IC 751: A NEW CHANGING LOOK AGN DISCOVERED BY NUSTAR

Variability of the line of sight column density (${N}_{{\rm{H}}}$) might be rather common in active galactic nuclei (AGNs; Risaliti et al. 2002; Bianchi et al. 2012; Torricelli-Ciamponi et al. 2014), and in the past decade several objects have been found to show eclipses of the X-ray source, both due to Compton-thick (CT; ${N}_{{\rm{H}}}\gtrsim {10}^{24}\;{\mathrm{cm}}^{-2}$) and to Compton-thin (${N}_{{\rm{H}}}\lt {10}^{24}\;{\mathrm{cm}}^{-2}$) material. Since the X-ray source is believed to be located very close to the supermassive black hole (SMBH), this variable absorption could be associated either with broad-line region (BLR) clouds, or with clumps in the molecular torus. In at least a few cases (e.g., Risaliti et al. 2009; Maiolino et al. 2010) these variations have been found to occur on timescales of days, and are consistent with being related to material in the BLR. Markowitz et al. (2014) have recently shown, by studying RXTE light curves of 55 AGNs, that for eight objects of their sample there seems to be variation of absorbing material also on longer timescales (months to years). They associated these changes in ${N}_{{\rm{H}}}$ with clumps in the molecular torus. Interestingly, none of the eclipses detected by Markowitz et al. (2014) were due to CT material.

So far variations in the ${N}_{{\rm{H}}}$ of the neutral absorber have been found in more than 20 AGNs including 1H 0419−577 (Pounds et al. 2004), Centaurus A (Beckmann et al. 2011; Rivers et al. 2011), ESO 323−G77 (Miniutti et al. 2014), H0557–385 (Longinotti et al. 2009), MR 2251–178, Mrk 348, Mrk 509 (Markowitz et al. 2014), Mrk 6 (Immler et al. 2003), Mrk 766 (Risaliti et al. 2011), Mrk 79 (Markowitz et al. 2014), NGC 1068 (Marinucci et al. 2016), NGC 1365 (Risaliti et al. 2005, 2007; Maiolino et al. 2010; Walton et al. 2014; Rivers et al. 2015b), NGC 3227 (Lamer et al. 2003), NGC 3783 (Markowitz et al. 2014), NGC 4151 (Puccetti et al. 2007), NGC 4388 (Elvis et al. 2004), NGC 4395 (Nardini & Risaliti 2011), NGC 4507 (Braito et al. 2013; Marinucci et al. 2013), NGC 454 (Marchese et al. 2012), NGC 5506 (Markowitz et al. 2014), NGC 6300 (Guainazzi 2002), NGC 7582 (Piconcelli et al. 2007; Bianchi et al. 2009; Rivers et al. 2015a), NGC 7674 (Bianchi et al. 2005), PG 2112+059 (Gallagher et al. 2004), UGC 4203 (Guainazzi et al. 2002; Risaliti et al. 2010), and SWIFT J2127.4+5654 (Sanfrutos et al. 2013). In most cases these variations are due to Compton-thin material, and only for a handful of sources is the varying absorber CT (i.e., ESO 323−G77, NGC 1068, NGC 1365, NGC 454, NGC 6300, NGC 7582, NGC 7674, UGC 4203). AGNs switching between Compton-thin and CT states are usually called changing look AGNs (e.g., Matt et al. 2003) because their spectral shape changes dramatically (from transmission-dominated to reflection-dominated). In the optical band, changing look AGNs are objects that transition from type 1 to type 1.8/1.9/2 (e.g., LaMassa et al. 2015), or from type 1.8/1.9/2 to type 1 (e.g., Shappee et al. 2014). In the following, we will refer to the X-ray classification only.

IC 751 ($z=0.0311$, D = 137 Mpc; Falco et al. 1999) is a type 2 AGN (Véron-Cetty & Véron 2010) in an edge-on spiral galaxy (de Vaucouleurs et al. 1991) that has not yet been studied in detail in the X-ray band. The source was reported in the 70 month Swift/Burst Alert Telescope (BAT) catalog (Baumgartner et al. 2013), and was observed by Nuclear Spectroscopic Telescope Array (NuSTAR) as part of the campaign aimed at following up on Swift/BAT-detected sources (M. Balokovic et al. 2016, in preparation). The interesting X-ray characteristics of this object triggered several follow-up observations with NuSTAR. We report here on the five NuSTAR observations of this source carried out between 2012 and 2014, two of which were performed jointly with XMM-Newton. The source switches from a CT to Compton-thin state between the different observations, and is the first changing look AGN discovered by NuSTAR. Following our X-ray spectral and temporal analysis, we put constraints on the location of the varying obscuring material. Throughout the paper, we consider a luminosity distance of the source of ${d}_{{\rm{L}}}=135$ Mpc, and adopt standard cosmological parameters (${H}_{0}=70\;\mathrm{km}\;{{\rm{s}}}^{-1}\;{\mathrm{Mpc}}^{-1}$, ${{\rm{\Omega }}}_{{\rm{m}}}=0.3$, ${{\rm{\Omega }}}_{{\rm{\Lambda }}}=0.7$). Unless otherwise stated, all uncertainties are quoted at the 90% confidence level.

IC 751 was observed five times by NuSTAR, twice jointly with XMM-Newton (PI F. Bauer), and two times by the Swift/X-Ray Telescope (XRT). Details about these observations are reported in Table 1.

