Detection of the high energy cut-off from the Seyfert 1.5 galaxy NGC 5273
Mayukh Pahari, I. M. McHardy, Labani Mallick, G. C. Dewangan, R. Misra
aa r X i v : . [ a s t r o - ph . H E ] J un MNRAS , 1–11 (2017) Preprint 6 April 2018 Compiled using MNRAS L A TEX style file v3.0
Detection of the high energy cut-off from the Seyfert 1.5galaxy NGC 5273
Mayukh Pahari , I. M. M c Hardy , Labani Mallick , G. C. Dewangan and R. Misra Inter-University Centre for Astronomy and Astrophysics, Pune, 411007, India School of Physics & Astronomy, University of Southampton, Highfield, Southampton SO17 1BJ, UK
ABSTRACT
We perform the
NuSTAR and
Swift /XRT joint energy spectral fitting of simul-taneous observations from the broad-line Seyfert 1.5 galaxy NGC 5273. When fittedwith the combination of an exponential cut-off power-law and a reflection model, ahigh energy cut-off is detected at 143 +96 − keV with 2 σ significance. Existence of suchcut-off is also consistent with the observed Comptonizing electron temperature whenfitted with a Comptonization model independently. We observe a moderate hard X-rayvariability of the source over the time-scale of ∼
12 years using
INTEGRAL /ISGRIobservations in the energy range of 20-100 keV. When the hard band count rate (6-20keV) is plotted against the soft band count rate (3-6 keV), a hard offset is observed.Our results indicate that the cut-off energy may not correlate with the coronal X-rayluminosity in a simple manner. Similarities in parameters that describe coronal prop-erties indicate that the coronal structure of NGC 5273 may be similar to that of thebroad-line radio galaxy 3C 390.3 and another galaxy MCG-5-23-16 where the coronalplasma is dominated by electrons, rather than electron-positron pairs. Therefore, thecoronal cooling is equally efficient to the heating mechanism keeping the cut-off energyat low even at the low accretion rate.
Key words: accretion, accretion disc — galaxies: Seyfert — black hole physics —X-rays: galaxies — galaxies: individual: NGC 5273
Although widely studied, the coupling between the accretiondisc and the Comptonizing corona as well as the the ori-gin of seed photons for Comptonization (Haardt & Maraschi1991; Liu et al. 1995; Zycki et al. 1995; Liu et al. 1999;Taam et al. 2012; Mallick & Dewangan 2017) is yet to befully understood in accreting systems harbouring a super-massive black hole, i.e., Active Galactic Nuclei (AGN). Onerobust way to predict the connection between the coro-nal properties in response to the seed photon flux is tomeasure the cut-off in the hard X-ray spectrum and thephoton power-law index. The spectral cut-off is directlyrelated to the Comptonizing electron temperature of thecorona while the power-law index depends on the inter-play between the electron temperature and the optical depth(Titarchuk & Lyubarskij 1995). The optical depth providesthe size of the Compton up-scattering region for a typicalelectron density. Therefore, measuring both the high en-ergy cut-off as well as photon power-law index can pro-vide much accurate description of the nature of Comp-tonizing corona and its geometry. Since high energy spec-tra ( > NuSTAR spectra, with its unprecedented sensitivityat high energy (the detection limit in 10-30 keV is 1 × − erg cm − s − ; Harrison et al. (2013)), has provided strongconstraints on the coronal properties as well as the cut-offenergy of a few AGN (Brennman et al. 2014; Fabian et al.2015; Balokovic et al. 2015; Matt et al. 2015).One possible mechanism for the existence of a very highenergy cut-off, which in turn implies a very high temperatureof the corona, is the inefficient coronal cooling process. Suchinefficient cooling can be caused by a low supply of seed pho-tons which is common in the low luminosity AGN (LLAGN)due to their very low mass accretion rate ( < >
210 keV (Madsen et al. 2015). On the other hand, brightAGN with high mass accretion rate ( >
10% of Eddingtonaccretion rate), e.g., SWIFT J2127.4+5654 show relatively c (cid:13) Pahari et al. lower cut-off at 108 ±
11 keV (Marinucci et al. 2014). Us-ing
Suzaku and
NuSTAR observations of one of the bright-est AGN IC 4329A ( ∼
46% of Eddington accretion rate;de La Calle Pe´rez et al. (2010)), the high energy cut-off isdetected in the energy range 170-200 keV (Brennman et al.2014) while Malizia et al. (2014) constrained the cut-off en-ergy at 152 +51 − keV using the simultaneous Swift /BAT and
INTEGRAL /IBIS observations. Variable cut-off energy isalso reported. For example, the cut-off energy detected inNGC 5506 have been reported few times: using simultane-ous
Swift /BAT and
INTEGRAL /IBIS observations duringthree epochs, Molina et al. (2013) constrained the cut-offenergy at 36 +5 − keV, 54 +14 − keV and 76 +27 − keV by keepingthe disc reflection fraction fixed at 0, 1 and 2 respectively.Using simultaneous XMM- Newton and BeppoSAX observa-tions, Bianchi et al. (2003, 2004) obtained the high energycut-off at 140 +40 − keV while using simultaneous Swift /XRTand
NuSTAR spectra, Matt et al. (2015) found the high en-ergy cut-off at 720 +130 − keV. Although we certainly cannotrule out the possibility of a variable cut-off energy detectedfrom a source at different epochs, however, if such variationis linked to the change in the mass accretion rate, it is un-likely that the cut-off energy would vary on short time-scales(from months to years).Such luminosity-dependence of the cut-off energy maynot be described by a simple relationship, rather complicatedby the dichotomy in physical and observational proper-ties like Compton-thin/Compton-thick circum-nuclear mat-ter and narrow-line Seyfert 1 (NLS1)/broad line. Never-theless, strong constraints on the cut-off energy have beenproven to be extremely crucial in understanding the accre-tion and radiation mechanism in AGN. Continuing the ef-forts to understand the hard X-ray spectral properties ofAGN, we present the spectral analysis of a low-luminosity,Seyfert 1.5 galaxy NGC 5273 in the energy range 0.8-75 keV.NGC 5273 is a broad-line, nearby AGN at a redshiftof 0.00362 (Bentz et al. 2014). The AGN is hosted by afaint-spiral galaxy (S0 morphology) at a distance of 16.5 ± ± × M ⊙ . Subtract-ing M32 template spectrum from the optical spectrum ofNGC 5273, Osterbrock and Martel (1993) found a weak H β emission line. However, based on the very large ratio of H α band H β b lines, they categorized the source as a Seyfert 1.9.However, in a recent paper Trippe et al. (2010) acquired op-tical spectra with the GoldCam spectrograph on the 2.1 mtelescope at Kitt Peak National Observatory (KPNO) andsignificantly detected both H α and H β lines after subtract-ing the E0 host galaxy contribution. Neither the narrowline region (NLR) nor BLR (broad line region) were foundto be reddened. Therefore, they classified NGC 5273 as aSeyfert 1.5 galaxy. Since the later claim of the Seyfert classi-fication was based on the improved instrumentation and bet-ter calibration than the former claim, we assume the galaxytype 1.5 in the rest of the paper.Using XMM-Newton observations of NGC 5273 on 14June 2002, Cappi et al. (2006) found the equivalent width ofthe Fe-K α emission line 226 ±
