EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 546 (October 2012) A&A, 546 (2012) A84 Full HTML
Free Access
Issue
A&A
Volume 546, October 2012
Article Number A84
Number of page(s) 8
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201220036
Published online 09 October 2012

© ESO, 2012

1. Introduction

A population of heavily obscured active galactic nuclei (AGN) at cosmological distances, which might be missed by conventional quasar surveys, has been postulated by AGN synthesis models of the X-ray background (XRB, e.g., Gilli et al. 2007; Treister et al. 2009a,b) and the super-massive black hole mass function in the local Universe (Marconi et al. 2004). Various infrared selections have been employed extensively for seaching for these objects in which strong re-radiation from obscuring dust is expected (Martínez-Sansigre et al. 2005; Alonso-Herrero et al. 2006; Daddi et al. 2007; Fiore et al. 2008, 2009; Bauer et al. 2010; Vignali et al. 2010; Alexander et al. 2011; Luo et al. 2011; Donley et al. 2012). Although X-ray observations should, in principle, also be effective for the search on account of the intrinsic X-ray loudness of AGN (relative to galaxy emission) and the penetrating power against obscuration, the low throughput of the existing X-ray telescopes limits the accessibility to high redshift. However, dedicated deep surveys with extremely long exposures, for example, in the Chandra Deep Field South (CDFS) conducted by XMM-Newton (Comastri et al. 2011) and Chandra (Giacconi et al. 2002; Xue et al. 2011) X-ray observatories now allow us to pursue this approach. Here, we present a study of X-ray selected heavily obscured active galaxies using the 3 Ms XMM-Newton survey of CDFS.

X-ray absorption is measured by the low energy cut-off of an X-ray spectrum, which moves to higher energies as absorbing column density increases. When NH approaches 1024 cm-2, the cut-off occurs above 10 keV. As demonstrated for nearby examples, such as NGC 4945 (Iwasawa et al. 1993), the Circinus Galaxy (Matt et al. 1999a), and NGC 6240 (Vignati et al. 1999), detection of emission above 10 keV plays a key role in discoveries of heavily obscured AGN in those galaxies with absorbing column density exceeding 1024 cm-2. This method works as long as the optical depth is not too large, that is, when a source becomes fully Compton thick with NH ≥ 1025 cm-2, Compton down-scattering suppresses the hard X-rays, leaving only reflected light, as observed in NGC 1068 (e.g., Matt et al. 1997). While a direct access is not possible for nearby objects with XMM-Newton, this crucial energy-band is redshifted into its bandpass for high redshift objects at z ≥ 2. Given the shape of an absorbed X-ray spectrum, a negative K-correction sustains the detectability of absorbed sources to high redshift. Utilizing these properties, we searched for the rest-frame 9−20 keV excess sources to identify heavily obscured AGN candidates in the sources detected in the XMM-CDFS field.

The cosmology adopted here is H0 = 70 km s-1 Mpc-1, ΩΛ = 0.72, ΩM = 0.28.

2. The sample

We selected sources from the first XMM-CDFS catalogue with conservative detection criteria: 176 sources that were detected at significance larger than 8σ in the 2−10 keV band and have X-ray spectra verified for use for a spectral analysis are available (details will be described in Ranalli et al., in prep.). Since the signal to noise ratio of individual spectra, obtained from the EPIC cameras of XMM-Newton, falls steeply above 7.5 keV, we set the lower bound of the redshift range of our sample to z = 1.7, for which rest-frame 20 keV corresponds to observed-frame 7.4 keV.

There are 47 objects with z > 1.7 for which spectral data are available from all the three EPIC cameras, pn, MOS1 and MOS2, apart from two objects which are located outside the field of view of the pn but within the two MOS cameras (see Table 1). Spectroscopic redshifts are available for 33 objects, while photometric redshifts were estimated by various papers (Luo et al. 2010; Cardamone et al. 2008; Rafferty et al. 2011; Wuyts et al. 2008; Santini et al. 2009; Taylor et al. 2009) for other 14 objects1.

For some objects with photometric redshifts, more constrained redshift estimates could be obtained using the Fe K feature in their X-ray spectra when they have a strongly absorbed X-ray spectrum. We use these X-ray redshifts for five sources (Sect. 3.2). As a result of X-ray redshift determination, one source (z = 1.60), which had the original photometric redshift z = 1.78, went out of the redshift range. This source (PID 352) is therefore excluded from the sample and will not be discussed further.

Table 1

Properties of the sample.

thumbnail Fig. 1

Distribution of redshifts of the 46 sources in the sample. PID 352 with X-ray determined redshift zX = 1.60 is excluded.

thumbnail Fig. 2

Distribution of the 46 objects in our sample in the rest-frame 10−20 keV luminosity vs. redshift plane. No correction for internal absorption has been made when calculating the luminosities. The median L10−20 is 0.9 × 1044 erg s-1.

Hereafter we use “PID” for the identification number of X-ray sources listed in Ranalli et al, and the basic information of the 46 objects in the sample is presented in Table 1. The redshift distribution of the sample is shown in Fig. 1. The median redshift is ˜z=2.5\hbox{$\tilde z=2.5$}. The background-corrected counts obtained from the the sum of the three EPIC cameras range from 400 to 8000 in the respective rest-frame 3−20 keV band, while the typical counts are ~1400. The typical source fraction of the total (source plus background) counts is ≃ 0.4 in both EPIC pn and MOS cameras.

