Comparison with other recent hard X-ray surveys

In the last 2 years, other papers have been published, or are in the referee process, concerning the population study of hard X-ray AGN. These studies made use of INTEGRAL/ISGRI or Swift/BAT data during different periods of these missions and with different survey areas. Sazonov et al. (2007) presented an all-sky survey obtained adding to the first 3 years of INTEGRAL data some private IN-TEGRAL observations aimed to cover the less exposed extragalactic regions of the sky. At Galactic latitudes |b| > 5^◦, they collected a sample of 66 non-blazar AGN detected at ≥ 4.8σ. The region including the quasar 3C 273 and the galaxy cluster Coma is the extragalactic area with highest ex-posure with INTEGRAL. The 4 Ms of data collected in this 4900 deg²(2500 deg²with at least 10 ks of effective exposure) have been used by Paltani et al. (2008) for a deep, narrow-field study of the AGN population, on a sample of 34 objects, of which 22 firmly detected AGN. The first 9 months of Swift/BAT observations have led to two studies, one on the all sky (|b|>15^◦, Tueller et al. 2008) and the other concentrated on a smaller region of∼7000 deg²(Ajello et al. 2008). Their samples include 102 and 24 AGN, respectively. The limiting sensitivity on the∼80% of the sky for the all-sky surveys is comparable to ours, whereas the INTEGRAL 3C 273/Coma-region survey goes deeper by a factor of∼1.6 in flux.

All these works find a fraction of absorbed AGN in the range 45–70% and confirm the paucity of Compton-thick AGN. The all-sky surveys detected 4–5 Compton-thick objects corresponding to a fraction ≤ 10% that might increase up to 20% if all the objects with unknown NHresulted to have absorptions larger than 10²⁴cm⁻². No Compton-thick AGN are detected in the areas selected by Paltani et al. (2008) and Ajello et al. (2008), and an upper limit of∼24% can be deduced.

A decrease of the fraction of absorbed AGN with luminosity has been observed in the population at energies below 20 keV (Ueda et al. 2003; Sazonov & Revnivtsev 2004; Treister & Urry 2005).

Sazonov et al. (2007) confirm this result obtaining a fraction of 66% for objects with luminosity log L(20−60 keV) <43.6 and 25% above this value. Our work finds 70% and 50%, respectively, and a higher normalisation of the luminosity function for absorbed sources below log L(20−40 keV) ∼ 43.4, but the significance of these results is not sufficient to draw a firm conclusion. The same applies to the works of Tueller et al. (2008) and Paltani et al. (2008), as the small sample sizes do not allow to confirm this dependence.

The luminosity functions derived by these studies can be well fitted by parameters that are at all consistent with those found by our work, except for the normalisation that appears to be lower for our luminosity function compared to the others (Paltani et al. 2008). However, when considering the uncertainties, the value we found is consistent with the one obtained by Paltani et al. (2008) and no uncertainty on this parameter is provided by Sazonov et al. (2007). In Tueller et al. (2008) there is a discrepancy between the normalisation value stated in the text and that used in the figure, making difficult a definite comparison. Paltani et al. (2008) made use of the high-luminosity tail of our sample in order to complement their luminosity function that is lacking the high-luminosity objects due to the relatively small volume sampled by their survey. The Swift log N−log S seems to have a lower normalisation with respect to the INTEGRAL ones and Tueller et al. (2008) claim that this discrepancy is compatible with being due to instrument calibration differences. However, Paltani et al. (2008) argue that calibration issues are not sufficient to justify this difference, which remains at the moment unexplained.