75 eV. They found moderateabsorption column density of the order of 10 cm − and in- terestingly the highest ratio between the 2-10 keV X-ray fluxand O III line flux (at 5700 ˚A) amongst all Seyfert 1 galax-ies. This diagnosis, based on the fact that [O III] luminosityis the isotropic indicator of the intrinsic luminosity, impliesthat NGC 5273 is a Compton-thin AGN. Using Suzaku (com-bining FI-XIS, BI-XIS, HXD/PIN) and
Swift /BAT observa-tions of NGC 5273 on 16 July 2013, Kawamuro et al. (2016)found the power-law index of 1.57 +0 . − . and absorption col-umn density N H of 2.60 +0 . − . × cm − . The Fe-K α lineequivalent width of 100 ±
20 eV is smaller by a couple of fac-tors than the earlier measurement. They showed that for thegiven power-law index and absorption column density, thetorus model of Ikeda, Awaki & Terashima (2009) underesti-mated the equivalent width of the Fe-K α line for all possiblevalues of the half-opening angle of the torus. Although thereason of such underestimation is not clear, the mismatchcould be due to the difference in the line of sight columndensity and the actual circum-nuclear matter density whichmostly lies out of observer line-of-sight but may contributeto the line emission. However, considering the complexityof AGN geometry, such relationship requires detail study toestablish.In this work, we carry out spectral analysis from Swift /XRT and
NuSTAR simultaneous observations ofNGC 5273. Hard X-ray lightcurve in the energy range 3-80 keV is strongly variable and the fractional rms variabilitydecreases at higher energy. Using the best fit model, consistsof a power-law with a high energy cut-off, a reflection anda partial covering absorption, we constrain the high energycut-off at 143 +96 − keV with the 2 σ significance. Combining344 pointing observations with INTEGRAL spanning over ∼
12 years, the source is detected with ∼ σ significances inthe shadowgram of INTEGRAL /ISGRI in the energy range20-100 keV with a total exposure of ∼
553 ks and the sourceshow hard X-ray variability over this time-scale. We find theabsorption column density and covering fraction of 2.36 +0 . − . × cm − and 0.97 +0 . − . respectively. Data reduction andanalysis procedure are provided in Section 2 while resultsfrom the spectro-timing analysis are summarized in Section3 and discussions and conclusions are provided in Section 4. NGC 5273 was observed by
NuSTAR on 14 July 2014 start-ing at 02:56:07 UT. While the total duration of the obser-vation lasted for ∼
39 ksec, the effective, on-source expo-sure was 21.2 ksec. The
NuSTAR data were collected us-ing the two focal plane telescopes (FPMA and FPMB) cen-tred roughly 1 arcmin away from the center of NGC 5273.Data are reduced using the
NuSTAR
Data Analysis Soft-ware nustardas v1.6.0 included in heasoft v6.19 andthe recent (as of 05 March 2017) calibration database
CALDB version 20161207 is used. Event files from bothtelescopes are filtered and depth corrections are appliedusing the nupipeline task ( version 0.4.5 ). Circular re-gions with the radius of 80 arcsec are chosen as source andbackground regions with the source region circle centred onNGC 5273 while the background region circle is chosen suchthat its centre is at least 5 arcmin away from the sourcecentre so that background selection is not contaminated bythe PSF wing of the source. Energy spectra, lightcurves in
MNRAS , 1–11 (2017) uSTAR view of NGC 5273 . N u S T A R ( F P M A + F P M B ) c oun t r a t e ( c t s / s ) Time (sec)3.0−80.0 keV . ( c oun t s / s ) S o ft − band . ( c oun t s / s ) H a r d − band . . . r a t i o H a r d / S o ft Time (sec)
Figure 1.
The 3-80 keV background-subtracted lightcurve from
NuSTAR is shown in the top panel where observations from FPMA andFPMB are combined. Bottom panel shows lightcurves in the soft band (3-6 keV; top), hard band (6-20 keV; middle) and the hardnessratio (defined as the ratio of 6-20 keV count rate to that of 3-6 keV; bottom) as a function of time. All lightcurves and hardness ratioare plotted with 500 sec bin size. Variability in both lightcurve and hardness ratio can be seen. different energy bands, response matrix files (rmf) and aux-iliary response files (arf) are extracted using the nuprod-ucts task. While lightcurves from FPMA and FMPB arecombined to increase the signal-to-noise ratio, responses andspectra are not combined to minimize the systematic effects.Instead they are fitted simultaneously by allowing to varycross-normalization factor between two modules.
Swift observed NGC 5273 on 14 July 2014 starting at03:55:00 UT.
Swift /XRT exposure lasted till ∼ Swift /XRT data are reduced using the xrtpipeline v.0.13.2 task. A 40 arcsec circular region is used to extract thesource spectrum, and a 40 arcsec circular region, far awayfrom source PSF wing, is used to extract the background spectrum using
XSELECT v 2.4d . The xrtmkarf task isused along with the exposure map to generate an auxiliaryresponse using the latest
Swift /XRT spectral redistributionmatrices for the current observation.
INTEGRAL took 344 pointing observations ofNGC 5273 with regular intervals between 2003 and2014 and we analyse all pointing observations of
INTE-GRAL /ISGRI in the energy range 20-100 keV. ISGRI, oneof two instruments in
INTEGRAL /IBIS, has a collectingarea of 2600 cm , is sensitive between 15 and 1000 keVwith maximum effective area between 20 and 100 keV(Ubertini et al. 2003). All archival data are processed andanalysed using the INTEGRAL Offline Science Analysis(OSA; Courvoisier et al. (2003)) package v. 10.2, the In-
MNRAS , 1–11 (2017)
Pahari et al. . . . H a r dne ss r a t i o ( H − S ) / ( H + S ) Total count rate in 3−20 keV (cts s −1 ) 0 0.2 0.4 0.6 0.8 1 . . H a r d band c oun t r a t e ( . − . k e V ) Soft band count rate (3.0−6.0 keV)
Figure 2.
