The demographics, obscuration, and accretion properties of X-ray AGN

Abstract

Most massive galaxies host a Supermassive Black Hole (SMBH) at their centre. SMBHs primarily grow by accreting surrounding material, releasing significant amounts of energy as radiation across all wavelengths. SMBHs experiencing such accretion phases are classified as Active Galactic Nuclei (AGN). AGN are believed to play an important role in the formation and evolution of their host galaxies as suggested by well-established correlations between the mass of the SMBH and properties of the host galaxy bulge. Therefore, building a complete census of AGN, which serve as tracers of SMBH growth, is crucial for understanding the mechanisms behind the AGN and galaxies parallel evolution.
However, most AGN are hidden behind clouds of gas and dust that absorb their emission, especially in the UV/optical regime. This obscuration introduces biases in our observations, leading to an underestimation of the intrinsic obscured population. X-ray photons, on the other hand, are less affected by obscuration than UV/optical photons, thus providing a less biased AGN sample. Moreover, the obscuration, characterised by the line-of-sight hydrogen column density N_H, imprints its signature on X-ray spectra. Analysing these spectra provides constraints on both N_H and the intrinsic X-ray luminosity Lx. With both parameters, it is possible to quantify the detection probability of a source as a function of its properties and, therefore, statistically infer the intrinsic AGN demographics from an observed sample.
In this thesis, I aim to quantify the impact of obscuration on AGN demographics and properties. For that purpose, I develop a new methodology that combines multi-wavelength information with X-ray spectral analysis to break down the degeneracy between obscuration and intrinsic X-ray luminosity measurements and, thereby, improving the constraints on the AGN physical properties. This approach involves fitting the mid-to-far Infrared (IR) Spectral Energy Distribution (SED) of a source to separate the AGN IR emission at 6 µm from the host galaxy IR contribution. The expected intrinsic X-ray luminosity is then computed using an empirical Lx −L_6µm relationship. The uncertainties from the SED fitting are convolved with the scatter of the Lx −L_6µm relationship to build a Lx normalisation prior for the X-ray spectral analysis. This approach yields robust measurements of the column density N_H, especially for the most obscured and Compton-Thick AGN (CTK, N_H > 10^24 cm^−2 ) with low photon counts X-ray spectra. I apply this methodology to a sample of 1744 AGN detected in the hard X-ray band (2-7 keV) within the Chandra COSMOS-Legacy field. Thanks to the well-defined X-ray selection function in this field, I measure the AGN space density as a function of Lx, N_H and z in a non-parametric way. The uncertainties of the different measurements are consistently combined within a fully Bayesian framework, yielding reliable results in agreement with previous studies. Additionally, this multi-wavelength guided X-ray spectroscopy approach significantly reduces the number of observed CTK AGN in the sample compared to simple X-ray spectroscopy. Similarly to the AGN space density analysis, I carefully account for survey sensitivity to measure the intrinsic fraction of CTK AGN, finding it to be 21.0 +16.1−9.9 % at z < 0.5. At higher redshifts, I established 3-σ upper limits for the AGN CTK fraction of ≲ 40%. These values are at the low end of previous estimations, suggesting a lower significance of the CTK AGN population than previously thought.
Furthermore, I present the analysis of two additional Chandra deep extragalactic X-ray fields, CDFS and AEGIS, for a total of 3882 X-ray-selected AGN. The focus of this work is the comparison of the accretion properties of obscured and unobscured AGN. The Eddington ratio λ Edd ∝ L_bol /M_SMBH is a fundamental characteristic of AGN, quantifying how fast an AGN is accreting relative to its maximum rate, beyond which radiation pressure exceeds gravitational pull and expels the accreting material. Outside of the local Universe, the masses of SMBHs M_SMBH are determined mainly from the broadening of their optical spectral lines, however, these lines are suppressed in obscured AGN. Therefore, I use the specific accretion rate (SAR) λ ∝ Lx /M⋆ as a proxy for the Eddington ratio. Indeed, the stellar mass of the host galaxy M⋆ can be measured by optical-to-IR SED fitting and is proportional to M_SMBH. By combining M⋆ , Lx and N_H measurements and their uncertainties in a Bayesian non-parametric approach, I measure the SAR distribution (SARD) for obscured and unobscured AGN for the first time up to z ∼ 3. The SARD represents the probability of a galaxy with mass M⋆ at redshift z hosting an AGN with a given SAR λ and obscuration N_H.
My results show that the SARD shapes of both unobscured and obscured AGN are broadly similar, as expected from the orientation unification model, and shift towards higher accretion rates with increasing redshift, indicating AGN ”downsizing”. However, the SARDs of unobscured AGN display a systematic offset towards higher accretion rates compared to the obscured population, resulting in a sharp decrease in the obscured AGN fraction at log λ break ∼−2 for z < 0.5, in agreement with local Universe observations. Nevertheless, as redshift increases, λ break shifts to higher values, and the AGN density in the unstable blow-out region of the λ−N_H plane increases. These results provide evidence for a radiation-regulated obscuration model coupled with increasing obscuration from the Interstellar Medium (ISM) of the host galaxy.
Overall, this thesis highlights the power of inference of X-ray analysis when coupled with multi-wavelength information to study the intrinsic demographics and properties of AGN across redshifts as a function of their obscuration. This work paves the way for future studies to unveil the nature of AGN obscurer and establish the AGN phase in the galaxy-SMBH evolutionary picture.