XRT dataproduct generator (Evans et al. 2009) . ESTIMATING THE DURATION OF THE ECLIPSEThe . −
10 keV light curve is shown in Fig. 1. In thefirst ∼
60 ks the object exhibits the flickering behaviour thatis common in AGN (Vaughan et al. 2003). The light curvethen depicts a steady drop in intensity followed by a rela-tively constant low-flux segment, ending with a steady riseto the pre-dip brightness. This latter part of the light curveis commensurate of the transient brightness curves seen inexoplanet systems made famous with
Kepler (Borucki et al.2010; Steffen et al. 2010).Determining the start and end time of ingress and egresscan be rather arbitrary based on the light curve alone. Thehardness ratio (HR) curve is used to estimate these time binsmore robustly. With the HR curve in
500 s bins we calcu-late the slope over 3 neighbouring data points before movingover one point and repeating the process until completed forthe entire curve. The results are then binned in . bins(Fig. 1). In this manner, we can determine at what time theslope fluctuations are no longer random. As evident in Fig. 1, the HR fluctuations are random in Segment 1 and 3, but con-sistently hardening in Segment 2 (ingress) and softening inSegment 4 (egress). This method is completely analyticalrather than relying on a by-eye approximation.The ingress and egress in the NGC 6814 light curve areremarkably similar in duration and depth, and the transi-tions last for approximately T i ∼ . . During the min-imum, which lasts approximately T C ∼ . , the sourceremains relatively hard and the brightness variations are con-sistent with the broadband flickering that was evident priorto ingress. SPECTRAL ANALYSIS OF THE
XMM-Newton
DATA4.1.
Spectral analysis of the RGS data
The RGS spectra are examined to determine the influenceof the warm absorbers known to exist in this system (Turneret al. 1992; Leighly et al. 1994). Spectra are created dur-ing the high- and low-flux periods, and the two flux-resolvedspectra were fitted together between . − .
85 keV . Fittingthe spectra with a power law plus black body to determinethe continuum shape and allowing for just a change in theconstant of normalization described the general shape of thespectra relatively well, but left several narrow residuals. Thefit statistic was C = 1357 for 776 dof . A warm absorber wasthen applied via the XSPEC implementation (Parker et al.2019) of the
SPEX (Kaastra et al. 1996) model
XABS (Steen-brugge et al. 2003).The addition of one warm absorber with free column den-sity ( N H ), ionization parameter ( ξ = L/nr , where n isthe density of the cloud at a distance r from the source ofionising luminosity L), and redshift changed the statistic by ∆ C = 240 for 3 additional parameters. The residuals wereimproved, but significant absorption-like residuals remainedaround .
75 keV and .
95 keV . The addition of a secondwarm absorber produced a good quality fit without any sub-stantial deviations in the residuals. The final fit to the fluxresolved spectra is C = 1031 for 770 dof . The model pa-rameters are shown in Table 1 and the fit residuals are shownin Fig. 2. These warm absorbers were applied to the EPICmodel for the broadband spectrum.The low-flux spectrum displays several excess deviationsfrom the described model, exhibiting emission lines notpresent in the high-flux data. A search for significant positiveresiduals (i.e. stepping a Gaussian through . − .
85 keV every ) results in 5 possible features (Fig. 2) at .
50 keV , .
57 keV , .
65 keV , .
94 keV , and .
445 keV , with eachline improving the fit by ∆ C > for additional freeparameters. 4.2. The EPIC-pn spectra
CLIPSING THE X- RAY EMITTING REGION IN
NGC 6814 3 − . − . − . . . . H a r dn e ss R a t i o ( H - S ) / ( H + S ) HighIngressLowEgress0 20000 40000 60000 80000 100000 120000Time [s] − . − . . . . S l o p e [ × − s − ] r r D c D X V K [s] b r i g h t n e ss T i T i T C
1 2 3 4
Figure 1.
The light curve and hardness ratio curve demonstrating the eclipsing event in NGC 6814.