Left panel shows the hardness intensity diagram of NGC 5273 using
NuSTAR . The hardness ratio is defined as H-S/H+Swhere H is the hard band (6-20 keV) count rate and S is soft band (3-6 keV) count rate and intensity is defined 3-20 keV count rate. Itmay be noted that with the change of total count rate, hardness ratio show some dispersion around best fit constant shown by horizontalblack line. Right panel shows flux-flux plot where the hard band count rate (H) is plotted against the soft band count rate (S). Whenfitted with a linear function and extrapolated, the fitted line (shown in black) intersect Y-axis (i.e., hard count rate axis) at the originwhich implies hard excess even at zero soft band count rate and the presence of two spectral components. I N T E G R A L /I S G R I c oun t r a t e ( − k e V ) Time (sec)
Figure 3.
Top panel shows
INTEGRAL /ISGRI image at thefield of view of NGC 5273 in the energy range 20-100 keV. Itis constructed by superimposing all pointing observations (344)of NGC 5273 made by the
INTEGRAL covering ∼
12 years. Thesource is detected (shown by the black circle) at the positionof NGC 5273 with 3.8 σ significance. Bottom panel shows the20-100 keV lightcurve from all pointing observations with IN-TEGRAL /ISGRI covering ∼
12 years. The lightcurve binsize is 5day. Although signal-to-noise ratio is poor, a variability in hardX-rays count rate is observable. strument Characteristics v 10.2 and the
ReferenceCatalogue v. 40.0 . Following standard procedure, imagesare created combining all science windows.
The 3-80 keV background-subtracted lightcurve from
NuS-TAR is shown in the top panel of Figure 1 where countsfrom both FPMA and FPMB are combined and 500 sec binsize is used. Lightcurve in the relatively soft band (3-6 keV),hard band (6-20 keV) and their ratio are shown in the bot-tom panel of Figure 1. At all time intervals, relatively harderband has higher count rate than the soft band. A significantvariability (of the order of ∼
15 per cent) is observed in thehardness ratio (see the bottom panel of Fig. 1), particularlyduring the second and third time intervals. Therefore, fluxvariability in soft and hard bands are not similar. In or-der to clarify this, we plot the hardness intensity diagram(HID). The hardness ratio is defined as (H-S)/(H+S) whereH is the background-subtracted hard band (6-20 keV) countrate and S is the background-subtracted soft band (3-6 keV)count rate and the intensity is defined as the 3-20 keV countrate. The resulting plot is shown in the left panel of Figure 2.A positive hardness ratio is observed which implies the hardcount rate is higher than the soft count rate and a scatteringin the hardness ratio around the best-fit constant (shown bya black horizontal line) may be noted. To test the number ofspectral components present in the current energy spectrain a model-independent way, we perform the flux-flux anal-ysis where the hard band flux (6-20 keV) is plotted againstthe soft band flux (3-6 keV) in the right panel of Figure 2.The plot shows a linear relationship between count rates intwo bands. The best fit does not pass through the origin,implying the existence of a separate hard component, butthe offset is not large and a fit through the origin is stillconsistent with the data (change in χ is -16 for the changein 1 degree of freedom). This can further be confirmed bythe energy spectral analysis.Top panel in Figure 3 shows the INTEGRAL /ISGRIimage in the field of view of NGC 5273 in the energy range20-100 keV. The image is constructed by superimposing all344 pointing with total effective exposure of ∼
553 ksec. 0.5counts/sec in the image corresponds to a 3 σ limit of anysource detection. It is clear that the source is detected withat least 3 σ significance at the position of NGC 5273 (shownby the black circle where the average count rate is ∼ ± MNRAS , 1–11 (2017) uSTAR view of NGC 5273 . . D a t a / M ode l r a t i o Energy (keV)Swift/XRTNuSTAR/FPMANuSTAR/FPMB −6 −5 −4 −3 E F E k e V ( P ho t on s c m − s − k e V − ) χ Energy (keV)
Figure 4.
Left panel shows the data to model ratio of NGC 5273 as a function of energy when
Swift /XRT (red circles),
NuSTAR /FPMA(black squares) and
NuSTAR /FPMB (grey triangles) spectra are fitted jointly with a redshifted power-law ( zpowerlw in xspec ) modifiedby the Galactic absorption. The fitted powerlaw index is 1.23 ± ∼ tbabs × zpcfabs × const × [cutoffpl+xillver] along with model components and residuals. INTEGRAL /ISGRI lightcurvespanning nearly 12 years is shown in the bottom panel ofFigure 3. In the lightcurve, the source is clearly detected anda hard X-ray variability over long term is clearly observable.
Although there is an indication of variability in the hardnessratio (i.e., spectral variability) and the presence of a sec-ond spectral component from the flux-flux plot, the detailtime-resolved spectroscopic study is strongly restricted bythe poor signal-to-noise. Hence we perform energy spectralanalysis integrated over the entire observation.
NuSTAR (3-79 keV) and
Swift /XRT (0.8-7 keV) spectra are jointly fitted.The restriction in the high and low energy end of the spec-trum in
Swift /XRT are due to poor signal-to-noise and cal-ibration uncertainties. Spectral fittings are performed using xspec v 12.9.0n . All errors are quoted here correspond to 2 σ significances unless mentioned otherwise. For better signal-to-noise, spectral channels are binned such that there wouldbe at least 60 counts per bin. While the cross-normalizationfactor of FPMA is kept fixed at 1, in case of FPMB and Swift /XRT, it is kept free to vary. From the best fit model,cross-normalizations of FPMB and
Swift /XRT are 1.05 ± ∼ ± ∼ Swift /XRT is found in the range3-20% (Marcotulli et al. 2017; Madsen et al. 2015) which areconsistent with our results. However, we may note that the
NuSTAR exposure was ∼
39 ksec while the
Swift /XRT expo-sure was ∼ Swift /XRT cross calibration constant fromunity due to the short-time scale X-ray flux variability ob-served from the source.We fit the spectra with a power-law modified by theGalactic absorption column density of 0.0092 × cm − (Kawamuro et al. 2016). The model provides an unaccept-able fit with the photon powerlaw index of 1.23 ± χ /dof = 2990/243 (12.3). The data to model ratio as afunction of energy is shown in the left panel of Figure 4.The residual spectrum show a strong negative residual be-low 2 keV which is mostly due to the underestimation of in-trinsic absorption. Residuals are also observed at ∼ zpcfabs in xspec . The fitting improves significantly with χ /dof = 374/241 (1.55). The Hydrogen column density andcovering fraction are found to be 2.47 +0 . − . × cm − and0.96 +0 . − . respectively. However, residuals at ∼ ∼ ∼
20 keVand tails off at higher energies. This indicates the presence ofCompton reflection hump along with a high energy cut-off.To account for the reflection, we use a physically motivated xillver reflection model (Garcia et al. 2013). The latestversion ( xillver-a-ec5 ) can calculate angle-resolved reflec-tion spectra for an incident power-law with the index as lowas 1 and having a cut-off as high as 1000 keV. The additionof the xillver model provides a significant improvement inthe fitting with χ /dof = 300/239 (1.26). While fitting, wetie the incident photon power-law index of xillver model tothat with zpowerlw model. We assume the presence of nohigh energy cut-off and therefore fixed the cut-off parameterof xillver to 1000 keV. The Fe abundance from the fittingis 1.2 ± +0 . − . erg s cm − )and consistent with the neutral value . The disc inclinationangle of NGC 5273 with respect to the observer line of sightis not known. While fitting, the xillver model prefers lowinclination. Fixing this parameter at high inclination (say65 ◦ ) resulting in poor fitting. Therefore, we fix this to 40 ◦ since the fit prefers the range of 28 ◦ -50 ◦ . The preference oflow inclination by the spectral model is also consistent withthe Seyfert type 1.5 and face-on nature of the host galaxy MNRAS , 1–11 (2017)
Pahari et al. × − . × − × − . × − C u t − o ff po w e r l a w no r m a li z a t i on Covering fraction +
68 95 99 . × − . × − . × − . × − C u t − o ff po w e r l a w no r m a li z a t i on Column density N H (10 cm −2 ) +689599 H i gh ene r g y c u t − o ff ( k e V ) Powerlaw photon index + 999568 −6 −6 −6 −6 −6 −6 H i gh ene r g y c u t − o ff ( k e V ) Reflection normalization (ph cm −2 s −1 ) +999568 . . . . C o v e r i ng f r a c t i on Column density N H (10 cm −2 ) +
68 95 99 H i gh ene r g y c u t − o ff ( k e V ) Chain Step
Figure 5.
