Compton-thick AGN in the NuSTAR era II: A deep NuSTAR and XMM-Newton view of the candidate Compton thick AGN in NGC 1358
We present the combined NuSTAR and XMM-Newton 0.6–79 keV spectral analysis of a Seyfert 2 galaxy, NGC 1358, which we selected as a candidate Compton thick (CT-) active galactic nucleus (AGN) on the basis of previous Swift/BAT and Chandra studies. According to our analysis, NGC 1358 is confirmed to be a CT-AGN using physical motivated models, at 3 confidence level. Our best-fit shows that the column density along the “line-of-sight” of the obscuring material surrounding the accreting super-massive black hole is N = [1.96–2.80] 10 cm. The high-quality data from NuSTAR gives the best constraints on the spectral shape above 10 keV to date on NGC 1358. Moreover, by combining NuSTAR and XMM-Newton data, we find that the obscuring torus has a low covering factor ( 0.17), and the obscuring material is distributed in clumps, rather than uniformly. We also derive an estimate of NGC 1358’s Eddington ratio, finding it to be 10, which is in acceptable agreement with previous measurements. Finally, we find no evidence of short-term variability, over a 100 ks time-span, in terms of both “line-of-sight” column density and flux.
Subject headings:galaxies: active – galaxies: nuclei – galaxies: individual (NGC 1358) – X-rays: galaxies
The Cosmic X-ray Background (CXB; i.e., the diffuse X-ray emission observed between 0.5 keV and 300 keV) is thought to be mainly produced by obscured and unobscured active galactic nuclei (AGN; e.g., Alexander03; Gandhi03; gilli07; Treister09). Compton-thick (CT-) AGNs (with absorbing column density N 10 cm, where is the Thomson cross section) are supposed to contribute up to 10% of the CXB intensity at its spectral peak ( 30 keV, Ajello08) and are expected to be numerous (up to 50% of the overall population of Seyfert 2 galaxies; see, e.g., risaliti1999). However, as of today CT-AGNs have never been detected in large numbers, their observed fraction in the local Universe being 5–10% (see, e.g., Burlon11; Ricci15), significantly below the predictions of different CXB models (20%–30%, see Ueda14, and references therein). Nevertheless, it has been suggested that the small observed fraction of heavily obscured AGN observed can be caused by the bias in detecting CT-AGN in X-rays, even sampling the energy range above 10 keV (see, e.g. Burlon11). Efforts to correct for this observational bias have recovered a fraction of 20 % of CT-AGN, under some assumptions (see, e.g., Burlon11; BNtorus; Ricci15).
In Compton-thick AGN, the spectrum is significantly suppressed at energies 10 keV (gilli07; koss2016) and the overall emission is dominated by the Compton hump at 30–50 keV. Consequently, CT-AGNs at redshifts 1 can be studied using one of the several facilities sampling the 0.5–10 keV energy range, such as XMM-Newton, Chandra, Swift-XRT and Suzaku (see, e.g., Georgantopoulos2013; Buchner2015; Lanzuisi2105), since the Compton hump of high- sources is redshifted in the energy range covered by these instruments. For sources in the local Universe ( 0.1), however, the proper characterization of heavily obscured AGN requires an X-ray telescope sensitive above 10 keV. Thanks to the launch of Nuclear Spectroscopic Telescope Array (hereafter, NuSTAR, harrison), which provides a two orders of magnitude better sensitivity than previous telescopes at these energies (e.g., INTEGRAL and Swift/BAT; Winkler2003; Barthelmy2005), we can characterize the physical properties of heavily obscured AGN with unprecedented accuracy (see, e.g., Balokovic14; puccetti14; Annuar15; Stefano2017; Ursini18). However, since a typical highly obscured AGN spectrum barely depends on the column density at 10 keV but varies considerably at 10 keV (see, e.g., gilli07), it is difficult to constrain the column density with NuSTAR alone. Consequently, XMM-Newton, as the best instrument in terms of effective area in 0.3–10 keV (10 times larger than Swift-XRT and 2 times larger than Chandra), is the ideal instrument to complement NuSTAR strength in characterizing heavily obscured AGNs.
Indeed, the study of single targets using NuSTAR or combining NuSTAR and other lower-energy X-ray observatories (e.g., XMM-Newton and Chandra) has already been shown to be strategical to characterize heavily obscured AGN, and understand their physical properties. For example, NGC 1448 was observed and identified as a CT-AGN in X-rays for the first time using NuSTAR and Chandra (Annuar17). The source was too faint (intrinsic 2–10 keV luminosity = 3.5–7.6 10 erg s) to be identified by Swift/BAT, even using its deepest 104 month maps, and was only detected in one out of five Swift-XRT observations. Another example is the analysis of NGC 1068 reported in Bauer15. In this work, the authors used NuSTAR to characterize with unprecedented quality this largely studied CT-AGN, putting much stronger constraints on the high-energy spectral shape of NGC 1068.