2.1. NuSTAR

2.2. XMM-Newton

2.3. Swift

We performed X-ray spectral analysis with XSPEC v.12.8.2 (Arnaud 1996). To all models we added a photoelectric absorption component (tbabs; Wilms et al. 2000) to take into account Galactic absorption in the direction of the source, fixing the value of the column density to ${N}_{{\rm{H}}}^{\;{\rm{G}}}\;=1.2\times {10}^{20}\;{\mathrm{cm}}^{-2}$ (Kalberla et al. 2005). In order to use ${\chi }^{2}$ statistics, NuSTAR FPMA/FPMB and XMM-Newton EPIC PN and MOS spectra were binned to have at least 20 counts per bin. Quoted errors correspond to a 90% confidence level (${\rm{\Delta }}{\chi }^{2}=2.7$). Given the low signal-to-noise of the Swift/XRT spectrum of Observation 4, we did not bin the spectrum, and applied Cash statistics (Cash 1979; cstat in XSPEC).

As a first step to shed light on the spectral variability, we used a model which considers reprocessed radiation from a slab to reproduce the X-ray spectrum of IC 751. Since the first two data sets lacked simultaneous coverage below 5 keV, and the Swift/XRT observation in Observation 4 did not have a very high signal-to-noise ratio, we used the two joint NuSTAR/XMM-Newton spectra (Observations 5 and 6, Section 3.1) to constrain the fundamental parameters (photon index, normalization of the reflection component, normalization of the Fe Kα line), and then used this information to fit the other observations (Observations 2, 3, and 4, Section 3.2), in order to constrain the value of the line of sight column density, ${N}_{{\rm{H}}}$, and the normalization of the X-ray primary emission. Observation 1 was not used due to its poor statistics. It was, however, possible to infer the 2–10 keV flux of the X-ray source during this observation (${3.1}_{-2.4}^{+0.4}\times {10}^{-13}\;\mathrm{erg}\;{\rm{s}}\;{\mathrm{cm}}^{-2}$), which is consistent with that of Observation 3.

3.1. Observations 5 and 6

The slab model we used to analyze the broadband X-ray spectrum of IC 751 includes: (i) a power law with a high-energy cutoff, which represents the primary X-ray emission; (ii) photoelectric absorption and Compton scattering, to take into account the line of sight obscuration; (iii) unabsorbed X-ray reprocessed radiation; (iv) a Gaussian line to reproduce the Fe Kα emission; (v) a cutoff power-law component to reproduce the scattered X-ray emission; (vi) a thermal plasma model, to take into account a possible contribution of the host galaxy to the X-ray spectrum below ∼2 keV. It must be stressed that the emission below 2 keV, which cannot be accounted for by the scattered component alone, could also be due to the blending of emission lines created by photo-ionization. However, due to the limited energy resolution of our data, in the following we will use a thermal plasma model. To reproduce photoelectric absorption and Compton scattering, we used the tbabs and cabs models, respectively. We took into account the reprocessed X-ray radiation using the pexrav model (Magdziarz & Zdziarski 1995), which assumes reflection from a semi-infinite slab. The value of the reflection parameter (R) was set to be negative in order to include only the reprocessed component. The values of the photon index (Γ), cutoff energy (${E}_{{\rm{C}}}$), and the normalization were fixed to the values of the cutoff power law, while the inclination angle of the observer with respect to the reflecting slab was set to $i=30^\circ$. This angle was selected considering the type 2 nature of IC 751, and assuming that the reflecting material is associated with the molecular torus. The width of the Fe Kα line (σ) was fixed to 40 eV, a value well below the energy resolution of XMM-Newton and NuSTAR, while the energy and normalization of the line were left free to vary. The width of the Fe Kα was chosen considering that the bulk of the line is created in material located in the BLR or in the molecular torus (e.g., Shu et al. 2010; Ricci et al. 2014a, 2014b). The scattered X-ray emission was taken into account by multiplying an additional unabsorbed cut off power law by a constant (${f}_{\mathrm{scatt}}$, typically of the order of a few percent), fixing all the parameters to those of the primary X-ray emission. A multiplicative constant to include possible cross-calibration offset between the different instruments was added, and was found to be typically $\lesssim 12\%$ of unity. In XSPEC our model is:

constant×tbabs_Gal×[ztbabs×cabs×cutoffpl + apec + pexrav + zgauss +f_scatt×cutoffpl].

The results of the spectral fitting are reported in Table 2. In both Observations 5 and 6 we found the source in a Compton-thin state, with a photon index of ${\rm{\Gamma }}\sim 1.9$ and a high-energy cutoff of ${E}_{{\rm{c}}}\gtrsim 200\;{\rm{keV}}$. The two observations show a change in ${N}_{{\rm{H}}}$ (Figure 2), with Observation 5 being less obscured than Observation 6 (${\rm{\Delta }}{N}_{{\rm{H}}}\sim {10}^{23}\;{\mathrm{cm}}^{-2}$, significant at a $\sim 2\sigma$ level). The two observations also show evidence of intrinsic flux variation, with the source being significantly brighter in Observation 5. The reflection component appears to be rather weak, with a value of $R={0.14}_{-0.13}^{+0.16}$ in Observation 6, while it is less constrained in Observation 5 due to the higher flux level. The fraction of scattered flux is found to be ${f}_{\mathrm{scatt}}\sim 0.5\%$ of the primary X-ray flux. The X-ray spectra and the ratio between the data and the best-fitting model are shown in Figure 1.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 2.  X-Ray Spectral Analysis—Slab Model

