• Aucun résultat trouvé

Study of the high-energy emission from AGN and its correlation to the other wavelengths

N/A
N/A
Protected

Academic year: 2022

Partager "Study of the high-energy emission from AGN and its correlation to the other wavelengths"

Copied!
224
0
0

Texte intégral

(1)

Thesis

Reference

Study of the high-energy emission from AGN and its correlation to the other wavelengths

SOLDI, Simona

Abstract

Plusieurs études des noyaux actifs de galaxies (AGN) avec le satellite INTEGRAL ont permis de caractériser les paramètres de l'émission de Comptonisation responsable du rayonnement X dur dans les galaxies de Seyfert et de calculer pour la première fois la fonction de luminosité d'AGN au-dessus de 20 keV. Le lien entre la variabilité de l'émission à haute énergie et celle à différentes longueurs d'onde a été étudié pour le quasar 3C273, après la mise à jour de la base de données incluant maintenant 40 ans de données du domaine radio jusqu'aux rayons gamma. Cette étude a permis de découvrir: une autre preuve de la présence de poussières dans l'émission infrarouge de 3C273, des propriétés temporelles très différentes entre l'émission X en dessous et au-dessus de 20 keV et la probable origine des rayons X durs par émission Compton inverse des mêmes électrons qui produisent l'émission optique par rayonnement synchrotron.

SOLDI, Simona. Study of the high-energy emission from AGN and its correlation to the other wavelengths. Thèse de doctorat : Univ. Genève, 2008, no. 4017

URN : urn:nbn:ch:unige-7007

DOI : 10.13097/archive-ouverte/unige:700

Available at:

http://archive-ouverte.unige.ch/unige:700

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

UNIVERSITÉ DE GENÈVE FACULTÉ DES SCIENCES

Département d’astronomie Professeur T. J.-L. Courvoisier

Study of the High-Energy Emission from AGN and its Correlation to the Other Wavelengths

THÈSE

présentée à la Faculté des sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences,

mention astronomie et astrophysique

par

Simona SOLDI

de

Magenta (Milan - Italie)

Thèse No4017

GENÈVE

Atelier de reproduction ReproMail 2008

(3)

CETTE THÈSE A FAIT L’OBJET DES PUBLICATIONS SUIVANTES:

voir Appendice A.

(4)

[...] salimmo sù, el primo e io secondo, tanto ch’i’ vidi de le cose belle

che porta ’l ciel, per un pertugio tondo.

E quindi uscimmo a riveder le stelle.

DANTE ALIGHIERI, La Divina Commedia

(5)
(6)

iii

Résumé en français

Le travail présenté dans cette thèse a été entièrement effectué au sein de l’ISDC, centre chargé de collecter les données scientifiques du satellite INTEGRAL. Lancé le 17 octobre 2002, INTEGRAL est une mission de l’Agence Spatiale Européenne qui a pour but l’étude des sources émettant des rayons X et gamma entre 5 keV et 10 MeV, au moyen de 4 instruments: un imageur en rayons X durs et gamma (IBIS), un spectromètre (SPI), un double moniteur en rayons X (JEM-X), et un moniteur optique (OMC).

Le sujet principal de cette thèse est l’étude des noyaux actifs de galaxies (AGN) dans les domaines X et gamma. Bien que la dénomination d’“AGN” regroupe des objets aux propriétés en apparence très différentes, différents modèles permettent de les décrire de façon unifiée. Les modèles les plus adaptés sont basés sur un trou noir super-massif qui accrète la matière environnante, et qui possède parfois des jets s’étendant sur des distances du pc au Mpc. De plus, l’émission X et gamma provenant de ces sources est produite par des processus différents selon le type d’AGN considéré. Par exemple, dans les AGN radio-silencieux (tels que les galaxies de Seyfert), les rayons X sont probablement le résultat d’un processus de Comptonisation thermique entre les électrons situés dans une couronne au-dessus du disque d’accrétion et les photons émis par le disque lui-même. Par contre, dans les AGN radio-émetteurs, c’est la radiation due au jet qui domine l’émission. En particulier, dans les blazars (c’est-à-dire des AGN pour lesquels le jet est orienté vers la Terre), les rayons X peuvent être produits par rayonnement synchrotron, ou par effet Compton inverse d’une population d’électrons relativistes accélérés dans le jet. L’émission X provient donc des régions les plus centrales des AGN, et représente une grande partie de la puissance totale émise par ces objets. Par conséquent, l’étude de l’émission X constitue une étape importante pour la compréhension de la physique des AGN.

Au cours de mon doctorat, j’ai examiné différents aspects – à la fois spectraux et temporels – de l’émission X et gamma des AGN. Au moyen des données d’INTEGRAL, j’ai étudié les propriétés spectrales de plusieurs galaxies de Seyfert et blazars, et j’ai analysé la contribution des AGN au bruit de fond cosmique dans le domaine des rayons X durs. Une autre partie de mon travail de doctorat concerne l’étude de la variabilité des AGN. J’ai analysé la variabilité du quasar 3C273 dans differents domaines d’énergie en utilisant une base de données qui inclut plus de 40 années d’observations. Je décris aussi les propriétés temporelles dans le domaine des rayons X durs d’un échantillon d’AGN que nous avons étudié grâce au satellite Swift.

Les résultats obtenus par INTEGRAL pour les galaxies de Seyfert sont en accord avec ceux des missions précédentes. Pour certaines sources, l’empreinte du processus de Comptonisation se mar- que dans le spectre en rayons X par une coupure exponentielle à & 50 keV. Pour l’AGN le plus brillant dans le domaine d’énergie étudié (la galaxie de Seyfert NGC 4151), nous avons pu contrain- dre différents paramètres physiques du processus de Comptonisation, notamment la température des électrons, l’épaisseur optique de la couronne, et la fraction d’émission reflétée par de la matière froide.

J’ai aussi eu l’opportunité de collaborer avec des collègues organisant des campagnes multi- longueurs d’onde pour l’étude des blazars pendant les phases d’émission très intense. En 2004 et 2005, INTEGRAL a participé à ces campagnes en observant les blazars S 0716+714 et 3C 454.3.

Dans le cas de 3C 454.3, par exemple, nous avons pu extraire le spectre en rayons X à partir des données JEM-X et IBIS, ce qui a permis – en parallèle avec les observations obtenues dans d’autres domaines d’énergie – de montrer que l’émission à haute énergie est produite par le refroidissement des électrons du jet par effet Compton inverse sur les photons produits par synchrotron, et non sur les photons produits à l’extérieur du jet.

(7)

La compréhension du bruit de fond cosmique dans le domaine des rayons X dépend fortement de la forme du spectre des AGN dans ce même domaine d’énergie, car ces derniers sont vraisemblablement à l’origine de cette émission. En particulier, il est probable qu’une fraction importante des AGN soient obscurcies à basse énergie (E <10 keV) à cause de l’absorption intrinsèque (NH>1024cm−2), mais fournissent une contribution importante à plus haute énergie (E ≃ 30 keV), là où le bruit de fond cosmique atteint son maximum. Le large champ de vue et la résolution angulaire d’IBIS ont fourni la première possibilité d’étudier les AGN au-dessus de 20 keV dans le ciel entier. Pour ce faire, nous avons utilisé les données IBIS des 15 premiers mois de la mission, ce qui a permis de détecter 71 AGN. En excluant les quelques blazars détectés, cet échantillon peut alors être considéré comme local et représente la partie la plus brillante de la fonction de luminosité des AGN. Ces AGN brillants sont responsables d’1% du bruit de fond cosmique en rayons X. A haute énergie (E > 10 keV), le rapport des sources ayant une absorption faible (NH > 1022cm−2) et celui des sources ayant une absorption élevée (NH > 1024cm2) aux sources n’ayant aucune absorption (NH < 1022cm2) sont plus grands qu’à plus basse énergie (E < 10 keV), et semblent être en accord avec les rapports prédits par les modèles les plus récents du bruit de fond cosmique en rayons X.