Upper left:
The
XMM-Newton and
Swift (filled circles) . −
10 keV light curves of NGC 6814. Prior to ∼
55 ks demonstrates the typical random fluctuations in AGN intensity curves.After that time, the light curve shows a constant dimming for ∼ . , followed by a relatively constant segment for ∼ . , and thenbrightening to its original level in ∼ . . Lower left:
With the HR curve in
500 s bins (top panel) we calculate the slope of 3 neighbouringdata points before moving over one point and repeating the process until completed for the entire curve, binning the results in . bins(bottom panel). Segment 1 and 3 show random changes in the slope. Segment 2 and 4 mark regions that are defined by progressive hardeningand softening, respectively. We take these segments to represent ingress and egress. Right:
Together the curves are consistent with a transienteclipsing event for which the speculative geometry is shown (the observer is viewing from the bottom of the page).
EPIC-pn spectra were created in each of the four time seg-ments representing the pre-eclipse, ingress, low-state, andegress (Fig. 3 and 4). The spectra corroborate the behaviourin the hardness ratio and light curves. The fractional vari-ability spectrum ( F var , Fig. 3), which illustrates the level ofvariability in each energy band (Edelson et al. 2002), con-firms the variations during ingress and egress are predom-inately at lower energies, and more achromatic during theminimum and pre-ingress periods. Moreover, the variabilityabove ∼ is comparable during the entire observation(Fig. 3).For the simplest model, we fitted the spectra with a singlepower law plus fixed warm absorbers, modified by a partialcoverer. Even with the power law and partial coverer com-ponents permitted to vary, this resulted in a relatively poorfit ( C = 2008 for 728 dof ), and demonstrated the need formore physical continuum models. Notwithstanding this, it was notable that the effects of the partial coverer was moreenhanced in the low-state than during pre-eclipse.The continuum was modelled assuming the blurred reflec-tion scenario (Ross & Fabian 2005). Here, some fractionof the corona illuminates the inner accretion disc producingbackscattered emission that is modified for Doppler, Special,and General relativistic effects. We use RELXILL
D (Garc´ıaet al. 2016; Jiang et al. 2019) to model the scenario, whichallows the density of the disc to be altered.The spectra are modified by the double warm absorber sys-tem that was found in the RGS analysis (Fig. 2). We deter-mined that refitting the warm absorbers to the pn data did notprovide a substantially better fit than simply using the RGSmeasured parameters. Therefore, the warm absorber param-eters were fixed to the RGS values throughout the analysis.An Fe K α emission line was present in the data, but at ener-gies slightly higher than . . A simple Gaussian profileis used to model this component. The line energy and width G ALLO , G
ONZALEZ & M
ILLER − − − C o un t ss − k e V − c m − C/dof = 1031 /
770 ModelHighLow0123 D a t a/ M o d e l D a t a/ M o d e l Figure 2.
The high-resolution RGS spectra show the presence ofa warm absorber and possible emission lines during the low-fluxinterval. The top panel shows the RGS spectra in the pre-eclipsehigh-flux state (Segment 1) and the deep minimum flux state (Seg-ment 3). Fitted to the data is a model including two warm absorbers(Table 1) and the residuals (data/model) are shown in the next twopanels. The low-flux level can be well described by simply renoma-lizing the high-flux model. However possible emission-like resid-uals are evident in the low-flux state (lower panel) at ∼ .
50 keV , .
57 keV , .
65 keV , .
94 keV , and .
