Muestras de galaxias activas, del infrarrojo a los Rayos X

DEPARTAMENTO DE FÍSICA MODERNA
UNIVERSIDAD DE CANTABRIA
INSTITUTO DE FÍSICA DE CANTABRIA
IFCA (CSIC-UC)
Muestras de Galaxias Activas,
del Infrarrojo a los rayos X
Memoria presentada por
Núria Castelló Mor
para optar al tı́tulo de Doctora por la Universidad de Cantabria
Santander, Febrero de 2014
DEPARTAMENTO DE FÍSICA MODERNA
UNIVERSIDAD DE CANTABRIA
INSTITUTO DE FÍSICA DE CANTABRIA
IFCA (CSIC-UC)
AGN samples,
from the Infrared to X-rays
A dissertation submitted in partial fulfillment of the requirements
for the degree of Doctor of Philosophy in Physics
by
Núria Castelló Mor
Declaración de Autorı́a
Xavier Barcons Jáuregui, Doctor en Ciencias Fı́sicas y Profesor de Investigación del
Consejo Superior de Investigaciones Cientı́ficas,
y
Francisco Carrera Troyano, Doctor en Ciencias Fı́sicas y Profesor Titular de Universidad de Cantabria,
CERTIFICAN que la presente memoria
Muestras de Galaxias Activas, del Infrarrojo a los rayos X
ha sido realizada por Núria Castelló Mor bajo nuestra dirección en el Instituto de Fı́sica
de Cantabria, para optar al tı́tulo de Doctor por la Universidad de Cantabria. Consideramos que esta memoria contiene aportaciones cientı́ficas suficientemente relevantes
como para constituir la Tesis Doctoral de la interesada.
En Santander, a 12 Febrero de 2014,
Xavier Barcons Jáuregui
Francisco Carrera Troyano
Als meus cabretes Jordi i Nuc
Acknowledgements
Es difı́cil entender la importancia de los agradecimientos de una tesis doctoral hasta
que no se ha terminado. En ese momento te das cuenta de cuánto tienes que agradecer
a tanta gente. Desde estas lı́neas pretendo expresar mi más sincero agradecimiento a
todas aquellas personas que durante estos años de trabajo han estado a mi lado, amigos, familia y compañeros, y que de una u otra forma han contribuido a que esta tesis
haya llegado a buen fin.
Debo agradecer de manera especial y sincera al Profesor Xavier Barcons -el jefe- por
aceptarme para realizar esta tesis doctoral bajo su dirección. Su apoyo y confianza en mi
trabajo y su capacidad para guiar mis ideas ha sido un aporte invaluable, no solamente
en el desarrollo de esta tesis, sino también en mi formación como investigadora. Quisiera
también agradecerle las veces que ha sabido escucharme, y que al hacerlo haya movido
cielo y mar para complacer a todos los implicados. Gràcies Xavier per la conversa, la
veritat es que em va ajudar molt. Es el momento de clamar que nunca podrı́a haber
realizado esta tesis sin Francisco Carrera, siendo para mi la versión presente de Xavier en
sus numerosas ausencias. Agradecerle su consejo, el apoyo y el ánimo que me brindó una
y otra vez. Les agradezco que me hayan abierto hace ya cinco años las puertas de su
grupo de investigación, dándome la oportunidad de tener una visión más amplia del
mundo de la investigación y descubrir cuánto me motiva. Espero que mi trabajo y
dedicación hayan estado a la misma altura que el suyo.
Me gustarı́a extender este agradecimiento a Lucia Ballo y Silvia Mateos. Poder trabajar con astrónomas como ellas ha sido una experiencia increı́ble y muy, pero que muy
gratificante. Agradecerle a Lucia su pragmatismo (¿sera porque te criaste con un lógico?), y a Silvia su patológica obsesión de controlar y entender todos los pasos y/o números que esconde cualquier razonamiento y/o procedimiento. En el proceso, he aprendido
el valor que tiene un trabajo meticuloso y cuánto cuesta conseguirlo. . . Quisiera también
agradecerles el haber echo de hermano mayor, por tantas horas de discusiones, códigos,
lamentos y demás, que llenaron muchos de los momentos de fatiga. Agradecerle a Silvia
también la posibilidad de visitar uno de los lugares mas hermosos de este mundo el
desolado desierto de Atacama.
De igual manera agradecer a Almudena Alonso-Herrero su desinteresada ayuda y su
apoyo durante el trabajo diario, mostrándose accesible en todo momento para resolver
las más variopintas cuestiones. Por su visión crı́tica de muchos aspectos cotidianos de
x
la vida, por su rectitud y lucha constante como mujer en un mundo de astrofı́sicos, por
sus consejos, que ayudan a formarte como persona e investigador. Aprendı́ con ella que
ser madre y astrónoma son dos trabajos perfectamente compatibles.
Likewise I would like to particularly thank to Professor Martin Ward for his hospitality and the help from his group during my visit to Durham. I came back enthusiastically
engaged with the research project, in love with Durham and thankful for your support.
I really appreciate your efforts very much.
Ha sido un verdadero placer trabajar con todos vosotros.
Mi agradecimiento al programa de becas FPI del Ministerio, a través del cual este
trabajo de investigación ha sido financiado en su mayor parte.
Mi agradecimiento también a toda la gente del grupo de Astronomı́a de Rayos X
del IFCA por todos los calóricos cafés. En especial gracias a mi compañero de fatigas
y querido amigo Anuar por introducir ese frikismo en el grupo y por no ser un STV
(¡ninguna dedicatoria estarı́a a su altura!); a Antonio por estar siempre disponible; a
la SUSI, Maite Ceballos, por atenta lectura de este trabajo y atinadas correcciones,
aportando el control de calidad necesario para que este trabajo sea una merecida tesis
doctoral; a Ester por esa tarde de cine; a Bea por su calor humano; y a Judit por su
optimista visión del mundo.
También quiero dar las gracias a todos los doctorandos (y algunos ya Doctores)
-Airam, Raúl, Luis, Anuar, Anaı̈s, Nicolas, Violeta, Claudia, Diego, Andrés- por las interesantes conversaciones, a veces incluso sobre ciencia, en las sobremesas, cafés, bares,
etc. que han hecho que todo haya sido bastante más entretenido. Una mención especial
merecen la casi-Doctora Biuse Casaponsa y el vasco David Moya del Instituto de Fı́sica
de Cantabria, por su inestimable amistad: lamento haberles taladrado tantas, y tantas
horas. Con todos ellos he compartido muchos de los mejores e inolvidables momentos
de esta etapa de mi vida. Espero no perder el contacto con ninguno de ellos.
I would like to give my special thanks to Dr. Jin Chichuan for providing me with
the big IDL code for producing the Balmer emission line analysis, and also to thank
him for introducing me to IDL programming and for his extremely valuable discussion
and suggestions about the work on the spectral analysis.
Gracias en general a toda la gente del IFCA por crear un ambiente de trabajo tan
sano, y en ocasiones, lleno de frikismo. En especial gracias a Pablo y Alvaro por intentar
poner el software de Astrofı́sica en el Grid y no desistir en el intento; a Ana y Enol
xi
por esos deliciosos cafés entre bebes. Ha sido un verdadero placer pasar estos años allı́ y
aprender de los mejores. Toda la gente que he conocido y los lugares que he podido
visitar durante mi doctorado han hecho de este mucho más que una mera experiencia
académica.
Per últim, y no per això menys important, voldria donar les gràcies a la meva familia
(al meu pare Jaume, a la meva mare Lourdes, al meu germà Xavier, a la meva germana Marta, a la meva padrina Maria) pel seu suport incondicional en tot moment. En
especial voldria donar les gracies a l’Emı́lia per venir fins a Santander sempre que l’hi
hem demanat. Als meus sogres agraı̈r-los que ho hagin deixat tot per venir a cuidar-nos
(tot i saber que nomes trobarien pluja). Gracias Eduardo por ocuparte de Nuc incluso
cuando aún ası́ no te encontrabas bien. Gràcies també a les meves cunyades i cunyats
per ensenyar-nos Cantabria. Gracias Javi por tus visitas, y sobre todo, por hacernos reir.