Afin d’étudier aussi les propriétés temporelles des AGN, j’ai mené une analyse de l’émission multi-longueurs d’onde et de la variabilité du quasar radio-émetteur 3C 273. Cet objet, qui émet du domaine radio jusqu’aux rayons gamma, est fréquemment observé. Nous avons mis à jour la base des données disponibles sur 3C 273, qui avait été créée et publiée en 1999, et qui inclut maintenant plus de 40 années d’observations, dont en particulier un intense suivi dans les rayons X au cours de ces 10 dernières années. Cette étude de variabilité nous a permis de découvrir une autre preuve de la présence au sein de 3C 273 de poussières émettant dans le domaine infrarouge, grâce à la mise en évidence d’une corrélation entre la radiation chauffante (dans les domaines optique et ultraviolet) et l’émission dans le domaine infrarouge proche. Suite à cette analyse de corrélation de l’émission due à différents domaines spectraux, nous avons pu conclure que ce sont probablement les mêmes électrons produisant l’émission optique lentement variable par rayonnement synchrotron qui produisent aussi les rayons X durs par émission Compton inverse. De plus, l’émission en rayons X présente des caractéristiques temporelles différentes en dessous et au-dessus de 20 keV, qui pourraient s’expliquer soit par 2 composantes d’émission séparées, soit par une seule composante décrite par au moins 2 paramètres indépendants. Cette question est l’une des nombreuses pour lesquelles des analyses futures seront nécessaires afin d’apporter une réponse définitive.

En parallèle, nous avons mené une étude préliminaire des 9 premiers mois des données récoltées par l’instrument BAT à bord du satellite Swift. Le but était de approfondir nos connaissances des propriétés temporelles de l’émission X dure des AGN. Le résultat le plus important de ce travail a été de montrer que les sources les plus absorbées se sont révélées être également les plus variables.

Cette tendance s’explique probablement par le fait que les sources les plus absorbées sont aussi les plus lumineuses; or la variabilité dans les rayons X durs est plus importante pour les sources les plus lumineuses.

Enfin, travailler à l’ISDC signifie aussi participer à diverses tâches d’intérêt commun, telles que les tests du programme d’analyse des observations INTEGRAL, et le contrôle des données scientifiques en temps (quasi-)réel en tant que “scientifique de garde” durant une semaine tous les trois mois. Ceci m’a donné l’occasion de découvrir l’émission X et gamma de sources qui ne sont pas le sujet de cette thèse (par exemple les binaires X), et d’établir des collaborations intéressantes qui ont aussi donné lieu à des publications.

(8)

Contents

I Introduction 1

1 Introduction 3

2 Active Galactic Nuclei 5

2.1 The standard picture . . . 5

2.2 The unified models . . . 9

2.3 The X-ray emission from AGN . . . 13

2.3.1 X-ray spectra of Seyfert 1 . . . 13

2.3.2 X-ray spectra of Seyfert 2 . . . 15

2.3.3 X-ray spectra of radio loud AGN . . . 16

2.4 The hard X-ray emission of AGN . . . 17

3 The high-energy missions 19 3.1 The INTEGRAL mission . . . . 19

3.1.1 Coded mask technique . . . 20

3.1.2 The spacecraft . . . 21

3.1.3 The INTEGRAL ground segment . . . . 24

3.1.4 INTEGRAL observation programme and strategy . . . . 25

3.1.5 The Offline Science Analysis software . . . 26

3.2 Other X-ray missions relevant to this thesis . . . 27

3.2.1 The BeppoSAX satellite . . . . 27

3.2.2 The RXTE satellite . . . . 27

3.2.3 The XMM-Newton satellite . . . . 28

3.2.4 The Swift satellite . . . . 28

II Hard X-ray emission from AGN as seen by INTEGRAL 29 4 Hard X-ray spectral properties of AGN 31 4.1 Early studies of AGN with INTEGRAL . . . . 31

4.2 Six AGN in the Galactic plane . . . 32

v

(9)

4.3 The bright Seyfert galaxy NGC 4151 . . . 35

4.4 INTEGRAL observations of blazars . . . . 37

4.4.1 Multiwavelength campaigns of blazars in outburst . . . 37

4.4.2 INTEGRAL and XMM-Newton observations of 3C 273 . . . . 39

4.5 Recent INTEGRAL studies of AGN . . . . 40

4.6 Future perspectives . . . 41

4.7 Scientific article on a sample of AGN in the Galactic plane . . . 43

5 Population studies of AGN in the hard X-rays 55 5.1 The cosmic X-ray background . . . 55

5.1.1 The hard X-ray background . . . 56

5.1.2 The obscured AGN and the AGN synthesis models . . . 57

5.1.3 The AGN spectral parameters . . . 58

5.1.4 The role of INTEGRAL . . . . 59

5.2 The first INTEGRAL AGN catalogue . . . . 59

5.2.1 Average properties of hard X-ray AGN . . . 59

5.2.2 After the first INTEGRAL AGN catalogue . . . . 62

5.2.3 Scientific article on the INTEGRAL AGN catalogue . . . . 63

5.3 The hard X-ray luminosity function of AGN . . . 75

5.3.1 Comparison with other recent hard X-ray surveys . . . 78

5.3.2 The synthesis of the CXB . . . 79

5.3.3 Future perspectives . . . 80

5.3.4 Scientific article on the luminosity function of INTEGRAL AGN . . . . 82

III AGN variability 93 6 X-ray variability of AGN 95 6.1 Medium X-ray variability . . . 95

6.2 Hard X-ray variability . . . 97

6.3 Correlation with other wavelengths . . . 98

7 The quasar 3C 273 101 7.1 The broad-band emission of 3C 273 . . . 102

7.1.1 The jet emission . . . 102

7.1.2 The IR emission . . . 103

7.1.3 The blue bump . . . 104

7.1.4 The high-energy emission . . . 105

8 The 3C 273 on-line database 107

(10)

CONTENTS vii

8.1 The new data . . . 107

8.1.1 Optical data . . . 108

8.1.2 X-ray data . . . 110

8.2 The new format . . . 115

9 The multi-wavelength variability of 3C 273 117 9.1 The amplitude of the variations . . . 117

9.1.1 Results on the variability amplitude . . . 120

9.2 The characteristic time scales of the variations . . . 124

9.2.1 The auto-correlation function analysis . . . 124

9.2.2 The structure function analysis . . . 128

9.2.3 Results on the characteristic time scales of the variations . . . 131

9.3 Cross-correlations and time delays . . . 136

9.3.1 Some of the previous works on 3C 273 correlations . . . 136

9.3.2 Our correlation study . . . 138

9.4 The X-ray spectral variability . . . 141

9.4.1 Photon index evolution . . . 141

9.4.2 The soft-excess . . . 143

9.4.3 The broad line at 6.4 keV . . . 143

9.5 The global picture . . . 145

9.6 Scientific article on 3C 273 variability . . . 146

10 Hard X-ray variability of AGN 163 10.1 A study with Swift/BAT . . . 163

IV Other scientific contributions 167 11 Scientist on duty 169 11.1 Data monitoring . . . 169

11.2 The announcements . . . 170

11.3 Science out of the barn . . . 171

V Conclusions 175

12 Conclusions 177

13 Acknowledgments 193

(11)

VI Appendix 197

A Publications 199

B Glossary 201

(12)

Part I

Introduction

1

(13)
(14)

Chapter 1

Introduction

A crucial discovery for extragalactic astronomy was made in 1963 by Schmidt (1963) and Greenstein

& Matthews (1963): the optical identification of the radio sources 3C 273 and 3C 48 and the mea- surement of their redshifts proved the existence of a new class of sources, apparently stellar-like but being at least as luminous as the known galaxies. As it became clear over the following years, the strong central activity of Active Galactic Nuclei (AGN) can show many different manifestations that are classified under various names (quasars, Seyfert galaxies, BL Lac, etc.), though originating from the same process. The large power emitted by AGN (Lbol & 1042erg s−1) is thought to be produced by accretion onto a supermassive black hole (MBH≈106−1010M).

