In our analysis, we have found a good description of the behavior of the SXLF from a combination of various ROSAT surveys. As explained in Sect. 3.1, our expression is for the total AGN population, including type 1 and type 2's and for the observed 0.5-2 keV band, because of the uncertainties of contents and evolution of AGNs in various spectral classes. These have to be assumed to find the best-bet K-corrected AGN evolution in the source rest frame. A detailed discussion of this aspect is beyond the scope of this paper. An approach for the problem is to make a population synthesis modeling, e.g., composed of unabsorbed and absorbed AGNs similar to those of Madau et al. (1994) and Comastri et al. (1995) (see also Gilli et al. 1999for a recent work). If one is constructing a model in a similar approach using our SXLF as a major constraint, what the model constructor should do is to calculate the expected SXLFs in the observed 0.5-2 keV band for all emission-line AGN populations (spectral classes) considered in the model (e.g. corresponding to different absorbing column densities) and then to compare the total model SXLF with our LDDE1/LDDE2 expressions. One version of our own models constructed using this approach has been shown in M99b. We do not recommend the use of the expressions in Appendix A. as the SXLF of unabsorbed AGNs for the reasons described there and Sect. 3.1.
We have found two versions of LDDE expressions consistent with our sample in the luminosity and redshift regime covered: one which produces of the 0.5-2 keV extragalactic CXRB (the lower etimate, see Sect. 4), and the other one which produces , as two relatively extreme cases on extrapolation. The real behavior is probably somewhere between these two. Note that we have only calculated the contribution to the CXRB for , where fits were made. Below this luminosity, we observe an excess (Fig. 3), which connects well with the SXLF of nearby Galaxies (Schmidt et al. 1996; Georgantopoulos et al. 1999), in the very local universe. This component has a local volume emissivity comparable or more to our sample AGNs in the 0.5-2 keV range and can contribute significantly to the soft CXRB. Because of the low luminosity, we can only detect this population in the very nearby universe in a large-area surveys like RASS. Deep small-area surveys would not give enough volume to detect them, since even the deepest part of RDS-LH can detect a galaxy only up to .
The X-ray emission of this low luminosity population is probably contributed by both star-formation and by low-activity AGNs (including LINERS). Although Georgantopoulos et al. (1999)'s analysis suggests that a major contribution is from Seyfert galaxies and LINERS even at these low luminosities, star-formation activity can also contribute significantly to the X-ray emission of these low-activity AGNs (see Lehmann et al. 1999a). As one extreme scenario, we assume that the X-ray emission from these low-luminosity sources is mostly from star-formation activity and their volume emissivity is assumed to evolve like the global star-formation rate (SFR; e.g. Madau et al. 1996; Connolly et al. 1997), the integrated intensity would be roughly 30-40% of the lower estimate of the CXRB intensity. Even if the evolution of these low-luminosity sources were PLE, we would not detect any of them at intermediate to high redshifts even in the deepest ROSAT Survey on the Lockman Hole. Therefore this picture is still consistent with the result that the RDS-LH did not find any starburst galaxies. If the above scenario is the case, the bahavior of the AGN component would need to be close to LDDE1 to allow room for a contribution from star-forming galaxies. In that case, the softer emission from star-formation activity could contribute to the [keV] excess of the CXRB spectrum and the total extragalactic 0.5-2 [keV] intensity could be closer to the upper estimate. If on the other hand, the large apparent local volume emissivity for the low-luminosity component is produced by the local overdensity and not representative of the average present-epoch universe (e.g. Schmidt et al. 1996is from a sample within 7.5 [Mpc]) and/or the X-ray evolution is slower than the global SFR (e.g. delayed formation of LMXB, White & Ghosh 1998), an LDDE2-like behavior for the AGN component may also be possible. A more detailed invetigation of the above scenarios and the exploration of other possibilities will be a topic of a future work.
One of the most interesting results is the evolution of luminous QSOs discussed in Sect. 5. A comparison of the evolution and the global star-formation rate is discussed in Franceschini et al. (1999), where it is proposed that the evolution of the volume emissivity of the luminous QSOs evolves like the star-formation rate (SFR) of early-type galaxies, while that of the total AGN population (from the LDDE1 and LDDE2 models) may evolve like the SFR of all galaxies. Another interesting feature is that we find no evidence for a rapid decline of the QSO number density at high redshift. The SSG95-like decrease at is marginally rejected. The difference may be caused by different selection criteria. SSG95 have selected QSOs by the Ly luminosity and their QSOs are representative of more luminous QSOs (). Recently Wolf et al. (1999) reported a similar tendency in their sample of QSOs from one of their CADIS fields, which typically have lower luminosities than the SSG95 sample. Our X-ray selected AGNs with have a seven times higher space density than SSG95 at and thus are sampling lower luminosity QSOs than SSG95. Thus if the behavior of our ROSAT -selected QSOs and those of Wolf et al. (1999) is really flat, this can be indicative of different formation epochs for lower and higher mass black holes. Adding more deep ROSAT surveys would enable us to trace the evolution in this regime with a better statistical significance. The upcoming Chandra and XMM Surveys would extend the analysis to lower-luminsoity objects at the highest redshifts as well as enabling us to give spectral information to separate the K-effect and the actual evolution of the number density.
© European Southern Observatory (ESO) 2000
Online publication: December 8, 1999