Using best fit spectra, correlation among different spectral parameters are explored. In all panels (except bottom right), 68%,95% and 99% confidence contours are shown in black, dark grey and light grey respectively. Top left panel shows cut-off power-law fluxand covering fraction (between 96% and 98%) are well-constrained and non-degenerate while top right panel shows absorption columndensity is observed within the expected range of 2.1-2.6 × cm − . Central left panel shows that high energy cut-off is well-constrainedbetween 90 and 220 keV within 2 σ limit while photon power-law index is high but constrained (1.76-1.84). Central right panel showsthat reflection component is well-constrained and non-degenerate with respect to the high energy cut-off. Bottom left panel shows thecovering fraction and absorption column density are well-constrained. Bottom right panel show the High energy cut-off as a function ofchain steps as observed from the Markov Chain Monte Carlo simulations. It is well constrained between 100-200 keV. (Trippe et al. 2010). Left panel of Figure 4 shows that be-yond 40 keV, the data to model ratio drops rapidly and sig-nificantly close to zero. Such a high energy roll-over in theratio spectra strongly indicates the presence of high energycut-off. Therefore, we replace the simple zpowerlw modelwith a cut-off power-law ( cutoffpl model in xspec ) whichis a power-law along with the high energy cut-off as a pa-rameter. While fitting, this parameter is kept free to vary and the high energy cut-off parameter from xillver modelis tied to it. This provide the best-fit model with χ /dof= 264/238 (1.11). The cut-off energy is well-constrained at143 +96 − keV. In order to check whether a redshifted cold ab-sorption model can describe the low energy spectra as goodas z pcfabs, we replace the z pcfabs with z tbabs model. Thebest-fit absorption column density is found to be 2.31 +0 . − . × cm − with χ /dof = 295/239 (1.23). An F-test between MNRAS , 1–11 (2017) uSTAR view of NGC 5273 spectral fitting with partially covering absorption model andthe simple cold absorption model favours the partial cover-ing absorption model with the F-test probability of 2.82 × − which is at least 3 σ significant. Additionally we do notfind any change in the photon index of the primary contin-uum while switching from the partial covering absorptionmodel zpcfabs to a simple cold absorption model ztbabs .Therefore, the steeper powerlaw photon index obtained fromthe present analysis compared to that found with XMM- Newton observation (Cappi et al. 2006) is either a more ac-curate measurement of the photon index due to the inclusionof hard X-ray spectra up to 70 keV or due to the change inthe nature of X-ray emitting corona. The spectrum alongwith the best-fit model, its components and the residualsare shown in the right panel of Figure 4. Best fit parametersalong with 2 σ error bars and X-ray fluxes in different energybands are provided in Table 1. From Table 1, we note thatX-ray flux increase up to 50 keV and then it drops signifi-cantly beyond 100 keV.To check whether fitted parameters are independentlyconstrained and non-degenerate, we produce contour plotsamong different spectral parameters at 68, 95 and 99 percent confidence levels and they are shown in Figure 5. Topleft panel of Figure 5 shows that the covering fraction ofthe absorber as predicted by the zpcfabs model is well con-strained between 95.5 per cent and 98.5 percent with respectto the variation of cut-off powerlaw normalization. Similarly,the absorption column density of partial covering is also well-constrained within the range 2.1-2.6 × cm − with re-spect to the cut-off powerlaw normalization which is shownin the top right panel of Figure 5. Therefore, if we assumethat zpcfabs truly describes the absorption that dominatesthe low energy part of the spectrum, then the column den-sity is moderate to low but the absorbing cloud may havelarge spatial extension so that it nearly blocks the entireline of sight of the observer, therefore strongly reducing low-energy/soft X-ray flux below 2 keV.Central left panel of Figure 5 shows the correlation be-tween the power-law index and high energy cut-off. Due tothe simplicity of the spectrum and the remarkable capa-bilities of NuSTAR in constraining the cut-off beyond theenergy range of its regular operation, the cut-off energyin NGC 5273 can be constrained in the range 90-350 keVwithin the 3 σ limit. It may be noted that even higher en-ergy cut-off was constrained previously using NuSTAR spec-tra for NGC 5506 and IC 4329A at the 3 σ lower limit of 350keV and 170 keV respectively. Within 3 σ limit, the pho-ton power-law index is found in the range 1.76-1.84 whichis steeper compared to earlier measurements using spec-tra which cover only up to 10 keV. Using XMM- Newton spectra, Cappi et al. (2006) reported the power-law indexof 1.4 ± Suzaku /XIS spectra of NGC 5273in 2013, Kawamuro et al. (2016) found the power-law indexof 1.57 +0 . − . which is steeper than the XMM- Newton mea-surement. However, in Figure 5 of Kawamuro et al. (2016),they used the power-law index in the range ∼ NuSTAR spectra. However, itmay also be noted that the source shows both flux vari-ability as well as a hint of spectral variability during the
NuSTAR observation. Therefore the change in the power-
Table 1.