Per últim (i ara sı́) grácies al Doctor Jordi Duarte per estar sempre al meu costat,
per ajudar-me a tirar endavant i no deixar que ho tiri tot per la borda; al Nuc li donc
les gràcies per aparèixer a la nostra vida, malgrat tots els moments inhumans que ens
planteja el seu creixement, no cap la possibilitat de no estimar-lo.
xii
xiv
Contents
xv
Contents
xvii
List of Figures
List of Tables
xix
Acronyms
xxi
xxiii
Notation and Conventions
Resumen
xxv
Summary
xxxiii
1. Introduction
1
1.1. Active Galactic Nuclei . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1.2. The AGN paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
1.3. AGN taxonomy and Unified model . . . . . . . . . . . . . . . . . . . . .
6
1.4. The AGN spectral energy distribution . . . . . . . . . . . . . . . . . . .
13
1.5. The search for obscured AGN . . . . . . . . . . . . . . . . . . . . . . . .
21
1.6. Motivation of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
2. Data
31
2.1. X-ray data: XMM-Newton . . . . . . . . . . . . . . . . . . . . . . . . . .
31
2.2. Optical data: SDSS DR7
. . . . . . . . . . . . . . . . . . . . . . . . . .
36
2.3. IR data: Spitzer/IRAC . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38
3. AGN missed by the BPT diagram
3.1. Motivation
41
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
3.2. Data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
3.3. Optical classification versus X-ray emission . . . . . . . . . . . . . . . .
45
xvi
CONTENTS
3.4. Optical versus X-ray properties . . . . . . . . . . . . . . . . . . . . . . .
3.5. An overview of the
missing-AGN
49
subsample . . . . . . . . . . . . . . .
54
3.6. X-ray Spectral Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . .
58
3.7. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . .
69
4. Non-Standard NLS1 galaxies
75
4.1. Motivation: non-standard NLS1 . . . . . . . . . . . . . . . . . . . . . . .
75
4.2. Sample and Data Preparation . . . . . . . . . . . . . . . . . . . . . . . .
76
4.3. Optical Spectral Properties . . . . . . . . . . . . . . . . . . . . . . . . .
77
4.4. Broadband SED Modelling . . . . . . . . . . . . . . . . . . . . . . . . .
81
4.5. Physical Properties of the NLS1 sample . . . . . . . . . . . . . . . . . .
87
4.6. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . .
95
5. IR power-law as an indicator of obscuration
5.1. Motivation
101
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.2. Data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3. X-ray spectral analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.4. Intrinsic absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.5. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6. Conclusions and Future Work
125
Glossary
131
References
133
List of Figures
1.1. AGN Unified model . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
1.2. AGN Spectral Energy Distribution . . . . . . . . . . . . . . . . . . . . .
14
1.3. IR Spectral Energy Distribution of Seyferts . . . . . . . . . . . . . . . .
15
1.4. AGN X-ray spectrum
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
1.5. Optical spectroscopic diagnostic diagrams . . . . . . . . . . . . . . . . .
25
3.1. Hα FWHM distribution . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
3.2. NELG sample: distribution of the angular distance separation . . . . .
44
3.3. NELG sample: FWHM distribution of Hα, Hβ, [N II], and [O III] . . .
45
3.4. NELG sample: BPT diagram . . . . . . . . . . . . . . . . . . . . . . .
48
3.5. NELG sample: distributions of HR, T , XO . . . . . . . . . . . . . . .
52
3.6. NELG sample: HR versus T and XO and T
. . . . . . . . . . . . . .
53
3.7. NELG sample: LX versus Hβ FWHM . . . . . . . . . . . . . . . . . .
55
3.8. missing-AGN: LHα /LX distribution . . . . . . . . . . . . . . . . . . . .
56
3.9. missing-AGN: optical spectra . . . . . . . . . . . . . . . . . . . . . . .
57
3.10. missing-AGN: R4570 distribution . . . . . . . . . . . . . . . . . . . . . .
58
3.11. missing-AGN: Γ2−10keV distribution . . . . . . . . . . . . . . . . . . . .
63
3.12. missing-AGN: hard X-ray best-fit power-law . . . . . . . . . . . . . . .
71
4.1. Hβ emission line fitting . . . . . . . . . . . . . . . . . . . . . . . . . . .
78
4.2. Balmer decrement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
81
4.3. optxagnf model geometry . . . . . . . . . . . . . . . . . . . . . . . . .
83
4.4. non-standard NLS1: Best-fit broadband SED . . . . . . . . . . . . .
89
4.5. optxagnf: bolometric luminosity . . . . . . . . . . . . . . . . . . . . .
90
4.6. optxagnf: black hole mass . . . . . . . . . . . . . . . . . . . . . . . . .
90
4.7. optxagnf: Eddington ratio . . . . . . . . . . . . . . . . . . . . . . . . .
91
4.8. optxagnf: optical-to-X-ray spectral index . . . . . . . . . . . . . . . .
91
4.9. optxagnf: X-ray bolometric correction . . . . . . . . . . . . . . . . . .
91
4.10. optxagnf: temperature of the soft X-ray component . . . . . . . . . .
92
LIST OF FIGURES
xviii
4.11. optxagnf: optical depth of the soft X-ray component . . . . . . . . . .
92
4.12. optxagnf: coronal radius Rcor . . . . . . . . . . . . . . . . . . . . . . .
92
4.13. optxagnf: intrinsic absorption . . . . . . . . . . . . . . . . . . . . . . .
93
4.14. Hβ FWHM vs. λEdd correlation . . . . . . . . . . . . . . . . . . . . . . .
94
4.15. Rest-frame broadband SED . . . . . . . . . . . . . . . . . . . . . . . . .
98
4.16. Broadband SED best-fit model . . . . . . . . . . . . . . . . . . . . . . .
99
5.1. Redshift distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.2. Angular separation distribution . . . . . . . . . . . . . . . . . . . . . . . 106
5.3. IRAC colour-colour diagram . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.4. Source net counts versus background counts . . . . . . . . . . . . . . . . 108
5.5. ID210=130,283,6,161: X-ray best-fit model . . . . . . . . . . . . . . . . 111
5.6. Γ, LX , NH , and FX distributions . . . . . . . . . . . . . . . . . . . . . . 112
5.7. Correlations between Γ and NH , LX or z . . . . . . . . . . . . . . . . . 113
5.8. ID210=66,144: X-ray bet-fit model and contour plots of Γ and NH . . . 114
5.9. Fraction of absorbed sources . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.10. NH (LX ) distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
List of Tables
1.1. AGN Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
1.2. AGN Selection Methods . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
3.1. Subsample description and Optical/X-ray properties . . . . . . . . . . .
46
3.2. Description of the XMM-Newton observations (and SDSS counterpart) .
60
3.3. missing-AGN subsample: best-fit (absorbed)power-law model . . . . .
64
3.4. missing-AGN subsample: best-fit parameters for the used SE models
65
3.5. Results of the fitting spectral data for the true-SF population . . . . .
67
3.6. Summary of the results of the XMM-Newton spectral analysis results. .
68
4.1. NLS1 sample: Optical Key parameters . . . . . . . . . . . . . . . . . .
80
4.2. Broadband SED: best-fit optxagnf model . . . . . . . . . . . . . . .
88
5.1. Best-fit models summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.2. X-ray properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.3. Compton-thick candidates . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.4. Absorbed/Unabsorbed classification at 1σ . . . . . . . . . . . . . . . . . 118
5.5. Summary of the X-ray properties . . . . . . . . . . . . . . . . . . . . . . 124
LIST OF TABLES
xx
Acronyms
S/N
signal-to-noise ratio.
