1.1. Brief Historical Summary of X-ray Deep Field Research
The CXRB was discovered by Giacconi et al. (1962) in a rocket flight originally designed to detect X-ray emission from the Moon; the CXRB was the first cosmic background discovered. The data showed a strong Galactic X-ray source (Sco X-1) and diffuse emission of approximately constant intensity from all directions observed during the flight. The first all-sky X-ray surveys with Uhuru and Ariel V in the 1970's revealed a high degree of CXRB isotropy leading researchers to the conclusion that the CXRB has to be mainly extragalactic. If it were comprised of discrete sources, such as AGN (e.g., Setti & Woltjer 1973), the number of sources contributing to the CXRB had to be very large: N > 106 sr-1 (Schwartz 1980). On the other hand, high-quality data from 1 showed that the shape of the 3-50 keV CXRB could be well fit by an isothermal bremsstrahlung model corresponding to an optically thin, hot plasma with kT 40 keV (Marshall et al. 1980). This was taken by some to suggest an origin of the CXRB in a hot intergalactic medium, until this possibility was ruled out by Compton-distortion constraints on the spectrum of the cosmic microwave background (see Section 5.1 of Fabian & Barcons 1992 and references therein).
When sensitive, high angular resolution imaging X-ray observations with Wolter telescopes became possible, the discrete nature of the CXRB became increasingly clear. Pointed observations of previously known AGN soon showed that, as a class, they are luminous X-ray emitters (e.g., Tananbaum et al. 1979). Deep Einstein observations resolved 25% of the 1-3 keV CXRB into discrete sources at fluxes down to 3 × 10-14 erg cm-2 s-1, a large fraction of which were identified as AGN (Giacconi et al. 1979). Deep 0.5-2 keV surveys with ROSAT to limiting fluxes of 10-15 erg cm-2 s-1 were for the first time able to resolve the majority ( 75%) of the soft CXRB into discrete sources (e.g., Hasinger et al. 1993, 1998). Extensive optical follow-up spectroscopy identified the bulk of these sources as AGN (e.g., McHardy et al. 1998; Schmidt et al. 1998; Zamorani et al. 1999; Lehmann et al. 2001), demonstrating that at least the 0.5-2 keV CXRB is predominantly due to accretion onto SMBH, integrated over cosmic time. The deep ROSAT surveys detected an AGN sky density ( 780-870 deg-2) larger than at any other wavelength and found evidence for luminosity-dependent density evolution of AGN (Miyaji, Hasinger & Schmidt 2000), contrary to the pure luminosity evolution which had been proposed for both optically and X-ray selected AGN (e.g., Boyle et al. 1993).
Deep surveys above 2-3 keV had to wait considerably longer than those at lower energies due to technological challenges. ASCA performed a number of medium-to-deep sky surveys to limiting 2-10 keV fluxes of 10-13 - 5 × 10-14 erg cm-2 s-1 (e.g., Georgantopoulos et al. 1997; Ueda et al. 1998; Cagnoni, della Ceca & Maccacaro 1998; Ishisaki et al. 2001). These surveys reached AGN sky densities of 10-100 sources deg-2. At higher densities these surveys were heavily confused due to the limited angular resolution of ASCA. An analysis of the spatial fluctuations in deep ASCA images probed the 2-10 keV source counts down to a flux limit of 2 × 10-14 erg cm-2 s-1, resolving 35% of the 2-10 keV CXRB (Gendreau, Barcons & Fabian 1998). Surveys in the 5-10 keV band were pioneered using BeppoSAX, which was well suited for such work because of its relatively large throughput at high energies and its significantly sharper point spread function at high energy compared with ASCA. These observations resolved 20-30% of the 5-10 keV CXRB (e.g., Comastri et al. 2001).
The above X-ray observations were interpreted in the context of CXRB population synthesis models based on unified AGN schemes (e.g., Madau, Ghisellini & Fabian 1994; Comastri et al. 1995; Gilli, Salvati & Hasinger 2001). These explain the CXRB spectrum using a mixture of obscured and unobscured AGN, folded with the corresponding luminosity function and its cosmological evolution. According to these models, most AGN spectra are heavily absorbed, and about 85% of the radiation produced by SMBH accretion is obscured by dust and gas (e.g., Fabian & Iwasawa 1999). These models predicted a significant number of heavily obscured AGN in deep hard X-ray fields, which had already been partly found in the ASCA and BeppoSAX surveys. They also predicted a substantial contribution from high-luminosity, obscured X-ray sources (e.g., type 2 quasars), which at that time had only scarcely been detected. However, these models were far from unique and contained a number of implicit assumptions. For instance, evolution of obscuration over cosmic time and the dependence of obscuration on intrinsic source luminosity were largely free parameters. Nevertheless, these models provided a working framework and predictions for deeper surveys, which could be tested with Chandra and XMM-Newton.