Best fit model parameters from the simultaneous fit-ting of
Swift /XRT and
NuSTAR energy spectra using the model tbabs × zpcfabs × const × [cutoffpl+xillver] . 2 σ errorsare quoted. N H, tbabs is the Galactic absorption colum density,N H, zpcfabs and f c, zpcfabs are the absorption column density andcovering fraction due to the zpcfabs model. Γ cutoffpl and E cutoff are the photon powerlaw index and the high energy cutoff due tothe cutoffpl model. Γ xillver and E cut xillver are the photon pow-erlaw index due to the primary emission and the cutoff energyin the xillver model. A Fe is the Iron abundance with respectto the Solar abundance, ξ, xillver is the ionization parameter and i is the disc inclination angle. F total , F cutoffpl and F xillver arethe total flux, the flux due to cutoffpl and xillver models re-spectively in the energy range 0.1-100 keV. F . − . , F . − . ,F . − . , F . − . , F . − . , F . − . are fluxes in theenergy range 0.3-3 keV, 3-6 keV, 6-20 keV, 20-50 keV, 50-100 keVand 100-150 keV respectively. Spectral FittedParameters valuesN H, tbabs [10 cm − ] 0.0092 (f)N H, zpcfabs [10 cm − ] 2.47 +0 . − . f c, zpcfabs +0 . − . Γ cutoffpl +0 . − . E cutoff [keV] 143 +96 − Γ xillver =Γ cutoffpl A Fe ± ξ, xillver [erg s cm − ] 0.98 +0 . − . E cut, xillver [keV] =E cutoff i [degrees] 40 ◦ (f)F total [10 − ergs s − cm − ] 6.49 +0 . − . F cutoffpl [10 − ergs s − cm − ] 4.84 +0 . − . F xillver [10 − ergs s − cm − ] 1.65 +0 . − . F . − [10 − ergs s − cm − ] 0.11 +0 . − . F − [10 − ergs s − cm − ] 1.76 +0 . − . F − [10 − ergs s − cm − ] 1.33 +0 . − . F − [10 − ergs s − cm − ] 1.95 +0 . − . F − [10 − ergs s − cm − ] 1.36 +0 . − . F − [10 − ergs s − cm − ] 0.43 +0 . − . χ /dof 264/238 law index can be intrinsic to the source owing to the changein the optical depth of the corona. With the variation ofthe cut-off energy, the normalisation of the reflection com-ponent is also found to be well constrained and shown in thecentral right panel of Figure 5. Because of the narrow Ironemission line and the ionisation parameter being consistentwith the neutral value, the reflection may take place in theouter disc, and the reflection model prefers a low inclination( < ◦ ). In the presence of Compton-thick, strong absorber(e.g., NGC 5643, NGC 4941, NGC 4102 etc. where measuredcolumn density is > cm − ; Kawamuro et al. (2016)), ahigh inclination angle with respect to the plane of the ac-cretion disc is naturally preferred to explain the stronglyabsorbed soft X-ray spectra. In case of NGC 5273, three in-dependent observations using XMM- Newton (0.9 ± × cm − ; Cappi et al. (2006)), Suzaku (2.60 +0 . − . × cm − ; Kawamuro et al. (2016)) and Swift /XRT+
NuSTAR (2.36 +0 . − . × cm − ) at three different epochs confirmthe presence of a Compton-thin absorber along the line ofsight in NGC 5273. The bottom left panel of Figure 5 showsthat the covering fraction and the absorption column densityare well constrained. MNRAS , 1–11 (2017)
Pahari et al.
To measure the coronal parameters, we replace cut-offpl model with a thermal Comptonization model comptt assuming the primary powerlaw spectrum is dueto the Comptonization of soft seed photons. For simplicity,parameters of xillver reflection model, obtained from thebest-fit using cutoffpl , are kept fixed except the normal-isation. Alternatively using the slab as well as the spher-ical geometry of the corona, which is a parameter of the comptt model, the temperature and the optical depth ofthe Comptonizing electron cloud are found to be 57 +18 − keVand 0.66 +0 . − . respectively. Therefore, the electron tempera-ture of the Comptonizing electron from the comptt modelis consistent with the range of theoretically predicted highenergy cut-off (Lightman & Zdziarski 1987) which we obtainfrom the cutoffpl model fitting. We compute the intrinsic fractional rms variability overthe measurement noise at different energy bin as thistechnique is proven to be useful to study the true in-trinsic variability of the source as well as to probepossible existence of variable components in the spec-tra (Vaughan et al. 2003; Markowitz & Edelson 2004;Mallick et al. 2016; Lobban et al. 2016; Mallick et al. 2017).Following Rodri´guez-Pascual et al. (1997); Vaughan et al.(2003), fractional rms variability amplitude F var is definedas p ( V − h σ x i ) / h x m i . Here V is the variance of thelightcurve at a given energy, h σ x i is the mean of the squareof errors of counts on each time bin and h x m i is the squareof the mean counts of the entire time-series. The errors onF var is computed following equation (B2) provided in Ap-pendix B in Vaughan et al. (2003). The rms spectrum ofNGC 5273 is shown with 1 σ error-bars in the left panel ofFigure 6. In the case of reflection-dominated spectra, thefractional rms variability usually decreases with the increasein photon energies as seen from few other Seyfert galax-ies, e.g., IC 4329A (Brennman et al. 2014), MCG 6-30-15(Ponti et al. 2004). Due to large error-bars, the decreasingtrend is not significant and the rms spectrum can be fit-ted using a constant model which is shown by the horizon-tal line. However, from 19-50 keV, a decrease in the frac-tional rms is observed from ∼
17 per cent to ∼
12 per centwhich is significant compared to the best-fit constant line.To check whether such a decrease in F var is real, we plotthe background-subtracted, FPMA and FPMB combinedlightcurves in the energy range 5-6 keV (black) and 19-50keV (red) respectively in the right panel of Figure 6. Fromthe visual inspection of both lightcurves, it is clear that the5-6 keV lightcurve shows larger variability compared to the19-50 keV lightcurve, particularly during the first, second,third and last time intervals. From the left panel of Figure6, a slight increase in F var is also observed in the energyrange 10-20 keV although it is not highly significant con-sidering the error-bars. With the increase in energy, suchincreasing trends/bump-like structure in the fractional vari-ability by a few percent is observed previously from MCG6-30-15, NGC 7469, 3C 390.3 and 3C 120 in the similarenergy range (Markowitz, Edelson & Vaughan 2003). How-ever, a significant drop in F var above 20 keV is prominentwhich may be caused by the existence of a high energy cut-off, or at the high energy we have more contribution from the reflected component, which will be less variable com-pared to the power-law component and effectively reducethe high energy variability.Although there is a drop in the fractional rms at higherenergy, the variability is still of the order of 10-12% in 19-50keV band with ∼
39 ksec exposure. A variability in 20-100keV is also visible from the
INTEGRAL /ISGRI lightcurveshown in the bottom panel of Figure 3. Taking into consider-ation the latest
INTEGRAL /IBIS survey (Bird et al. 2016)and rescaling the flux of a close detected AGN, the 2 σ up-per limit of 20-100 keV flux for NGC 5273 is 8.7 × − ergs s − cm − . With NuSTAR , we obtain a higher flux (2 σ lower limit) of 3.31 × − ergs s − cm − in 20-100 keV.Therefore, such a difference in hard X-ray flux measurementby INTEGRAL and
NuSTAR , may be caused by the vari-ability in the hard X-ray emission of the source.