The obscuration observed in AGN across the electromagnetic spectrum, from the X-ray, to the optical and infrared, is usually explained with a pc-scale, torus-like structure of dust and gas (see, e.g., Natureastro2017). Consequently, in the past two decades several tori models, based on Monte Carlo simulations, have been developed to characterize CT-AGN X-ray spectra (Matt1994; Shinya09; MYTorus2009; BNtorus; Liu14; Furui16; Borus). All these models assume a continuous distribution of the obscuring material, but with different assumption on the geometry of the torus. In particular, in the models proposed by Shinya09, BNtorus and Borus, the half opening angle of torus, i.e., the torus covering factor, is a free parameter, thus allowing to put constraints on the typical tori geometry. Given the intrinsic complexity of these models, and the multiple free parameters involved, using them in full capacity requires high-quality X-ray spectra, with excellent statistics on a wide energy range, i.e., between 2 and 100 keV: as of today, such requirements can be satisfied only by a joint NuSTAR and XMM-Newton observation.
In this work we present the results of a deep, 50 ks joint NuSTAR and XMM-Newton observation of NGC 1358, a nearby Seyfert 2 galaxy and a CT-AGN candidate. The paper is organized as follows: in Section 2 we present the selection technique that brought us to classify NGC 1358 as a new candidate CT-AGN, and we report the NuSTAR and XMM-Newton data reduction and spectral extraction process. In Section 3, we describe the different models, both phenomenologicals and physicals, which have been used to fit the spectra, and the results of the spectral analysis. In Section LABEL:discussion we compare our results with previous ones, derive the source Eddington ratio and discuss the constraints on the geometry and clumpiness of the obscuring materials. All reported errors are at 90% confidence level, if not otherwise stated. Standard cosmological constants are adopted as follows: = 70 km s Mpc, = 0.0 and = 0.73.
2. Observation and Data Analysis
NGC 1358 (z 0.013436, Theureau1998), is a Seyfert 2 galaxy detected in the 100-month BAT catalog (with a 7.8 significance; Segreto et al. 2018 in prep.), a catalog of 1000 AGN detected by Swift-BAT in the 15–150 keV band.
In marchesi2017APJ, we describe a technique developed to select highly obscured AGN candidates from the BAT sample, using the following criteria:
Lack of a 0.5–2.4 keV, ROSAT/RASS (Boller16) counterpart. For objects located outside the Galactic plane (i.e., having Galactic latitude 10 ), the lack of ROSAT counterparts already implies a minimum AGN column density log(N) 23 (see, e.g., Figure 2 in koss2016).
Seyfert 2 galaxy optical classification, i.e., the source must have an optical spectrum without broad (FWHM 2000 km s) emission lines. It has been shown (see, e.g., marchesi2016, and references therein) that Seyfert 2 galaxies are more likely to be obscured than Seyfert 1 ones. Furthermore, there are no known Seyfert 1 galaxy that are Compton thick111there are sources which are Compton-thick but with ambiguous activity classification, e.g. NGC 424 (a.k.a. Tololo0109-383) (see, e.g., Ricci15).
Low redshift ( 0.04). Due to selection effects, the vast majority of BAT-selected CT-AGNs are detected in the nearby Universe: for example, 47 out of 55 CT-AGNs reported in Ricci15 are located at 0.04.
Following these criteria, we obtained a snapshot (10 ks) Chandra observation for a sample of seven sources, and we performed a first measurement of their fundamental spectral parameters, particularly the power law photon index, , and the column density, N. NGC 1358 was found to be the most obscured object in our sample, having “line-of-sight” column density N = 1.05 10 cm, thus making it a candidate CT-AGN, although only at a 1 confidence level, due to the low-quality of the Chandra spectrum.
To further investigate this new candidate CT-AGN we proposed for a joint deep NuSTAR (50 ks) and XMM-Newton (48 ks) follow-up observation, which was accepted in NuSTAR Cycle 3 (proposal ID 3258, PI: Marchesi). We report a summary of the two observations in Table 1.
|Instrument||Sequence||Start Time||End Time||Exposure Time||Count Rateaafootnotemark:|
|XMM-Newton||0795680101||2017-08-01T17:05:27||2017-08-02T06:03:10||48||0.980.05 0.910.05 3.680.15|
The reported NuSTAR net count rates are those of the FPMA and FPMB modules between 3–79 keV, respectively. The reported XMM-Newton net count rates are those the MOS1, MOS2 and pn modules in 0.6–10 keV, respectively.