(1)	(2)	(3)	(4)	(5)	(6)
Model Parameters	Obs. 2	Obs. 3	Obs. 4	Obs. 5	Obs. 6
	(2012 Oct 28)	(2013 Feb 04)	(2013 May 23)	(2014 Nov 28)	(2014 Nov 30)
Photon index Γ	1.96a	1.74a	1.73a	${1.93}_{-0.12}^{+0.13}$	${1.89}_{-0.17}^{+0.17}$
Cutoff energy ${E}_{{\rm{C}}}$ (keV)	≥222a	≥222a	≥222a	$\geqslant 192$	$\geqslant 222$
Reflection param. R	0.02a	0.27a	0.28a	$\leqslant 0.17$	${0.14}_{-0.13}^{+0.16}$
Col. density ${N}_{{\rm{H}}}$ (${10}^{22}\;{\mathrm{cm}}^{-2}$)	${199}_{-67}^{+75}$	${121}_{-27}^{+29}$	${32}_{-6}^{+9}$	${38}_{-3}^{+3}$	${49}_{-6}^{+7}$
Fraction of scattered component ${f}_{\mathrm{scatt}}$ (%)	${17.3}_{-10.5}^{+9.0}$	0.9a	0.9a	${0.5}_{-0.2}^{+0.2}$	${0.6}_{-0.2}^{+0.3}$
Temperature kT (keV)	0.94b	0.94b	1.01a	${0.82}_{-0.09}^{+0.15}$	${0.94}_{-0.19}^{+0.07}$
Energy (Fe Kα) (keV)	6.25a	6.36a	6.25a	${6.42}_{-0.09}^{+0.08}$	${6.36}_{-0.11}^{+0.10}$
Norm. (Fe Kα) (${10}^{-6}$ photons $\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	2.1a	1.9a	1.5a	${1.4}_{-1.0}^{+1.1}$	${1.3}_{-0.8}^{+0.8}$
EW(Fe Kα) (eV)	${281}_{-218}^{+43}$	${489}_{-26}^{+955}$	${130}_{-65}^{+617}$	${60}_{-44}^{+33}$	${108}_{-39}^{+64}$
${F}_{\;2-10}^{\;\mathrm{obs}}$ (${10}^{-12}\;\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	${0.7}_{-0.5}^{+0.1}$	${0.22}_{-0.08}^{+0.01}$	${0.62}_{-0.06}^{+0.02}$	${1.11}_{-0.12}^{+0.07}$	${0.56}_{-0.09}^{+0.03}$
${F}_{2-10}$ (${10}^{-12}\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	${22}_{-15}^{+3}$	${3.4}_{-1.2}^{+0.2}$	${2.1}_{-0.2}^{+0.1}$	${5.9}_{-0.6}^{+0.4}$	${3.8}_{-0.6}^{+0.2}$
${F}_{\;10-50}^{\;\mathrm{obs}}$ (${10}^{-12}\;\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	${4.2}_{-2.4}^{+0.5}$	${3.5}_{-1.6}^{+0.1}$	${3.9}_{-0.3}^{+0.3}$	${5.2}_{-0.6}^{+0.3}$	${3.5}_{-0.7}^{+0.2}$
${F}_{10-50}$ (${10}^{-12}\;\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	${22}_{-13}^{+3}$	${6.7}_{-3.1}^{+0.2}$	${4.7}_{-0.4}^{+0.4}$	${7.3}_{-0.8}^{+0.4}$	${5.2}_{-1.0}^{+0.3}$
$\mathrm{log}{L}_{2-10}$ ($\;\mathrm{erg}\;{{\rm{s}}}^{-1}$)	43.7	42.9	42.7	43.1	42.9
$\mathrm{log}{L}_{10-50}$ ($\;\;\mathrm{erg}\;{{\rm{s}}}^{-1}$)	43.7	43.2	43.0	43.2	43.1
${\chi }^{2}/\mathrm{dof}$	31.3/27	84.7/95	131.9/111c	256.8/257	224.8/176

Notes. The table reports the model parameters obtained by fitting the X-ray spectra of the five epochs with the slab model: constant×tbabs_Gal×[ztbabs×cabs×cutoffpl + apec + pexrav + zgauss + constant×cutoffpl] in XSPEC; the luminosities reported here are the intrinsic (i.e., absorption-corrected) values. For details see Section 3.1.

^aValue of the parameter left free to vary within the uncertainties of Observation 6. ^bValue of the parameter fixed. ^cPoissonian statistics applied for Swift/XRT dta. The uncertainties reported for the Fe Kα EW correspond to the 68% confidence interval.

Download table as:  ASCIITypeset image

3.2. Observation 2, 3, and 4

To further investigate the structure of the absorbers, and to disentangle the torus absorption from that caused by clouds in the line of sight, assuming an homogeneous torus, we used the MYTorus model¹⁹ (Murphy & Yaqoob 2009). The MYTorus model considers absorbed and reprocessed X-ray emission from a smooth torus with a half-opening angle ${\theta }_{\mathrm{OA}}$ of 60°, and can be used for spectral fitting as a combination of three additive and exponential table models. These tabulated models include the zeroth-order continuum²⁰ (mytorusZ), the scattered continuum (mytorusS), and a component which contains the fluorescent emission lines (mytorusL).