We present the
NuSTAR and
Swift /XRT joint analysis ofobservations in the energy range 0.8-75 keV of a low-mass,Seyfert 1.5 galaxy NGC 5273. The hardness ratio (H-S/H+S;H: 6-20 keV count rate and S: 3-6 keV count rate) as a func-tion of total flux (3-20 keV count rate) is relatively flat whichis often seen from Seyfert galaxies at hard-band dominated,high flux level (Connolly et al. 2014; Lamer et al. 2003). Theplot of hard band (6-20 keV) flux as a function of soft band(3-6 keV) flux shows a hard offset which can be interpretedas the presence of a second spectral component in the hardband (Taylor, Uttley & McHardy 2003). The Seyfert classi-fication of NGC 5273 as 1.5 is also consistent with our X-rayspectral analysis where the reflection model prefers the in-clination angle between 28 ◦ and 50 ◦ with respect to the discnormal. The model preferred inclination angle is also con-sistent with two facts : (1) a low to moderate absorptioncolumn density obtained from our joint X-ray spectral anal-ysis and previous reports and (2) large & variable equivalentwidth of Iron emission line observed in different epochs withXMM- Newton and
NuSTAR compared to what is predictedby the Torus model from Ikeda, Awaki & Terashima (2009).These two facts imply that there may exist a large reservoirof cold absorption material which is not intercepted by theobserver’s line of sight and responsible for a large equiva-lent width of Iron emission line. Similar scenario was alsoproposed by Trippe et al. (2010) using the optical data.From the energy spectral analysis, we find a moder-ate to low absorption column density which is consistentwith earlier measurements using XMM-
Newton and
Suzaku and we also obtain a reasonably high covering fraction 95-98 per cent within the 3 σ limit. Although we have usedpartial covering absorption model, the soft X-ray emis-sion from intermediate and type 2 Seyfert galaxies is dom-inated by the scattered emission from the photo-ionisedmedium (see, e.g., Turner et al. (1997); Awaki et al. (2000,2008)). In NGC 5273, the scattering fraction is ∼ NuSTAR spectra observed in2014, when we replace xillver with a zgauss model at ∼ MNRAS , 1–11 (2017) uSTAR view of NGC 5273
105 20 50 . . . F v a r Energy (keV) 0 10 . . . . . N u S T A R c oun t r a t e ( c t s s − ) Time (sec)
Figure 6.
Left panel shows the fractional rms variability amplitude (F var ) as a function of photon energies, computed using background-subtracted, combined FPMA and FPMB lightcurves. When fitted with a constant (shown by a black line), the F var in 19-50 keV energyrange shows significant decrease then the best-fit. To justify this, we compare lightcurves in the 5-6 keV (black circle) and 19-50 keV (redsquare) energy bands in the right panel. During first three and last time intervals, the larger dispersion in 5-6 keV count rate is clearlyobserved compared to that observed from the 19-50 keV lightcurve. keV, we obtain equivalent width of the Fe emission line tobe 103 ±
23 eV which is consistent with the
Suzaku mea-surement performed with data taken in 2013. Therefore, on ashort-time scale, the Fe line flux do not change which is con-sistent with earlier studies (McHardy et al. 1999). However,from the spectral modelling of XMM-
Newton and
Suzaku observations taken with a gap of 11 years (2002 and 2013respectively), the equivalent width of the Fe-k α line is foundto be ∼
230 eV and ∼
90 eV respectively which imply that onlong time-scale, Fe emission line flux may change and there-fore, the geometry/position of the Fe-line emitting regionchanges during both epoch of observations. The origin of thenarrow Fe-k α line could be the reprocessing of coronal X-rayphotons from the inner part of the circum-nuclear matter orthe outer accretion disc while the change in the equivalentwidth of Fe emission line is possibly due the change in under-lying continuum flux while the reprocessed flux remains un-changed. Comparing the hard X-ray (2-10 keV) flux betweenthe XMM- Newton observation from Cappi et al. (2006) andthe present
NuSTAR observation, we find that the
NuSTAR flux is nearly 2-3 times higher than the XMM-
Newton flux.As a consequence, we find the low equivalent width of theFe emission line with the
NuSTAR spectrum compared tothat of the XMM-
Newton spectrum. Therefore, our best-fit spectral model along with the detection of narrow Fe lineand its low equivalent width are consistent with the primaryemission from an intrinsically and moderately variable hotcorona and a constant reflected emission from the outer disc,the circum-nuclear matter or both.As shown by Kawamuro et al. (2016), the torus modelof Ikeda, Awaki & Terashima (2009) largely underestimatesthe equivalent width of the Fe-K α line compared to what weobserve with XMM- Newton , Suzaku and the current analysisusing
