The evolution of the Compton thick fraction and the nature of obscuration for AGN in the Chandra Deep Field South

TL;DR

This study analyzes high-redshift AGN in the Chandra Deep Field South using advanced X-ray spectral models to investigate the evolution of the Compton thick fraction and the obscuration geometry of AGN over cosmic time.

Contribution

It applies new spectral models to high-z AGN data, revealing the evolution of the Compton thick fraction and obscuration geometry, and reports 29 new Compton-thick AGN detections.

Findings

The covering factor correlates with NH, indicating more obscured AGN are more geometrically buried.

Approximately 20% of sources are likely completely buried with near 4 pi coverage.

The intrinsic Compton thick fraction increases from about 20% locally to 40% at higher redshifts.

Abstract

(Abridged) We present the results from the X-ray spectral analysis of high-z AGN in the CDFS, making use of the new 4Ms data set and new X-ray spectral models from Brightman & Nandra, which account for Compton scattering and the geometry of the circumnuclear material. Our goals are to ascertain to what extent the torus paradigm of local AGN is applicable at earlier epochs and to evaluate the evolution of the Compton thick fraction (f_CT) with z, important for XRB synthesis models and understanding the accretion history of the universe. In addition to the torus models, we measure the fraction of scattered nuclear light, f_scatt known to be dependant on covering factor of the circumnuclear materal, and use this to aid in our understanding of its geometry. We find that the covering factor of the circumnuclear material is correlated with NH, and as such the most heavily obscured AGN are in fact also the most geometrically buried. We come to these conclusions from the result that f_scatt decreases as NH increases and from the prevalence of the torus model with the smallest opening angle as best fit model in the fits to the most obscured AGN. We find that a significant fraction of sources (~ 20%) in the CDFS are likely to be buried in material with close to 4 pi coverage having been best fit by the torus model with a 0\degree opening angle. Furthermore, we find 41 CTAGN in the CDFS using the new torus models, 29 of which we report here for the first time. We bin our sample by z in order to investigate the evolution of f_CT. Once we have accounted for biases and incompleteness we find a significant increase in the intrinsic f_CT, normalised to LX= 10^43.5 erg/s, from \approx 20% in the local universe to \approx 40% at z=1-4.