4.1. Standard MYTorus

The analysis we carried out using the slab model (Section 3) showed that ${N}_{{\rm{H}}}$ is highly variable, so that it cannot be associated with a smooth absorber alone. This is also confirmed by the fact that applying the smooth MYTorus model to all the X-ray spectra available, setting the values to be the same for the different observations, results in a chi-squared of ${\chi }^{2}=2628.2$ for 707 dof. Also considering different normalizations of the direct and scattered component or different values of the column density of the scattering and absorbing material fails to reproduce the X-ray spectrum, resulting in values of the reduced chi-squared of ${\chi }_{\nu }^{2}\gt 2$. We therefore used an alternative approach to take into account variable absorption by combining the non-varying torus absorption [mytorusZ(Tor)] with what we define as the cloud absorption. The geometry of the absorber we assume is shown in Figure 3. We adopted a model which includes the three components of MYTorus plus a collisionally ionized plasma, and power law to reproduce the scattered component, similar to what was done in Section 3. In order to take into account the variable absorber, we used an additional obscuring multiplicative component [mytorusZ(Cloud, $\;{N}_{{\rm{H}}}^{{\rm{C}}}$ )], where ${N}_{{\rm{H}}}^{{\rm{C}}}$ is the column density of the cloud. In XSPEC the syntax of our model is:

Zoom In Zoom Out Reset image size

constant×tbabs_Gal × [mytorusZ(Cloud) × mytorusZ(Tor) × zpowerlaw + mytorusS + apec + gsmooth(mytorusL) +f_scatt × zpowerlaw].

The free parameters of MYTorus are the photon index Γ, the equatorial column density of the torus ${N}_{\;{\rm{H}}}^{\;{\rm{T}}}$, and the inclination angle of the observer ${\theta }_{{\rm{i}}}$. We convolved the fluorescent emission lines of MYTorus using a Gaussian function (gsmooth in XSPEC) to take into account the expected velocity broadening. We fixed the width of the lines to a FWHM of $2000\;\mathrm{km}\;{{\rm{s}}}^{-1}$, consistent with the average value obtained for 36 AGNs at $z\lt 0.3$ by the Chandra/HEG study of Shu et al. (2010). The model was also multiplied by a constant to take into account cross-calibration between the different spectra.

We simultaneously fitted the five sets of observations discussed above. The values of ${\theta }_{{\rm{i}}}$, ${N}_{{\rm{H}}}^{\;{\rm{T}}}$, Γ, and the normalization of the scattered component (${n}_{\mathrm{refl}}$, which includes the Compton hump) were left free to vary, and tied to be constant for all observations. The normalization of the fluorescent lines was fixed to ${n}_{\mathrm{refl}}$. The normalization of the primary power-law component (${n}_{\mathrm{po}}$), the value of ${N}_{{\rm{H}}}^{{\rm{C}}}$, and that of ${f}_{\mathrm{scatt}}$ were left free to have independent values for different observations.

The results obtained by this torus model are reported in the upper part of Table 3. The fit yields a chi-squared of ${\chi }^{2}=774.8$ (for 688 dof), and the primary X-ray emission has a value of the photon index consistent with that found using the slab model. As with the slab model, we find a clear variation of the line of sight column density, with the clouds having values of the column density spanning between $1.5\times {10}^{24}$ and $5\times {10}^{21}\;{\mathrm{cm}}^{-2}$. This approach also confirms a significant change in the value of ${f}_{\mathrm{scatt}}$ between the five observations, which varies between $\sim 4\%$ and $\sim 0.4\%$. These variations in ${f}_{\mathrm{scatt}}$ are interpreted as being due to a partially covering absorber in the line of sight. Contrary to what we obtained using the slab model, applying MYTorus we do not find a significant variation of the line of sight column density between Observation 5 and Observation 6. The model used, with the parameters set to those obtained for Observation 2, is shown in Figure 4.

Zoom In Zoom Out Reset image size

Table 3.  X-Ray Spectral Analysis—Torus Model

MYTorus
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Observation	${N}_{\;{\rm{H}}}^{\;{\rm{C}}}$	${n}_{\mathrm{po}}$	Γ	${N}_{\;{\rm{H}}}^{\;{\rm{T}}}$	${\theta }_{{\rm{i}}}$	${n}_{\mathrm{refl}}$	kT	${f}_{\mathrm{scatt}}$
	(${10}^{22}\;{\mathrm{cm}}^{-2}$)	(${10}^{-3}\;\mathrm{ph}\;{\mathrm{keV}}^{-1}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)		(${10}^{24}\;{\mathrm{cm}}^{-2}$)	(deg)	(${10}^{-3}\;\mathrm{ph}\;{\mathrm{keV}}^{-1}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	(keV)	(%)
2 [2012.82]	${150}_{-15}^{+19}$	${3.83}_{-0.42}^{+0.43}$	${1.88}_{-0.04}^{+0.01}$	${4.76}_{-0.27}^{+0.09}$	${60.3}_{-0.2}^{+0.2}$	${1.14}_{-0.11}^{+0.15}$	${0.93}_{-0.04}^{+0.04}$	${4.2}_{-0.8}^{+0.8}$
3 [2013.10]	${111}_{-7}^{+8}$	${2.10}_{-0.17}^{+0.18}$	//	//	//	//	//	${1.6}_{-0.5}^{+0.5}$
4 [2013.39]	${5.1}_{-3.1}^{+3.0}$	${0.94}_{-0.06}^{+0.06}$	//	//	//	//	//	${3.2}_{-1.2}^{+1.5}$
5 [2014.91]	${0.58}_{-0.05}^{+0.08}$	${1.83}_{-0.05}^{+0.07}$	//	//	//	//	//	${0.4}_{-0.1}^{+0.1}$
6 [2014.92]	${0.49}_{-0.05}^{+0.10}$	${0.75}_{-0.04}^{+0.04}$	//	//	//	//	//	${0.8}_{-0.1}^{+0.1}$
MYTorus—Decoupled Mode
Observation	${N}_{\;{\rm{H}}}^{\;{\rm{C}}}$	${n}_{\mathrm{po}}$	Γ	${N}_{\;{\rm{H}}}^{\;{\rm{T}}}(Z)$	${N}_{\;{\rm{H}}}^{\;{\rm{T}}}(S,L)$	${n}_{\mathrm{refl}}$	kT	${f}_{\mathrm{scatt}}$
	(${10}^{22}\;{\mathrm{cm}}^{-2}$)	(${10}^{-3}\;\mathrm{ph}\;{\mathrm{keV}}^{-1}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)		(${10}^{24}\;{\mathrm{cm}}^{-2}$)	(${10}^{24}\;{\mathrm{cm}}^{-2}$)	(${10}^{-3}\;\mathrm{ph}\;{\mathrm{keV}}^{-1}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$)	(keV)	(%)
2 [2012.82]	${147}_{-48}^{+82}$	${4.71}_{-2.45}^{+6.29}$	${1.98}_{-0.07}^{+0.08}$	${0.37}_{-0.02}^{+0.02}$	${6.0}_{-2.2}^{+3.5}$	${1.22}_{-0.36}^{+0.50}$	${0.93}_{-0.11}^{+0.05}$	${4.1}_{-2.0}^{+3.4}$
3 [2013.10]	${105}_{-26}^{+32}$	${2.43}_{-0.90}^{+1.60}$	//	//	//	//	//	${1.6}_{-0.7}^{+0.9}$
4 [2013.39]	$\leqslant 18$	${1.20}_{-0.30}^{+0.45}$	//	//	//	//	//	${2.7}_{-1.8}^{+3.4}$
5 [2014.91]	${0.64}_{-0.07}^{+0.17}$	${2.08}_{-0.03}^{+0.05}$	//	//	//	//	//	${0.4}_{-0.1}^{+0.1}$
6 [2014.92]	${11.1}_{-4.9}^{+5.0}$	${1.28}_{-0.27}^{+0.38}$	//	//	//	//	//	${0.6}_{-0.3}^{+0.2}$