2dF
Two-degree-Field Galaxy Redshift Survey.
AGN
Active Galactic Nuclei.
BBB
Big Blue Bump.
BH
Black Hole.
BLR
Broad Line Region.
BLRG
Broad-Line Radio-Galaxy.
BLS1
Broad Line Seyfert 1.
BPT
Baldwin, Philips & Terlevich.
CCD
Charge-Coupled Device.
CDF
Chandra Deep Field (X-rays).
COMBO-17
Classifying Objects by Medium-Band Observations in 17 Filters.
COSMOS
Cosmic Evolution Survey.
far-IR
far InfraRed.
FOV
Field of View.
FR
Fanaroff-Riley.
FSRQ
Flat Spectrum Radio Quasars.
FWHM
Full Width at Half Maximum.
HEASARC
High Energy Astrophysics Science Archive
Research Center.
HR
Hardness Ratio.
xxii
Acronyms
ILR
Intermediate Line Region.
IR
InfraRed.
IRAC
InfraRed Array Camera.
IRAS
InfraRed Astronomical Satellite.
ISM
InterStellar Medium.
LINERs
Low-Ionisation Nuclear Emission-line Galaxies.
MIPS
Multiband Imaging Photometer (on board
NASA’s Spitzer satellite).
near-IR
near InfraRed.
NELG
Narrow Emission Line Galaxies.
NLR
Narrow Line Region.
NLS1
Narrow Line Seyfert 1.
PSF
Point Spread Function.
RL
Radio-Loud.
RQ
Radio-Quiet.
SDSS
Sloan Digital Sky Survey.
SED
Spectral Energy Distribution.
SF
Star-Forming.
SMBH
SuperMassive Black Hole.
UV
UltraViolet.
WISE
Wide-field Infrared Survey Explorer.
XRB
cosmic X-Ray Background.
Notation and Conventions
List of selected symbols
MBH
Lbol
LEdd
λEdd
Av
rg
κ2−10keV
NH,Gal , NH,z
Ṁ
αOX
E(B − V)
RL
z
R4570
LX
M
T
Γ
XO
black hole mass
bolometric luminosity
Eddington luminosity
Eddington ratio
extinction in the V band
gravitational radius
hard X-ray bolometric correction
hydrogen (HI+H2 ) column density
(Galactic and intrinsic, respectively)
mass accretion rate
optical-to-X-ray spectral index
colour excess between the B and V band
radio-loudness
redshift
relative strength of the Fe II multiplets
rest-frame 2-10 keV luminosity
solar mass
thickness parameter
X-ray photon index
X-ray-to-optical flux ratio
Throughout this dissertation, we have adopted a concordance ΛCDM model as a
parametrisation of the Big Bang cosmological model in which the universe has a dark
energy density of ΩΛ = 0.73 and a matter density of Ωm = 0.27 with a Hubble constant
of H0 = 71 km s−1 Mpc−1 (Spergel et al., 2007, 2003).
From radio to infrarred, the energy spectral index, α, is defined such that fν ∝ ν α .
The X-ray photon index Γ is, on the other hand, defined as FE ∝ E −Γ where FE is the
photon flux per unit energy (E being the photon energy).
NOTATION AND CONVENTIONS
xxiv
Resumen
Los núcleos activos de galaxias (AGN, del inglés Active Galactic Nuclei ), descubiertos ya hace más de cincuenta años, son fuentes de luminosidades extremas, cuyas
propiedades permiten adentrarnos en un universo más joven. Se trata de objetos estudiados históricamente por sus extremas y exóticas propiedades, cuya aparente relación
entre su crecimiento y la formación de las galaxias ha extendido recientemente el interés por su estudio. Prueba de dicha relación son las correlaciones observadas entre
la masa del agujero negro y las propiedades del bulbo de la galaxia que lo alberga,
ası́ como las similitudes entre las evoluciones temporales de la formación estelar y la
actividad de acreción en las galaxias. La densidad de AGN nos da, no sólo información
acerca de la importancia relativa de la actividad de acreción en el cómputo global de
energı́a producida en el universo, sino que además impone importantes limitaciones a
la formación inicial de estructuras y en la ulterior evolución de las galaxias.
La explicación más aceptada para dar cuenta de las altas luminosidades de los
AGN es la acreción de material sobre un agujero negro central muy masivo (106−10
M , SMBH del inglés Supermassive Black Hole). La gran cantidad de energı́a generada
por la caı́da de materia en el potencial gravitatorio de un agujero negro supermasivo
hace que el disco de acreción se encuentre a una temperatura de hasta 105 K, lo que
da lugar a la intensa emisión en el óptico y ultravioleta caracterı́stico de los núcleos
activos. Esta radiación acaba emergiendo en todo el espectro electromagnético (particularmente en rayos X) debido a procesos de reprocesado que ocurren cerca del disco
de acreción.
El corazón del AGN formado por el agujero negro y el disco de acreción, se encuentra
rodeado por regiones de nubes de material responsable de las propiedades espectroscópicas. En las nubes de la llamada región de las lı́neas anchas (BLR, del inglés Broad Line
Region) se originan lı́neas con un perfil ensanchado (anchuras a mitad de altura de
& 1000 km/s) debido a las elevadas turbulencias que adquiere el material por su proximidad al SMBH. Situadas a un radio mayor se encuentran las nubes de material
responsable de las lı́neas estrechas (NLR, del inglés Narrow Line Region). Mientras
RESUMEN
xxvi
que los perfiles ensanchados son propios de los núcleos activos de tipo 1, los perfiles
más estrechos son observados tanto en espectroscopı́a de AGNs de tipo 1 como en tipo 2.
Miller & Antonucci (1983) introdujeron el modelo unificado para tratar de explicar
las diferencias observadas en el espectro de los distintos tipos de galaxias activas. La
primera explicación que se dió a la ausencia de lı́neas anchas en el espectro de AGN de
tipo 2 fue simplemente que estas galaxias carecı́an de BLR. Sin embargo, el descubrimiento de lı́neas anchas en el espectro de luz polarizada de la galaxia Seyfert NGC 1068
de tipo 2 hizo cambiar de forma radical esta idea, y sentó la base de los actuales modelos de unificación, que explican la ausencia de lı́neas anchas en el espectro de los AGNs
de tipo 2 como debida a efectos de orientación con respecto a nuestra lı́nea de visión:
intrı́nsecamente serı́an el mismo tipo de objeto que los AGNs de tipo 1, pero observados
con inclinaciones diferentes.
A pesar de que los modelos de unificación sólo pueden ser probados en detalle en el
universo local, con la instrumentación de la que se dispone hoy en dı́a (HST, Spitzer,
Chandra, XMM-Newton) el estudio de los AGN a alto redshift (desplazamiento al rojo)
hace posible plantearse la validez del modelo unificado a distancias cosmológicas (hasta
un redshift 3 o superior). Los AGN en particular constituyen un laboratorio privilegiado para estudiar la evolución de las galaxias debido a la extrema luminosidad de sus
núcleos. Ası́, son numerosos los trabajos sobre núcleos activos en campos profundos
con el objetivo de caracterizar sus propiedades observacionales y contrastarlas con lo
estudiado en el universo local. Uno de los principales problemas de los cartografiados
profundos de galaxias activas es la fracción de AGN oscurecidos (principalmente de
tipo 2) frente a los no oscurecidos. Si éste porcentaje varı́a dependiendo del tipo de instrumentación utilizada ya en el universo local, esto se hace crı́tico cuando se trata de
cuantificar este cociente a alto redshift. La principal limitación en estudiar la población
de AGN, en especial los oscurecidos, es que pueden no aparecer como tales en muchos
cartografiados.