The emission recorded from AGN extends in some cases from radio to gamma-rays, covering the whole electromagnetic spectrum. However, astronomy had to wait for the development of the appropriate technology in order to open the window to observations at high energies up to several hundred GeV, where measurements from Earth are prevented due to the atmospheric absorption.

Thanks to balloon and rocket experiments in the 1960s, the quasar 3C 273 was the first extragalactic object to be detected in the X-rays in 1969 (Bowyer et al. 1970). The first X-ray satellite, UHURU, launched in 1970, has been followed by many X- and gamma-ray missions that have flown to space providing us with many answers and more questions about the physics of AGN.

Not only a large part of the power emitted by AGN is radiated in the X-ray band but this emis- sion also maps the innermost regions of these sources, as suggested by the fast variations observed in this energy range. All this makes the study of X-rays from AGN a fundamental step towards the understanding of the AGN phenomenon. Looking in particular at the hard X-ray emission, much progress has been made in the past years thanks to observations with the EXOSAT, Ginga, BeppoSAX and CGRO satellites, allowing the identification of the main spectral and timing properties of AGN above 20 keV. However, many questions are still open and wait for an answer. For example, it is necessary to better characterise the shape of the continuum emission, to constrain the position of the high-energy cut-off and quantify the amount of Compton reflection and the location and geometry of the reflector. The dependence of these properties on the AGN type, on the intrinsic absorption, or on the accretion rate is an important test for the unified models, according to which these differences are fully explained by different viewing angles, and for the understanding of the cosmic X-ray back- ground and its synthesis around the 30 keV peak. The amplitude and the time scales characteristic for the variations in the hard X-rays and their relation with the emission at other wavelengths represent an important tool to study the underlying emission process and the geometrical and physical properties of the emitting region. These open issues are currently addressed by a few high-energy satellites that have been launched shortly before or during the period of my PhD: INTEGRAL, Swift and Suzaku.

3

(15)

My PhD work has concentrated on the study of the hard X-ray emission from AGN. In the first part of this thesis, an introduction to AGN is given, presenting the general characteristics of these sources and focussing on their X-ray spectral properties. Then the INTEGRAL mission is presented in detail as a large part of this work is based on data collected by its hard X-ray imager IBIS. A brief summary of other missions relevant for this work is also provided.

There are different ways to approach the study of AGN hard X-ray emission, looking into the spectral or temporal aspects, using single sources or samples of objects, limiting the investigation to the X-ray band or analysing the relation with other wavelengths. I had the great chance to walk along many of these paths during my PhD and the results of this journey are presented in the second and third part of this thesis.

The studies centered on INTEGRAL data are described in the second part. First, I present a study of few bright Seyfert galaxies (among them the Circinus galaxy and NGC 4151) plus Centaurus A, with the aim of investigating the signatures of thermal Comptonisation, e.g. a high-energy cut-off and reflection, in their spectra. In radio loud AGN, the hard X-ray emission is instead thought to be related to the jet and to be due to either synchrotron or inverse Compton (or both) processes, depending on the class type. The INTEGRAL observations of two blazars in outburst and of the radio loud quasar 3C 273, together with the measurements at other wavelengths, are described and discussed.

Thanks to the large field of view of INTEGRAL, many AGN have been serendipitously detected and it has been possible to collect the first INTEGRAL AGN catalogue and to study the global spectral properties of an hard X-ray selected sample of AGN. The energy domain above 20 keV provides us with an unbiased view of the AGN population with intrinsic absorption≤1025cm−2. The distribution of the absorption and the dependence of AGN spectral properties on this parameter are important questions to be addressed for our understanding of the cosmic X-ray background (CXB) and its origin.

The contribution of INTEGRAL AGN to the CXB has been studied with this sample and this work is concluding the second part of my thesis.

The third part is dedicated to the study of variability in AGN. An update of the on-line multiwave- length database of 3C 273, first created in 1999, was needed, considering the large amount of X-ray data that have become available in the last years. We have worked to this update and taken advantage of this collection of data covering 40 years of observations to study the multiwavelength variability of 3C 273, with special emphasis on the X-ray properties and their connection to the emission in other bands. I include in the third part also a work performed on a sample of AGN detected in first 9 months of operations by the hard X-ray detector BAT on board Swift. The large field of view of BAT and the observation strategy of Swift make BAT the most suitable instrument to study the variability of hard X-ray sources.

The work presented in this thesis has been completely carried out at the former INTEGRAL Science Data Centre, now evolved into ISDC Data Centre for Astrophysics. Being at the data centre of a satellite is a very enriching experience, especially for a young scientist. First it gives you the possibility to widen your scientific horizons, learning about what is going on in the hard X-ray sky in near-real time and beyond the emission from your favourite objects. This is mainly possible thanks to the experience as Scientist on Duty described in the fourth part of this thesis. Then, you have the opportunity to build up collaborations thanks to the discovery of interesting events during your duty time or because of the knowledge of the data analysis software you acquired in the place where the software is partly developed and completely integrated and tested. Finally, you have the great chance to work every day in an international environment, which helps to make you grow scientifically and personally.

(16)

Chapter 2

Active Galactic Nuclei

Since this thesis discusses mainly results on the high-energy emission of Active Galactic Nuclei (AGN), a brief introduction to these objects is presented in this Chapter, which makes no pretension to being complete. After describing the main observational properties of the different AGN types, the standard picture (which explains the similar characteristics of different classes) and the unified models (which try to explain the differences between different classes) are presented. Then the X-ray emission from AGN is described in more detail, concentrating on the spectral aspects and on the pre- INTEGRAL knowledge of the hard X-ray properties of AGN. The X-ray timing properties of AGN are discussed in Chapter 6.

2.1 The standard picture

The AGN zoo

The term Active Galactic Nucleus (AGN) is generically used to indicate a large variety of objects, as for example Seyfert galaxies, blazars, quasars, Quasi-Stellar Objects (QSO) and Low-Ionisation Narrow Emission-line Regions (LINER). From an observational point of view, they are apparently point-like sources characterised by a non-stellar spectrum and they are at cosmological distances. A large fraction of them show strong variability and extend their emission along the whole electromag- netic spectrum. Besides these general characteristics, one can observe a large variety of properties at the origin of the different AGN classifications.

