A multifrequency view of Active Galactic Nuclei

Author: Ignacio Ordovás Pascual | Supervisors: Silvia Mateos, Francisco J. Carrera

PhD Thesis PDF | Link to UCrea repository

Introduction

Active Galactic Nuclei (AGN) present emission through the entire electromagnetic spectrum characterized by a very high and persistent luminosity, a non-thermal Spectral Energy Distributions (SED) and variability in time scales from years to minutes, the latter indicating that this emission takes place in very compact regions. These characteristics indicate that the origin of this emission is due to the accretion of material to a supermassive black hole (SMBH) whose mass ranges between 10^6 and 10^9 solar masses and is located in the nucleus of the active galaxy. Mainly, in the UV/optical, AGN are divided into type-1 sources, that have broad emission lines (with full width at half maximum FWHM > 1500 km/s), and type-2 AGN, with-out detected broad emission lines. These differences are explained with what is known as the Unified Model of AGN. This model, in its simplest version, interprets these differences as orientation effects. The way to explain this is with the existence of a toroidal structure of dust and gas surrounding the SMBH: the torus. Depending on the inclination of the torus equatorial plane with respect to our line-of-sight, if the innermost regions of the AGN (as the Broad Line Region; BLR) intercepts the torus, it is obscured and therefore we classify the AGN as type-2. If the line-of-sight does not intercept the torus, the AGN emission is unobscured and therefore we classify the AGN as type-1. This model is able to explain the observed properties for most AGN. However, there is a significant fraction of sources whose properties are discordant at different frequencies of the electromagnetic spectrum, and therefore cannot be explained with the simplest version of the orientation-based Unified Model.

In this Thesis we aim to study the relationship between the extinction in the UV/optical and the absorption in X-rays of type-1 AGN. The characterization of the obscuration of the nuclear emission is an important aspect in the study of AGN since it is necessary to recover their intrinsic properties. It also allows us to understand the properties of the environment surrounding the SMBH. Today there are still many unresolved questions about obscuration in AGN and hence the validity of torus-based models as well. Studies that use samples as unbiased as possible are needed in order to bring light into this issue.

Planning and methodology

In this Thesis we have conducted a study focusing on type-1 AGN from the Bright Ultra-hard XMM-Newton Survey (BUXS). This is a survey composed by 255 AGN selected in the hard X-rays 4.5-10 keV). This selection allows to obtain a sample of type-1 AGN that reduces the strong bias against heavily absorbed AGN with hydrogen column densities of NH 10^23 cm^-2 (the NH is indicative of the absorption in X-rays). As in this Thesis we work only with type-1 AGN, the selection in hard X-rays should not introduce significant bias in the sample as objects with these high NH levels are not expected to present broad lines in the UV/optical. We have good quality XMM-Newton X-ray data for this sample, as well as UV/optical spectra from the Sloan Digital Sky Survey (SDSS) public survey and dedicated observations at several ground-based telescopes for their optical counterparts. We have high quality information for the UV/optical spectra, since the spectroscopic identification of the BUXS AGN is almost complete ($>$99 per cent of identified sources, with 172 type-1 and 80 type-2 AGN).

In this Thesis we conducted two different studies. In the first one we have two type-1.9 AGN, but whose X-ray spectra indicates that in the line-of-sight the absorption is low. According to the simplest version of the Unified Model of AGN, objects with high optical extinction should also have a high X-ray absorption, so they must be explained in some different way. To determine the origin of the mentioned discrepancy in these two objects, we carried out a detailed study dividing the AGN and host galaxy emission, so we could obtain the optical extinction and information about broad and narrow emission lines at UV/optical frequencies. We used VLT/X-Shooter spectra with high spectral resolution and wavelength coverage from the UV to the Near InfraRed. We also have high quality spectra in the X-rays obtained with the XMM-Newton observatory. We have investigated different scenarios to uncover the origin of the discordant optical and X-ray classifications.

In the second work, we have carried out a statistical study of a representative sample of 132 type-1 AGN. For this sample we have excluded all AGN with intrinsic luminosities in the 2-10 keV band with log( L(2-10 keV) / erg/s ) < 42, to increase the contrast between AGN and the host galaxy. In addition, we analyse only objects with $z<1$, to ensure that the\NH is measured with robustness. This is because with increasing $z$ we lose coverage of the X-ray emission at lower energies, which is where it is best to constrain the NH.

We analysed the optical and X-ray spectra to obtain the optical extinction (Av) and X-ray absorption (NH). In addition, we have obtained parameters such as the AGN intrinsic luminosity, the flux of certain emission lines and the gas velocity dispersion in the BLR. In the case of the optical spectra, after correcting from Galactic extinction, we separated the host galaxy starlight and the AGN emission. The models used for the host galaxy emission are selected using SED fits. The extinction model of the Small Magellanic Cloud of Gordon et al. 2003 is applied to the AGN emission, which is the most appropriate to reproduce the optical extinction of the nuclear emission of AGN. With this, we can isolate the nuclear emission and obtain the optical extinction. By fitting the wavelength regions with emission lines using a power law in addition to Gaussian profiles, we obtained the flux and width in velocity of the broad and narrow emission lines of the AGN.

With respect to the analysis of the spectra in X-rays, we model their emission by fitting different contributions (power law continuum, gas absorption, hot gas emission, etc.), obtaining the X-ray absorption and the intrinsic luminosity of the AGN. For the X-ray spectra of the two AGN with discordant classifications, the fits have been made in this Thesis, while the spectra of the second work have been fitted by members of our group, and so in this Thesis we have worked with the models already fitted. Combining all the available XMM-Newton observations for each object allows us to determine the X-ray absorption with great precision, since we are able to measure the NH up to NH = 10^19 cm^-2, whereas in previous studies normally set a minimum of NH = 10^20 cm^-2 when no absorption is detected. This is critical for our study since type-1 AGN generally present low X-ray absorption.