NuSTAR for all values of the torus half-opening an-gle. The reason for under-prediction is the assumption thatthe column density of the cloud along the observer’s lineof sight (N H ) is equal to the equivalent column density ofthe circumnuclear matter that is responsible for the narrowFe emission line. However, because of the Compton-thin na-ture of the source, this assumption may not be valid in thecase of an inclination very close to or higher than the half opening angle of the torus with respect to the disc plane.The moderate/low absorption column density as well as theSeyfert classification of the source imply that the observer’sline of sight may pass through the outer skin of the torus sothat the line of sight column density could actually be dif-ferent (which could be low) than the actual circumnuclearmatter density (may be higher) which is responsible for thenarrow Fe emission line and is not overlapped by our line-of-sight. Because of this difference, the observed equivalentwidth of the narrow Fe-k α line is higher than the model-predicted value if we assume that the location of the circum-nuclear matter is the reprocessing site. Using optical data,Trippe et al. (2010) also found a similar scenario where theobserved column density may arise from dust-free warm ab-sorber while the actual column density, derived from cold ab-sorption, is significantly higher than the line-of-sight columndensity. Therefore, in such complicated scenario, the appli-cability of the torus model by Ikeda, Awaki & Terashima(2009) is not clearly understood. In an alternate scenario,the equivalent width of the narrow Fe emission line can varybetween few eV to a few keV depending on how strongly theprimary continuum is absorbed.Since the mass of NGC 5273 is well-known (4.7 ± × M ⊙ ) from the reverberation mapping, the Edding-ton luminosity is 5.9 ± × ergs s − . Using the Suzaku and
Swift /BAT flux in 2-10 keV, Kawamuro et al.(2016) found the Eddington ratio of 0.0032 assuming thebolometric ratio of 10. Using similar procedure with 2-10keV
NuSTAR flux, we find the Eddington luminosity frac-tion of 0.0093. Using our analysis, L bol /L Edd is found tobe ∼ < − implies that the source can be categorised as a low-luminosity AGN (LLAGN). In LLAGN, the coronal elec-tron temperature is usually expected to be high due to theinefficient cooling of the corona caused by the low supplyof the seed UV/soft X-ray photons (Yuan & Narayan 2014;Xie et al. 2010; Bottcher 2001). Therefore, the cut-off energyis thought to be high. Despite the low Eddington fraction,this does not hold true for NGC 5273 as presented here. Us-ing the broadband spectral modelling, we constrain the highenergy cut-off at 143 +96 − keV with the 2 σ significance. We MNRAS , 1–11 (2017) Pahari et al. also perform the Markov Chain Monte Carlo (MCMC) simu-lation with 10 trials which also constrain the cut-off energybetween 110 keV and 250 keV which is shown in the bottomright panel of Figure 5. Our study indicates that LLAGN canhave moderate to low cut-off energy. Surprisingly two othersources show low cut-off energy despite of the low accretionrate. Using joint Suzaku / NuSTAR broadband spectral anal-ysis of the broad-line radio galaxy 3C 390.3, Lohfink et al.(2015) detected a high energy cut-off at 117 +18 − keV when thesource accretes at ∼
1% of Eddington luminosity. They ob-tained the photon power-law index of 1.71 +0 . − . and the coro-nal electron temperature of 30 +32 − keV. These parametersthat describe coronal properties of 3C 390.3 are qualitativelysimilar to that obtained from the energy spectral analysis ofNGC 5273 presented here. However, it may be noted thatthe core of NGC 5273 is not detected in radio with the Eu-ropean VLBI Network (EVN) at 1.6 GHz and 5 GHz abovethe threshold of few hundreds of µ Jy (Panessa & Giroletti2010). Therefore, the coronal properties of a radio galaxyand a radio-silent AGN could be very similar. Exploring theconnection between the radio emission and coronal prop-erties in AGN is beyond the scope of the present work.Using the
NuSTAR and
Swift /XRT simultaneous observa-tions of the radio-quiet galaxy MCG-5-23-16, Zoghbi et al.(2017) found the cut-off energy in the range 107-152 keV andshowed that the cut-off energy increases with the increase inthe hard X-ray flux. Although the source is obscured, MCG-5-23-16 is accreting at ∼ ∼ × − erg cm − s − (Zoghbi et al. 2014), the blackhole mass of ∼ × M ⊙ (Wandel & Mushotzky 1986)and the redshift of 0.0085 (Wandel & Mushotzky 1986)).The photon power-law index varies in the range 1.78-1.85.These values are qualitatively similar to that obtained fromNGC 5273. Zoghbi et al. (2017) argued that due to verysmall compactness ratio between the corona and the seedphoton source, the coronal plasma is possibly dominated byelectrons, rather than electron-positron pairs. As a conse-quence, the coronal cooling efficiency increases while coro-nal heating rate remains same. A similar explanation can beapplied to the case of NGC 5273 where the low/moderatecut-off is observed at the low Eddington fraction. Not onlyLLAGN, but high luminosity AGN also show a low energycut-off. Low energy cut-off is expected since the adequatesupply of seed photons cool the corona efficiently. For exam-ple, low cut-off energy ∼
108 keV is observed from SWIFTJ2127.4+5654 which accrete 18-22% of Eddington luminos-ity (Marinucci et al. 2014). Very low energy cut-off ( < > AstroSat where si-multaneous observations using three different instruments(SXT, LAXPC and CZTI) in 0.2-150 keV energy range(Singh et al. 2014; Yadav et al. 2016; Vadawale et al. 2016)would be extremely useful to find narrower constraints onthe high energy cut-off as well as the low energy absorptionproperties.
We thank the referee for constructive comments and sug-gestions. We are thankful to
NuSTAR team for making datapublicly available. This research has made use of the
NuS-TAR
Data Analysis Software (NuSTARDAS) jointly de-veloped by the ASI Science Data Center (ASDC, Italy)and the California Institute of Technology (USA).
INTE-GRAL /ISGRI and
Swift /XRT data obtained through theHigh Energy Astrophysics Science Archive Research Centeronline service, provided by the NASA/Goddard Space FlightCenter.