2.1. NuSTAR Observation
NGC 1358 was observed by NuSTAR on 2017 August 1 (ObsID 60301026002): the net exposure time is 50 ks. The observation actually took place in a 96.9 ks time-span and was divided in 16 (3 ks) intervals. The non-exposed time between each interval is when the target is occulted by the Earth.
The NuSTAR data are derived from both focal plane modules, FPMA and FPMB. The raw files are calibrated, cleaned and screened using the NuSTAR nupipeline script version 0.4.5. The NuSTAR calibration database (CALDB) used in this work is the version 20161021. The ARF, RMF and light-curve files are obtained using the nuproducts script.
For both modules, the source spectrum is extracted from a 25 circular region, corresponding to 40% of the encircled energy fraction at 10 keV, centered on the source optical position. We then extract a background spectrum for each module, choosing a 30 circular region located nearby the outer edges of the field of view, to avoid contamination from NGC 1358. We group the NuSTAR spectra with a minimum of 15 counts per bin with task. The signal of both modules is 3 in 3–79 keV band.
2.2. Xmm-Newton Observation
The XMM-Newton observation was taken quasi-simultaneously to the NuSTAR one starting 12 hours after the NuSTAR one, but ending at the same time (due to the gaps between observing intervals in NuSTAR). XMM-Newton data have been reduced using the Science Analysis System (SAS; SAS) version 16.1.0. 13 ks of XMM-Newton modules MOS1 and MOS2 and 30 ks of pn observations were affected by a strong background flare, therefore we decided to exclude that part of observation from our analysis. Consequently, the total net XMM-Newton exposure time of our observation is 101 ks. The source spectra are extracted from a 15, corresponding to 70% of the encircled energy fraction at 1.5 keV, circular region, while the background spectra are from a 80 circle located nearby the source. We visually inspected the XMM-Newton image to avoid contamination to the background from sources nearby NGC 1358. All three modules, MOS1, MOS2 and pn are jointly used in the spectral modeling, and their normalization are tied together assuming their cross-calibration uncertainties are marginal.
3. Spectral Modeling Results
We use XSPEC (Arnaud1996) v12.9.1 to fit the spectrum and rely on the statistic for the optimization of the spectral fit. The photoelectric cross section for all absorption components used here are derived from Verner1996, adopting an element abundance from Anders1989. The Galactic absorption column density is N = cm (Kalberla05). The metal abundance is fixed to Solar.
Following a standard approach in analyzing heavily obscured AGN, we first fit our data using different phenomenological models, particularly the pexrav one (pexrav). We then move to more accurate self-consistent models, based on Monte Carlo simulations, which are specifically developed to treat the spectra of heavily obscured AGN: the physical models we use in this work are MYTorus (MYTorus2009) and borus02 (Borus). We report the results of our analysis in the following sections.
3.1. Phenomenological Models
3.1.1 Absorbed power law
We initially fit our data with a simple phenomenological model, comprising a power law (zpowerlw in XSPEC) absorbed by intervening gas modeled with (zphabs). We also add a Gaussian (zgauss) to model the Fe K fluorescent emission line (E = 6.4 keV); we assume the line to be narrow, fixing the line width to 50 eV, since there is no statistical improvement in fits if the parameter is left free to vary. We also add a second, unabsorbed power law, to model the fractional AGN emission, which is not intercepted by the torus on the “line-of-sight”, and/or the scattering emission that is deflected, rather than absorbed by the obscuring material. Here, and elsewhere in the paper, the photon index of the scattered component is tied to the one of the main power law. The scattered component is usually less than 5–10% of the main one (see, e.g., Marchesi2018). We denote this fraction as , and we model it with a constant (). Furthermore, we add to the fit a thermal component, namely mekal (mekal), to model the soft excess observed below 1 keV, and potentially due to either star-formation processes and/or thermal emission from a hot interstellar medium. The temperature and the relative metal abundance in mekal are both left free to vary.
The first model (hereafter, “Model A”), in XSPEC nomenclature, is therefore:
where represents the cross calibration between different instruments, noted as . In our fits, the cross-calibration between different modules of the same instrument is fixed to 1. phabs is applied here to model the Galactic absorption.