La forma más eficaz de buscar signos de actividad nuclear en galaxias cercanas es
mediante espectroscopı́a en el óptico o en el infrarrojo cercano. El método de clasificación espectral más utilizado en el óptico es el diagrama de Baldwin, Philips y Terlevich,
conocido como diagrama BPT (Baldwin et al., 1981). Este diagrama está basado en la
comparación de la intensidad de diferentes lı́neas de emisión, lo cual proporciona información sobre el tipo de continuo ionizante que las produce. De esta forma, es posible
distinguir entre fuentes con un continuo de naturaleza estelar, donde el gas está ionizado
por los brotes de formación estelar, y fuentes con un continuo tipo ley de potencias con
xxvii
fuerte componente en rayos X, como el que se genera por la acreción de material sobre
un SMBH, caracterı́stico de los AGN. Las lı́neas de emisión de los espectros integrados
desafortunadamente pueden estar tan diluı́das por la luz de la galaxia subyacente que
pueden no detectarse con suficiente señal/ruido, o bien los cocientes de lı́neas se ven
afectados de tal forma por otros procesos (en particular la formación estelar) que la
clasificación de los objetos resulta incorrecta. Ası́ pues, en el óptico una parte de las
galaxias clasificadas como normales pueden ser realmente AGN oscurecidos o de baja
luminosidad, lo cual debe ser tenido en cuenta a la hora de censar la población de este
tipo de núcleos activos.
Gracias a su alto poder de penetración, los rayos X duros producidos muy cerca
del agujero negro supermasivo en AGN (2-10 keV) deberı́an ser el rango de energı́as
más fiables para detectar actividad de origen no estelar en galaxias. En las galaxias
donde no se tiene una visión directa del sistema en acreción (i.e. AGN de tipo 2),
parte de la radiación emitida es oscurecida por el gas y polvo que circunda el corazón
del AGN. Esta absorción de los rayos X, en mayor medida de los blandos (<2keV)
que de los duros, hace que un porcentaje no despreciable de AGN de tipo 2 quede sin
ser identificado a través de su emisión en rayos X, la mayorı́a de ellos, objetos muy
oscurecidos (los llamados Compton thick ).
El infrarrojo medio parece ser otra región del espectro electromagnético clave en el
estudio de AGNs. Los fotones emitidos en el óptico/UV por las regiones más internas
del AGN son absorbidos por los granos de polvo que rodean el sistema de acreción,
reemitiendolos isotrópicamente en el infrarrojo. La escasa interacción que sufren los
fotones infrarrojos (i.e. baja extinción) permite que en objetos oscurecidos se puedan
observar componentes espectrales reprocesadas pero que reflejan inequı́vocamente la
actividad en zonas cercanas al núcleo que las que observarı́amos en el óptico o incluso, en
los rayos X. El principal obstáculo de las técnicas de selección en esta zona del espectro
electromagnético es la dificultad que existe en discernir entre la emisión infrarroja
procedente de la reemisión del polvo calentado por el AGN y la debida a formación
estelar en las propias galaxias.
Aunque todos los cartografiados de AGNs, que se basan en la detección de alguna o
varias caracterı́sticas espectrales, son técnicas válidas, ninguno de ellos carece de efectos
de selección.
En esta tesis doctoral hemos pretendido abordar el estudio de muestras de AGN,
tanto a escala local como a distancias cosmológicas, con el objetivo de caracterizar los
parámetros fı́sicos esenciales en dichas galaxias activas y determinar ası́ la naturaleza de
RESUMEN
xxviii
las distintas muestras de núcleos activos que fueron seleccionadas en distintas bandas
del espectro electromagnético.
En la primera parte de esta tesis, que incluye los Capı́tulos 3 y 4, hemos seleccionado
una muestra de galaxias activas mediante la obtención de cocientes de lı́neas de emisión
en el óptico con la finalidad de caracterizar, por un lado, la naturaleza de las galaxias
activas ópticamente clasificadas y, por otro lado, la de aquellas galaxias cuyos cocientes
de lı́neas son consistentes con regiones de alta formación estelar, pero que por otro
lado muestran inequı́vocamente la presencia de un AGN cuando se estudian en rayos
X. Esta pesquisa concluye con el estudio de otro subgrupo de galaxias activas, las
llamadas Seyfert 1 de lı́neas estrechas (NLS1, del inglés Narrow Line Seyfert 1). Estos
objetos comparten algunas de las caracterı́sticas espectrales en el óptico de los AGN de
tipo 1, pero presentan lı́neas de Balmer más estrechas (FWHM<2000 km/s), además
de una emisión muy intensa en Fe II y O I. En rayos X las NLS1 tienen propiedades
que también las diferencian de los AGNs no oscurecidos, como un marcado exceso
de emisión en rayos X blandos (0.5-2 keV) y variabilidad en escalas de tiempo muy
cortas. Estas propiedades distintivas en rayos X se atribuyen principalmente a sus
altos cocientes de Eddington (Lbol /LEdd ∼1-10) y a las bajas masas de sus agujeros
negros con respecto a los de los AGNs de tipo 1 estándar (MBH ∼ 106−7 M ). Las
condiciones extremas en las que viven estos objetos pueden arrojar nuevas pistas sobre
los parámetros fundamentales que dirigen la actividad nuclear.
En la parte final de la tesis, Capı́tulo 5, hemos hecho uso del rango infrarrojo mediocercano. Debido a la reemisión en el infrarrojo de la radiación primaria del núcleo activo
se deduce que una manera eficaz de detectar el mayor número posible de AGN oscurecidos es posiblemente con observaciones en el infrarrojo medio. En esta parte de la tesis
hemos llevado a cabo un estudio sobre la eficiencia de selección de AGN absorbidos en
el infrarrojo medio con respecto a una selección en los rayos X duros.
Los objetivos de esta tesis pueden ser resumidos en los siguientes puntos:
1. Estudiar los desacuerdos entre las propiedades de rayos X de una muestra de
galaxias con lı́neas de emisión estrechas y su clasificación óptica basada en el
diagrama de diagnóstico BPT.
2. Caracterizar la naturaleza de aquellas galaxias cuya emisión óptica es consistente
con un origen térmico (formación estelar) pero que sin embargo, sus propiedades
en rayos X manifiestan la presencia de un núcleo activo.
3. Determinar las distintas componentes espectrales responsables de la emisión a
lo largo del espectro electromagnético (desde el óptico hasta los rayos X) de una
muestra de NLS1, con el objetivo de entender porque algunos de ellos no presentan
xxix
un exceso de emisión en rayos X blandos y otros carecen de emisión Fe II en el
óptico, ası́ como estudiar cualquier dependencia con otros parametros (Lbol /LEdd ,
Hβ FWHM).
4. Comparar la distribución de la absorción intrı́nseca de una muestra de AGNs cuya
emisión en el infrarrojo medio-cercano es una ley de potencias, con la distribución
de una muestra seleccionada en los rayos X duros.
5. Determinar si éste método de selección de AGNs (núcleos activos cuya emisión
en el infrarrojo es una ley de potencias) es eficaz en seleccionar AGNs absorbidos
(oscurecidos).
Para alcanzar los objetivos planteados anteriormente nos hemos valido de numerosas herramientas: fotometrı́a, espectroscopı́a, ajuste espectral en rayos X, ajuste de
distribuciones espectrales de energı́a, etc. No obstante, los tres estudios estan enfocados
principalmente hacia la explotación del rango de rayos X por ser una de las bandas de
energı́a clave en la identificación de AGNs oscurecidos.
La estrategia adoptada para el primer estudio (la naturaleza de una muestra de
galaxias cuya clasificación en el óptico difiere de la de rayos X) consistió en seleccionar
una muestra de galaxias con lı́neas estrechas en el espectro óptico. Los cocientes de
estas fuertes lı́neas de emisión estan ı́ntimamente relacionadas con la fuente de ionización pudiendo ası́ determinar si esta tiene un origen de carácter no estelar o no. En los
diagramas de cocientes de lı́neas se observa cómo se establece una clara diferencia entre
los AGN y las galaxias con alta tasa de formación estelar. Ası́, estas últimas presentan
siempre unos cocientes entre las lı́neas de alta excitación y las de recombinación mucho
más reducidos. Estos diagramas son de gran importancia debido precisamente a que
permiten diferenciar rápidamente galaxias con formación estelar frente a AGN. Por
otro lado, una luminosidad en rayos X de & 1042 erg s−1 puede únicamente generarse
mediante procesos de acreción.