REFERENCES
Awaki, H., Terashima, Y., Higaki, Y., Fukazawa, Y. 2009, PASJ,61, 317Awaki, H., Anabuki, N., Fukazawa, Y., Gallo, L. C., Ikeda, S.,Isobe, N. et al. 2008, PASJ, 60, 293Awaki, H., Ueno, S., Taniguchi, Y., Weaver, K. A. 2000, ApJ,542, 175Balokovic´, M., Matt, G., Harrison, F. A., Zoghbi, A., Ballantyne,D. R. et al. 2015, ApJ, 800, 62Bentz, M. C., Horenstein, D., Bazhaw, C., Manne-Nicholas, E. R.et al. 2014, ApJ, 796, 8Bianchi, S., Balestra, I., Matt, G., Guainazzi, M., Perola, G. C.,2003, A&A, 402, 141Bianchi, S., Matt, G., Balestra, I., Guainazzi, M., Perola, G. C.,2004, A&A, 422, 65Bird, A. J., Bazzano, A., Malizia, A., Fiocchi, M., Sguera, V.,Bassani, L. et al. 2016, ApJS, 223, 15Bottcher, M. 2001, ApJ, 553, 960Brenneman, L. W., Madejski, G., Fuerst, F., Matt, G., Elvis, M.,Harrison, F. A. et al. 2014, ApJ, 788, 61Cappi, M., Panessa, F., Bassani, L., et al. 2006, A&A, 446, 459Connolly, S. D., McHardy, I. M., Dwelly, T. 2014, MNRAS, 440,3503Courvoisier T. J.-L., et al., 2003, A&A, 411, L53de La Calle Pe´rez, I., Longinotti, A. L., Guainazzi, M., Bianchi,S., Dovciak, M., Cappi, M., Matt, G., Miniutti, G., et al. 2010,A&A, 524, A50Fabian, A. C., Lohfink, A., Kara, E., Parker, M. L., Vasudevan,R., Reynolds, C. S. 2015, MNRAS, 451, 4375Garcia J., Dauser T., Reynolds C. S., Kallman T. R., McClintockJ. E., Wilms J., Eikmann W., 2013, ApJ, 768, 146Haardt, F., Maraschi, L. 1991, ApJ, 380, L51Harrison, F. A., Craig, W. W., Christensen, F. E., Hailey, C. J.,Zhang, W. W., Boggs, S. E., et al. 2013, ApJ, 770, 103Ikeda, S., Awaki, H., & Terashima, Y. 2009, ApJ, 692, 608Kalberla, P. M. W., Burton, W. B., Hartmann, Dap, Arnal, E.M. et al. 2005, A&A, 440, 775Kawaguchi, T., Shimura, T., Mineshige, S. 2001, ApJ, 546, 966Kawamuro, T., Ueda, Y., Tazaki, F., Terashima, Y., Mushotzky,R., 2016, ApJ accepted, arXiv:1604.07915Keek, L. and Ballantyne, D. R. 2016, MNRAS, 456, 2722Lamer G., McHardy I. M., Uttley P., Jahoda K., 2003, MNRAS,338, 323Lightman, A. P., & Zdziarski, A. A. 1987, ApJ, 319, 643Liu, B. F., Taam, R. E., Qiao, E., Yuan, W. 2015, ApJ, 806, 223Liu, B. F., Yuan, W., Meyer, F., Meyer-Hofmeister, E., Xie, G.Z. 1999, ApJ, 527, L17Liu, B. F., Xie, G. Z., Ji, K. F. 1995, Ap&SS, 225, 299Lobban, A. P., Vaughan, S., Pounds, K., Reeves, J. N. 2016, MN-RAS, 457, 38Lohfink, A. M., Ogle, P., Tombesi, F., Walton, D., Balokovic´, M.,Zoghbi, A. et al. 2015, ApJ, 814, 24MNRAS , 1–11 (2017) uSTAR view of NGC 5273 Malizia, A., Molina, M., Bassani, L., Stephen, J. B., Bazzano, A.,Ubertini, P., Bird, A. J. 2014, ApJ, 782, 25Mallick, L., Dewangan, G. C., McHardy, I. M., Pahari, M., 2017preprint (arXiv:1703.06432)Mallick, L. and Dewangan, G. C., 2017 preprint(arXiv:1702.08383)Mallick, L., Dewangan, G. C., Gandhi, P., Misra, R., Kembhavi,A. K. 2016, MNRAS, 460, 1705Marinucci A. et al., 2015, MNRAS, 447, 160Marinucci A. et al., 2014, MNRAS, 440, 2347Markowitz, A. and Edelson, R. 2004, ApJ, 617, 939Markowitz, A., Edelson, R., Vaughan, S. 2003, ApJ, 598, 935Madsen, K. K., Harrison, F. A., Markwardt, C. B., et al. 2015,ApJS, 220, 8Marcotulli, L., Paliya, V. S., Ajello, M., Kaur, A., Hart-mann, D. H., Gasparrini, D. et al. 2017, accepted in ApJ,arXiv:1703.10657Matt, G., Balokovic´, M., Marinucci, A., Ballantyne, D. R., Boggs,S. E. et al. 2015, MNRAS, 447, 3029McHardy, I. M., Papadakis, I. E. and Uttley, P. 1999, NuPhS, 69,509Molina, M., Bassani, L., Malizia, A., Stephen, J. B., Bird, A. J.,Bazzano, A., Ubertini, P. 2013, MNRAS, 433, 1687Nicastro, F., Piro, L., De Rosa, A., Feroci, M. et al. 2000, ApJ,536, 718Osterbrock, D. E. and Martel, A. 1993, ApJ, 414, 552Pahari, M., Misra, R., Dewangan, G. C., Pawar, P. 2015, ApJ,814, 158Panessa, F. and Giroletti, M. 2010, ASPC, 427, 377Ponti, G., Cappi, M., Dadina, M., Malaguti, G. 2004, A&A, 417,451Rodri´guez-Pascual, P. M., Alloin, D., Clavel, J., Crenshaw, D.M., Horne, K. et al. 1997, ApJS, 110, 9Singh, K. P., Tandon, S. N., Agrawal, P. C., Antia, H. M., Man-chanda, R. K., Yadav, J. S. et al. 2014, SPIE, 9144, 1Taam, R. E., Liu, B. F., Yuan, W., Qiao, E. 2012, ApJ, 759, 65Taylor, R. D., Uttley, P., McHardy, I. M., 2003, MNRAS, 342,L31Titarchuk, L. and Lyubarskij, Y. 1995, ApJ, 450, 876Tonry, J. L., Dressler, A., Blakeslee, J. P. et al. 2001, ApJ, 546,681Trippe, M. L., Crenshaw, D. M., Deo, R. P., Dietrich, M. et al.2010, ApJ, 725, 1749Tully, R. B., Shaya, E. J., Karachentsev, I. D., et al. 2008, ApJ,676, 184Turner, T. J., George, I. M., Nandra, K., Mushotzky, R. F. 1997,ApJ, 488, 164Ubertini, P., Lebrun, F., Di Cocco, G., Bazzano, A., Bird, A. J.et al. 2003, A&A, 411, L131Vadawale, S. V., Rao, A. R., Bhattacharya, D., Bhalerao, V. B.,Dewangan, G. C. et al. 2016, proc. SPIE, 9905, 1Vaughan, S., Edelson, R., Warwick, R. S., Uttley, P. 2003, MN-RAS, 345, 1271Wandel, A. and Mushotzky, R. F. 1986, ApJL, 306, L61Xie, F., Niedz´wiecki, A., Zdziarski, A. A., Yuan, F. 2010, MNRAS,403, 170Yadav, J. S., Agrawal, P. C., Antia, H. M., Verdhan Chauhan,Jai, Dedhia, Dhiraj, Katoch, Tilak et al. 2016, proc. SPIE,9905, 1Yuan, F. and Narayan, R. 2014, ARA&A, 52, 529Zoghbi, A., Matt, G., Miller, J. M., Lohfink, A. M. et al. 2017,ApJ, 836, 2Zoghbi, A., Cackett, E. M., Reynolds, C., Kara, E., Harrison, F.A., Fabian, A. C. et al. 2014, ApJ, 789, 56Zycki, P. T., Collin-Souffrin, S., Czerny, B. 1995, MNRAS, 277,70 This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000