Since the exposure time for each source varies, the observed flux in the observed-frame 1−4 keV band, which is shared by all the sources with various redshifts, is given in Table 1 as an objective measure of source brightness. Median values of the observed frame 1−4 keV flux, f1−4, the rest-frame 2−10 keV and 10−20 keV luminosities, L2−10, and L10−20, are 2.5 × 10-15 erg s-1 cm-2, 9.1 × 1043 erg s-1, and 8.7 × 1043 erg s-1, respectively. These luminosities are corrected for the Galactic extinction, NH = 9 × 1019 cm-2 (Dickey & Lockman 1990). Figure 2 shows how the objects in our sample are distributed in the L10−20 − z plane. The spread of the 10−20 keV luminosity is relatively narrow with a logarithmic dispersion of 0.3 (or a factor of ~2).

thumbnail Fig. 3

X-ray colour–colour diagram, based on the data obtained from the XMM-Newton EPIC cameras, where s, m and h are the detector-response-corrected photon counts in the rest-frame bands of 3−5 keV, 5−9 keV and 9−20 keV, respectively. The four categories, V, A, M and U and their boundaries are indicated. Our reference heavily obscured AGN, PID 144, is labeled in the diagram. The red dashed-line indicates the evolution track of the X-ray colour when a power-law of Γ = 1.8 is modified by various absorbing colmun. The crosses mark log NH values 21, 22, 22.5, 23, 23.3, 23.5, 23.7, 23.85 and 24 (cm-2) from the bottom-right to the upper-left along the track. The X-ray colours estimated for the nearby, heavily obscured AGN, NGC 6240, NGC 4945 and NGC 1068 are also plotted. Note that NGC 4945 has a large value of h/m = 6.8, which is outside of the frame (see text for details of these sources). X-ray spectra of the two sources (PID 84 and 366) with h/m ~ 1.4, located in the M and U intervals, respectively, are described in text (Sect. 3.1).

3. Results

3.1. X-ray colour analysis

For selecting sources with various degrees of absorption, three rest-frame energy bands: s (3−5 keV); m (5−9 keV); and h (9−20 keV), are defined and two X-ray colours: s/m and h/m are computed. At energies above 3 keV, little contribution from soft X-ray emission originating from the extranuclear region is expected. As the intrinsic continuum slope in the 3−20 keV band is not expected to vary wildly between objects, absorption would be the main driver of changes in the X-ray colours. For the adopted rest-frame energy range, these X-ray colours are sensitive to column densities larger than NH ≃ 1022 cm-2.

Since our objects have a wide range of redshift (1.7−3.8 in z), these X-ray colours are derived using photon spectra, i.e., spectral data corrected for the detector response and the Galactic absorption as a function of the rest-frame energy. The correction method employed here is practically the same as that used in the XMM-COSMOS spectral stacking analaysis (Iwasawa et al. 2012). The photon counts in each band are the weighted mean of the three EPIC cameras, where we adopted the signal-to-noise ratio in the rest-frame 3−20 keV band as the weight. The two X-ray colours, s/m and h/m, for individual sources are listed in Table 1, and the colour–colour diagram is shown in Fig. 3.

With the two X-ray colours, a column density range of log NH = 22−24 (cm-2) can be probed, as s/m covers the lower NH regime and h/m does the higher. In Fig. 3, a locus of spectral evolution when a power-law continuum of photon index Γ = 1.8 is modified by various absorbing column of log NH between 21 and 24 (cm-2) is drawn. As the s/m represents softness of a spectrum below 9 keV, objects at the bottom-right in Fig. 3 are populated by sources with little absorption. The s/m colour moves to the left as absorption increases. Two divisions were made along the s/m axis, at s/m = 0.6 and 1.1. In the lowest interval, the model locus turns upwards as increasing absorption at log NH ≥ 23.5 (cm-2) and a few sources indeed spread towards higher h/m values, which indicates an excess of 9−20 keV emission.

PID 144 (z = 3.70) is a previously known, heavily obscured AGN with an X-ray absorbing column of NH ~  (0.6−0.9)  × 1024 cm-2 (Norman et al. 2002; Comastri et al. 2011), located in this interval. We take this object with h/m = 1.46 as the reference and sources that have h/m larger than this object were classified as 9−20 keV excess sources.

According to the three intervals along s/m and two intervals along h/m, four zones, V: Very absorbed; A: Absorbed; M: Modestly absorbed; and U: Unabsorbed, are defined in the colour–colour diagram, as shown in Fig. 3. The degree of absorption thus increases in the order of U, M, A, and V, and typical column densities for these X-ray colour categories would be log NH of ≤ 22, 22.7, 23.4, and 23.8 (cm-2), respectively.

For a comparison, the X-ray colours of nearby, well-studied heavily obscured AGN, NGC 6240 (NH ~ 2 × 1024 cm-2), NGC 4945 (NH ~ 5 × 1024 cm-2), and NGC 1068 (NH ≥ 1025 cm-2) were computed, based on the spectra presented in Vignati et al. (1999), Guainazzi et al. (2002), and Matt et al. (1997), respectively, obtained from the BeppoSAX observations (see Fig. 3). In these low luminosity systems, non AGN components, e.g., a circumnuclear starburst, flaring X-ray binaries (e.g., Brandt et al. 1996), can make a significant contribution to their spectra in the lower energy range, altering the s/m colour in particular, more than in high luminosity AGN like our sample. Despite of this spectral complexity, h/m serves as a good indicator of strongly absorption seen in sources like NGC 6240 and NGC 4945. The h/m colour moves back to a lower value for a fully Compton thick source, e.g., NGC 1068, but it still remains in a zone of hard spectrum sources.

