Astron. Astrophys. 347, 424-433 (1999)

1. Introduction

The cosmic X-ray background (XRB) above keV is the result of the integrated emission of discrete sources, since the contribution of any intergalactic hot medium must be negligible (Wright et al. 1994). In the soft X-ray band from 0.5 to 2 keV the largest fraction of the XRB has already been resolved into sources (Hasinger et al. 1998), most of which turned out to be broad line active nuclei (Schmidt et al. 1998), i.e. Quasi Stellar Objects (QSOs) and Seyfert galaxies of type 1. The spectra of these sources are however too steep to reproduce also the hard XRB at several tens of keV, where the bulk of the energy resides, and a population of objects with flatter spectra is therefore required.

The most popular synthesis models of the XRB are based on the so-called unification schemes for Active Galactic Nuclei (AGNs), where the orientation of a molecular torus surrounding the nucleus determines the classification of the source. At a zeroth-order approximation level, sources observed along lines of sight free from the torus obscuration should have unabsorbed X-ray spectra and optical broad lines (type 1 AGNs), while sources seen through the torus should have absorbed X-ray spectra and appear as narrow line objects in the optical (type 2 AGNs, e.g. Seyfert 2 galaxies).

In this framework type 2 AGNs provide a natural class of sources with X-ray spectra flattened by absorption. The intrinsic X-ray luminosity function (XLF) of type 2 objects is unknown and has been usually assumed to be the same as the one derived for type 1s (e.g. Boyle et al. 1993), apart from a normalization factor. The cosmological evolution has also been taken identical for type 1s and type 2s. Under these assumptions it has been shown that the broad band 3-100 keV spectrum of the XRB can be reproduced by an appropriate mix of unabsorbed and absorbed AGNs (Matt & Fabian 1994; Madau et al. 1994; Comastri et al. 1995, hereafter Co95). The number ratio R of type 2 to type 1 objects, as well as the distribution of the absorbing column densities , are key parameters of the models; these have been assumed to be independent of redshift and of intrinsic source luminosity, and have been treated as free parameters in the fitting procedure. Since the overall parameter space of the models is quite large and a good fit to the XRB can be obtained with different set of values, it is important to compare the model predictions with the largest number of observational constraints. Indeed, Co95 showed that the source counts in the 0.5-2 keV and 2-10 keV energy bands, as well as the redshift distributions, could successfully be reproduced by their model.

Very recently an additional set of observational constraints has become available. Deep surveys from ROSAT have extended our knowledge to the low luminosity part of the AGN XLF (Miyaji et al. 1999a, hereafter Mi99a). Contrary to previous results (Boyle et al. 1993; Page et al. 1996; Jones et al. 1997) a pure luminosity evolution (PLE) of AGNs with redshift is no longer consistent with the data, and a luminosity dependent density evolution (LDDE) is required. From the X-ray data of an optically selected sample of Seyfert galaxies Risaliti et al. (1999) have determined the distribution for local Seyfert 2 galaxies, pointing out that a significant fraction of sources have columns exceeding cm^-2 and are therefore completely thick to Compton scattering. The R ratio between type 2s and type 1s has been determined in the local Universe for low luminosity AGNs, i.e. Seyfert galaxies (Maiolino & Rieke 1995), while the existence of a relevant number of high luminosity absorbed sources, the so-called QSO 2s, which is a basic assumption of previous models, is still uncertain (Akiyama et al. 1998). An observational constraint to the QSO 2 number density can be obtained from the infrared source counts. Indeed, QSO 2s are expected to have strong infrared counterparts, since the dust present in the torus should re-emit in the IR band the nuclear radiation absorbed by the gas. The ultraluminous infrared galaxies (ULIRGs) discovered by IRAS are the only local objects with QSO-like bolometric luminosities (Soifer et al. 1986; Kim & Sanders 1998). Thus, even if all ULIRGs were powered by a hidden AGN, the local QSO 2s could not be more numerous than ULIRGs. Finally, source counts in the 5-10 keV band have been derived for the first time by the BeppoSAX satellite with the HELLAS survey (Giommi et al. 1998; Comastri et al. 1999).

In the present paper we test the standard synthesis model to verify if it remains compatible with the new data. These data leave still some latitude to important parameters of the model, and various choices are possible to fit the XRB equally well. However, in all cases we find moderate but consistent evidence that at least some of the standard assumptions have to be relaxed: extra hard spectrum AGNs are needed at intermediate or high redshifts, in addition to those expected in the usual scenario. The additional sources could be analogous to local Seyfert 2s, if they evolve faster than type 1s, or they could be other astrophysical sources not yet enlisted among the contributors to the XRB. We discuss the observations which could distinguish between the alternatives.

Throughout this paper the deceleration parameter and the Hubble constant are given the values and km s^-1 Mpc^-1.

© European Southern Observatory (ESO) 1999

Online publication: June 30, 1999
helpdesk.link@springer.de