Left: Observational constraints on the X-ray luminosity function of AGN in two redshift intervals, z=3-4 and z=4-5. The coloured shaded regions are parametric model fits to the data. The width of the shaded regions measures the 1σ uncertainties. The black symbols (circles or triangles) are non-parametric (binned) estimates of the AGN space density. The plot shows that X-ray selected AGN experience strong evolution from redshift z=3.5 to z=4.5. For example, the space density of AGN with luminosities L_X<10⁴⁵erg/s drop by about one order of magnitude between the two redshifts. At brighter luminosities the evolution of AGN is milder. Right: Observational constraints on the contribution of AGN to the UV photon-rate density needed to keep the Universe ionised at any given redshift. The shaded regions are hydrogen ionisisng photo-rates based on the parametric X-ray luminosity functions plotted on the left panel. For simplicity we assume a photon escaping fraction of unity, i.e. ignoring obscuration effects close to the supermassive black hole. The thick black line in the plot shows the photon rate density required to keep the Universe ionised at any redshift. The ratio between the shaded region and the black line is presented in the inset plot. It shows that AGN alone cannot reionise the Universe at redshifts z>4. More details in Georgakakis et al. 2015.

The whereabouts of the first supermassive black holes in the early Universe, with masses in excess of 10⁷ in solar units, is an open question in current astrophysical research. The formation process of such monsters when the Universe was still young is still under debate. Moreover, the interplay between the growth of the first black holes and the formation of proto-galaxies is still not well understood. The formation of the first black holes also has cosmological implications. During their growth phase such black holes are believed to emit hard UV photon radiation, which may contribute of even dominate the UV photon field needed to reionise the Universe at early epochs.

The first step for addressing the issues above is to constrain the space density evolution of AGN at high redshift, z>3, i.e. when the Universe was younger than 2Gyr. Selection of AGN at X-ray wavelengths is advantageous for such studies, because X-ray photons are less sensitive to obscuration effects, compared for example, to UV/optical selection methods. Challenges for X-ray selection methods are (i) the compilation of large samples of high redshift AGN that span a wide-range of accretion luminosities and (ii) the determination of reliable redshifts for individual sources, particularly since a sizable fraction of X-ray sources in current X-ray surveys are too faint for follow-up spectroscopy.

Work is currently in progress to provide robust constraints on the AGN space density as a function of redshift and accretion luminosity at high redshift by (i) combining wide-area/shallow and pencil-beam/deep X-ray surveys to have sufficiently good coverage of the L_X-z plane, (ii) using multiwavelength photometric data to measure photometric redshifts for X-ray sources that are too faint for follow-up spectroscopy and (iii) developing a Bayesian approach to propagate uncertainties, in e.g. photometric redshift measurements or accretion luminosity estimations, in the AGN space density calculations. Recent results of this approach include the determination of z=3-5 X-ray luminosity function by Georgakakis et al. 2015.