Two sources, PID 84 and PID 366, have h/m values similar to the reference PID 144 but softer s/m colours (see Table 1). An inspection of their spectra shows that PID 84 has a moderately absorbed spectrum with NH ≃ 1 × 1023 cm-2 as expected for the M category, while PID 366 in the U interval shows a relatively soft spectrum but with a deficit at the rest-frame 7−10 keV (observed 2.2−3 keV range), causing the large value of h/m. This could be attributed to a strong Fe K edge caused by absorption of NH ~ 6 × 1023 cm-2, where a spectral complexity might play a role to mask the strong absorption.

Table 2

Properties of the four X-ray colour categories, V, A, M, and U.

The sources in the V and A categories are absorbed by NH of a few times of 1023 cm-2 or larger, so that prominent Fe K features in the form of an emission line or an absorption edge can be observed. This offers a possibility to derive a reliable X-ray spectroscopic redshift. There are five objects in the two categories with only photometric redshifts. X-ray redshift (zX) were obtained for these five objects and their X-ray colours were recomputed assuming the new redshifts. Details of the X-ray redshift measurements are described in Sect. 3.2.

Basic information on the sources in the four categories is given in Table 2 and their stacked, rest-frame 3−20 keV spectra are shown in Fig. 4, which demonstrates representative spectral shapes for respective categories. Note that exceptionally hard 10−20 keV spectrum of V compared to the other three, indicating that large absorption column (NH ~ 1024 cm-2) are affecting the sources in this category (Table 2).

thumbnail Fig. 4

Rest-frame 3−20 keV stacked spectra for the four categories, defined in Fig. 3. The vertical axis is in arbitrary unit of flux density. Only the XMM-Newton data were used. The spectral stacking is a straight sum of individual sources while a weighted mean of the available EPIC data, based on the signal to noise ratio, is taken for each source. Number of sources, typical redshift and luminosity of each category can be found in Table 2. For a reference, the spectral slope of the U category spectrum is α ≃ 0.8, where FE ∝ E−α, i.e., photon index Γ ≃ 1.8.

Table 3

X-ray redshift measurements.

3.2. X-ray redshift measurements

Redshift was measured with X-ray spectra for five objects, PID 30, 64, 116, 245, and 352, which have only photometric redshifts (Table 3). These objects are too faint in the optical band to obtain a reliable spectroscopic redshift. The photometric redshifts reported by various authors for each object spread over a significant range, while they can serve as a guide for a redshift range to be searched in. The Fe K features imprinted in their absorbed spectra gave improved accuracy in the redshift measurements. The Chandra data from the 4 Ms (Xue et al. 2011) and the ECDFS (Lehmer et al. 2005) observations were also added to the analysis for improving the spectral quality. The observed-frame 0.5−7 keV spectra of these sources are shown in Fig. 5, except for PID 352, details of which will be reported in a separate paper. They are photon spectra combining XMM-Newton and Chandra data for displaying purpose only. All the spectral results presented hereafter were obtained by fitting spectral datasets from different cameras jointly.

The redshift determination was principally driven by the Fe K edge, which is assumed to arise from cold medium and thus at the energy of 7.1 keV, as it is normally statistically more robust feature than the line. Two exceptions are PID 116 and PID 352, for which the Fe K emission-lines, detected at 2.5σ with EW ≈ 0.15 keV and 4.5σ with EW ≈ 0.30 keV, respectively, were used to obtain the redshifts assuming the rest-frame line energy of 6.4 keV emitted from cold matter.

This assumption may not be true for PID 116, which is a luminous submillimetre galaxy detected at 870 μm with LABOCA (LESS 9, Wardlow et al. 2011; Biggs et al. 2011). As the X-ray detected luminous infrared galaxies with LIR ~ 1013 L⊙ at z > 2 in the COSMOS field appears to show high-ionization Fe K emission, e.g., Fe xxv at 6.70 keV and/or Fe xxvi at 6.97 keV as inferred from a spectral stacking analysis (Iwasawa et al. 2012), the emission line of PID 116 could also be either of these high-ionization lines. In this case, the redshift would be z = 3.96 or z = 4.16, when it was identified with Fe xxv and Fe xxvi, respectively. The former value is close to the photometric redshift derived by Luo et al. (2010) and the latter to the secondary solution of Wardlow et al. (2011, see Table 3).

As shown in Table 3, zX are found to lie within the range of various photometric redshifts.

3.3. 9–20 keV excess sources

The spectra of the seven objects in the V category are shown in Figs. 5 and 6. For displaying purpose, EPIC pn, EPIC MOS1, MOS2, and the Chandra data from the ACIS detector are combined together. All the spectra show spectral discontinuities at 6−7 keV, indicating an Fe K line and/or a deep Fe K absorption edge, in agreement with strong absorption.

To estimate the absorbing column density of these sources, an absorbed power-law model with photon index of Γ = 1.8 is fitted (Table 4). The absorption model with the Wisconsin cross section (Morrison & McCammon 1983) and the one with the effects of Compton scattering taken into account, PLCABS (Yaqoob 1997), give consistent results on NH. Whilst the absorption cut-off is slightly modified when Compton scattering is taken into account, the 7−20 keV spectrum, shaped by an Fe K edge, remains unchanged in shape for the NH range of our sources, NH ≤ 1024 cm-2 (although the flux is further suppressed). This also applies to recently developed more sophisticated X-ray spectral models (e.g., Ikeda et al. 2009; Murphy & Yaqoob 2010). As our NH fits are mainly driven by the data in the Fe K edge band, it can be understood that both absorption models gives similar NH, given the data quality of the spectra. However, since Compton scattering reduces the continuum level further compared to the case where the scattering effect is not taken into account, the absorption correction factor for estimating an intrinsic continuum luminosity would be larger. This effect also depends on the geometry of the absorber (e.g., Matt et al. 1999b) which is not known for our objects. In Table 4, we give absorption-correction factors for a spherical absorber. These factors could go up by a factor of ~2 as the covering factors decreases down to that of a disk-like geometry for the relevant range of NH.

thumbnail Fig. 5

X-ray spectra of the four objects (PID 30, 64, 116, 245) whose redshifts were determined using the Fe K features. The Chandra ACIS-I data, obtained in the deep CDFS (4 Ms) and the ECDFS observations, were combined with the XMM-Newton data. The rest-frame 6.4 keV and 7.1 keV which would be observed with those redshifts are indicated by the dotted lines. PID 30, 64, and 245 are heavily obscured sources in the V category while PID 116 is a source in the A category.

