Selection of Compton-thick AGN from a hard photometric sample using XMM-Newton observations

Abstract

We present a selection technique to detect Compton-thick (CT) active galactic nuclei (AGNs) in the 3XMM/SDSS-DR7 cross-correlation. A subsample of 3481 X-ray sources that are detected in the hard band (2-8 keV) and have photometric redshifts constitute our parent sample. We first applied an automated spectral-fitting procedure to select highly absorbed sources (N_H > 10^23 cm^-2). We found 184 highly absorbed candidates. Then, we performed the Bayesian Monte Carlo Markov chains (MCMCs) selection technique to find CT AGNs. We also tested the MCMC selection technique by applying Monte Carlo simulations. We found that the method is accurate at 90 percent independently of the nature of the underlying source. Our sample contains 52 bona fide CT AGNs. The CT AGNs were selected to have a range > 0.75 of probability of being CT when either fitting with the two models Torus and MYTorus. About 75 percent of CT AGNs in our sample had probabilities > 90 percent. From the spectral analysis, we significantly found an anticorrelation between the equivalent width of the neutral Fe K_{\alpha} line and the X-ray luminosity at 2-10 keV, the so-called X-ray Baldwin effect.