Investigating the population of growing supermassive black holes using X-rays surveys combined with multi-wavelength information

Abstract

In this Thesis I investigate the population of Active Galactic Nuclei (AGN), which trace the growth of supermassive black holes, using a new X-ray survey combined with multi-wavelength imaging and spectroscopy. I use my samples to quantify the numbers of AGN in the extra-galactic sky and investigate their obscuration properties, presenting an updated view of the AGN population and shedding light on how it changes over cosmic time.

In Chapter 3 I present my work on creating the Extra-galactic Serendipitous Swift Survey (ExSeSS), providing a new well-defined sample constructed from the observations performed using the Swift X-ray Telescope. The ExSeSS sample consists of 79,342 sources detected in the medium (1 − 2 keV), hard (2 − 10 keV) or total (0.3 − 10 keV) energy bands, covering 2086.6 deg² of sky across a flux range of ƒ0.3−10keV ∼ 10⁻¹⁵ − 10⁻¹⁰ erg s⁻¹ cm⁻². Using this new ExSeSS sample, I present measurements of the differential number counts of X-ray sources as a function of 2- 10 keVflux that trace the population ofAGNin a previously unexplored regime. I obtain a good agreement between the ExSeSS measurements and previous, higher energy data from NuSTAR and Swift/BAT when taking into account the varying column density of the ExSeSS sample as well as the X-ray spectral parameters of each of the samples we are comparing to. I also find discrepancies between the ExSeSS measurements and AGN population synthesis models, indicating a change in the properties of the AGN population over this flux range that is not fully described by current models at these energies, hinting at a larger, moderately obscured population at low redshifts (z ≲ 0.2) that the models are not currently taking into account.

In Chapter 4 I present my work on creating a cross-matched sample of ExSeSS with optical and infrared counterparts within the Legacy Survey 8 (LS8): FoExSeSS. This multi-wavelength information enables the identification and removal of 3, 734 stars and allows me to create a large, clean sample of 59, 812 X-ray selected AGN, of which 47, 546 have reliable multi-wavelength counterparts. I find stellar contamination to be an insignificant contribution to my differential number count results; removing them from my sample does not alter the overall shape of the distribution. A major benefit of creating my cross-matched FoExSeSS sample is the identification of photometric redshifts for all my sources. In addition, I compile spectroscopic redshifts (primarily from SDSS and DESI) which are available for 15.1% of the ExSeSS-LS8 sources, enabling an assessment of the accuracy of the photometric redshifts and providing the most redshift measurements for these sources.

I present updated measurements of the differential number counts, categorising them by redshift, to test my results from Chapter 3. I find that FoExSeSS is in good agreement with the population synthesis models for low redshifts (z ≤ 1.0) but beyond z ≥ 1.0 the measurements deviate from the Gilli et al. (2007) model, but remain consistent with Ueda et al. (2014) until z ≥ 2.0, where FoExSeSS then diverges to having greater numbers than the models predict. This contradicts my initial prediction in Chapter 3, where I hypothesized it was a low redshift population causing the excess in the number counts at high X-ray fluxes in ExSeSS. My updated results in Chapter 4 lead me to conclude that the population synthesis models are currently not taking into account a population of moderate-to-high luminosity AGN at z = 1.0 − 3.0.

In Chapter 5 I present results from the follow-up optical spectroscopic survey of some of the brightest, hard X-ray selected sources in ExSeSS (as well as hard X-ray sources from the 4XMM sample) performed as part of SDSS-V. I compare this sample to the follow-up performed by SDSS-V on eROSITA sources to see if there are any differences in the optical spectroscopic properties of the populations of hard X-ray selected sources (2 − 10 keV for ExSeSS and 4XMM) and soft X-ray selected sources (0.6 − 2.3 keV for eROSITA).

The whole eROSITA sample consists of 64.5 ± 0.96% QSOs (6,107), 12.4 ± 0.66% GALAXYs (1,178) and 23.1±0.85% STARs (2186), based on the optical spectroscopic classifications. In contrast, the Swift/XMM sample is composed of 82.0 ± 2.40% QSOs (808), 14.7 ± 2.21% GALAXYs (145) and 3.2 ± 1.11% STARs (32). I also extract a sub-sample from the eROSITA sample that matches the X-ray flux and rmagnitude distributions of the Swift/XMM sample, which brings these numbers into closer agreement; 79.0 ± 1.82% QSOs (1545), 12.9 ± 1.47% GALAXYs (251) and 8.02 ± 1.25% STARs (174). However, the eROSITA sub sample still has ∼ 5% more stars than the Swift/XMM sample. These findings suggest that the hard selection criteria do not significantly alter the observed population characteristics. I also create composite spectra for two of the aforementioned classes (QSOs and GALAXYs) for both the Swift/XMM and eROSITA samples. QSOs in the Swift/XMM sample (with 𝛼� = 2.46 ± 0.004 for wavelengths less than 3900Å) have slightly, but significantly, bluer spectra than those in the matched eROSITA sample (with 𝛼� = 2.37 ± 0.005 for wavelengths less than 3900Å). In contrast, the composite GALAXY spectra created from the two individual samples were identical.

In creating this new, unique sample of X-ray identified AGN, with matching optical and IR photometric/spectroscopic data, I have produced updated constraints on the number counts of the AGN population over cosmic time and investigated the optical spectroscopic properties of hard X-ray selected sources in a previously unexplored parameter space. The analysis and results that I present in this thesis indicate the need to update current models of the AGN population.