Table 4

X-ray absorption in the 9−20 keV excess sources.

The column densities given in Table 4 were obtained, assuming the observed emission is transmitted light through an absorber. However, the hard X-ray colour exhibited by these objects could also result from a reflection-dominated spectrum of a Compton thick source. This is probably the case for PID 114, in which a strong Fe K line is detected (see below and Table 5), whereas the apparently moderate column density is inferred from the absorption model for the poor quality continuum spectrum (Table 4). PID 252 has the hardest spectrum in terms of the hard X-ray colour h/m (Fig. 6) although no obvious Fe line is seen. For these two objects, a pure reflection spectrum from cold matter, modelled by pexrav (Magdziarz & Zdziarski 1995) or pexmon (Nandra et al. 2007), provides a comparable fit to their spectra, compared to the absorption model. This indicates that these two objects might be Compton thick AGN with a larger NH than that given in Table 4, e.g., ~1025 cm-2.

Fe K emission is detected at ~2σ or larger significance in these objects except for PID 252 (Table 5, see also Comastri et al. 2011). The spectrum of PID 252 does not show clear Fe K emission with EW ≤ 0.5 keV (2σ upper limit of a narrow line at 6.4 keV). This weak-line source may be a high-redshift analogue of Mrk 231, a Compton thick AGN with a weak Fe K line in the local Universe (e.g., Braito et al. 2004; Gallagher et al. 2002; Iwasawa et al. 2011, and other references therein). The large EW observed in PID 114 (Table 5) agrees with an Fe K line expected from a reflection-dominated spectrum from cold medium, giving a support to the possibility of a Compton thick source.

thumbnail Fig. 6

X-ray spectra of the four objects (PID 114, 144, 180, 252) which complete the seven objects of the V category in addition to the three objects shown in Fig. 5. The rest-frame 6.4 keV and 7.1 keV are indicated by the dotted lines that were computed assuming the spectroscopic redshifts for respective objects.

Table 5

Fe K line equivalent widths of the V category objects.

4. Discussion

4.1. X-ray selection of heavily obscured AGN

Seven heavily obscured active galaxies with NH ≥ 0.6 × 1024 cm-2, including one previously known source (PID 144, Norman et al. 2002; Comastri et al. 2011), were selected by the rest-frame X-ray colour selection, primarily utilizing the excess emission in the 9−20 keV band relative to emission at lower energies. Two of them (PID 114 and PID 252) are possibly Compton thick AGN with a reflection-dominated spectrum. Given the limited bandpass available from XMM-Newton, this selection can be applied only for high redshift objects, but the a posteriori checks showed that this selection is reliable for sources with spectra of reasonable quality, and can pick up strongly absorbed sources with near Thomson-thick opacity.

This method is a pure X-ray selection, and these seven objects compose a sample of heavily obscured, moderate-luminosity quasars with L10−20 ~ 1044 erg s-1, selected by the hard X-ray emission above 10 keV beyond the local Universe.

There are various reports in the literature on Compton thick AGN candidates in the CDFS using whole or part of the Chandra 4 Ms data (e.g., Norman et al. 2002; Mainieri et al. 2005; Tozzi et al. 2006; Fiore et al. 2008; Gilli et al. 2011; Feruglio et al. 2011; Luo et al. 2011; Brightman & Ueda 2012; Fiore et al. 2012). Some of them lie in the redshift range of our sample, although they are expectedly faint and just a few of them entered in our sample of relatively bright sources. Fiore et al. (2012) investigated high-redshift sources at z > 3 in CDFS and selected several heavily obscured AGN. Their E537 (=PID 245), M5390 (=PID 144), M8273 (PID 180), M3320 (=PID107) and M4302 (=PID 120) are in our sample. Our results on the spectra of these sources agree except for PID 120 for which only moderate absorption of NH =1.6-0.6+0.5×1022\hbox{$ = 1.6^{+0.5}_{-0.6}\times 10^{22}$} cm-2 is found. The smaller NH value for PID 245 obtained by us (Table 4) is explained by the lower redshift adopted for this source: the X-ray redshift zX = 2.68 (Sect. 3.2, Table 3), instead of the photometric redshift z = 4.29 from the GOODS-ERS (Grazian et al. 2011) adopted by Fiore et al. (2012), which a close inspection of the X-ray/optical/infrared images suggests to be the redshift for another galaxy near the X-ray source.

thumbnail Fig. 7

Distribution of absorbing column density NH, obtained by fitting an absorbed power-law to the EPIC spectra. The lowest bin represents the number of objects with no detection of absorption. The typical error bar of each bin is ± 1.