The hard X-ray luminosity function of AGN 79 5.3.2 The synthesis of the CXB

The study of the space density and evolution of the population of absorbed AGN, together with the analysis of the dependence of their absorption distribution on luminosity and redshift, are the keys to resolve their contribution to the cosmic X-ray background. The INTEGRAL and Swift surveys show that the fraction of absorbed AGN detected in the hard X-rays is higher than that observed at lower energies (∼30 % in the 2–20 keV band; Ueda et al. 2003; Sazonov & Revnivtsev 2004), as expected due to the more unbiased view of the absorbed AGN population at hard X-rays. The Seyfert 2 : Seyfert 1 ratio is instead lower, about 1:1, compared to that measured in optical surveys and expected in the standard unified AGN models. In the Sloan Digital Sky Survey for example, this ratio has been found to be equal to 1 for low luminosity objects, whereas at high luminosities it assumes a value between 2.5 and 8 depending on the definition of Seyfert 2 (Hao et al. 2005). A ratio 4:1 is compatible with the average opening angle of ionisation cones in Seyfert galaxies (Maiolino & Rieke 1995 and references therein). The decrease of the fraction of absorbed AGN with the luminosity could indicate that actually it is necessary to modify the unified models to take into account that the geometry and size of the dust torus could depend on luminosity (and maybe redshift, see Sect. 5.1.2).

A possibility would be that the structure of the torus is modified by the high radiation pressure of more luminous AGN (Lawrence & Elvis 1982).

Another important result of these works is the lack of a substantial number of Compton-thick AGN detected. Even though more objects could be identified as Compton-thick among those with unknown NH, the fraction of Compton-thick objects detected at hard X-rays is significantly lower than the 50% expected from optical-X-ray crossed works (Risaliti et al. 1999; Guainazzi et al. 2005).

This could imply that the missing Compton-thick objects have to be found either among the heavily absorbed AGN (10²⁵ −10²⁶cm⁻²), which are too faint to be detected by these hard X-ray surveys, or at higher redshifts, not investigated by the INTEGRAL and Swift surveys. The latter possibility is supported by La Franca et al. (2005) and Ballantyne et al. (2006) that measured an increasing fraction of absorbed AGN at increasing redshifts, whereas other works did not confirm this trend (Ueda et al.

2003; Treister & Urry 2005).

Nevertheless, the hard X-ray AGN population seems to have the right characteristics to explain the shape and normalisation of the CXB around 30 keV. Sazonov et al. (2007) used the absorption distribution of the INTEGRAL AGN combined with a typical shape of their X-ray spectrum and a pure luminosity evolution of the AGN population, and found that in this way they can account for the shape and the amplitude of the cosmic hard X-ray background. Furthermore, the results of the hard X-ray surveys are marginally compatible with the predictions of the synthesis model of Gilli et al.

(2007a, see Sect. 5.1.2). The fraction of both Compton-thin and Compton-thick AGN is expected to vary with the limiting flux of the survey as shown in Fig. 5.11 (left panel). Even though early Swift measurements are consistent with Gilli et al. predictions (Markwardt et al. 2005), the fraction of Compton-thick objects observed by INTEGRAL and that reported by the more recent Swift work of Tueller et al. (2008)¹ would be marginally consistent only if it included all the objects with unknown N_H, whereas the fraction of Compton-thin AGN seems to be overestimated by the model. The model of Gilli et al. was shown to be very sensitive to the choice of the average slope of the X-ray continuum used for the modelling of the AGN spectra. In fact, when a Gaussian distribution of slopes with

1In this work 80 out of 82 Seyferts have measured absorption and the fractions of absorbed and Compton-thick AGN are 49% and 6%, respectively. If the 2 Seyfert galaxies with unknown NHare assumed to be Compton-thick or not, these fractions vary in the range 48–50% and 6–9%, respectively, in agreement with INTEGRAL results.