Ası́, los objetos de nuestra muestra tenı́an que cumplir algunos requisitos adicionales,
como tener un espectro óptico con suficiente calidad para poder determinar los cocientes de lı́neas, además de tener detectadas las cuatro lı́neas de emisión involucradas en el
diagrama de diagnóstico BPT (Hβ, [O III]λ5700, Hα, [N II]λ6584). El último requisito
limita nuestra muestra a fuentes con un redshift máximo de 0.4. Además, los objetos
de la muestra tienen que haber sido detectados en rayos X duros. Una vez adquiridos
los datos (tanto ópticos como de rayos X), la clasificación óptica basada en el diagrama
de diagnóstico BPT se confronta con la luminosidad emitida en los rayos X duros, ob-
RESUMEN
xxx
tenida a partir del modelo del espectro en rayos X, además de cuantificar y caracterizar
otros parámetros clave. Se ha utilizado el catalogo Sloan Digital Sky Suervey (DR 7)
para los espectros ópticos, y el catálogo de fuentes de rayos X de XMM-Newton (DR
3) para los datos de rayos X.
Para el estudio de la población de NLS1 se adopta un modelo que pueda reproducir
de manera coherente la emisión desde el óptico hasta los rayos X, y se confronta la componente responsable del exceso de emisión en los rayos X blandos con los parametros
que describen el motor principal del AGN (cociente de Eddington, masa del agujero
negro, Hβ FWHM), comparándolos ası́mismo con los equivalentes en una población
tı́pica de AGN de tipo Seyfert 1.
Para el último estudio, la estrategia a seguir involucra el rango infrarrojo mediocercano, y nos hemos valido de observaciones ultra-profundas en el CDF-S (del Inglés
Chandra Deep Field South) llevadas a cabo por el observatorio espacial XMM-Newton.
Cuando se hace referencia a los diferentes métodos de selección y clasificación de AGN,
el rango infrarrojo resulta ser clave, junto con los rayos X y el óptico. La importancia
del infrarrojo en el estudio y caracterización de AGN reside en la estructura toroidal de
polvo ópticamente gruesa predicha por el modelo unificado. El polvo asociado a este toro reprocesa la radiación de alta energı́a emitida por el núcleo activo, reemitiéndola en
el infrarrojo, haciendo del infrarrojo una ventana a través de la cual es posible estudiar
las propiedades de este tipo de estructuras, ası́ como la identificación de los AGN absorbidos, posiblemente aquellos perdidos por los cartografiados de rayos X. Esto es posible
ya que la emisión térmica reprocesada por el polvo constituye la principal fuente de continuo en el infrarrojo medio, siendo también una contribución importante al infrarrojo
cercano en las galaxias activas. La metodologı́a propuesta selecciona en el infrarrojo
una muestra de objetos tal que su densidad espectral de energı́a en este banda es una
ley de potencias. Además de estar detectados en los cuatro filtros del detector IRAC a
bordo del observatorio Spitzer, los objetos de la muestra deben ser lo suficientemente
brillantes como para tener un espectro de rayos X con la suficiente calidad (&8σ) como
para medir su absorción intrı́nseca. Una vez seleccionada la muestra y adquiridos los
datos espectrales, el criterio de seleción en el infrarrojo adoptado para este estudio se
pone frente a la distribución de la absorción intrı́nseca y la fracción de AGN absorbidas.
Las conclusiones que resultan de este trabajo de tesis son las siguientes: en primer
lugar, muchos de nuestros resultados ofrecen nuevo soporte observacional al modelo
unificado. En segundo lugar, encontramos que el diagrama de diagnóstico BPT pierde
aproximadamente un 15 % de AGN. Además, estos núcleos activos, extraviados por el
xxxi
diagrama BPT, son compatibles con ser todos ellos galaxias Seyfert 1 de lı́neas estrechas, i.e. NLS1. En tercer lugar, caracterizamos en detalle la emisión, desde el óptico
hasta los rayos X (∼10keV), de las NLS1, dos de las cuales carecen de una de las propiedades inherentes a este tipo de objetos, el llamado exceso de emisión en los rayos
X blandos y otras dos no presentan emisión de Fe II en la banda óptica. Se deduce de
este trabajo que la población de NLS1 y Seyfert 1 normales no son poblaciones disjuntas, estando estas cuatro fuentes a medio camino entre ambas poblaciones. Por último,
presentamos datos observacionales de AGN detectados en rayos X y en el infrarrojo
medio-cercano que nos permiten especular sobre la posibilidad de que una selección en
el infrarrojo medio-cercano pudiera aportar una fracción mayor de AGN absorbidos.
Los resultados presentados en esta tesis podrı́an extenderse a través de las siguientes
lı́neas:
1. Comparación con otras muestras. A pesar que las muestras estudiadas tanto en
el Capı́tulo 3 como en el Capı́tulo 5, son estadı́sticamente representativas, ambos
estudios deberı́an repetirse para otras muestras seleccionadas de una manera más
homogénea. Esto harı́a posible obtener conclusiones acerca de la población global
de NLS1 que son clasificadas como galaxias con formación estelar por un lado,
ası́ como determinar la validez del método de selección de AGN oscurecidos en el
infrarrojo, por el otro.
2. Comparación con otros métodos de selección. Para estudiar la población de AGN
oscurecidos nos hemos centrado únicamente en fuentes detectadas en ambas bandas espectrales, infrarrojo y rayos X. Este trabajo podrı́a extenderse considerablemente con el estudio de la absorción de una muestra de galaxias seleccionadas
en el infrarrojo sin detección en los rayos X.
3. Incrementar la muestra de NLS1. Para que nuestras conclusiones sean estadı́sticamente válidas, deberı́amos recopilar una muestra mayor de NLS1 sin exceso de
emisión en los rayos X blandos. Resulta también importante estudiar la posible
relación entre la variabilidad y el exceso de emisión en los rayos blandos. Estas investigaciones son necesarias para obteener una visión global acreca de la relación
entre las NLS1 y las galaxias Seyfert de tipo 1.
RESUMEN
xxxii
Summary
“... a taxonomic system,
based not necessarily on the most easily observed parameters
but rather on the most physically relevant quantities
and some theoreticl picture, is required
in order to make substantial progress in our interpretation of AGN”
Goodrich, R.W.
Deep X-ray surveys have shown that around 80% of the cosmic X-ray background
is due to resolved extra-galactic X-ray sources, the bulk of which are Active Galactic
Nuclei (AGN). Active Galactic Nuclei are among the most fascinating objects in the
Universe. In the heart of these objects, large amounts of energy are released and they
are impressive manifestations of supermassive black holes, which in turn are probably amongst the most intriguing objects of astrophysics. Apart from the supermassive
black hole and an accretion disk surrounding it, warm dust and gas are important constituents of AGN. The nuclei are thought to play an important role in the formation
of galaxies, and in the formation of cosmological structures.