Seyfert galaxies were the first AGN to be discovered in the 1940s as bright, point-like nuclei in some spiral galaxies (Seyfert 1943). Different subclasses are identified based on the width of the strong emission lines present in their optical spectra. Seyfert 1 objects show a strong continuum from IR to X-rays with some emission lines broadened up to few thousand km s−1 of full width at half maximum and other narrow lines. Seyfert 2 present only narrow lines and have a weaker continuum emission. Objects with intermediate properties between the Seyfert 1 and 2 types have been classified as types 1.5, 1.8, etc. For example, in Seyfert 1.5 the broad and narrow components of the Hβhave similar intensities, in Seyfert 1.8 the broad components of Hαand Hβare still detectable even though very weak, and in Seyfert 1.9 only the Hαbroad component can be detected.

To a first approximation, quasars can be considered as the more luminous equivalent of Seyfert galaxies. In quasars, though, the host galaxy is often much fainter than the active nucleus and there- fore it is difficult to resolve. About 10% of the quasars show strong radio emission, which can be

5

(17)

compact or show the double-lobe structure typical of radio galaxies.

The blazar class includes BL Lacertae objects and Optically Violent Variable (OVV) quasars.

They are characterised by a non-thermal, highly variable, continuum emission that is believed to be relativistic beamed, and they show radio emission. BL Lac objects have lower luminosities and much weaker, if any, emission lines than OVV. Among the radio quasars, a further distinction is done between flat-spectrum radio quasars (FSRQ) and steep-spectrum radio quasars (SSRQ) depending on the value of the spectral index in the radio band. This classification reflects the different extension of the radio emitting region in the two classes: a flat spectrum indicates a self-absorbed synchrotron emission from a compact region, whereas a steeper spectrum reflects the dominance of the emission from the radio lobes over the nuclear emission.

Low-Ionisation Narrow Emission-line Regions (LINER) are the least luminous and most common AGN, being observed in about 40% of the nearby galaxies (Heckman 1980; Ho et al. 1997). They are characterised by lines with low ionisation and as narrow as the narrow-lines in Seyfert galaxies, but some of them show also weak broad lines (e.g. Eracleous & Halpern 2001).

Many other classes and subclasses of AGN have been identified, but the general properties of all these objects can be described by a common scenario that involves the presence of a supermassive black hole with mass of MBH≈(106−1010) M, accreting matter from its surroundings and producing bolometric luminosities L&1042erg s−1.

The central black hole

The presence of a black hole in the centre of AGN is the most plausible possibility to explain several observational facts. Variations on very short time scales (below one hour) have been observed in some AGN and can be used to set an upper limit to the size of the central source. In fact, there cannot be substantial changes in a system of size l on time scales shorter than t = l/c, where c is the speed of light. This implies l < 1014cm for tvar < 1 hour. It is necessary though to take into account that, if the source is moving at relativistic speed (as suggested by the apparent superluminal motion observed in some AGN), then the measured variability time scale is shorter than that in the rest frame of the source. This would imply that the size of the most rapidly variable sources is underestimated with this simple calculation.

On the other hand, the high luminosities measured in AGN indicate that accretion rather than nuclear burning is the source of all this energy. In fact, the efficiency of energy conversion (in terms of rest energy) for fusion of hydrogen into helium (η∼0.7%) is more than an order of magnitude less than that available from accretion into a compact object (∼ 10%, e.g. Frank et al. 1992). Therefore, for an object with luminosity L=1047erg s−1, the rate at which the mass is processed in the source:

M˙ = L

c2η (2.1)

results to be 250 My−1 for nuclear burning. A much lower and more plausible mass rate of 18 My−1is needed for accretion. When one assumes that accretion is the most likely process to power AGN, a simple argument can give a rough estimate of the mass of the central source based on the Eddington luminosity. The Eddington luminosity is defined as the maximum luminosity achievable through spherical accretion onto a mass M and it corresponds to the equilibrium condition between gravitation and radiation pressure on electrons:

LEdd = 4πGM mpc

σT ≃1.3×1046 M

108M

!

erg s−1 , (2.2)

(18)

The standard picture 7 where mp is the proton mass and σT is the Thomson cross-section. When considering a luminosity L=1047erg s−1, a source with mass M=8×108Mis required.

Therefore, the small sizes obtained with constraints imposed by variability and the high masses deduced from the observed luminosities indicate the compact nature of the AGN central source. Even though it had been proposed that a dense cluster of massive stars or a single star with M = 108M

could be the source of energy of an AGN, these hypotheses were abandoned mainly because the first system would evolve too fast and the second would not be stable. On the contrary, a black hole is a stable compact source whose existence is guaranteed by general relativity. In favor of the presence of a supermassive black hole in the centre of AGN are the strong gravitational potentials suggested by the width of some optical and UV emission lines, corresponding to velocities of the order of 2,000–

100,000 km s−1and interpreted as Keplerian velocities. A single, compact rotating object might also be the easiest explanation to the presence of the well collimated jets up to Mpc scales observed in many AGN.

The accretion disc

If the presence of a black hole fuelled by accretion is well accepted, it is still not completely estab- lished how the accreting matter is distributed around the black hole. In order to conserve the angular momentum and to consider interactions between particles, the matter cannot fall into the black hole on direct paths, therefore it is commonly believed that the matter forms a disc-like structure. In the standard accretion-disc model (Shakura & Sunyaev 1973), the matter is supposed to follow at any radius a Keplerian orbit (v(MBH/R)1/2) forming a geometrically thin (thickness/radius <<1) and opaque, i.e. optically thick, disc. The transporting mechanism of angular momentum outwards is probably due to viscosity produced by turbulent effects (as in the case of the shear viscosity caused by the differential rotation and parametrised withα; Shakura & Sunyaev 1973) and magnetic insta- bilities (Frank et al. 1992). The heat energy produced in the disc is released as radiation and can be locally described by black body emission with an effective temperature:

T (R)=T(R/R)−3/4 (2.3)

where R is the radius at which the radiation is emitted, Ris the radius of the compact object and T

is defined as:

TMBHM˙ R3

!1/4

(2.4) where MBH is the mass of the compact object and ˙M is the mass accretion rate. Since the disc has different temperatures as a function of the radius, the resulting disc spectrum is a superposition of black body spectra with a wide range of effective temperatures.

The disc emission is in general believed to be at the origin of the blue bump, the emission excess observed in optical-UV spectra of AGN (Shields 1978). However, several problems have been found when trying to apply this standard model to AGN (see Koratkar & Blaes 1999 for a review). The major issue concerns the observed quasi-simultaneous variability of the optical and UV continuum which is in contradiction with variations internal to the disc due to the temperature gradient, which are supposed to propagate on viscous time scales (Courvoisier & Clavel 1991). Furthermore, the low polarization observed in the optical-UV (Antonucci et al. 1996), the need of additional structures (as for example an external X-ray heating source or a corona; Haardt et al. 1994) and the complex relation between the variations in the optical-UV and X-ray bands (Uttley 2005) suggest that the real

(19)

geometry and physics of accretion in AGN is more complicated than that proposed in the standard accretion-disc model.

Even though they have not been explored yet as much as the standard model, in addition to the geometrically thin, optically thick one, different kinds of discs have been proposed to describe the accretion of matter into a black hole, e.g., among the thick accretion-disc models, the radiation tori (see e.g. Frank et al. 1992) and the ion tori (Rees et al. 1982).

An alternative scenario to fuel the black hole involves matter accreted in clumps, rather than through an accretion disc (Courvoisier & Türler 2005; Ishibashi & Courvoisier 2008). The interaction of these clumps at∼100 Schwarzschild radii generates optically thick shocks producing the optical- UV emission, whereas optically thin shocks closer to the black hole give origin to the X-rays.