Figure 5.11: Left:Dependence of the fraction of absorbed (upper line) and Compton-thick (lower line) AGN on the limit flux of the X-ray survey, as predicted by the model of Gilli et al. (2007a). The squares and the stars refer to early measurement obtained bySwift(Markwardt et al. 2005), in particular those with solid error bars are the measured values and those with dotted error bars assume that the unidentified sources are obscured. The circles indicate the measurements obtained byINTEGRAL(Beckmann et al. 2006b) and their error bars give the range of the values obtained when assuming that none or all the sources with unknownNHare Compton-thick. Right: The 20–40 keVlog N−log S predicted by the model of Gilli et al. (2007a) when assuming different values for the average slope of the AGN X-ray spectra and including or not as much Compton-thick AGN as Compton-thin (continuous and dashed lines). The shaded area indicate thelog N−log S measured by Beckmann et al. (2006b).

average value 1.8 and dispersion 0.2 is assumed, the CXB can be well represented without need of a Compton-thick population (Gilli et al. 2007b). On the other hand, the chosen average value of 1.9 is in better agreement with observational constraints. When the 20–40 keV log N−log S predicted by this model, assuming different average slopes and Compton-thick contributions, is compared to that we have obtained with INTEGRAL, the curve corresponding to a slope of 1.9 is in better agreement with the measurements, even though it does not put tight constraints (Fig. 5.11, right panel).

5.3.3 Future perspectives

Among the more than 200 sources discovered by INTEGRAL, about 50% have been identified and classified thanks to observations in other energy bands, typically optical and soft X-rays. 50 of them have revealed to be AGN (Bird et al. 2007; Bodaghee et al. 2007). Among the still unidentified objects or those AGN that are missing a measurement of the NH, one can expect to find a number of absorbed AGN. In order to better constrain the upper limits given above for the fraction of absorbed and Compton-thick AGN in the hard X-ray surveys, it is necessary to continue with the follow-up of these sources. With this aim, we have applied and obtained Chandra observing time during the last two years to identify some of the new sources detected in the deep INTEGRAL field around 3C 273. The first observations have been already performed and will be object of a future work. Yet, a recent study on soft X-ray follow-ups of 34 new INTEGRAL AGN and AGN candidates (Malizia et al. 2007) has shown a fraction of absorbed objects of∼50% and only 3 Compton-thick candidates (9%), confirming the distribution of absorption in the local AGN population as discussed above.

The hard X-ray luminosity function of AGN 81 With the INTEGRAL and Swift observations performed up to now, we are able to resolve into sources only up to 2.5% of the flux of the CXB above 20 keV, due to the high-flux limits of these surveys. Considering that the fraction of absorbed, and in particular Compton-thick, sources is in-creasing when the flux limit decreases (Fig. 5.11, left panel), deeper surveys are needed to improve our knowledge of the number-flux distribution of local AGN at low fluxes. Besides the 3C 273 re-gion, INTEGRAL has performed deep observations in two other extragalactic fields, one centered on the XMM Large Scale Structure survey field and the other close to the north ecliptic pole, where observations are still on-going and will continue in AO–6. The exposure accumulated in these two regions at the end of 2009 will be about 3 Ms and 4 Ms, respectively. Even though a comparison with the results obtained up to now will be very useful, these surveys will at maximum reach the same flux limit as the 3C 273 region has now. Much more INTEGRAL observing time should be concentrated on a well defined extragalactic field in order to obtain a significant improvement of the flux sensitivity and of the scientific outcome of the survey. On the other hand, thanks to its larger field of view when compared to ISGRI and to its observational strategy (randomized by the re-pointing during gamma-ray burst events), Swift/BAT will provide us with the deepest all-sky, hard X-gamma-ray survey, before the next generation satellites. Considering the sensitivity reached by the 9-months survey (Tueller et al.

2008), one can roughly estimate that after a total of 7 years Swift will be able to reach a flux limit of 1.6×10⁻¹¹erg cm⁻²s⁻¹in the 14–195 keV band on the whole sky.

Among the future missions, XEUS, Simbol-X and maybe EXIST will play a fundamental role to study the AGN population responsible for the hard X-ray background. Both XEUS and Simbol-X will use grazing incidence optics and imaging detectors in a very long focal length telescope, thanks to formation flying configuration. Simbol-X is a project of the French and Italian space agencies, with German participation, that has just entered a phase B development aiming to a launch around 2014.