Note. Model parameters obtained by simultaneously fitting the 15 X-ray spectra of IC 751 (divided in five epochs). The table reports (1) the observation number, (2) the cloud column density, (3) the normalization of the primary power-law emission, (4) the photon index of the primary X-ray emission, (5) the equatorial column density of the torus, (6) the inclination angle of the observer (see Figure 3), (7) the normalization of the reflection component, (8) the temperature of the collisionally ionized plasma, and (9) the fraction of scattered unabsorbed emission. The upper part of the table refers to the MYTorus model in his original formulation (with the addition of a cloud of neutral material; see Section 4.1), while the lower part reports the results obtained by using MYTorus in the decoupled mode (Section 4.2). In the lower part column (5) is the torus column density of the mytorusZ component for ${\theta }_{{\rm{i}}}(Z)=90^\circ$, and (6) is the column density of the mytorusS and mytorusL components, assuming ${\theta }_{{\rm{i}}}(S,L)=0^\circ$ and ${\theta }_{{\rm{i}}}(S,L)=90^\circ$.

Download table as:  ASCIITypeset image

We found an equatorial column density of ${N}_{{\rm{H}}}^{\;{\rm{T}}}\simeq \;4.8\times {10}^{24}\;{\mathrm{cm}}^{-2}$ and an inclination angle of the observer of ${\theta }_{{\rm{i}}}\simeq 60\_\_AMP\_\_fdg;3$, close to grazing incidence. The line of sight column density in the toroidal geometry assumed by MYTorus can be obtained by

$$\begin{eqnarray}&&{N}_{{\rm{H}}}^{{\rm{T}}}{({\theta }_{{\rm{i}}})={N}_{{\rm{H}}}^{{\rm{T}}}(1-4{\mathrm{cos}}^{2}{\theta }_{{\rm{i}}})}^{\displaystyle \frac{1}{2}},\end{eqnarray} \tag{ 1 }$$

which implies that for IC 751 ${N}_{{\rm{H}}}^{{\rm{T}}}({\theta }_{{\rm{i}}})\simeq 6.4\times {10}^{23}\;{\mathrm{cm}}^{-2}$, a value larger than the lowest column density inferred by adopting the slab model. However, given the dependence of the column density on the inclination angle, and the problems associated with the toroidal geometry for inclination angles close to the edges (see discussion in Yaqoob 2012), the uncertainty associated with this value is large.

4.2. Decoupled MYTorus

X-ray observations have shown that besides a highly variable line of sight column density, IC 751 also presents significant flux variability of the primary X-ray source on timescales of days to months both in the soft and hard X-ray bands (see Tables 2 and 3). As illustrated in Figure 5 (filled points), the observed flux of IC 751 varies by a factor of 4 in the 2–10 keV band, and by a factor of 1.6 in the 10–50 keV band. When considering intrinsic fluxes (i.e., absorption-corrected, empty points in Figure 5), the amplitude of the variability is larger: the X-ray source varies by a factor of ∼5 both in the 2–10 and 10–50 keV bands. The average observed fluxes in the 2–10 keV and 10–50 keV bands are $6.4\times {10}^{-13}\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$ and $4.0\times {10}^{-12}\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$, respectively. The intrinsic average nuclear fluxes in the two bands are $7.4\times {10}^{-12}\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$ (2–10 keV) and $9.3\times {10}^{-12}\;\mathrm{erg}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}$ (10–50 keV), which correspond to k-corrected average luminosities of $\mathrm{log}({L}_{2-10}/\;\mathrm{erg}\;{{\rm{s}}}^{-1})=43.22$ and $\mathrm{log}({L}_{10-50}/\;\mathrm{erg}\;{{\rm{s}}}^{-1})=43.32$. The 12 μm rest frame luminosity of IC 751 could be estimated using WISE by linearly interpolating the fluxes in the W3 (11.56 μm) and W4 (22.09 μm) bands. The mid-IR flux of IC 751 is dominated by the AGNs, since $W1-W2\gt 0.8$ (Stern et al. 2012). We found that the X-ray luminosity is in agreement with the 12 μm  luminosity $[\mathrm{log}({L}_{12\mu {\rm{m}}}/\;\mathrm{erg}\;{{\rm{s}}}^{-1})=43.70$], as expected from the well known mid-IR/X-ray correlation (e.g., Gandhi et al. 2009, Asmus et al. 2015, Stern 2015).