4.2. Absorbed AGN fraction

In Fig. 3, when the model locus is used as a guide, 10 objects appear to have unobscured X-ray sources, i.e., their X-ray absorption is NH < 1022 cm-2. That is, ~3/4 of our sample objects host signicantly obscured active nuclei. Fitting to the individual spectra verifies the above assessment with 12 objects having NH values smaller than 1022 cm-2, and gives the NH distribution shown in Fig. 7. The distribution of log NH (cm-2) is nearly flat between 22−24, although the two objects (PID 114 and 252) possibly move up to log NH > 24 (cm-2). For their typical 2−10 keV intrinsic luminosities [(0.8−5) × 1044 erg s-1], these active galaxies at z ~ 2.5 can all be considered to emit at quasar luminosity, and 74 ± 8 per cent of them are absorbed X-ray sources. We have estimated this absorbed AGN fraction using a Bayesian approach and the binomial distribution (Wall & Jenkins 2008) with a 68 per cent confidence interval (Andreon, priv. comm.). It should be noted that, since the fraction of Compton thick AGN is not constrained, this value is considered to be the lower limit of the absorbed AGN fraction.

We compared our findings with the predictions of the XRB synthesis model by Gilli et al. (2007). Our sample spans the 2−10 keV flux range (2−54)  × 10-15 erg cm-2 s-1. Since the sensitivity of the XMM-CDFS observations strongly varies across the field, we computed the model predictions at the 2−10 keV limiting flux, f2−10lim=4×10-15\hbox{$f^{\rm lim}_{2-10}=4\times 10^{-15}$} erg cm-2 s-1, which returns the same AGN surface density of our sample, i.e. 46 sources at z > 1.7 distributed over a ~0.27 deg2 area. The predicted obscured fraction (defined as the number of AGN with log NH > 22 over the total number of AGN in the sample) is 0.54 ± 0.06, smaller than the observed value of 0.74 ± 0.08.

In the local Universe, the Swift/BAT and INTEGRAL surveys show that absorbed sources (with NH > 1022 cm-2) consist ~55 per cent of hard X-ray selected AGN (e.g., Burlon et al. 2011, and references therein). It is also found that this fraction depends on X-ray luminosity, and at the luminosity matched to our sample, the fraction is 21 ± 8 per cent (Burlon et al. 2011, see also Ebrero et al. 2008). The absorbed quasar fraction (L2−10int≥1044\hbox{$L^{\rm int}_{2-10}\geq 10^{44}$} erg s-1) in our sample is higher than that of the local Universe, suggesting a positive evolution with redshift, as found in the previous work by La Franca et al. (2005), Treister & Urry (2006) and Ebrero et al. (2008). No evolution of the obscured AGN fraction was assumed in Gilli et al. (2007), yet the prediction comes close to the observation at z > 1.7 as discussed above. However, we note that the luminosity dependence of the obscured AGN fraction assumed in Gilli et al. (2007) appears to be shallower than the observations (Hasinger 2008; Brusa et al. 2010; Burlon et al. 2011), and it overestimates the obscured fraction of quasi-stellar objects (QSOs) with L2−10 > 1044 erg s-1 in the local Universe by a factor of ~2.5. This excess number assumed for local obscured QSOs then compensates the lack of a redshift evolution of their fraction in the model.

Contrary to the high-luminosity AGN, no strong evidence for a redshift dependence of the obscured AGN fraction at luminosities <1044 erg s-1 has been found. Gilli et al. (2010), for instance, showed that the increasing trend of the absorbed fraction as observed by Hasinger (2008) for AGN with L2−10 ≤ 1044 erg s-1, can be accounted for by the K-correction effect, and is instead consistent with a non-evolving intrinsic absorbed fraction. Here we suggest that the obscured fraction increases with redshift only for luminous QSOs. The different behaviours in obscured fraction between low- and high-luminosity AGN may reflect their distinct accretion mechanisms, as argued in literature (Hasinger 2008; Hopkins et al. 2008; Hickox et al. 2009): merger-driven accretion for luminous AGN (e.g., Menci et al. 2008) and secular accretion for less luminous AGN, possibly mirroring their respective drivers of star formation (e.g., Elbaz et al. 2011). This may not be the whole story but qualitatively explains the different bahaviours between AGN of the low and high luminosity ranges. If all QSOs originate from a major merger of gas-rich galaxies (e.g., Sanders et al. 1988), the increase of merger rate at high redshift (with ∝ (1 + z)2, e.g., Xu et al. 2012) naturally sees an increase in number of QSOs. A merger causes gas channelling to the nuclear region (Barnes & Hernquist 1991). This concentration of gas and the chaotic geometry left by a merger would lead to a high probability of the nuclear region to be seen obscured (e.g., Hopkins et al. 2006; but see Shawinski et al. 2012) until the radiation pressure of the buried QSO sweeps it away. In the context of this evolutionary scenario alone, the obscured fraction of QSOs is expected to be constant at all redshift, given the short duration of the QSO lifetime (≤ 108 yr, Hopkins et al. 2005). The evolution we observed is probably driven by the increase in the gas fraction of a galaxy towards high redshift (e.g., Carilli et al. 2011), combined with the efficient inflow induced by a merger. A higher gas fraction of merger progenitor galaxies means more gas to be transported to the nuclear region to form heavier obscuration. This would result in a longer duration of the obscured phase, which can be translated to a higher obscured fraction of the QSO population at high redshift. At the same time, the elevated gas density by a merger increases the efficiency of star formation leading to a starburst (e.g., Barnes & Hernquist 1991; Elbaz et al. 2011). Kinetic energy injection from a starburst may help to maintain the obscuration by inflating gaseous wall around AGN (e.g., Fabian et al. 1998). Conversely, the lack of mergers may explain the little evolution of the obscured fraction in lower luminosity AGN. The gas fraction of galaxies hosting them also increases towards high redshift in the same way as for high-luminosity AGN. However, without a major merger, the gas reservoir is not transported to the nuclear region rapidly. This means that the nuclear obscuration condition remains little affected regardless the amount of gas contained in a galaxy (hence redshifts). The gas content is instead consumed to form stars over galaxy-wide as a secular process, and the feeding to the black hole from a large-scale disk remains relatively inefficient.

