Searching for quasars at z > 3 is one of the key elements of studying the growth history of supermassive black holes and the evolution of massive galaxies in the Universe. For this purpose, multiwavelength observations of a large number of quasars are needed. Particularly valuable are X-ray observations: the intensive X-ray emission is directly connected with powerful processes of accretion onto the black hole. In order to improve our fairly poor knowledge of the evolution of the X-ray luminosity function of quasars at z > 3, larger X-ray samples of distant quasars are needed.

We have made an attempt to obtain a large sample of luminous X-ray quasars at z > 3 in the fields of the serendipitous XMM-Newton 3XMM-DR4 (Watson et al., 2009) survey at Galactic latitudes |b| > 20° using photometric data from SDSS (Eisenstein et al., 2011), 2MASS (Skrutskie et al., 2006) and WISE (Wright et al., 2010). The 3XMM-DR4/SDSS overlapping area is ~300 sq. deg, exceeding by several times the areas covered by known samples of distant X-ray quasars such as Champ (Kalfountzou et al., 2014) and XMM-XXL (north) (Menzel et al., 2016). The sample of Kalfountzou et al. (2014) covers ≈33 sq. deg and includes 87 X-ray selected optically bright (detected by SDSS) quasars. Half of them have photometric redshifts only. The spectroscopic program of (Menzel et al., 2016) has revealed 61 X-ray objects at z_spec > 3 in a 18 sq. deg subfield of XMM-XXL (north). Our sample thus provides an opportunity to study a population of rare high-luminosity, distant quasars and complements recent studies of less luminous high-redshift quasars conducted using deep small-area surveys (e.g., Onoue et al., 2017; Ricci et al., 2017; Vito et al., 2017).

It should be noted that SDSS spectroscopic data provide redshift measurements for 33 thousand optically selected quasars at z_spec > 3 (Alam et al., 2015) over 11,000 sq. deg in the sky. However, existing wide-area X-ray surveys are too shallow to study these objects. It is only the forthcoming all-sky X-ray survey by eROSITA aboard the Spektrum-Roentgen-Gamma observatory (Pavlinsky et al., 2011; Merloni et al., 2012) that will be capable of finding X-ray counterparts for a substantial fraction of known quasars at z_spec > 3.

Based on broadband photometry, we obtained photometric redshift estimates (z_phot) and compiled a catalog of 903 candidates for distant quasars with z_phot > 2.75 (Khorunzhev et al., 2016). The catalog includes 515 known quasars (with spectroscopic redshifts, of which 266 are at z_spec > 3) and 388 new quasar candidates. Most (80%) of the spectroscopic redshifts have been obtained by SDSS (Alam et al., 2015). The remaining z_spec have been gathered from various sources (Flesch, 2015). The new quasar candidates selected by z_phot constitute a substantial addition to the spectroscopic sample. Thus, if most of these candidates prove to be quasars at z > 3, the existing 3XMM-DR4/SDSS sample of distant quasars will be enhanced by a factor of ~1.5.

Spectroscopic verification is necessary to determine the accuracy of our z_phot estimates and assess the purity of our selection of quasar candidates. We have thus started a spectroscopy identification program for the new quasar candidates (Khorunzhev et al., 2017). Some initial results of this program are reported below.

K16 considered point SDSS sources at Galactic latitudes |b| > 20° that have an X-ray counterpart in 3XMM-DR4¹. In addition, we used near- (2MASS) and medium-infrared (WISE) photometry if available. To keep sources with reliable photometry and to get rid of M-dwarfs, we applied the following condition: (1)δmz′<0.2 & i′−z′<0.6 , where i′ and z′ are the PSF magnitudes in the appropriate SDSS bands and δmz′<0.2 is the corresponding error. This is a well-known technique (e.g., Richards et al., 2002; Wu et al., 2012; Skrzypek et al., 2015) for separating M-dwarfs from distant quasars. Stars have i′ − z′ > 0.8, while quasars have i′ − z′ < 0.4. This color remains approximately constant up to a redshift of ≈5.5, until the Lyα line passes from i′ to z′. The color of quasars then becomes i′ − z′ ≈ 2.

We then performed a broadband energy distribution fitting using the EAZY software (Brammer et al., 2008) to obtain photometric redshift estimates. We made two fitting iterations for each object using libraries of quasar (by various authors) and star (Pickles, 1998) templates. Those objects with (2)χstar2/χqso2>1 &zphot>2.75, finally constituted our catalog of 903 candidates for distant quasars selected by photometric redshift. At the same time, our procedure missed 63 known quasars at z_spec > 3 (see Figure 1). Note that we lowered the threshold of selection from z_phot = 3.0 to 2.75 to achieve reasonable selection completeness at z ~ 3.

The completeness of our catalog in the investigated fields relative to existing spectroscopic catalogs of quasars (SDSS DR12 Alam et al. 2015 and The Half Million Quasars Flesch 2015) with z_spec > 3 is about 80%. The normalized median absolute deviation of photometric redshift estimates for the spectroscopically confirmed quasars (Δz = |z_spec − z_phot|) is σ_{Δz/(1+zspec)} = 0.07, while the catastrophic outlier fraction is η = 9% (when Δz/(1+z_spec) > 0.2).

We have performed a quasi-random spectroscopic survey of quasar candidates from the K16 catalog at the 1.6-m AZT-33IK telescope (Kamus et al., 2002; Denisenko et al., 2009) equipped with the low- and medium-resolution ADAM spectrograph (Afanasiev et al., 2016; Burenin et al., 2016). A number of additional observations were conducted with the SCORPIO I (Afanasiev and Moiseev, 2005) spectrograph at the 6-m BTA telescope.

The obtained spectra of dosens of quasars at z ~ 3 and especially the discovery of one of the most distant X-ray selected quasars (3XMM J125329.4+305539) at z_spec = 5.08 demonstrate that the AZT-33IK telescope of the Sayan Solar Observatory equipped with the new ADAM spectrograph is well suited for identification of distant quasars. This telescope is planned to be one of the instruments employed for optical support of the upcoming all-sky X-ray survey by the Spektrum-Roentgen-Gamma observatory (with its eROSITA and ART-XC telescopes, Pavlinsky et al., 2011; Merloni et al., 2012).

We have demonstrated that the existing sample of distant X-ray quasars can be substationally increased using publicly available X-ray (XMM-Newton) and optical photometry (SDSS) data, complemented by a relatively low-cost spectroscopic identification program.

We are continuing our spectroscopic verification program for K16 candidates using the AZT-33IK and BTA telescopes and intend to obtain a complete spectroscopic sample at least for brighter X-ray sources from the K16 catalog.

GK provided key contributions to the compilation of the catalog, spectroscopic observations, data analysis and writing of this paper. SS supervised the research at all stages. RB contributed to the aquisition and processing of spectroscopic data. ME organized spectroscopic observations at AZT-33IK.