Zoom In Zoom Out Reset image size

In order to improve our constraints on the absorbing material, and to study its evolution on shorter timescales, we analyzed the XMM-Newton EPIC/PN and NuSTAR light curves of IC 751 in different energy bands. We extracted XMM-Newton EPIC/PN light curves in the 0.3–10, 0.3–2, and 2–10 keV bands, with bins of 1 ks. NuSTAR FPMA light curves were extracted in the 3–79, 3–20, and 20–60 keV bands with bins of 6 ks. We also analyzed the variability of the hardness ratio, defined as ${HR}=H-S/H+S$, where H and S are the fluxes in the soft (0.3–2 and 3–20 keV) and hard (2–10 and 20–60 keV) bands, respectively.

For all the observations we performed a ${\chi }^{2}$ test in order to assess the variability of the hardness ratio and the flux in the three different energy bands. We considered the flux or the hardness ratio to be variable if the minimum confidence level was $p\leqslant 1\%$. To constrain the amplitude of the variability, we used the rms variability amplitude (${F}_{\mathrm{var}}$; see Equation (10) and B2 of Vaughan et al. 2003). NuSTAR light curves of Observation 2 show significant variability both in the broad and in the soft bands, while the other NuSTAR observations do not show any sign of short-term variability. The XMM-Newton light curve of Observation 5 shows significant flux variation in the hard band, while the flux is consistent with being constant during Observation 6. The largest value of the rms variability amplitude is found for the NuSTAR 3–20 keV light curve of Observation 2 (${F}_{\mathrm{var}}=34\pm 7\%$). The hardness ratio is significantly variable only for the NuSTAR light curve of Observation 2 ($p\simeq 0.5\%$, ${F}_{\mathrm{var}}=59\pm 16\%$), which shows a hardening of the spectrum in the last ∼6 ks of the observation.

The 104 month 14-195 Swift/BAT light curve does not show evidence of significant long-term variability ($p\sim 6\%$) on a timescale of 25 Ms.

We also carried out a time-resolved spectral analysis of the longest NuSTAR observation (Observation 3) by splitting it in five time intervals with similar lengths. We applied the slab model described in Section 3, leaving the normalization of the primary X-ray emission and the line of sight column density free to vary, and found that the parameters obtained are consistent within their 90% confidence interval between all observations. Leaving the value of ${f}_{\mathrm{scatt}}$ free to vary significantly improves the fit only for the second (${\rm{\Delta }}{\chi }^{2}\simeq 8$) and third (${\rm{\Delta }}{\chi }^{2}\simeq 7$) segments, and results in values consistent with those obtained for the other segments and for the whole observation.

The X-ray observations of IC 751 presented here show clear evidence of changes in the line of sight column density, with the X-ray source being obscured by Compton-thick $[\mathrm{log}({N}_{{\rm{H}}}/\;{\mathrm{cm}}^{-2})\sim 24.3{\rm{]}}$ material in two observations, and by Compton-thin material $[\mathrm{log}({N}_{{\rm{H}}}/\;{\mathrm{cm}}^{-2})\simeq 23.60{\rm{]}}$ in three observations. This result is confirmed adopting both the slab (Section 3, top panel of Figure 6) and the torus (Section 4, bottom panel of Figure 6) X-ray spectral models. In particular, by using a physical torus model, we were able, assuming a smooth and azimuthally symmetric torus, to disentangle the intrinsic obscuration associated with a non-varying absorber from the column density of the varying absorber. Variations in the observed line of sight column density might be related either to: (i) intrinsic variation of the absorbing material, caused by moving clouds; (ii) changes in the intensity of the nuclear radiation, which would cause a variation in the ionization state of the absorbing material. Although the observations during which the highest values of the column density were also found to correspond to the stage in which the source was more luminous, for a low-density photo-ionized absorber there might be delay between the flux variation and the response of the absorber, so that the second scenario cannot be completely discarded. In the following, we will, however, assume that changes in ${N}_{{\rm{H}}}$ are related to clouds eclipsing the X-ray source, as found for several other changing look AGNs (e.g., Risaliti et al. 2005).

Zoom In Zoom Out Reset image size

Following Risaliti et al. (2007) and Marinucci et al. (2013), the distance of the cloud from the X-ray source (R_c) can be estimated by considering that the size of the source and that of the cloud are similar (${D}_{c}\simeq {D}_{{\rm{s}}}$), and that the transverse velocity is given by the ratio between the size of the source (D_s) and the crossing time T_cr: ${V}_{k}={D}_{s}/{T}_{\mathrm{cr}}$. It must be remarked that in Observation 2 we found possible evidence of partial covering, which would imply that ${D}_{c}\lt {D}_{s}$. By applying the slab model, we found that the cloud covers 83% of the X-ray source. The value of the covering factor was larger when adopting the torus model ($\sim 96\%$). Given the rather large values of the covering factor, taking the possible difference between D_c and D_s into account does not significantly affect our results. Assuming that the cloud is moving with a Keplerian velocity, we obtain:

$$\begin{eqnarray}&&{R}_{c}=\displaystyle \frac{{{GM}}_{\mathrm{BH}}}{{V}_{k}^{2}}=\displaystyle \frac{{{GM}}_{\mathrm{BH}}{T}_{\mathrm{cr}}^{2}}{{D}_{s}^{2}}.\end{eqnarray} \tag{ 2 }$$