A seemingly ubiquitous feature of AGN is their prodigious X-ray emission. This
means that X-ray surveys are an efficient way to detect large numbers of AGN, despite
the fact that only a few percent of the bolometric luminosity of AGN is emitted in the
X-ray band. Progressively deeper blank field X-ray surveys have detected higher and
higher sky densities of point sources, the vast majority of which are AGN that generate
the cosmic X-ray background. However, there is an apparent paradox: the detected
AGN in the 1-10 keV energy range do not reproduce the spectrum of the cosmic X-ray
background in the 1-30 keV energy range. The much harder spectrum of the cosmic
SUMMARY
xxxiv
X-ray background in that range means that heavily obscured AGN are thought to emit
the bulk of the hard X-ray background, although the major contribution at the soft-Xray background come from AGN of type 1. To compute the summed AGN spectrum
some assumptions on population parameters are mandatory, such as the evolution of
the type 2/type 1 ratio, the hard X-ray luminosity function, as well as the intrinsic
absorption distribution. In the local Universe, around 80% of the nuclei of Seyfert
galaxies show substantial obscuration in their X-ray spectra. These obscured AGN are
expected to outnumber unobscured AGN by a ratio of 4-5:1, or even higher at higher
redshift z ∼2-3. Although the detected fraction is significantly lower, 2-4:1; the exact
value is still not well constrained as the observational estimates on this fundamental
parameter are significantly biased. This might be because these sources have low luminosities, or because they are predominantly at higher redshift and have thus escaped
the shallow hard X-ray surveys. Although the latter scenario seems to predict a lower
fraction of obscured AGN versus unobscured AGN by a ratio of 2-3:1, a significant
fraction of the obscured AGN population remains undetected, and many questions are
still open. Are absorbed AGN more/less common in the local Universe than at earlier
times? Are the obscured AGN intrinsically similar systems to unobscured AGN (as
suggested by the unification schemes), or are they actually very different objects? Did
the peak in obscured AGN activity occur at the same epoch (z ∼ 2) as the peak in
unobscured quasar activity? The properties of individual AGN in the local Universe
can be studied in detail because of their brightness. However, if we are to understand
the high-z AGN population, particularly the obscured objects, then large numbers of
faint AGN must be studied. For this we require to study deep X-ray surveys collecting
large AGN sample of distant AGN.
The prime interest of this inquiry lies in ascertaining the nature of galaxy samples
selected at different wavelength bands: optical, infrared and X-ray. The main objective
of this work is to determine, on the one hand, whether both optical and infrared criteria
are effective at selecting type-2 AGN (i.e. absorbed AGN with an intrinsic absorbing
column density NH > 1022 cm−2 ) versus X-ray selected surveys; and draw, on the other
hand, a comprehensive and consistent picture of a rare, but more interesting population
of Seyfert 1 with narrower lines, i.e NLS1, where the otherwise ubiquitous soft-X-ray
excess as well as the Fe II optical emission are not detected in the observed data for
some optical- and X-ray-detected sources.
This dissertation begins with an introduction to the field of AGN and the methods
that have been used to detect AGN focusing on type 2 Seyfert galaxies (Chapter 1). It
is meant to provide background information on these objects and to convey the motiva-
xxxv
tion for the research presented in this thesis. More detailed disquisitions on AGN can
be found in the books by Robson (1996), Peterson (1997) and Osterbrock & Ferland
(2006), as well as in Beckmann & Shrader (2012). This thesis work is based mainly on
observations made with the XMM-Newton observatory, and complemented with data
obtained at other optical and infrared facilities (Chapter 2).
Chapter 3 starts with a study of a large sample of Narrow Emission Line Galaxies (NELGs) from the Sloan Digital Sky Survey (SDSS) with an X-ray counterpart
detected by the XMM-Newton. The main goal for this inquiry is understanding the
physical cause of why a fraction of galaxies exhibit optical line spectroscopy diagnostics compatible with star formation, yet have X-ray properties that are indicative of
an AGN. It must relate to the demographics of AGN, black hole growth, and galaxy
evolution. In contrast with previous works, the main finding of our work is that a
NLS1 core can be identified in a large fraction of X-ray luminous galaxies but optically
diagnosed as starforming, being ∼25% of hard X-ray-selected AGN. This work leads
to study a population of NLS1 galaxies in order to achieve a comprehensive picture
of this kind of galactic nuclei and to understand their role in the AGN framework. In
Chapter 4, we have studied their broadband SED from optical to X-rays, to understand
the lack of the soft-X-ray component, as well as the missing Fe II optical emission in a
small fraction of NLS1.
Chapter 5 is focused on the IRAC criteria of Donley et al. (2012) to evaluate
the effectiveness on selecting type 2 AGN in the infrared. While absorbed AGN are
rather elusive in surveys of distant galaxies (including those at X-ray energies), the midinfrared waveband has the advantage that the AGN’s primary emission is isotropically
re-emitted in the mid-infrared. We found that the IR power-law method is efficient in
finding X-ray-absorbed sources at any AGN luminosity.
Finally, in Chapter 6, we conclude by summarizing our findings.
SUMMARY
xxxvi
CHAPTER 1
Introduction
1.1.
Active Galactic Nuclei
Astronomers have been aware of the existence of Active Galactic Nuclei (AGN)
since the begining of the 20th century, although at that time their extragalactic nature
was not known. It was not until the work of Seyfert (1943), that the study of these
emission line objects began. The term ‘Active Galactic Nucleus‘ refers to the existence
of energetic phenomena in the central region of galaxies which cannot be solely due to
star formation. Their luminosities range from the nuclei of some nearby galaxies emitting about 1040 erg s−1 to distant quasars emitting more than 1048 erg s−1 . The huge
amount of emitted energy is widely spread over the entire electromagnetic spectrum,
often peaking in the UltraViolet (UV), but with significant luminosity in the X-ray
and InfraRed (IR) bands. Most of this energy output is of non-thermal (non-stellar)
origin. In comparison, normal galaxies have bolometric luminosities . 1042 erg s−1 and
the bulk of their luminosity is emitted in the visible or IR band, essentially produced
by stars. All these properties are indicators of powerful physical mechanisms acting
at the centers of active galaxies producing such highly energetic phenomena in a very
compact region. Clearly, nuclear processes in the cores of stars cannot account for
the enormous AGN energy output, although sometimes an AGN at the cosmological
distance does appear star-like (i.e. QSO). A fundamental feature of most AGN is the
strong time-variability on time scales of years and sometimes on time scales of days,
hours, or even minutes, observed in their optical/UV/X-ray emission, implying that
the spatial scale of the main engine of AGN is within the order of light-days. Interpretation of the variability is still open to discussion at some level, but some basic ideas
are widely accepted: mass accretion rate onto the SuperMassive Black Hole (SMBH)
is likely a fundamental parameter driving the AGN variability.
INTRODUCTION
2
The different observational properties of the AGN arise, to a large degree, from
their intrinsically anisotropic geometry and radiation pattern, from absorption as well
as from relativistic effects. Further empirical evidence for the existence of an accreting
SMBH has accumulated over the years. As extremely luminous and distant objects,
AGN are also unique probes of the Universe at early stages and thus useful cosmological
tools.
1.2.
The AGN paradigm
A variety of AGN types and properties can be found in the literature, but all of them
are believed to share the same basic phenomena at their centers. In the standard model
of AGN, the accretion disk surrounding the SMBH is formed by the cold and warm
matter close to it. Dissipative process heat up the accretion disk and transport matter
inwards and angular momentum outwards. A corona of very energetic electrons forms
above the accretion disk and inverse-Compton scatters photons up to X-ray energies.
The first notably distinct observational feature of AGN is the presence of emission
lines in the optical and UV spectrum. Such features arise in ionised gas regions further
away from the SMBH: the Broad Line Region (BLR) and the Narrow Line Region
(NLR). A large fraction of the AGN’s radiation may be obscured by molecular gas and
dust surrounding the BLR and the central engine, the so-called torus. In the standard
Unification model, the torus is the main ingredient which blocks both the BLR and the
primary X-ray continuum as we will discuss later. In what follows, the AGN’s principal
components of this currently best-accepted standard scheme will be discussed briefly.
A visual representation of the AGN paradigm is presented in Figure 1.1.
1.2.1.