445 keV , with each line im-proving the fit by ∆ C > for additional free parameters were linked between the epochs, but the normalisation wasleft free to examine for variability.Attempting to fit all four spectra in a self-consistent man-ner, we allow only the photon index and power law flux tovary between epochs, which are expected to vary on suchtime scales. The reflection fraction ( R ) was linked indi-cating that the ratio of reflected-to-continuum flux was notchanging. This could be examined in future modelling, butthe rather constant F var spectra (Fig. 3) suggests modestspectral variability during the high- and low-flux intervals.This model resulted in a rather poor fit to the data with C = 2198 for 737 dof .A partial coverer was applied to the central X-ray emis-sion component only (i.e. RELXILL
D). This was a markedimprovement to the fit. The best fit was obtained when thepartial coverer had a fixed column density and ionisation pa-rameter, but the covering fraction ( C f ) was permitted to varybetween epochs (Fig. 4 and Table 2). In this manner, the cov-ering fraction changed from C f ≈ prior to the eclipse and C f ≈ . in the minimum flux state. The final fit statisticwas C = 1029 for 731 dof Allowing the partial coverer to be neutral rather thanionised resulted in a poorer fit ( ∆ C = 22 for 1 fewer param-eter). Allowing the column density to vary between epochs Table 1.
The RGS data fitted with a double warm absorber anda phenomenological continuum of a blackbody and power law. In
XSPEC terminology: tbabs × WA1 × WA2 × ( bb + po ) . Model Component Model Parameter Parameter Value xabs log ξ/ erg cm − s − . +0 . − . N H / cm − . +16 . − . v out / km s − +262 − xabs log ξ/ erg cm − s − . +0 . − . N H / cm − . ± . v out / km s − +245 − bb kT e / eV 111 ± / − . ± . po Γ 1 . ± . / − ± const F low /F high . ± . rather than the covering fraction also resulted in a poorer fit( C = 1537 for 731 dof ).We considered if the partial covering results could bedependent on the assumed X-ray continuum. To testthis we replaced the blurred reflection model with a soft-Comptonisation model for the continuum. This is anothercommonly used X-ray model that attributes the soft-excessemission to an optically thick warm corona that is situatedover the disc (Magdziarz et al. 1998; Czerny et al. 2003;Ballantyne 2020). We adopt the model previous used inPetrucci et al. (2018), which incorporates two NTHCOMP components, one for the traditional optically-thin, hot coronaand the other for the warm corona. As with the blurred re-flection model, the continuum is modified by warm absorbersand a ∼ .
45 keV
Gaussian profile is included. Various com-binations could be attempted to describe the intrinsic vari-ability. We found the the simplest scenario was to link thewarm corona parameters, but allow the hot corona to varyfrom segment-to-segment.A partial coverer is again added and the covering frac-tion is free to vary (Fig. 4 and Table 2) . The partial cov-erer column density and ionisation parameter are very simi-lar to what was found with the blurred reflection continuum.The covering fractions were offset by about +20 per cent ateach interval compared to the reflection model, but the rel-ative change between intervals was the same as in the re-flection model. Despite using a different continuum model,the partial covering parameters were very similar. The best-fit soft-Comptonisation model resulted in a fit statistic of C = 1131 for 728 dof DISCUSSION AND CONCLUSIONS
CLIPSING THE X- RAY EMITTING REGION IN
NGC 6814 5
Table 2.
The partial covering model applied to two different continuum scenarios, Comptonization and blurred reflection. In
XSPEC termi-nology the Comptonization model is: tbabs × ztbabs × WA1 × WA2 × ( zxipcf × ( nthcomp soft + nthcomp hard ) + zgauss ) and the blurred reflection model is: tbabs × WA1 × WA2 × ( zxipcf × ( cflux × cutoffpl + const × cflux × relxillD ) + zgauss ) . Parameter values without uncertainties are fixed during the fitting. Continuum Model Model Parameter ValueModel Component Parameter All High Ingress Low EgressComptonization ztbabs N H / cm − . +2 . − . zxipcf N H / cm − . +0 . − . log ξ/ erg cm − s − . +0 . − . f cov . +0 . − . . +0 . − . . ± .
01 0 . +0 . − . nthcomp soft Γ 2 . +0 . − . kT e / eV 113 +10 − kT bb / eV 3Norm / − cm − s − keV − . +0 . − . nthcomp hard Γ 1 . ± .
03 1 . +0 . − . . ± .