For the statistical study of type-1 AGN we obtained the individual probability distributions of \Av and \NH for each object, being the first study that uses these distributions for the analysis of the obscuration in AGN. This, along with the good quality of the available data, allows us to conduct a much more robust investigation than in previous studies. With the obtained parameters from each spectra at each frequency, we can better understand the relationship between the classification of AGN and their obscuration, and hence we can provide possible explanations for those objects whose optical and X-ray classification does not match. We also compared the extinction in the optical and the absorption in the X-rays to explore how these quantities are correlated in a complete sample.

Original contributions

The two AGN with discordant classification show a different origin of the discrepancy. In one case, the most likely explanation is that the medium that obscures the nuclear emission has more dust than gas compared to the Galactic dust-to-gas relation. This causes us to observe more extinction in the UV/optical than in X-rays. As for the second object, the dust-to-gas ratio is compatible with a relative composition such as the Galactic. We found that the discrepancy is due to the effects of dilution of the nuclear emission by stellar emission, since the relationship between the masses of the host galaxy and the black hole is much higher than expected compared to the typical relation in the literature. We have ruled out in both cases that they are AGN Compton-thick. We have also ruled out that the flux of the broad lines is intrinsically weak compared to the typical broad and narrow line fluxes ratios that AGN usually present.

The analysis of the representative sample of type-1 AGN allowed to obtain that the fraction of optically reddened and X-ray absorbed objects is significantly higher than in the sample that we have complete classification (those with z<0.2) that in the full sample of type-1 AGN (40-50 per cent versus 20 per cent). The fraction of objects for the complete sample that are absorbed type-1 AGN but are unreddened and viceversa is 32 per cent, that is ~2 times higher than the one in the literature. This indicates that, without working with complete samples, the obtained results are biased against highly obscured objects. With increasing z, we lose obscured sources as they are misclassified as type-2 AGN, as their broad emission lines are not covered in their UV/optical spectrum.

In our study we have obtained an increasing relation between subtype of AGN (from type-1.0 to type-1.9) and obscuration, both in the optical range and in X-rays. In the case of optical extinction, Av gradually rises with the subtype, while the X-ray absorption rises steeply from type-1.0/2/5 to type-1.8/9. Although there may be different scenarios that can contribute to the broad and narrow emission lines relative flux, that defines the optical classification, in our analysis we have obtained that extinction is the main contribution to the subclassification of AGN. We finally conclude that both Av and NH have a direct relation with the subtype of AGN.

We have compared the optical extinction estimation methods based on the Halfa and Hbeta broad emission line fluxes (Balmer decrement) with our estimates based on the analysis of the UV/optical continuum emitted by the accretion disk. While the mean intrinsic ratios we obtain are consistent with those used in the literature, the Av measurements are not very correlated. These results are in agreement with studies that support that the intrinsic broad Halfa and Hbeta emission line flux ratio can vary from one object to another depending on the physical conditions of the material where the lines are emitted. Therefore, the analysis performed in this Thesis indicates that Balmer decrement estimations of Av should be taken with caution.

As for the Av-NH ratio, which accounts for the relative amount of gas and dust, we have found that there is no clear relationship between optical extinction and X-ray absorption, so we cannot directly extrapolate a certain level of absorption in X-rays from the extinction in the optical and vice versa. For approximately half of the sample, those AGN with low \Av and NH, their dust-to-gas ratio is biased due to systematic errors. This implies that for an important part of the sample it is not possible to estimate the dust-to-gas ratio, even with high quality data. In previous studies the dust-to-gas is found to be compatible or below the Galactic ratio, while only a minority $3-9 per cent) has higher amounts of dust with respect to gas compared to the Galactic ratio. In our case we found a minimum of 47 per cent with higher dust-to-gas ratio objects than the Galactic for the complete sample of type-1. If only type-1.0/2/5 AGN are considered, as other studies do, we obtain a minimum of 32 per cent of objects with higher dust-to-gas ratios than the Galactic. This higher fraction of dust-to-gas ratios higher than the Galactic standard compared to previous studies might be due misclassification of type-1 objects, AGN selection methods that miss low reddened sources or because this analysis is not formally performed in these works.

Conclusions

In this thesis we have shown that the discrepancies between the optical and X-ray classifications observed for a significant fraction of AGN may be caused by diverse factors. For this reason, those AGN with an optical classification that indicates high optical extinction but whose X-ray spectrum has low X-ray absorption cannot be considered a physical family of objects. In the detailed study of the two sources we discussed several contributions as high dust-to-gas ratios or dilution of the nuclear emission by stellar emission that can explain AGN to have a low or not absorbed X-ray spectrum and UV/optical classification indicating high extinction.

In the type-1 AGN sample that we studied statistically, we obtained that type-1 samples without Halfa spectral coverage for all sources are incomplete for type-1.8/9. This means that these samples are biased against high Av and high NH. We have proven than the intrinsic broad Halfa and Hbeta flux ratio is very wide, affecting the Balmer decrement Av estimations. We have obtained that the amount of X-ray absorption and the optical extinction do not present a clear relationship, favouring models that assume that the bulk of the X-ray absorption is produced in a different structure than the optical extinction, or alternatively that dust-to-gas ratios can be very different from source to source.