The central engine
Nowadays it is widely accepted that the heart of an AGN is an accreting SMBH
at the center of its host galaxy. Throughout the text, we refer to SMBH to designate
black holes with masses estimated to be greater than approximately 106 M . A black
hole is an extreme object predicted by Einstein’s general relativity, the name of which
suggests that is totally invisible. It was proposed that if an object’s mass is so huge
that its escape velocity exceeds the speed of light c, then no light will be observed from
it, i.e. totally black to all types of detectors. Once the black hole is formed, only three
properties will be able to be known1 : mass (MBH ), charge (QBH ) and spin (a). A key
parameter of a black hole is its gravitational radius (rg = GMBH /c2 ), which serves
1
According the value of these parameters the black hole can be classified into the following four
types: a Schwarzschild black hole (QBH = 0, a = 0), a Kerr black hole (QBH = 0), a ReissnerNordstrom black hole (a = 0) and a Kerr-Newman black hole
1.2. The AGN paradigm
3
low power
high power
radio-loud (RL) AGN
BL Lac
FSRQ
BLRG,
Type I
QSO
BLRG
jet
NLRG,
Type II
QSO
NLRG
reﬂected
absorbed
radio-quiet (RQ) AGN
Seyfert 2
d
mitte
trans
red
scatte
dusty absorber
accretion disc
electron plasma
black hole
broad line region
narrow line region
Seyfert 1
Figure 1.1: Artist’s conception of the AGN paradigm in the standard model. According to the unified scheme, the type of AGN depends on the viewing angle, on
whether or not the AGN produces a significant jet emission, and on how powerful the
central engine is. Image credit: Marie-Luise Menzel
as a natural unit of distance of the gravitational field around it. Whilst the event
horizon defines the surface for which no electromagnetic wave or particle can escape
(this is Rs = 2rg , for a non-rotating Schwarzschild black hole), the last stable circular
orbit defines the region within which no stable circular orbit can exist (risco = 6rg ,
for a non-rotating Schwarzschild black hole). The infall of material onto a black hole
converts gravitational potential energy into kinetic and thermal energy, resulting in
strong radiative emission over the entire waveband from radio to hard X-rays. The
efficiency in the conversion of the accreted mass into energy, i.e. the fraction of mass
that is converted into radiation, can be described by the accretion efficiency . Thus,
INTRODUCTION
4
if the mass accretion rate is Ṁ , then the radiated power is
L = Ṁ c2
(1.1)
Estimates of accretion efficiency in luminous AGN are in excess of 0.01 (see for example
Ballo et al., 2012). However, synthesis models for the cosmic X-Ray Background (XRB)
which is contributed mainly by AGN call for an average of of the order of 0.1 or larger.
The basic idea of accretion process is that during the in-fall of matter a decrease of
potential energy of a fluid element has to be compensated with an increase of kinetic energy (energy conservation). If part of the kinetic energy becomes internal energy (which
means an increase of the gas temperature), there are processes that can dissipate this
internal energy and emit radiation. Note that, the continuum emission in the AGN core
can be understood as the signature of the accretion disk (see Section 1.4), and thus a
superposition of many black body spectra of a continuous range of temperatures. In a
simple model the luminosity of the central object can be estimated assuming a stationary, spherically symmetric fully-ionised accreting flow mainly composed by hydrogen.
The effects of the radiation pressure due to Thomson electron scattering can become
important on the accreting gas when there is copious radiation being produced. In this
scenario the accreting flow has an upper limit, known as the Eddington accretion rate
ṀEdd , where the gravitational force inwards equals the continuum radiation force outwards. Thus, the maximum luminosity allowed is the so-called Eddington luminosity,
LEdd ' 1.3 × 1046
M
[erg s−1 ]
108 M
(1.2)
and this is an upper limit for a stationary source, i.e. if L LEdd , the radiation pressure stops accretion. The ratio between actual bolometric luminosity and Eddington
luminosity is called the Eddington ratio (i.e. λEdd = L/LEdd ), which represents the
relative importance between radiation pressure and gravity. Therefore, assuming that
an AGN is accreting at the Eddington limit, the minimum mass required for a given
luminosity L is
M∼
= 8 × 105
LEdd
44
10 erg s−1
M
(1.3)
This implies minimum masses for accreting SMBH: for sources with luminosities of
1046 − 1048 erg s−1 , a central mass of the order of 108 − 1010 M is necessary.
In a more realistic scenario, a spherical cloud of in-falling matter may not be a
1.2. The AGN paradigm
5
valid description of the flow, so the above LEdd is only a rough approximation. For
example, the accreted material may contain heavier elements than hydrogen, and the
material may not be fully ionised, thus the actual cross-section may be much bigger
than the Thomson scattering cross-section σT . Despite these caveats in LEdd , λEdd
has turned out to be one of the most fundamental parameters in understanding the
accretion process around black holes. One interesting issue concerns sources having
super-Eddington accretion flows λEdd > 1, like Narrow Line Seyfert 1 (NLS1) as a
peculiar class of type 1 AGN (see Section 1.3.1). In such cases the radiation pressure
is too high to be overcome by gravity, and so blows away at least part of the accreting
material, forming a disk wind. In Chapter 4, a set of unusual sources, which are often
super-Eddington sources, are discussed.
1.2.2.
The disk-corona system
Over the innermost part of the accretion disk, there is a corona of electrons that
are responsible for the inverse Compton scattering of the photons emitted from the
disk. While the presence of a corona is broadly accepted as an interpretation of the
hard X-ray continuum and in some cases the “soft X-ray excess” component (an extra
emission below ∼2 keV often observed in AGN, see Section 1.4), the precise nature
and geometry of this corona and its coupling to the accretion disk remains unclear
despite the amount of observational data accumulated, but very likely magnetic fields
have an important role. In particular, its thermal or non-thermal character has not
been settled. This component has an essential role in the quest of whether the same
correlations which apply for galactic black holes are also valid for AGN and viceversa,
e.g. whether the NLS1 are the supermassive black hole analogues of stellar mass black
hole X-ray binaries in their high/soft state (Dewangan et al., 2007).
1.2.3.
Torus
According to the standard model, AGN are thought to have a dusty environment
surrounding an optically and X-ray-bright accretion disk that extends from 1 to 100
pc. The simplest geometry for this material is believed to be distributed in a torusshaped structure centered in the SMBH and to contain both cold gas and dust (see
Figure 1.1). The torus provides anisotropic obscuration of the central region: viewing
the torus pole-on the observer has unobscured view of the accretion disk and its hard
radiation; on the other hand, the central engine’s radiation is blocked from direct view
when observing the torus edge-on. In such cases the radiation is reprocessed by the
dust via absorption and re-emission, typically in the IR band where both pole-on and
edge-on should show very similar spectra.
INTRODUCTION
6
A variety of torus models for the infrared emission of AGN have become available in
the literature over the last decade. They include radiative transfer models using smooth
or clumpy dust and hydrodynamic models (e.g.
Fritz et al., 2006, Nenkova et al.,
2008, Schartmann et al., 2008). However, the torus dynamical origin, and especially
its vertical support, presents a serious challenge which has not yet been settled (for an
overview of different AGN torus models see Hoenig, 2013).
1.2.4.
Broad and Narrow Line Regions
The BLR is a region of clouds of gas where broad permitted lines observed in the
optical/UV band arise, presenting significant turbulent motions which emerge the gravitational potential of the SMBH. The deep gravitational potential is responsible for the
dynamical broadening of the observed lines, whose velocities span from 1000 km s−1
up to 30000 km s−1 . The size of the BLR can be estimated by reverberation mapping
of the broadened lines (Dietrich et al., 1994) and ranges between ∼ 0.01 and ∼ 0.1
pc. The lack of broad forbidden lines means that the BLR typical density has to be
> 108 cm−3 , so collisional de-excitation prevents atoms to emit forbidden lines.
The narrow permitted and forbidden emission lines come from clouds of gas with
smaller turbulence than the ones in the BLR (a few hundreds of km s−1 ) and with lower
densities (103 -106 cm−3 ). A narrower width and lack of variability led early on to the
conclusion that they emanated from a region that was much more distant (∼100 pc)
and kinematically separated from that of the broad lines.
1.2.5.