It will have imaging and spectral capabilities in the 0.5–100 keV range, with an angular resolution of

∼8” and sub-µCrab sensitivity at 20–40 keV. Thanks to these capabilities, Simbol-X is expected to be able to resolve about 65% of the CXB in the 20–40 keV band (Pareschi & Ferrando 2005). XEUS²is a major project currently under study by the European and Japanese space agencies, with a possible launch date in 2018. This X-ray observatory will be designed to have 2–5 arcsec angular resolution and a large effective area of 5 m² at 1 keV, making it around 200 times more sensitive than XMM (Hasinger et al. 2006). Even though the current specifications require that the 0.1–15 keV band is covered by XEUS instruments, the goal is to extend the Wide Field Imager range up to∼50 keV. This would make XEUS able to study the hard X-ray properties of AGN as a function of redshift up to z = 10, going far beyond the local population studied by INTEGRAL and Swift. EXIST is a hard X-ray, imaging all-sky deep-survey project of NASA that has been proposed for the “Beyond Einstein”

program but has not been selected in the first round in 2007. Nevertheless, it will be considered for

“Decadal Survey”, with decisions expected in late 2009 or early 2010. The mission design is based on the well understood coded-aperture technique (as used in e.g. INTEGRAL and Swift), and will not include “risky” components such as moving parts or active cooling. EXIST would have two main instruments covering the 3–600 keV range, with 6^′resolution over a 154^◦×65^◦field of view above 10 keV and an area of the high-energy detector 20 times larger than the ISGRI one. In comparison to the shallow surveys in the 20–150 keV range provided by Swift and INTEGRAL, EXIST with its improved sensitivity will for the first time allow to detect 10,000s of AGN up to a redshift of 0.5, and

2In July 2008, the International ray Observatory (IXO) was announced to the astronomical community as a joint X-ray mission with participation from the European, American and Japanese space agencies. The IXO mission supersedes the XEUS mission concept.

above for bright objects.

5.3.4 Scientific article on the luminosity function of INTEGRAL AGN

The details of this research are published in the Astrophysical Journal as Beckmann et al. (2006b), a copy of which is provided on the next pages:

• Beckmann V., Soldi S., Shrader C.R., Gehrels N. and Produit N., The hard X-ray 20–40 keV AGN luminosity function

The Astrophysical Journal, 2006, Vol. 652, pages 126–135

The hard X-ray luminosity function of AGN 83

THE HARD X-RAY 20Y40 keV AGN LUMINOSITY FUNCTION V. Beckmann,¹S. Soldi,^{2, 3}C. R. Shrader,^{4, 5}N. Gehrels,⁴and N. Produit²

Received 2006 March 1; accepted 2006 June 28

ABSTRACT

We have compiled a complete extragalactic sample based on25,000 deg²to a limiting flux of 3;10¹¹ergs cm² s¹ (7000 deg²to a flux limit of 10¹¹ergs cm² s¹) in the 20Y40 keV band withINTEGRAL. We have con-structed a detailed exposure map to compensate for effects of nonuniform exposure. The flux-number relation is best described by a power law with a slope of ¼1:660:11. The integration of the cumulative flux per unit area leads tof20Y40 keV¼2:6;10¹⁰ergs cm²s¹sr¹, which is about 1% of the known 20Y40 keV X-ray background. We present the first luminosity function of AGNs in the 20Y40 keV energy range, based on 38 extragalactic objects detected by the imager IBIS-ISGRI on boardINTEGRAL. The luminosity function shows a smoothly connected doubleYpower-law form with an index of1¼0:8 below and2¼2:1 above the turnover luminosity ofL¼2:4; 10⁴³ ergs s¹. The emissivity of all INTEGRALAGNs per unit volume is W20Y40 keV(>10⁴¹ ergs s¹)¼2:8; 10³⁸ ergs s¹ h³₇₀ Mpc³. These results are consistent with those derived in the 2Y20 keV energy band and do not show a significant contribution by Compton-thick objects. Because the sample used in this study is truly local ( ¯z¼ 0:022), only limited conclusions can be drawn for the evolution of AGNs in this energy band.