The Broad and Narrow Line Regions

The presence of both broad and narrow, optical and UV emission lines in the spectra of AGN suggests the existence of two different regions where these lines are emitted, which are indicated as the Broad (BLR) and Narrow Line Regions (NLR), respectively. The distribution and composition of the matter in these regions is quite uncertain, but they are usually represented as a large number of optically thin clouds photoionised by the UV radiation and therefore emitting the observed lines. The broad line region extends from 0.01–0.1 pc for Seyfert 1 and up to 1 pc for the brightest quasars, as deduced from the study of broad-line variability. Considering the absence of broad forbidden lines, as [OIII], one can infer an electron density of≥108cm−3and a typical gas velocity of 3,000–10,000 km s−1can be deduced from the width of the lines (Netzer 1990). The narrow line region is instead formed by gas with lower velocities of 300–1000 km s−1, as indicated by the line width. Unlike the broad lines, no correlation is observed between the narrow-line variations and the variability of the continuum emission suggesting that the NLR is situated much farther out than the BLR. In Seyfert galaxies, the NLR has sizes of 100–300 pc and seems to reach up to a few kpc diameter for bright quasars.

The jet

If radio lobes had been observed in radio galaxies since 1953 (Jennison & Das Gupta 1953), only in 1966 Rees proposed that they could be powered through jets emitted from the galaxy instead of being ejected and completely detached, as previously thought. Over the years and especially thanks to the development of radio interferometry techniques, jets have revealed to be a common component not only in radio galaxies but in radio loud AGN in general. They extend on scales from parsec to hundreds of kpc, are highly collimated and can have a continuous appearance or can present knots.

The low-luminosity radio galaxies exhibit two symmetric jets, whereas the more luminous objects show only one jet, or two with very different intensities, as an effect of orientation and Doppler boosting. The observation of apparent superluminal motions in the jet of many AGN (Whitney et al.

1971) is also interpreted as due to the relativistic speed of the jet plasma combined with small angles between the jet axis and the line of sight.

The production, acceleration and collimation of such powerful structures are still not understood.

At present, the most popular models are based on magnetohydrodynamics (e.g. Blandford 2003), involving the presence of strong electromagnetic fields at the central engine that coupled with differ- ential rotation can convert the rotational kinetic energy into an outflow. The energy can be extracted from both the rotating black hole and the accreting matter.

Many questions remain still open, as for example what leads to the formation of jets with such

(20)

The unified models 9 different strengths as in radio loud and radio quiet objects. In fact, radio quiet AGN do not show evidence of this component on kpc scale, but a pc feature, corresponding to an unresolved point source or a jet, has been observed in a handful of objects (Blundell et al. 1996; Ulvestad et al. 2005), suggesting that weak jets might be present but quenched for some reason and therefore usually not observable in radio quiet sources. Among the solutions proposed to interpret the radio loud and radio quiet dichotomy, Wilson & Colbert (1995) suggested a correlation between the radio loudness and the black hole spin, with faster rotating black holes having larger radio power. Another possibility proposed by Capetti & Balmaverde (2006) relates the radio loudness to the surface brightness profile in elliptical host galaxies. In fact, in early-type galaxies radio loud AGN seem to be associated only to galaxies with “core” profiles (i.e. with evidence of an unresolved optical nuclear source), whereas radio quiet objects are hosted only by “power-law” galaxies (i.e. with shallow brightness profiles and no core). The surface brightness profile could be related to the formation history of the galaxy, with core galaxies (and therefore radio loud AGN) being the result of one major merger and power-law galaxies (and radio quiet AGN) forming rather from a series of minor mergers.

The radio to infrared non-thermal emission in AGN is now commonly interpreted as synchrotron emission from the well-collimated and relativistic plasma in the jet as well as from the core. In the most common scenario (e.g. Blandford & Konigl 1979, Marscher & Gear 1985), this plasma is continuously flowing from the AGN core and instabilities in this flow can generate shock waves propagating down the jet and accelerating particles to relativistic speeds. The electrons accelerated in the jet magnetic field produce the observed synchrotron emission, beamed in the forward direction into a cone of half opening angle 1/Γ, whereΓis the bulk Lorentz factor of the jet. These electrons are thought to be responsible also for the X-ray and gamma-ray emission via inverse Compton cooling (see Sect. 2.3.3).

2.2 The unified models

From the observational point of view, a simplified classification of the AGN population can be based just on two parameters1: the radio loudness and the width of the emission lines. Following different definitions, a radio loud object has a radio flux at 5 GHz at least 10 times larger than the optical- B one or has a radio luminosity larger than 1033erg s−1Hz−1 (Stocke et al. 1992). These objects represent only the 10–15% of the entire AGN population, which is dominated by radio quiet sources.

Depending on the presence or absence of broad emission lines, AGN are usually classified as type 1 or type 2, respectively, in analogy with the early Seyfert 1 – Seyfert 2 distinction.

The dust torus

Observations of Seyfert galaxies have revealed the presence of intermediate objects, between type 1 and type 2, that show only weakly detectable broad line components or important variations of the line width. Therefore, it has been proposed that both AGN types have the same physical nature and that the differences are due to varying amounts of absorbing material along the line of sight. This led

1Narrow Line Seyferts 1 (NLS1) galaxies seem to escape this simple classification. In fact, NLS1 are characterised by strong narrow lines and small widths of the broad Balmer lines (<2000 km s−1), like in Seyfert 2, but their strong optical and ultraviolet FeII emission and their ratio of optical to X-ray luminosity identify them rather as Seyfert 1. Many of the characteristics of NLS1 are understood as a consequence of their super-Eddington accretion rates (Collin & Kawaguchi 2004).

(21)

Figure 2.1: Schematic view of an AGN, where the main components are indicated (Urry & Padovani 1995).

The unified models propose that all these components (apart maybe for the jet) are present in all AGN and that the different AGN types correspond to different orientations of the AGN with respect to the observer.

to the introduction of a dust shell in the standard AGN picture, having any possible geometry, from a 4πshell with varying thickness to an angle-dependent axisymmetric structure or a patchy distribution of dust. The idea of a torus-like structure was motivated by the discovery by Antonucci & Miller (1985) of broad lines in the Seyfert 2 galaxy NGC 1068 when observed in linearly polarized light.

They argued that the broad lines in the spectrum are produced by reflection of radiation from the nuclear region. Historically, this study (afterwards supported by the observation of ionisation cones in a number of Seyferts; e.g. Wilson et al. 1993) was fundamental for the formulation of the unified model of AGN that considers the observational properties of the different AGN classes as a result of an orientation effect between the observer and the torus (Antonucci 1993). In this scheme, AGN have a dusty torus extending from 1 to 100 pc from the centre and type 1 AGN are believed to be objects of which the observer can view the inner regions within the torus opening angle (Fig. 2.1). In these objects both the BLR and NLR are visible. On the other hand, when the line of sight crosses the torus, the BLR appears to be hidden and this results in type 2 AGN spectra. This unification, first proposed for Seyfert galaxies, was then extended to include the observational differences of type 1 and type 2 objects also among radio loud AGN, with the radio galaxies being considered as obscured radio loud quasars.