04 1 . ± . kT e / keV 100 kT bb / eV 3Norm / − cm − s − keV − . +0 . − . . +0 . − . . ± .
00 9 . ± . zgauss E/ keV 6 . ± . σ/ eV 141 +24 − Norm / − ph . cm − s − . +0 . − . . +1 . − . . +0 . − . . +1 . − . Blurred reflection zxipcf N H / cm − . ± . ξ/ erg cm − s − . +0 . − . f cov < .
01 0 . ± .
03 0 . ± .
02 0 . ± . cflux log F/ erg cm − s − − . +0 . − . − . +0 . − . − . +0 . − . − . +0 . − . cutoffpl Γ 1 . +0 . − . . +0 . − . . +0 . − . . +0 . − . const R . +0 . − . relxillD q in . +0 . − . q out R b / r g a/ [cJ / GM ] 0 . i/ ◦ +1 − log ξ/ erg cm − s − . +0 . − . log N H / cm − A Fe . +0 . − . zgauss E/ keV 6 . ± . σ/ eV 137 +21 − Norm / − ph . cm − s − . +0 . − . . +0 . − . . +0 . − . . +1 . − . Similar to exoplanet light curves, the transient light curvein NGC 6814 can be interpreted as an occultation event, inthis case, of the primary X-ray source by an orbiting glob-ule. Such events have been reported previously (Risaliti et al.2007, 2011; Turner et al. 2018), but this may be the first timea rapid occultation is captured in its entirety with spectral andtemporal data. The symmetry in the transient light curve in-dicates the obscurer is rather uniform and possibly a singlecloud. The rapid time scales of the eclipse place the cloudclose to the black hole, and the fact the dip does not reachzero brightness implies the obscurer only partially covers theX-ray source.The illustration in Fig. 1 highlights the situation anddemonstrates the parameters that can be estimated from thespectral and temporal measurements. The depth of theeclipse is energy-dependent (Fig. 3), with the low-energy X-rays diminishing to ∼ per cent of the pre-eclipse bright- ness and the high-energy X-rays to ∼ per cent. There issome indication the shape of the eclipse may also differ – athigher energies, the transitions between time points might besmoother than at lower energies. This may be an indicationthe source size is energy dependent and that the high-energyX-rays are more centrally compact. Such limb darkening be-haviour is common in stellar and exoplanet transient curves.We modeled the . −
10 keV spectra from each of thetime segments (see Fig. 1) simultaneously in a self-consistentmanner. We tested different continuum models, whichyielded similar results, but for ease of presentation, here wewill discuss the results assuming the intrinsic X-ray emis-sion is described by ionised blurred reflection (Ross & Fabian2005). The continuum emission was also modified by twonon-variable ionized (warm) absorbers and a Gaussian pro-file at . ± .
01 keV . This primary X-ray emission wasthen obscured by a partial coverer (Holt et al. 1980; Tanaka G
ALLO , G
ONZALEZ & M
ILLER − − − C o un t ss − k e V − c m − HighIngressLowEgress0.3 1 3 10Energy [keV]0 . . . . . . F r a c t i o n a l V a r i a b ili t y , F v a r HighIngressLowEgress
Time [s] N o r m a li s e d t o p r e - ec li p s e c oun t r a t e Figure 3.
The spectral changes show the effects of the transientabsorber and energy dependence on the eclipse.
Top:
The EPIC-pn spectra are created in the four time segments shown in Fig. 1.The spectra are remarkably similar during ingress and egress, andclearly harder when the source is dimmer.