Subject headinggs: galaxies: active — galaxies: Seyfert — gamma rays: observations — surveys — X-rays: galaxies

Online material:color figures

1. INTRODUCTION

The Galactic X-ray sky is dominated by accreting binary sys-tems, while the extragalactic sky shows mainly active galactic nuclei (AGNs) and clusters of galaxies. Studying the popula-tion of sources in X-ray bands has been a challenge ever since the first observations by rocketborne X-ray detectors (Giacconi et al.

1962). At soft X-rays (0.1Y2.4 keV ), deep exposures byROSAT have revealed an extragalactic population of mainly broad-line AGNs, such as type 1 Seyfert and quasars ( Hasinger et al. 1998;

Schmidt et al. 1998). In the 2Y10 keV range, surveys have been carried out with theAdvanced Satellite for Cosmology and As-trophysics(ASCA; e.g., Ueda et al. 2001),XMM-Newton(e.g., Hasinger 2004), andChandra(e.g., Brandt et al. 2001) and have shown that the dominant extragalactic sources are more strongly absorbed than those within theROSATenergy band. For a sum-mary on the deep X-ray surveys below 10 keV see Brandt &

Hasinger (2005). At higher energies the data become more scarce.

Between a few keV and1 MeV, no all-sky survey using im-aging instruments has been performed to date. TheRossi X-Ray TimingExplorer(RXTE) sky survey in the 3Y20 keV energy band revealed about 100 AGNs, showing an even higher fraction of absorbed (NH>10²²cm²) sources of about 60% (Sazonov

& Revnivtsev 2004).

The International Gamma-Ray Astrophysics Laboratory (INTEGRAL; Winkler et al. 2003) offers an unprecedented

>20 keV collecting area and state-of-the-art detector

electron-ics and background-rejection capabilities. Notably, the imager IBIS, with an operating range from 20 to 1000 keV and a fully coded field of view of 10;10, now enables us to study a large portion of the sky. A first catalog of AGNs showed a similar fraction of absorbed objects to the RXTE survey ( Beckmann et al. 2006a). The Burst Alert Telescope ( BAT) of the Swift mission (Gehrels et al. 2004) operates in the 15Y200 keV band and uses a detector similar to IBIS-ISGRI, but provides a field of view about twice the size. The BAT data of the first 3 months of the mission provided a high Galactic latitude survey, including 44 AGNs ( Markwardt et al. 2005). Within this sample a weak anticorrelation of luminosity versus intrinsic absorption was found, as previously found in the 2Y10 keV band ( Ueda et al. 2003;

La Franca et al. 2005), revealing that most of the objects with luminositiesL_X>3;10⁴³ ergs s¹show no intrinsic absorption.

Markwardt et al. (2005) also pointed out that this luminosity cor-responds to the break in the luminosity function.

Related to the compilation of AGN surveys in the hard X-rays is the question of which sources form the cosmic X-ray back-ground (CXB). While the CXB below 20 keV has been the focus of many studies, the most reliable measurements in the 10Y 500 keV range have been provided by theHigh Energy Astronom-ical Observatory(HEAO-1), launched in 1977 ( Marshall et al.

1980). The most precise measurements provided by the UCSD/

MIT Hard X-Ray and Gamma-Ray instrument (HEAO-1A-4) show that the CXB peaks at an energy of about 30 keV ( Marshall et al. 1980; Gruber et al. 1999). The isotropic nature of the X-ray

MIT Hard X-Ray and Gamma-Ray instrument (HEAO-1A-4) show that the CXB peaks at an energy of about 30 keV ( Marshall et al. 1980; Gruber et al. 1999). The isotropic nature of the X-ray