Micro-lensing (e.g., Chartas et al. 2002, 2009), occultation studies (Risaliti et al. 2009), and large-amplitude rapid X-ray variability have shown that the size of the X-ray source is ${D}_{{\rm{s}}}\simeq 10\;{r}_{{\rm{g}}}$, where ${r}_{{\rm{g}}}={{GM}}_{\mathrm{BH}}/{c}^{2}$. Assuming that ${D}_{{\rm{s}}}=10\;{r}_{{\rm{g}}}$, we obtain

$$\begin{eqnarray}&&{R}_{c}=\displaystyle \frac{{{GM}}_{\mathrm{BH}}{T}_{\;\mathrm{cr}}^{2}}{{10}^{2}{R}_{\;{\rm{G}}}^{2}}\simeq 2\;\mathrm{pc}\;{M}_{8.5}\;{R}_{10}^{-2}\;{T}_{10}^{2},\end{eqnarray} \tag{ 3 }$$

where T₁₀ is the crossing time in units of 10 days ($8.64\times {10}^{5}$ s) and ${M}_{8.5}={M}_{\mathrm{BH}}/{10}^{8.5}\;{M}_{\odot }$.

The black hole mass of IC 751 has recently been obtained by the study of the stellar velocity dispersion as part of work aimed at constraining the characteristics of Swift/BAT-selected AGNs in the optical band (M. J. Koss et al. 2016, in preparation), and is $\mathrm{log}({M}_{\mathrm{BH}}/{M}_{\odot })\simeq 8.5$. By using Equation (3), and considering that (i) no significant variation was found during the 50 ks of NuSTAR Observation 3 (over a total of 100 ks), and that (ii) the shortest interval in which a variation of the CT material is evident is between Observation 3 and Observation 4, which were carried out 108 days apart, we can say that the distance of the cloud is between ${R}_{\mathrm{min}}=0.027$ pc (∼32 light days) and ${R}_{\mathrm{max}}\simeq 230$ pc.

Optical reverberation-mapping studies have shown that the radius of the BLR scales with the square root of the luminosity (e.g., Kaspi et al. 2005). According to Kaspi et al. (2005), considering the Hβ lags and the results obtained averaging different observations of the same object, the radius of the BLR is given by

$$\begin{eqnarray}&&\displaystyle \frac{{R}_{{\rm{BLR}}}}{10\;\mathrm{lt}-\mathrm{days}}=0.86\times {\left(\displaystyle \frac{{L}_{2-10}}{{10}^{43}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{0.532}.\end{eqnarray} \tag{ 4 }$$

For the average 2–10 keV luminosity of IC 751, we find ${R}_{{\rm{BLR}}}=11.7$ light days, which implies that the obscuring clouds responsible for the variation of ${N}_{{\rm{H}}}$ are beyond the emission-weighted average radius of the BLR. Using a similar approach, it is possible to put constraints on the location of the hot inner wall of the torus, which is also known to scale with the square root of the luminosity (e.g., Suganuma et al. 2006, Kishimoto et al. 2011). Following Tristram & Schartmann 2011 (Figure 4 of their paper), the inner radius of the hot dust (${R}_{{\rm{NIR}}}$), obtained from K-band reverberation, can be approximated by:

$$\begin{eqnarray}&&\mathrm{log}\displaystyle \frac{{R}_{{\rm{NIR}}}}{1\;\mathrm{pc}}=-23.10+0.5\mathrm{log}{L}_{14-195},\end{eqnarray} \tag{ 5 }$$

where ${L}_{14-195}$ is the 14–195 keV luminosity (in $\;\mathrm{erg}\;{{\rm{s}}}^{-1}$). At 12 μm, interferometric studies (Tristram & Schartmann 2011; see also Burtscher et al. 2013) have shown that the size of the mid-IR-emitting region (${R}_{{\rm{MIR}}}$) in AGNs can be estimated by

$$\begin{eqnarray}&&\mathrm{log}\displaystyle \frac{{R}_{{\rm{MIR}}}}{1\;\mathrm{pc}}=-21.62+0.5\mathrm{log}{L}_{14-195}.\end{eqnarray} \tag{ 6 }$$

The 70 month averaged 14–195 keV luminosity of IC 751 is $\mathrm{log}({L}_{14-195}/\mathrm{erg}{{\rm{s}}}^{-1})=43.47$ (Baumgartner et al. 2013), which corresponds to ${R}_{{\rm{NIR}}}=0.04$ pc (∼48 light days) and ${R}_{{\rm{MIR}}}=1.3$ pc. From this we can conclude that the absorbing material could be related to the outer BLR, to clumps in the molecular torus, or even to material located at a further distance from the SMBH.