In summary, we present a result of a rest-frame 9−20 keV selection of heavily obscured AGN at z > 1.7, using the deep XMM-CDFS survey, and also show that the fraction of absorbed AGN at high luminosity may be higher at high redshift than in the local Universe. In the near future, a further advance in this area of research will benefit from even deeper observations of deep fields with Chandra and XMM-Newton, while NuSTAR and Astro-H which will provide us with useful templates and insights at lower redshifts. It is also useful to standardize various X-ray spectral models of strongly absorbed systems with improved physics incorporated for the community to share with.

1

The photometric redshift adopted in this paper are taken from Ranalli et al. (in prep.), in which the choices among various photometric redshift estimates are described in detail.

Acknowledgments

This research made use of the data obtained from XMM-Newton and the Chandra X-ray Observatory. K.I. thanks support from Spanish Ministerio de Ciencia e Innovación (MICINN) through the grant (AYA2010-21782-C03-01). W.N.B. thanks the NASA ADP grant NNX10AC99G. We acknowledge financial contribution from the agreement ASI-INAF I/009/10/0.

References

  1. Alexander, D. M., Bauer, F. E., Brandt, W. N., et al. 2011, ApJ, 738, 44 [NASA ADS] [CrossRef] [Google Scholar]
  2. Alonso-Herrero, A., Pérez-González, P. G., Alexander, D. M., et al. 2006, ApJ, 640, 167 [NASA ADS] [CrossRef] [Google Scholar]
  3. Balestra, I., Mainieri, V., Popesso, P., et al. 2010, A&A, 512, A12 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  4. Barnes, J. E., & Hernquist, L. E. 1991, ApJ, 370, L65 [NASA ADS] [CrossRef] [Google Scholar]
  5. Bauer, F. E., Yan, L., Sajina, A., & Alexander, D. M. 2010, ApJ, 710, 212 [NASA ADS] [CrossRef] [Google Scholar]
  6. Biggs, A. D., Ivison, R. J., Ibar, E., et al. 2011, MNRAS, 413, 2314 [Google Scholar]
  7. Braito, V., Della Ceca, R., Piconcelli, E., et al. 2004, A&A, 420, 79 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Brandt, W. N., Iwasawa, K., & Reynolds, C. S. 1996, MNRAS, 281, L41 [NASA ADS] [CrossRef] [Google Scholar]
  9. Brightman, M., & Ueda, Y. 2012, MNRAS, 423, 702 [NASA ADS] [CrossRef] [Google Scholar]
  10. Brusa, M., Civano, F., Comastri, A., et al. 2010, ApJ, 716, 348 [NASA ADS] [CrossRef] [Google Scholar]
  11. Burlon, D., Ajello, M., Greiner, J., et al. 2011, ApJ, 728, 58 [NASA ADS] [CrossRef] [Google Scholar]
  12. Carilli, C. L., Walter, F., Riechers, D., et al. 2011 [arXiv:1105.1128] [Google Scholar]
  13. Cooper, M. C., Yan, R., Dickinson, M., et al. 2012, MNRAS, 425, 2116 [NASA ADS] [CrossRef] [Google Scholar]
  14. Cardamone, C. N., Urry, C. M., Damen, M., et al. 2008, ApJ, 680, 130 [NASA ADS] [CrossRef] [Google Scholar]
  15. Comastri, A., Ranalli, P., Iwasawa, K., et al. 2011, A&A, 526, L9 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  16. Daddi, E., Alexander, D. M., Dickinson, M., et al. 2007, ApJ, 670, 173 [NASA ADS] [CrossRef] [Google Scholar]
  17. Dahlen, T., Mobasher, B., Dickinson, M., et al. 2010, ApJ, 724, 425 [NASA ADS] [CrossRef] [Google Scholar]
  18. Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215 [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  19. Donley, J. L., Koekemoer, A. M., Brusa, M., et al. 2012, ApJ, 748, 142 [NASA ADS] [CrossRef] [Google Scholar]
  20. Ebrero, J., Carrera, F. J., Page, M. J., et al. 2009, A&A, 493, 55 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  21. Elbaz, D., Dickinson, M., Hwang, H. S., et al. 2011, A&A, 533, A119 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  22. Fabian, A. C., Barcons, X., Almaini, O., & Iwasawa, K. 1998, MNRAS, 297, L11 [NASA ADS] [CrossRef] [Google Scholar]
  23. Feruglio, C., Daddi, E., Fiore, F., et al. 2011, ApJ, 729, L4 [NASA ADS] [CrossRef] [Google Scholar]
  24. Fiore, F., Grazian, A., Santini, P., et al. 2008, ApJ, 672, 94 [NASA ADS] [CrossRef] [Google Scholar]
  25. Fiore, F., Puccetti, S., Brusa, M., et al. 2009, ApJ, 693, 447 [NASA ADS] [CrossRef] [Google Scholar]
  26. Fiore, F., Puccetti, S., Grazian, A., et al. 2012, A&A, 537, A16 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  27. Gallagher, S. C., Brandt, W. N., Chartas, G., Garmire, G. P., & Sambruna, R. M. 2002, ApJ, 569, 655 [NASA ADS] [CrossRef] [Google Scholar]
  28. George, I. M., & Fabian, A. C. 1991, MNRAS, 249, 352 [NASA ADS] [CrossRef] [Google Scholar]
  29. Giacconi, R., Zirm, A., Wang, J., et al. 2002, ApJS, 139, 369 [NASA ADS] [CrossRef] [Google Scholar]
  30. Gilli, R., Comastri, A., & Hasinger, G. 2007, A&A, 463, 79 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  31. Gilli, R., Comastri, A., Vignali, C., Ranalli, P., & Iwasawa, K. 2010, X-ray Astronomy 2009, Present Status, Multi-Wavelength Approach and Future Perspectives, 1248, 359 [NASA ADS] [Google Scholar]
  32. Gilli, R., Su, J., Norman, C., et al. 2011, ApJ, 730, L28 [NASA ADS] [CrossRef] [Google Scholar]