Peculiar objects questioning the unified model

The unification model for type 1 and 2 objects is supported by the observations indicating that in general type 2 AGN have stronger absorbed X-ray spectra, a redder optical continuum and stronger

(22)

The unified models 11 spectral features from the host galaxy (Wilkes 2004) than type 1. On the other hand, the structure of the absorbing material seems to be more complex than a simple dust torus. In fact, a few sources have shown significant variations of the X-ray spectral shape, in some cases consistent with a temporary switching off of the nuclear source (e.g. NGC 4051, Matt et al. 2003) or with a clumpy torus where the absorption can strongly vary on relatively short time scales of days (e.g. NGC 1365, Risaliti et al.

2007). In addition some Seyfert 1 show strong intrinsic X-ray absorption (e.g. ESO 323-G077 with NH = 6.6×1022cm−2; Malizia et al. 2007), while some Seyfert 2 do not show intrinsic absorption.

Among the latter, NGC 3147 has been found to have an X-ray spectrum typical of Seyfert 1, even though it is optically classified as type 2. Bianchi et al. (2008) have proposed that the BLR of this object is not just hidden, but completely missing, maybe due to the relatively small accretion rate of this AGN, which might be insufficient for the BLR outflow to form. This indicates that at least for this object the key parameter to describe the observational characteristics is not the orientation but an intrinsic property (as for instance the mass accretion rate) that prevents the formation of the BLR.

The unification of radio loud AGN

When a jet is observable, the axisymmetric structure of the standard AGN model implies a large variety of the observed properties (and so classifications) as a function of the observing angle. A unification of the radio loud AGN has therefore been proposed based on the direction of the jet axis with respect to the line of sight. In this scenario, it is believed that radio loud AGN in which the jet is observed end-on belong to the blazar class, whereas ordinary radio loud quasars are viewed at larger angles from the jet axis and radio galaxies have jets seen in the plane of the sky. The particular jet orientation in blazars is responsible for the high luminosities (up to 1048−49erg s−1, when isotropic luminosity is assumed) and fast variability observed in these objects and for the apparent superluminal motions in the jet, all properties that can be explained as due to plasma moving at relativistic speeds towards the observer along the line of sight.

The blazar sequence

The spectral energy distributions (SED) of radio loud AGN present in general two main broad peaks (inνFν representation), one in the IR to X-ray range and the second one in the MeV to GeV band, depending on the AGN type (Figs. 2.6). These two peaks are often interpreted as due to the same population of relativistic electrons, accelerated in the jet, emitting synchrotron radiation at lower energies and scattering some seed photons through inverse Compton processes at higher energies. In the synchrotron self-Compton models (SSC), the seed photons for the inverse Compton process are assumed to be those emitted by synchrotron, whereas in the external Compton models (EC) the seed photons come from a source external to the jet, as for example the accretion disc (directly or first reprocessed by the ambient medium), the broad line region or the dusty torus (see von Montigny et al.

1995 for a review).

Fossati et al. (1998) have carried out a study on the SEDs of a sample of blazars and proposed a unification of the different blazar subclasses based on a continuum of properties that seems to be determined only by the blazar luminosity. The three blazar subclasses considered are the high- frequency-peaked BL Lac (HBL), the low-frequency-peaked BL Lac (LBL) and the flat-spectrum radio quasars (FSRQ), ordered by increasing luminosities. Fossati et al. (1998) noticed that the fre- quency of the two SED peaks decreases with increasing luminosity, the distance between the two peaks being about constant (left panel in Fig. 2.2). In addition, the ratio between the inverse Comp-

(23)

Figure 2.2:Left:The average SED of different blazar subclasses as found by Fossati et al. (1998). The points represent the data and the lines are analytical approximations. Going from the lower to the upper curves, HBL, LBL and FSRQ SEDs are shown. Right: Blazar SEDs for different values of black hole mass and accretion rate, as predicted by the model of Ghisellini & Tavecchio (2008). For FSRQs (upper thick lines), the accretion rate was assumed constant (L/LEdd=0.3) andMvaried as indicated. For BL Lac (lower thin lines), a mass of M=109Mwas chosen and the accretion rate (i.e.L/LEdd) was varied as labelled.

ton and the synchrotron peak fluxes increases with increasing luminosity. Ghisellini et al. (1998) explained this sequence as due to an increasing importance of the external radiation field going from HBL to FSRQs. In HBL the electron cooling would be less efficient and dominated by synchrotron processes, determining the extension of the spectrum to higher energies and a low contribution to the inverse Compton. The weakness of the external photon field in BL Lac objects is confirmed by the quasi-absence of emission lines. On the other hand, in FSRQs an intense radiation field external to the jet would cool down faster the electrons (shifting the SED peaks to lower energies) and mainly through inverse Compton radiation.

The discovery of blazars with broad lines but a HBL-type SED (e.g. Padovani et al. 2002, Maraschi et al. 2008) and of extremely luminous HBL (e.g. H1517+656; Beckmann et al. 1999), and the in- creasing evidence that LBL are more numerous than HBL (which would contradict blazar luminosity functions if LBL are more luminous than HBL; Padovani 2007) have questioned over the years the validity of this blazar sequence. Recently, retrieving some ideas already proposed by Maraschi &

Tavecchio (2003), Ghisellini & Tavecchio (2008) have elaborated a new approach to the problem and proposed that the SED properties and jet power in blazars can be represented by two parameters, the black hole mass M and the accretion rate ˙M (together determining the disc luminosity, Ldisc). Assum- ing that the jet power scales with ˙M, that the BLR exists only above a critical value of Ldiscand that its distance from the black hole scales as L1/2disc, and that the location in the jet where most of the emission is produced scales with M, Ghisellini & Tavecchio (2008) explain the blazar sequence as follows. In FSRQs, the large black hole mass and high accretion rate determine the presence of powerful disc and jet and position the BLR at large distances from the centre, where the BLR-jet interaction gives origin to Compton-dominated blazars (right panel in Fig. 2.2). With evolution, the accretion rate decreases and therefore the disc radiates less efficiently and the BLR becomes smaller. As a consequence, in some cases the most active region of the jet can still interact with the BLR and in others it is farther

(24)

The X-ray emission from AGN 13 out with respect to the BLR radiation field. This would explain the existence of broad-line blazars with HBL-type SED. When ˙M decreases below a critical value (Ldisc/LEdd = 3×10−3), the BLR is too weak and too small2to interact with the jet and this represents the situation in BL Lac objects.

2.3 The X-ray emission from AGN

After the first two years of observations of the satellite Ariel V in 1974–1976, it had been clear that X-ray emission is a common characteristic of most AGN (Elvis et al. 1978) and that a large part of their radiation power is emitted in this energy band. In addition, the fast variations of the AGN flux in the X-rays (down to hour time-scales; Grandi et al. 1992) indicate that they are associated with the innermost regions of the nucleus, close to the central black hole. All this assigns to the X-rays a crucial role for the understanding of the AGN phenomenon.

The properties of the X-ray emission are different for radio quiet and radio loud objects, as the strong jet present in the latter establishes the main emission process also at high energies. Within the radio quiet population, type 1 and type 2 objects present spectra whose differences can in general be explained by the unified model, i.e. as determined by the presence of dusty absorbing gas along the line of sight.

2.3.1 X-ray spectra of Seyfert 1

The X-ray spectra of Seyfert 1 galaxies extend from 0.1 to a few hundred keV and are characterised by the following components (Fig. 2.3).

Primary emission. To the first order, the intrinsic continuum of Seyfert galaxies is a power law, extending from 1–2 to a few hundred keV, and with typical photon index ranging between 1.8 and 2.