We report in Table 3.1.2 the best-fit results for the simple phenomenological model applied to the joint NuSTAR–XMM-Newton spectrum. The best-fit photon index is = 1.14; the column density is N = 0.95 cm. While the best-fit reduced of model A is statistically acceptable, being = /degree of freedom (d.o.f. hereafter) = 256/240 = 1.07, standard absorption components in XSPEC, such as zphabs, fail to characterize the spectral complexity of heavily obscured AGN like NGC 1358 properly. Therefore, a more physical model needs to be applied.
3.1.2 Including a reflection component
Obscured AGNs X-ray spectra have historically been modeled using the pexrav model (pexrav). pexrav is used to model an exponentially cut-off power law spectrum reflected from neutral slab. We first test the model with a pure reflector by setting the reflection scaling factor in pexrav to be = -1: this models a heavily obscured (N cm) source whose spectrum is dominated by the reflection from the “back-side” of the torus. The fit shows that photon index is = and /d.o.f = 349/242. Such a large reduced suggests that a pure reflector is not enough to describe the spectrum. Therefore, we follow the method described in, e.g., Ricci11 by using the complete pexrav model, which includes an intrinsic cut-off power law by setting the reflection scaling factor to be greater than 0.
The model in XSPEC is described as follows:
The components are those described previously in Section 3.1.1, except for the main power law, which is replaced by pexrav. The inclination angle , i.e., the angle between the axis of the AGN (normal to the disk) and the observer “line-of-sight”, which is a free parameter in pexrav, is fixed at = 60 (i.e. cos = 0.5): we find no significant change in the best-fit statistic and in the other parameters when allowing to vary. The cut-off energy of pexrav is fixed at 500 keV, to be consistent with the MYTorus model, which we will extensively discuss in the following section.
When leaving the reflection scaling factor in pexrav free to vary, we obtain a best-fit value of R 4 (such a large reflection scaling factor is also found by Ricci11, in heavily obscured AGN), although we are not able to put any constraint on the parameter 90% confidence uncertainties. Such a result would point towards a “reflection-dominated” scenario, where most of the observed emission comes from the reflected component, while the direct emission from the accreting SMBH is absorbed by the heavily obscuring material along the “line of sight”. A larger scaling factor can also be interpreted as the geometry of reflected material is more like a torus rather than a disk 222as a larger value of the scaling factor represents a larger amount of reflected material, thus the reflected material is more torus-like rather than disk-like geometrically.. Since is not constrained when left free to vary, we decided to complete our spectral analysis fixing the reflection scaling parameter to = 4. Here we are modeling a process that the intrinsic emission together with the reflection from the “back-side” are obscured by the same circumnuclear material.
We report in Table 3.1.2 the best-fit parameters for the analysis of the joint NuSTAR–XMM-Newton spectra using model B. The photon index is = 1.59. The best-fit column density is N = 0.76 cm. In agreement with what we found using Model A, the source is near the threshold of CT-AGN. We present the unfolded NuSTAR and XMM-Newton spectrum of NGC 1358, fitted with the model B and ratio between data and model, in figure 1.
In summary, both phenomenological models suggest that obscuration is near the Compton-thick threshold, such that the source cannot be confirmed as CT-AGN at 3 confidence level. However, the photon indices obtained above are far from the typical value observed in AGN ( 1.8, see, e.g., marchesi2016), showing that some components may not be well described by the above phenomenological models. Therefore, more physically motivated models are needed to describe the spectra and extract the physical and geometrical properties of NGC 1358.
|(coupled)||(decoupled face on)||(decoupled edge on)|
|333 = is the cross calibration between NuSTAR and XMM-Newton.||1.06||1.13||1.12||1.13||1.17||1.16|
|N444“line-of-sight” column density in phenomenological models in cm||0.95||0.76||…||…||…||…|
|norm555normalization of components in different models at 1 keV in photons keV cm s. 10||0.03||0.04||0.13||0.20||1.61||1.26|
|666angle between the axis of the torus and the edge of torus in degree, where the covering factor = cos().||…||…||…||…||…||84.0|
|kT777temperature in the thermal component mekal in keV.||0.49||0.49||0.58||0.57||0.58||0.52|
|abund888abundance in the thermal component mekal.||0.05||0.07||0.03||0.04||0.11||0.05|
|F999Flux between 2–10 keV in erg cm s.||4.18||4.09||4.03||4.03||3.84||3.87|
|F101010Flux between 10–40 keV in erg cm s.||8.22||8.68||8.48||8.55||8.51||8.51|
|L111111Intrinsic luminosity between 2–10 keV in erg s. 121212Intrinsic luminosity between 10–40 keV in erg s.|