  33. Grazian, A., Castellano, M., Koekemoer, A. M., et al. 2011, A&A, 532, A33 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  34. Guainazzi, M., Matt, G., Brandt, W. N., et al. 2000, A&A, 356, 463 [NASA ADS] [Google Scholar]
  35. Hasinger, G. 2008, A&A, 490, 905 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  36. Hickox, R. C., Jones, C., Forman, W. R., et al. 2009, ApJ, 696, 891 [Google Scholar]
  37. Hopkins, P. F., Hernquist, L., Martini, P., et al. 2005, ApJ, 625, L71 [Google Scholar]
  38. Hopkins, P. F., Hernquist, L., Cox, T. J., et al. 2006, ApJS, 163, 1 [Google Scholar]
  39. Hopkins, P. F., Hernquist, L., Cox, T. J., & Kereš, D. 2008, ApJS, 175, 356 [NASA ADS] [CrossRef] [Google Scholar]
  40. Ikeda, S., Awaki, H., & Terashima, Y. 2009, ApJ, 692, 608 [NASA ADS] [CrossRef] [Google Scholar]
  41. Iwasawa, K., Sanders, D. B., Teng, S. H., et al. 2011, A&A, 529, A106 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  42. Iwasawa, K., Koyama, K., Awaki, H., et al. 1993, ApJ, 409, 155 [NASA ADS] [CrossRef] [Google Scholar]
  43. Iwasawa, K., Mainieri, V., Brusa, M., et al. 2012, A&A, 537, A86 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  44. La Franca, F., Fiore, F., Comastri, A., et al. 2005, ApJ, 635, 864 [NASA ADS] [CrossRef] [Google Scholar]
  45. Lehmer, B. D., Brandt, W. N., Alexander, D. M., et al. 2005, ApJS, 161, 21 [NASA ADS] [CrossRef] [Google Scholar]
  46. Luo, B., Brandt, W. N., Xue, Y. Q., et al. 2010, ApJS, 187, 560 [NASA ADS] [CrossRef] [Google Scholar]
  47. Luo, B., Brandt, W. N., Xue, Y. Q., et al. 2011, ApJ, 740, 37 [NASA ADS] [CrossRef] [Google Scholar]
  48. Magdziarz, P., & Zdziarski, A. A. 1995, MNRAS, 273, 837 [NASA ADS] [CrossRef] [Google Scholar]
  49. Mainieri, V., Rigopoulou, D., Lehmann, I., et al. 2005, MNRAS, 356, 1571 [NASA ADS] [CrossRef] [Google Scholar]
  50. Marconi, A., Risaliti, G., Gilli, R., et al. 2004, MNRAS, 351, 169 [NASA ADS] [CrossRef] [Google Scholar]
  51. Martínez-Sansigre, A., Rawlings, S., Lacy, M., et al. 2005, Nature, 436, 666 [NASA ADS] [CrossRef] [Google Scholar]
  52. Matt, G., Guainazzi, M., Frontera, F., et al. 1997, A&A, 325, L13 [NASA ADS] [Google Scholar]
  53. Matt, G., Guainazzi, M., Maiolino, R., et al. 1999a, A&A, 341, L39 [NASA ADS] [Google Scholar]
  54. Matt, G., Pompilio, F., & La Franca, F. 1999b, New A, 4, 191 [Google Scholar]
  55. Menci, N., Fiore, F., Puccetti, S., & Cavaliere, A. 2008, ApJ, 686, 219 [NASA ADS] [CrossRef] [Google Scholar]
  56. Morrison, R., & McCammon, D. 1983, ApJ, 270, 119 [NASA ADS] [CrossRef] [Google Scholar]
  57. Murphy, K. D., & Yaqoob, T. 2009, MNRAS, 397, 1549 [NASA ADS] [CrossRef] [Google Scholar]
  58. Nandra, K., O’Neill, P. M., George, I. M., & Reeves, J. N. 2007, MNRAS, 382, 194 [NASA ADS] [CrossRef] [Google Scholar]
  59. Norman, C., Hasinger, G., Giacconi, R., et al. 2002, ApJ, 571, 218 [NASA ADS] [CrossRef] [Google Scholar]
  60. Popesso, P., Dickinson, M., Nonino, M., et al. 2009, A&A, 494, 443 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  61. Rafferty, D. A., Brandt, W. N., Alexander, D. M., et al. 2011, ApJ, 742, 3 [NASA ADS] [CrossRef] [Google Scholar]
  62. Sanders, D. B., Soifer, B. T., Elias, J. H., et al. 1988, ApJ, 325, 74 [NASA ADS] [CrossRef] [Google Scholar]
  63. Santini, P., Fontana, A., Grazian, A., et al. 2009, A&A, 504, 751 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  64. Schawinski, K., Simmons, B. D., Urry, C. M., Treister, E., & Glikman, E. 2012, MNRAS, 425, L61 [NASA ADS] [Google Scholar]
  65. Silverman, J. D., Mainieri, V., Salvato, M., et al. 2010, ApJS, 191, 124 [NASA ADS] [CrossRef] [Google Scholar]
  66. Szokoly, G. P., Bergeron, J., Hasinger, G., et al. 2004, ApJS, 155, 271 [NASA ADS] [CrossRef] [Google Scholar]
  67. Taylor, E. N., Franx, M., van Dokkum, P. G., et al. 2009, ApJS, 183, 295 [NASA ADS] [CrossRef] [Google Scholar]
  68. Tozzi, P., Gilli, R., Mainieri, V., et al. 2006, A&A, 451, 457 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  69. Treister, E., & Urry, C. M. 2006, ApJ, 652, L79 [NASA ADS] [CrossRef] [Google Scholar]
  70. Treister, E., Urry, C. M., & Virani, S. 2009a, ApJ, 696, 110 [NASA ADS] [CrossRef] [Google Scholar]
  71. Treister, E., Virani, S., Gawiser, E., et al. 2009b, ApJ, 693, 1713 [NASA ADS] [CrossRef] [Google Scholar]
  72. Vignati, P., Molendi, S., Matt, G., et al. 1999, A&A, 349, L57 [NASA ADS] [Google Scholar]
  73. Vignali, C., Alexander, D. M., Gilli, R., & Pozzi, F. 2010, MNRAS, 404, 48 [NASA ADS] [Google Scholar]
  74. Wall, J. V., & Jenkins, C. R. 2003, Practical Statistics for Astronomers (CUP), 21 [Google Scholar]
  75. Wardlow, J. L., Smail, I., Coppin, K. E. K., et al. 2011, MNRAS, 415, 1479 [NASA ADS] [CrossRef] [Google Scholar]
  76. Wuyts, S., Labbé, I., Schreiber, N. M. F., et al. 2008, ApJ, 682, 985 [NASA ADS] [CrossRef] [Google Scholar]
  77. Xu, C. K., Zhao, Y., Scoville, N., et al. 2012, ApJ, 747, 85 [NASA ADS] [CrossRef] [Google Scholar]
  78. Xue, Y. Q., Luo, B., Brandt, W. N., et al. 2011, ApJS, 195, 10 [NASA ADS] [CrossRef] [Google Scholar]
  79. Yaqoob, T. 1997, ApJ, 479, 184 [NASA ADS] [CrossRef] [Google Scholar]