At high energy, an exponential cut-off is often observed between 60 and 300 keV. This emission is believed to be produced in a two-phase accretion disc, where soft photons from a cold (kT < 50 eV) thick disc are Comptonised by a hot (kT ∼ 100 keV) thermal electron gas in a thin corona located above the disc (Haardt & Maraschi 1993). The soft photons with initial energy Ei are upscattered to an energy that can be approximated by Ef =eyEi, where y(4kT/mec2)max(τ, τ2) is the Compton parameter,τis the optical depth to Compton scattering and T is the electron temperature. Forτ >0.01 and y << 10 (i.e. non-relativistic electrons), this actually results in a power law spectrum up to a thermal cut-off at EkT3kT , determined by the cut-off in the thermal distribution of the electrons (Pozdniakov et al. 1983; Mushotzky et al. 1993).

Reflection components. An important result of the Japanese mission Ginga was the discovery of a hump above 5 keV superposed to the continuum X-ray emission in Seyfert galaxies (Piro et al. 1990).

This feature is interpreted to be due to the primary X-rays that are reflected (i.e. Compton scattered) by optically-thick cold material subtending a large solid angle to the X-ray source. The exact shape of the reflection component varies with the geometry and chemical composition of the reflector, but in general it has a peak around 30 keV, where the reflection efficiency reaches a maximum. The nature of the reflecting medium is still uncertain, possibly being the accretion disc itself (Zdziarski et al. 1990) or the inner edge of the absorbing gas (Ghisellini et al. 1994) or a wind (Elvis 2000). A reflection

2For MBH=109Mand Ldisc/LEdd=3×10−3, the radius of the BLR is found to be RBLR=0.02 pc, following Ghisellini

& Tavecchio (2008).

(25)

Figure 2.3: Typical total X-ray spectrum (upper black line) with its main components of a type 1 AGN.The primary continuum is a power law with a high-energy cut-off and absorbed at low energies by warm gas. A soft-excess, a cold reflection component and the iron Kαemission line at 6.4 keV are also shown.

component from a warm reflector is also observed in some X-ray spectra of Seyfert galaxies and it has the same spectral shape of the incident emission.

The iron line. In most Seyfert galaxies an iron emission line at 6.4 keV is observed, with a typical equivalent width of 100–200 eV. This emission corresponds to the Fe Kαtransition (n = 2 to n = 1) of cold iron and is attributed to fluorescence of some cold material illuminated by an X-ray source.

The reflection hump and the Fe line are consistent with the idea that the primary X-ray emission is reprocessed and both these features are usually observed in AGN spectra and thought to be connected.

The line often shows two components, a broad and a narrow one. The broad line is thought to originate in the inner part of the accretion disc, and it varies following the continuum variations with almost zero delays. Instead, the narrow line (with widths ≤ 1000 km s−1; Risaliti & Elvis 2004) does not follow the variations of the continuum, suggesting that it origins on a farther reflector, maybe the absorbing circumnuclear gas (Matt et al. 1996). There is evidence in several objects of a broad, red wing on the Fe line (left panel in Fig. 2.4) that is commonly interpreted as due to relativistic broadening and gravitational redshifting produced by a spinning black hole in the core of the AGN (e.g. Fabian & Miniutti 2005). An alternative model, involving complex absorbers with different ionisation states, has been shown to represent well the Fe-line shape with its red wing without need of a relativistically-blurred component, at least in some objects (e.g. MCG−6−30−15, Miller et al.

2008, and NGC 3783, Reeves et al. 2004).

The soft-excess. Many AGN show a prominent excess of emission below 2 keV with respect to the extrapolation of the power law continuum (e.g. Porquet et al. 2004). This so-called soft-excess can be well fitted with a black body with temperatures of 0.1–0.2 keV for a large range of black hole masses, 106−9M(Walter et al. 1994). These temperatures are too high to be explained by the standard accretion-disc model (Shakura & Sunyaev 1973) as thermal emission from the disc, unless one assumes for example super-Eddington luminosities. The soft-excess could instead be due to

(26)

The X-ray emission from AGN 15

Figure 2.4: Left: the 3–8 keV data fromSuzaku/XIS divided by a power law model show the presence of a broad Fe emission line in the Seyfert 1 MCG63015 (Miniutti et al. 2007). Right: 0.8–2.5 keV spectrum fromChandra/HETGS of the Seyfert 1 NGC 3783 (black curve). The numerous absorption features on the source continuum emission are well represented by a two-phase absorber (red curve; Krongold et al. 2003).

Comptonisation of the UV-disc photons, but its almost constant temperature would be difficult to explain, considering that it should be related to the disc temperature, depending in turn on the black hole mass. A more natural explanation for this constant temperature would therefore come from atomic processes, smeared out by high velocities or relativistic effects. In this frame, different models have been proposed, suggesting that the soft-excess origins as reflected X-ray emission (relativistic- blurred) on a disc partially ionised by the incident flux (Ross & Fabian 2005), or as relativistic-blurred absorption from a disc wind, producing a large absorption feature in the spectrum at 1–2 keV and therefore an apparent soft-excess below 1 keV (Gierli´nski & Done 2004).

Warm absorbers. Thanks to ASCA observations first and more recently to the high resolution spectra obtained by XMM-Newton and Chandra, it has been possible to observe strong absorption features in the X-ray emission of AGN. The spectral characteristics of these features associate them to warm absorbers probably in the form of outflowing gas. The geometry, the location and the physical conditions of the absorbing gas are still under debate. For the best studied case of NGC 3783 (right panel in Fig. 2.4), a two-phase absorbing medium with different temperatures and ionisation states, but in pressure equilibrium, might be required (Krongold et al. 2003), whereas Gonçalves et al. (2006) suggested for the same object a single medium with different temperatures producing a stratification of the ionisation structure. Studying the variations of the ionisation parameters, Nicastro et al. (2007) proposed that in NGC 4051 the warm absorber is associated to outflows from the disc at a distance of .5 light days consistent with the high-ionisation BLR, whereas in NGC 5548 the absorber would be located at larger radii (>0.2 pc), rather consistent with the torus region.

2.3.2 X-ray spectra of Seyfert 2

In the unified models, the differences between the X-ray spectra of type 1 and type 2 AGN can be accounted for by considering the presence of absorbing/scattering material on the line of sight. This is supported by the observations, showing that the X-ray emission from Seyfert 2 can be represented with the same components found in Seyfert 1, but modified by a certain amount of absorbing material.

(27)

Figure 2.5: Examples of X-ray spectra for an unabsorbed Seyfert 1 galaxy, IC 4329A (left panel, Zdziarski et al. 1996), and the absorbed (NH2×1022cm−2) Seyfert 1.9 MCG–5–23–16 (right panel, Reeves 2007).

In the X-rays obscuration is due to photoelectric absorption and Compton scattering. Depending on the amount of absorbing material in the line of sight (parametrised with the hydrogen column density NH, in atoms cm−2 units), part of the AGN X-ray emission can be absorbed between 1 and 10 keV when NH<1.5×1024cm−2, or even up to a few tens of keV, if 1024< NH<1025cm−2(Fig. 2.5).

The iron line equivalent width is found to depend on the absorbing column density, showing larger values for more obscured sources (Risaliti & Elvis 2004). This is interpreted as due to the fact that the line is, at least in part, emitted by the reflecting material along free lines of sight, whereas the continuum at the line energy is strongly dependent on the NHand is more absorbed than the line.