Middle:
The fractionalvariability spectra show the degree of variability during the differ-ent segments. During ingress and egress, the variability is clearlydominated by the low-energy emission, which would be consistentwith increasing absorption. During the high- and low-flux states,the variability is similar suggesting that the nature of the fluctua-tions are probably alike in the high and low-flux intervals (i.e. theintrinsic nature has not changed). Above ∼ , the variabilityis similar at all flux levels indicating the eclipse has little effect atthese highest energies. Bottom:
Light curves in the . − . (black), − (green), − (blue) show the depth ofthe eclipse is energy dependent as would be expected because ofthe column density of the cloud. There is indication the shape ofthe transient may also differ as a function of energy (e.g. see thetransitions in the ingress segment). et al. 2004) that was of constant column density and ion-ization parameter. The normalization (brightness) and pho-ton index ( Γ ) of the power law continuum were free to varybetween segments as is typical in Seyfert galaxies. The re-flection fraction ( R ) was linked indicating the relative frac-tion of reflected-to-continuum emission remained constant.For the partial coverer, only the covering fraction ( C f ) var-ied between the segments. This fit was acceptable yield-ing a C-statistic of C = 1029 for 731 degrees of freedom.The partial coverer could be described as having N H =(11 . ± . × cm − and ξ = (12 . +0 . − . ) erg s cm − .Prior to ingress (Segment 1), the covering fraction was C f < . . During ingress (Segment 2) and egress (Segment 4), C f = 0 . ± . and . ± . , respectively. The max-imum covering fraction ( C f = 0 . ± . ) was returnedduring the minimum flux (Segment 3).The mass of the black hole in NGC 6814 is M BH =1 . × M (cid:12) (Bentz & Katz 2015). Following (Turneret al. 2018) (section 6), assuming the obscurer is moving ona Keplerian orbit there is a relationship between the orbit andobscurer properties such that: r / = ( GM ) / L X ∆ TN H ξ . Thetotal duration of the eclipse (from onset of ingress to end ofegress) is ∆ T ≈ . and the X-ray luminosity prior toeclipse is measured to be L X ≈ × erg s − . This wouldplace the partial coverer at r ≈ r g = 4 . × cm .The electron density of the cloud is n = L X /r ξ =8 . +0 . − . × cm − and from the measured column den-sity we obtain a cloud diameter of approximately D C =1 . +0 . − . × cm . The Keplerian velocity of the partialcoverer is V K = D C /T i ≈ × km s − . The density iscomparable to the electron densities estimated for broad-lineregion (BLR) “clouds” and the velocity and distance are alsoin agreement with the inner BLR (Bentz et al. 2009; Netzer1990; Arav et al. 1998).The narrow Fe K α emission line in NGC 6814 exhibitsa relatively constant flux within uncertainties ( ∼ ± percent) in the four different segments implying that it is notaffected by the partial coverer. Its equivalent width doeschange in accordance with continuum flux changes (i.e.largest equivalent width during the low-flux interval). Wemodelled the line with a Gaussian profile and found it at E = 6 . ± .
01 keV and with σ = 136 +21 − eV . The re-sulting F W HM = 320 +49 − eV , which renders a velocity ofabout (15 ± × km s − . This is compatible with thelocation of the obscurer so we may have an example of ob-scuration and re-emission from the same region.Comparing the duration of ingress to the duration of thelow-flux interval provides an estimate of the X-ray sourcesize: D X = D C × T C /T i = 4 . +0 . − . × cm = 26 +3 − r g .The value is completely consistent with what is normally es-timated or assumed for the size of the corona ( ∼ r g ) (Al- CLIPSING THE X- RAY EMITTING REGION IN
NGC 6814 7 − − − − C o un t ss − k e V − High Ingress Low Egress D a t a/ M o d e l ( B a c k g r o und ) . . . D a t a/ M o d e l ( R e fl ec t i o n ) C/dof = 1030 / . . . D a t a/ M o d e l ( C o m p t o n i s a t i o n ) C/dof = 1069 /
732 0.5 1 3 10Energy [keV] 0.5 1 3 10Energy [keV] 0.5 1 3 10Energy [keV]
Figure 4.
Partial covering models applied to different continuum scenarios.
Top row:
The source and background pn spectra between . −
10 keV during each segment (labeled on top).
Upper middle row:
The residuals from fitting the background data.
Lower middle row:
The resulting residuals from the ionised blurred reflection model described in the text and in Table 2.