All Tables

Table 1

Properties of the sample.

In the text
Table 2

Properties of the four X-ray colour categories, V, A, M, and U.

In the text
Table 3

X-ray redshift measurements.

In the text
Table 4

X-ray absorption in the 9−20 keV excess sources.

In the text
Table 5

Fe K line equivalent widths of the V category objects.

In the text

All Figures

thumbnail Fig. 1

Distribution of redshifts of the 46 sources in the sample. PID 352 with X-ray determined redshift zX = 1.60 is excluded.

In the text
thumbnail Fig. 2

Distribution of the 46 objects in our sample in the rest-frame 10−20 keV luminosity vs. redshift plane. No correction for internal absorption has been made when calculating the luminosities. The median L10−20 is 0.9 × 1044 erg s-1.

In the text
thumbnail Fig. 3

X-ray colour–colour diagram, based on the data obtained from the XMM-Newton EPIC cameras, where s, m and h are the detector-response-corrected photon counts in the rest-frame bands of 3−5 keV, 5−9 keV and 9−20 keV, respectively. The four categories, V, A, M and U and their boundaries are indicated. Our reference heavily obscured AGN, PID 144, is labeled in the diagram. The red dashed-line indicates the evolution track of the X-ray colour when a power-law of Γ = 1.8 is modified by various absorbing colmun. The crosses mark log NH values 21, 22, 22.5, 23, 23.3, 23.5, 23.7, 23.85 and 24 (cm-2) from the bottom-right to the upper-left along the track. The X-ray colours estimated for the nearby, heavily obscured AGN, NGC 6240, NGC 4945 and NGC 1068 are also plotted. Note that NGC 4945 has a large value of h/m = 6.8, which is outside of the frame (see text for details of these sources). X-ray spectra of the two sources (PID 84 and 366) with h/m ~ 1.4, located in the M and U intervals, respectively, are described in text (Sect. 3.1).
In the text
thumbnail Fig. 4

Rest-frame 3−20 keV stacked spectra for the four categories, defined in Fig. 3. The vertical axis is in arbitrary unit of flux density. Only the XMM-Newton data were used. The spectral stacking is a straight sum of individual sources while a weighted mean of the available EPIC data, based on the signal to noise ratio, is taken for each source. Number of sources, typical redshift and luminosity of each category can be found in Table 2. For a reference, the spectral slope of the U category spectrum is α ≃ 0.8, where FE ∝ E−α, i.e., photon index Γ ≃ 1.8.

In the text
thumbnail Fig. 5

X-ray spectra of the four objects (PID 30, 64, 116, 245) whose redshifts were determined using the Fe K features. The Chandra ACIS-I data, obtained in the deep CDFS (4 Ms) and the ECDFS observations, were combined with the XMM-Newton data. The rest-frame 6.4 keV and 7.1 keV which would be observed with those redshifts are indicated by the dotted lines. PID 30, 64, and 245 are heavily obscured sources in the V category while PID 116 is a source in the A category.
In the text
thumbnail Fig. 6

X-ray spectra of the four objects (PID 114, 144, 180, 252) which complete the seven objects of the V category in addition to the three objects shown in Fig. 5. The rest-frame 6.4 keV and 7.1 keV are indicated by the dotted lines that were computed assuming the spectroscopic redshifts for respective objects.
In the text
thumbnail Fig. 7

Distribution of absorbing column density NH, obtained by fitting an absorbed power-law to the EPIC spectra. The lowest bin represents the number of objects with no detection of absorption. The typical error bar of each bin is ± 1.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.