A change in the Compton-thin absorber is also found between Observation 5 and Observation 6 using the slab spectral model and MyTORUS in its decoupled mode. The two observations were carried out about 48 hr apart, which means (applying Equation (3)) that for the Compton-thin material the $R\lesssim {R}_{\mathrm{max}}=0.08$ pc (∼95 light days). If the Compton-thin and CT clouds are located at the same distance from the X-ray source, then the regions where the varying absorbers are located are between 32 and 95 light days. This would imply that the absorber is consistent with being located either in the BLR or in the inner side of the dusty torus. Recent work has shown that the Fe Kα might also arise in this region (Gandhi et al. 2015; Minezaki & Matsushita 2015). Assuming that the cloud has about the same size of the X-ray source, i.e., $\sim 10\;{r}_{{\rm{g}}}$ ($\sim 4.7\times {10}^{14}$ cm), and that its column density is $1.5\times {10}^{24}\;{\mathrm{cm}}^{-2}$, the density of the cloud would be $n\sim 3.2\times {10}^{9}\;{\mathrm{cm}}^{-3}$. This value is in agreement with that expected for the BLR clouds (e.g., Peterson 1997). A BLR origin for the varying absorber in IC 751 would fit what has been found so far for other changing look AGNs, several of which show absorbers compatible with being part of the BLR (e.g., Maiolino et al. 2010; Risaliti et al. 2010; Burtscher et al. 2015).

Thanks to its broadband coverage, NuSTAR is a very powerful tool to study obscuration in AGNs, and it has been shown to be fundamental to well constrain the line of sight column density (e.g., Arévalo et al. 2014; Bauer et al. 2015; Gandhi et al. 2014; Annuar et al. 2015; Koss et al. 2015; Lansbury et al. 2015). Studying type II quasars, Lansbury et al. (2015) have shown that the estimates of ${N}_{{\rm{H}}}$ obtained by NuSTAR can be 2.5–1600 times higher than previous constraints from XMM-Newton and Chandra. This shows that in the absence of high-quality broadband observations, it would be possible to miss changing look events for weak sources. Another clear example is given by the recent detection of an unveiling event in NGC 1068 (Marinucci et al. 2016), which would have been missed by observations carried out below 10 keV. Repeated NuSTAR observations of obscured sources might therefore uncover a significant number of new changing look events. Burtscher et al. (2015) have recently reanalyzed the relation between ${N}_{{\rm{H}}}$ and the optical obscuration ${A}_{{\rm{V}}}$, and found that in several cases the deviation of ${N}_{{\rm{H}}}/{A}_{{\rm{V}}}$ from the Galactic value is due to variable absorption. This would imply that ideal targets to study occultations of the X-ray source are objects showing a large deviation from the Galactic ${N}_{{\rm{H}}}/{A}_{{\rm{V}}}$ value.

We reported here on the spectral analysis of five NuSTAR observations of the type 2 AGN IC 751, three of which were combined with XMM-Newton or Swift/XRT observations in the 0.3–10 keV range. IC 751 is the first changing look AGN (i.e., an object that has been observed both in a Compton-thin and a CT state) discovered by NuSTAR. We find that the X-ray source was obscured by CT material during the first two observations, while its line of sight obscuration is found to be Compton-thin during the following observations, which implies that absorption varies on timescales of $\lesssim 3$ months. Changes of the line of sight column density are also found on a timescale of ∼48 hr (${\rm{\Delta }}{N}_{{\rm{H}}}\sim {10}^{23}\;{\mathrm{cm}}^{-2}$). While we cannot constrain the location of the absorber precisely, by considering the lack of spectral variability during the longest NuSTAR observation we can infer the minimum distance to be further than the emission-weighted average radius of the BLR. Assuming that the varying Compton-thin and CT clouds are located at the same distance from the X-ray source, then the material is located between 32 and 95 light days. The absorber could therefore be related to either the external part of the BLR or to the inner part of the dusty torus, although the BLR origin might be slightly favored since the density of the clouds is found to be consistent with the value expected for BLR clouds. By adopting a physical torus X-ray spectral model, we are able to disentangle the column density of the non-varying absorber (${N}_{{\rm{H}}}\sim 3.8\times {10}^{23}\;{\mathrm{cm}}^{-2}$) from that of the varying clouds $[{N}_{{\rm{H}}}\quad \sim$(1–150)$\quad \times \quad {10}^{22}\;{\mathrm{cm}}^{-2}$], and to put constraints on the column density of the reprocessing material (${N}_{{\rm{H}}}\;\sim 6\times {10}^{24}\;{\mathrm{cm}}^{-2}$). We found that the X-ray source is highly variable in both the 2–10 keV and 10–50 keV bands. Future observational campaigns on IC 751 in the X-ray band will be able to improve the constraints on the location of the varying absorber, and confirm or not whether it is related to clouds in the BLR as has been found for several objects of this class.

We thank the anonymous referee for comments that helped us to improve the quality of our manuscript. CR acknowledges Marko Stalevski and Sebastien Guillot for fruitful discussions, and Chin-Shin Chang for her comments on the manuscript. This research has made use of the NuSTAR Data Analysis Software (NuSTARDAS) jointly developed by the ASI Science Data Center (ASDC, Italy) and the California Institute of Technology (Caltech, USA). We acknowledge financial support from the CONICYT-Chile grants "EMBIGGEN" Anillo ACT1101 (CR, ET, FEB, PA), FONDECYT 1141218 (CR, FEB), FONDECYT 1140304 (PA), Basal-CATA PFB-06/2007 (CR, ET, FEB), NuSTAR subcontract 44A-1092750 (WNB), the Swiss National Science Foundation and Ambizione fellowship grant PZ00P2 154799/1 (MK), and the Ministry of Economy, Development, and Tourism's Millennium Science Initiative through grant IC120009, awarded to The Millennium Institute of Astrophysics, MAS (FEB). This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, of data obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), provided by NASA's Goddard Space Flight Center, and of the SIMBAD Astronomical Database which is operated by the Centre de Données astronomiques de Strasbourg.

Facilities: NuSTAR - The NuSTAR (Nuclear Spectroscopic Telescope Array) mission, Swift - Swift Gamma-Ray Burst Mission, XMM-Newton - .