Bottom row:
The resulting residualsfrom the soft Comptonization model described in the text and in Table 2. ston et al. 2020; Chartas et al. 2017; Wilkins & Gallo 2015;Gallo et al. 2015; Wilkins et al. 2014; Risaliti et al. 2009).There are other interesting aspects of the eclipse that areobserved. The RGS spectra are generated for the high- andlow-flux intervals (Segment 1 and 3, respectively). The high-resolution grating data are well fitted with two warm ab-sorbers that are then applied to the CCD spectra. The dif-ference between the low- and high-flux states can largely beattributed to a change in normalization, but there are narrowemission features that appear during eclipse. These could beattributed to some scattered emission from the partial covereror from emission that originates at large distances from theblack hole that is only evident when the continuum bright-ness is suppressed (e.g. in the narrow-line region, starburstregion, or torus) (Strickland et al. 2004; Longinotti et al.2019; Buhariwalla et al. 2020).The detection of rapid eclipsing events in AGN light curvesare powerful tools for determining properties of the absorber,the BLR, and of the primary X-ray source. Here, we have reported the discovery of what appears to be a single cloudpassing in front of the central engine. The results show thatrelativistic reflection and partial covering are both natural tothe accretion flow and necessary for accurate modeling, noteffects that naturally oppose each other. The properties of theabsorber imply a BLR origin. The size of the X-ray source iscompact and consistent with expectations (Alston et al. 2020;Chartas et al. 2017; Wilkins & Gallo 2015; Gallo et al. 2015;Wilkins et al. 2014; Risaliti et al. 2009). The eclipse showsevidence of energy-dependent effects, which may lead to un-derstanding limb darkening in the corona.Such events are difficult to detect in the stochastic be-haviour of AGN light curves, but they are probably not rare.At least some Seyfert 1.5s may offer advantageous viewingangles: high enough to intercept the BLR in a manner thatcan give eclipses, but low enough to avoid being blocked by atorus. NGC 6814 is probably an excellent target for witness-ing such an event. The designation of NGC 6814 as an inter-mediate Seyfert 1.5 implies we are seeing the AGN at higher G
ALLO , G
ONZALEZ & M
ILLER inclinations (i.e. more edge-on). Estimates place the BLR inNGC 6814 at an inclination of − degrees (Rosenblattet al. 1994), which is consistent with our X-ray measureddisc inclination ( +1 − degrees). From our line-of-sight, theBLR crosses the X-ray source, but through relatively modestobscuration from the torus.Data in current X-ray archives can be used to search forsimilar episodes. Long, uninterrupted observations of well-selected sources can be studied more extensively using cur-rent missions, and studied even better with future missions. ACKNOWLEDGEMENTSWe thank the referee for providing a helpful report. The XMM-Newton project is an ESA Science Mission with in-struments and contributions directly funded by ESA Mem-ber States and the USA (NASA). This work made use of datasupplied by the UK Swift Science Data Centre at the Univer-sity of Leicester, data from the NuSTAR mission, a projectled by the California Institute of Technology, managed bythe Jet Propulsion Laboratory, and funded by the NationalAeronautics and Space Administration. L.C.G and A.G.G.are support by NSERC and the CSA. J.M.M. is supported byNASA funding, through
