2004A&A...424..545P

We use Virtual Observatory (VO) tools to identify optically faint, obscured (i.e., type 2) active galactic nuclei (AGN) in the two Great Observatories Origins Deep Survey (GOODS) fields. By employing publicly available X-ray and optical data and catalogues we discover 68 type 2 AGN candidates. The X-ray powers of these sources are estimated by using a previously known correlation between X-ray luminosity and X-ray-to-optical flux ratio. Thirty-one of our candidates have high estimated powers (L_x>10⁴⁴erg/s) and therefore qualify as optically obscured quasars, the so-called ``QSO 2''. Based on the derived X-ray powers, our candidates are likely to be at relatively high redshifts, z∼3, with the QSO 2 at z∼4. By going ∼3mag fainter than previously known type 2 AGN in the two GOODS fields we are sampling a region of redshift - power space which was previously unreachable with classical methods. Our method brings to 40 the number of QSO 2 in the GOODS fields, an improvement of a factor ∼4 when compared to the only 9 such sources previously known. We derive a QSO 2 surface density down to 10<sup>–15</sup>erg/cm<sup>2</sup>/s in the 0.5-8keV band of >330deg<sup>–2</sup>, ∼30% of which is made up of previously known sources. This is larger than current estimates and some predictions and suggests that the surface density of QSO 2 at faint flux limits has been underestimated. This work demonstrates that VO tools are mature enough to produce cutting-edge science results by exploiting astronomical data beyond ``classical'' identification limits (R≲25) with interoperable tools for statistical identification of sources using multiwavelength information.