The primary continuum emission of Seyfert 2 in the X-rays is also well represented by a power law, on average harder than in Seyfert 1 (see Section 5.2.1) and with a high-energy cut-off in the same range of energies as in Seyfert 1. A reflection component is present as well; in particular for Compton- thick objects (i.e. NH>1024cm−2) no direct emission but only the reflected and the scattered continua are observable below 10 keV. Soft-excess emission is present in Seyfert 2 and could be associated to the warm gas confining the BLR rather than to the disc (Risaliti 2002). In fact, in type 2 AGN the emission from the disc, if present, is expected to be absorbed like the power law component.

2.3.3 X-ray spectra of radio loud AGN

The presence of a jet dominates the properties of radio loud AGN over several energy decades. As seen in Sect. 2.2, the spectral energy distribution is characterised by two broad peaks where most of the emission is produced through synchrotron and inverse Compton processes. Differently from radio quiet AGN, the X-ray emission of radio loud AGN is therefore thought to be of non-thermal origin and it is located around the minimum between the two peaks of the SED, where both synchrotron and inverse Compton can contribute. In particular, in FSRQs, the X-ray emission corresponds to the beginning of the Compton peak (Fig. 2.6, top left panel) and in LBL it corresponds to the transition between the synchrotron and the Compton peaks (Fig. 2.6, top right panel) In HBL, the synchrotron emission usually peaks in the soft X-ray band (Fig. 2.6, bottom left panel), but the peak can move up to 100 keV during flaring states, as observed, for example, in Mrk 501 in April 1997 by BeppoSAX (Pian et al. 1998; Tavecchio et al. 2001). This makes the X-ray studies of these objects very valuable to understand what is the relative importance of the synchrotron and inverse Compton processes.

(28)

The hard X-ray emission of AGN 17

Figure 2.6:Examples of spectral energy distributions of radio loud AGN.Top left:SED of the FSRQ 3C 454.3 (Pian et al. 2006). Top right: SED of the LBL BL Lac (Guetta et al. 2004). Bottom left: SED of the HBL 1E S1959+650 (Tagliaferri et al. 2003).Bottom right:SED of the radio galaxy Centaurus A (Steinle 2006).

A contribution from thermal Comptonisation to the spectra of radio loud AGN cannot be excluded and might be hidden or diluted by the much stronger non-thermal component. Especially in radio galaxies the results are more controversial, with the detection of high-energy cut-offs (Molina et al.

2007), narrow Fe lines (Rothschild et al. 2006) but no or weak reflection components. There could be a transition leading from blazars to radio galaxies to Seyferts, in which the thermal Comptonisation becomes more and more important over the non-thermal emission and reprocessing features, like the reflection component and the Fe line, slowly appear and become stronger (Grandi et al. 2006).

2.4 The hard X-ray emission of AGN

After having described the broad-band X-ray emission from AGN, it is interesting to have a closer look at the hard X-ray part of the spectrum, describing the knowledge that we have acquired since the first observations 40 years ago and focusing on the open questions that INTEGRAL and other present and future high-energy satellites can address (see Chapter 3). A clarification is needed at this point on the meaning of hard X-ray emission throughout this thesis: we consider as hard X-rays the emission

(29)

above∼20 keV and up to several hundred keV.

After the first successful X-ray observations of an extragalactic object, the quasar 3C 273, with rocket experiments in 1969 (Bowyer et al. 1970), the first satellite completely dedicated to X-ray astronomy, UHURU, was launched in December 1970 and detected 10 AGN up to 20 keV (Forman et al. 1978). At hard X-rays, the number of known X-ray AGN and the quality of the available spectra increased significantly over the years with OSO-7 (Mushotzky et al. 1976), HEAO-1 (Rothschild et al. 1983) and EXOSAT (Turner & Pounds 1989). Before the launch of Ginga, the available X-ray spectra of AGN were well represented by simple power laws with typical photon index of 1.7, in some cases absorbed at low energies (e.g. Turner & Pounds 1989). In February 1987, Ginga started its 6 years of operations, with a significant increase in sensitivity with respect to previous satellites.

Ginga observations of AGN in the 1–30 keV range led to the important discovery of spectral features associated to reprocessing of cold matter: the Fe emission line and the reflection hump (Piro et al.

1990).

Before the launch of the Compton Gamma-Ray Observatory (CGRO) satellite in April 1991, the X-ray spectra of Seyferts were thought to be of non-thermal origin and extend in some cases up to the MeV band (as for example for NGC 4151; Perotti et al. 1981). Thanks to the CGRO/OSSE instrument operating above 50 keV, major progresses were accomplished in the understanding of the hard X-ray emission from AGN. For the first time high-energy cut-offs at few hundred keV were detected for several Seyfert galaxies (Zdziarski et al. 1995) and the high signal-to-noise of the data allowed a finer modelling of the spectra with more physical models than the phenomenological power law.

Thermal Comptonisation models applied to the data resulted in quantitative estimates of the plasma temperature (kT ∼50–150 keV) and Compton parameter (y ≤ 1; Zdziarski et al. 2000). In addition, hard X-ray Seyfert spectra were found to be steeper (Γ∼2.0−2.2) than those at lower energies and, in particular, Seyfert 1 spectra appeared to be softer than Seyfert 2’s.

A further improvement was achieved with the BeppoSAX satellite, operating between April 1996 and April 2002, thanks to the improved sensitivity and the large band coverage (0.1–300 keV) pro- vided by the three narrow-field instruments. BeppoSAX collected spectra for more than 100 Seyfert galaxies, allowing to confirm OSSE results and to put better constraints on the parameters describing the X-ray spectra. In particular, the simultaneous coverage of the spectrum from soft to hard X-rays enabled a deeper study of the reflected component (e.g. Risaliti 2002, Perola et al. 2002).

In spite of the important progresses made by the hard X-ray astronomy in the last decades, many questions are still open and support the need of more studies at hard X-rays. If the phenomenological components of the hard X-ray spectrum of AGN seem to have been identified (a power law, a high- energy cut-off, a reflection hump), the parameters characterising them and the physical model to represent them are still not constrained. This is very important not only to reconstruct the geometry and physics related to the X-ray emission in AGN (see Chap. 4) but also for the study of the cosmic X-ray background and the AGN contribution to it (see Chap. 5). In addition, a comparative study of the hard X-ray properties of Seyfert 1 and Seyfert 2 galaxies is a very important, complementary test for the unified models. In fact, the hard X-ray band provides an unbiased view of the AGN population, not being affected by absorption effects (provided that NHis smaller than a few 1024cm2).

Références

Documents relatifs

Correlation relationships between the physical parameters of granite (bulk density, porosity, ultrasonic P-wave velocity) and its basic mechanical parameters (compressive

Dans le chapitre 3, nous étudions les relations lead/lag à haute fréquence en utilisant un estimateur de fonction de corrélation adapté aux données tick-by-tick9. Nous illustrons

(a) Normalized amplitude of the cross correlation of seismic noise at periods between 5 and 10 s as a function of time and azimuth determined from the network analysis in

Keywords: Thin films, TiO 2 –ZnO, Sol–gel, Anatase, Brookite, Optical properties, Structural properties,

tion spectra has normally been made assuming purely sharp electronic levels (S.E.L.) and, therefore, neglecting the electron-lattice coupling, which gives. rise to more

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

The relaxation effects were included by performing separate SCF calculations for the initial and final states and taking the energy difference to determine

Numerical and experimental analysis of the correlation between EIT data eigenvalues and two-phase flow phase fraction... Numerical and Experimental Analysis