Chandra and
XMM-Newton guestobserver programs.REFERENCES
Alston, W. N., Fabian, A. C., Kara, E., et al. 2020, NatureAstronomy, 4, 597, doi: 10.1038/s41550-019-1002-xArav, N., Barlow, T. A., Laor, A., Sargent, W. L. W., & Blandford,R. D. 1998, MNRAS, 297, 990,doi: 10.1046/j.1365-8711.1998.297004990.xArnaud, K. A. 1996, in Astronomical Society of the PacificConference Series, Vol. 101, Astronomical Data AnalysisSoftware and Systems V, ed. G. H. Jacoby & J. Barnes, 17Ballantyne, D. R. 2020, MNRAS, 491, 3553,doi: 10.1093/mnras/stz3294Bentz, M. C., & Katz, S. 2015, PASP, 127, 67,doi: 10.1086/679601Bentz, M. C., Walsh, J. L., Barth, A. J., et al. 2009, ApJ, 705, 199,doi: 10.1088/0004-637X/705/1/199Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977,doi: 10.1126/science.1185402Buhariwalla, M. Z., Waddell, S. G. H., Gallo, L. C., Grupe, D., &Komossa, S. 2020, ApJ, 901, 118,doi: 10.3847/1538-4357/abb08aCash, W. 1979, ApJ, 228, 939, doi: 10.1086/156922Chainakun, P., & Young, A. J. 2017, MNRAS, 465, 3965,doi: 10.1093/mnras/stw2964Chartas, G., Krawczynski, H., Zalesky, L., et al. 2017, ApJ, 837,26, doi: 10.3847/1538-4357/aa5d50Czerny, B., Nikołajuk, M., R´o˙za´nska, A., et al. 2003, A&A, 412,317, doi: 10.1051/0004-6361:20031441den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001,A&A, 365, L7, doi: 10.1051/0004-6361:20000058Edelson, R., Turner, T. J., Pounds, K., et al. 2002, ApJ, 568, 610,doi: 10.1086/323779Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2009, MonthlyNotices of the Royal Astronomical Society, 397, 1177,doi: 10.1111/j.1365-2966.2009.14913.x Fabian, A. C., Zoghbi, A., Ross, R. R., et al. 2009, Nature, 459,540, doi: 10.1038/nature08007Gallo, L. C., Tanaka, Y., Boller, T., et al. 2004, MNRAS, 353,1064, doi: 10.1111/j.1365-2966.2004.08248.xGallo, L. C., Wilkins, D. R., Bonson, K., et al. 2015, MNRAS,446, 633, doi: 10.1093/mnras/stu2108Garc´ıa, J. A., Fabian, A. C., Kallman, T. R., et al. 2016, MNRAS,462, 751, doi: 10.1093/mnras/stw1696Holt, S. S., Mushotzky, R. F., Becker, R. H., et al. 1980, ApJL,241, L13, doi: 10.1086/183350Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1,doi: 10.1051/0004-6361:20000036Jiang, J., Fabian, A. C., Dauser, T., et al. 2019, MNRAS, 489,3436, doi: 10.1093/mnras/stz2326Kaastra, J. S., & Bleeker, J. A. M. 2016, A&A, 587, A151,doi: 10.1051/0004-6361/201527395Kaastra, J. S., Mewe, R., & Nieuwenhuijzen, H. 1996, in UV andX-ray Spectroscopy of Astrophysical and Laboratory Plasmas,411–414Kaastra, J. S., Mehdipour, M., Behar, E., et al. 2018, A&A, 619,A112, doi: 10.1051/0004-6361/201832629Leighly, K., Kunieda, H., Tsusaka, Y., Awaki, H., & Tsuruta, S.1994, ApJ, 421, 69, doi: 10.1086/173626Longinotti, A. L., Kriss, G., Krongold, Y., et al. 2019, ApJ, 875,150, doi: 10.3847/1538-4357/ab125aMagdziarz, P., Blaes, O. M., Zdziarski, A. A., Johnson, W. N., &Smith, D. A. 1998, MNRAS, 301, 179,doi: 10.1046/j.1365-8711.1998.02015.xMiller, L., Turner, T. J., Reeves, J. N., et al. 2010, MNRAS, 403,196, doi: 10.1111/j.1365-2966.2009.16149.xMizumoto, M., Done, C., Hagino, K., et al. 2018, MNRAS, 478,971, doi: 10.1093/mnras/sty1114Mukai, K., Hellier, C., Madejski, G., Patterson, J., & Skillman,D. R. 2003, ApJ, 597, 479, doi: 10.1086/378318
CLIPSING THE X- RAY EMITTING REGION IN