Relativistic jet
Finally, a very important component of the AGN phenomenon is a relativistic jet
formed by the emission of very energetic particles spiraling around strong magnetic
fields along the poles of the disk. These relativistic jets are responsible for the radio
emission in AGN.
1.3.
AGN taxonomy and Unified model
The taxonomy of AGN is complex and somewhat confusing because the fundamental
physical differences between different types of AGN are not clear. However, there are
some generally-recognized AGN classes based on radio emission and optical spectral
properties which are summarized in Table 1.1 and discussed below. It must be noticed
1.3. AGN taxonomy and Unified model
7
Table 1.1: Classification of AGN based on radio emission and optical spectral properties.
RadioLoudness
AGN type
subType
X-ray
Broad
Narrow
Obscured
Balmer Line
Balmer Line
< 10%
3
3
> 90%
7
3
NLS1
< 10%
3
3
RQ
LINER
3
7
3
QSO2
type-1
7
3
3
.............................................................................................
type-1
7
3
3
quasar3
type-2
3
7
3
1
RL
FSRQ
7
3
3
blazar
BL Lacs
7
7
7
Seyfert
type-1
type-2
1
Radio-loud objects represent a small percentage, ∼10-20%, of all AGN.
2
Radio-quiet type-2 QSO are very hard to identify.
3
Within quasar population, a small minority (∼5-10%) of these sources are strong radio sources.
that the X-ray properties of the objects are normally not taken into account in this
classification scheme.
Any model for AGN must be able to explain the great variety of AGN types and
their properties. Since the discovery of AGN, evidence has been accumulating that
their emission is not isotropic and the reasons for the anisotropy have been mainly
attributed to obscuration by dust or gas and relativistic beaming. As a consequence,
the belief grew that the large variety of AGN types resulted from a family of intrinsically similar objects in different orientations with respect to the observer’s line of sight.
The models with this underlying scenario became known as the unified schemes. The
current interpretation of the Unified model is given in the Section 1.3.2.
1.3.1.
The AGN zoo
The first rough division arises from the comparison of the radio flux to the optical
flux, according to the parameter called radio-loudness. One definition calls an object
Radio-Loud (RL) when
RL = log
f5GHz
fB
≥1
(1.4)
where f5GHz is the monochromatic radio flux at 5 GHz and fB is the monochromatic
optical flux in the B band, centered at the wavelength λ =4400Å. An object which is
Radio-Quiet (RQ) is, then, not necessary radio-silent. RQ sources represent the bulk
of all AGN (∼90%). In the samples used in this thesis, little attention is paid to their
radio properties, and RL objects are likely a small fraction of them.
INTRODUCTION
8
An alternative classification, independent from the previous one and valid for RL
and RQ AGN, is made according to the optical spectroscopy which is described below.
One has to keep in mind that the AGN phenomenon was first defined based on observational properties (i.e. in the optical band). Here we describe briefly the different
AGN types most relevant for this thesis as they are usually classified in the literature.
•
Seyfert galaxies
Seyfert (1943) realised that there is a class of sources similar to NGC 1068, whose
host galaxy has a compact and bright nucleus emitting strong, high-ionisation lines,
including several forbidden lines. The hydrogen lines are broader than the forbidden lines, like oxygen [O II]λ3727, [O III]λλ4959,5007, and nitrogen [N II]λλ6548,6584.
Khachikian & Weedman (1974) realised that there are actually two sub-types of Seyfert
galaxies depending on the visibility of broad component in the permitted lines, namely
type 1 and type 2 (hereafter Seyfert 1 and Seyfert 2, respectively). The intermediate
types of Seyfert (such as 1.5, 1.8 and 1.9) were afterwards introduced by Osterbrock
(1981), according to the presence and relative strength of a broad base in the Hα and
Hβ lines.
Seyfert galaxies are the most common class of AGN observed in the nearby Universe.
The identification as a Seyfert galaxy is nowadays based on the spectral signature of
the AGN core: it qualifies as such if the optical spectrum shows highly ionised emission
lines, unable to arise via ionisation from stars. In the continuum emission, a superposition of the host galaxy and the AGN core are observed. In Seyfert 2 galaxies
the AGN core is usually less dominant with respect to the surrounding galaxy than
in Seyfert 1 objects. This is one reason why it is generally more difficult to pick up
type 2 galaxies based on their optical spectra. The spectrum of both classes appears to
lack the features of stellar absorption lines, and often exhibits (specially in Seyfert 1)
a non-thermal continuum which further distinguishes them from the stellar spectra.
As we will see in Section 1.5, one of the most used methods to select Seyfert galaxies
is based on highly-ionised emission line ratios in the optical/UV band. In Chapter 3 one
of the most used AGN selection method based on optical spectroscopy will be checked
against their X-ray emission properties.
•
NLS1
NLS1 galaxies are a particular class of AGN with a diversity of properties. In
order to frame the scope of Chapter 4 we need to review the phenomenological def-
1.3. AGN taxonomy and Unified model
9
initions and properties of NLS1. Throughout the text, we use the term Broad Line
Seyfert 1 (BLS1) to designate Seyfert 1 galaxies with hydrogen line widths larger than
2000 km s−1 . In the local Universe, type 1 AGN are dominated by BLS1s spanning a
wide range of bolometric luminosities (44. log (Lbol /erg s−1 ) .47), black hole masses
(6. log (MBH /M ) .9) and Eddington ratios (-2. log (Lbol /LEdd ) .1). These objects
show broad recombination emission lines with typical widths of the order of several
thousands of km s−1 . In this framework, NLS1 galaxies are a peculiar class of Seyfert 1
with optical spectral properties similar to those of BLS1, except for having narrower
(but still broader in comparison with forbidden emission lines) Balmer lines and strong
optical Fe II emission which from part of the classification criteria of Osterbrock & Pogge
(1985). The classical and the most commonly used definition of NLS1s (Goodrich, 1989)
relies on phenomenological criteria:
- FWHM of the Hβ line < 2000 km s−1
- [O III]λ5007/Hβ ratio < 3
The criterion on [O III]/Hβ< 3 appears to be fulfilled by all AGN that do not suffer
severe extinction, as an indicator of a non-stellar ionizing mechanism (Véron-Cetty
et al., 2001), and therefore does not separate BLS1 from NLS1.
The vast majority of NLS1 also display strong Fe II features, and that has been
also used as a distinctive criterion to identify NLS1. As far as the presence of Fe II
multiplets is concerned, these are also present in almost all type 1 AGN from the UV
to the infrared -and it can also appear in the polarized flux of type 2 AGN when a
hidden broad-line region can be seen- (Boroson & Green, 1992, Rodrı́guez-Ardila et al.,
2000).
Therefore it is primarily the first property, on the width of the Hβ emission line, what
determines whether an AGN can be considered as a NLS1 candidate, as opposed to
a BLS1 or to a type-2 AGN (Osterbrock & Pogge, 1987). Although the ratio R4570
-defined as the ratio between the flux of the Fe II multiplets and the flux of the Hβ
emission line (see Eq. 3.5 in Chapter 3)- used to quantify the strength of the Fe II emission, is expected to be larger than 0.5 for NLS1 galaxies, some groups have pointed
out that NLS1s might not be universally strong iron emitters, but faint Hβ sources
(Véron-Cetty et al., 2001), and some of them could display very faint or non-detectable
Fe II emission (Zhou et al., 2006). Once again, this criterion does not appear to be a
universal feature of all individual NLS1.
The most extreme characteristics of the NLS1 class are seen in the X-ray domain
(Gallo et al., 2006). The 2-10 keV spectral slope is usually steeper in NLS1 than in
INTRODUCTION
10
BLS1 (Brandt et al., 1997, Leighly, 1999). A strong and variable soft excess emission
below ∼1 keV (George et al., 2000, Turner et al., 1999) is more frequently present than
in BLS1 (Leighly, 1999). These strong soft X-ray emission could be associated with
high Eddington ratios, while the extreme variability on short-time scales observed in
the soft X-rays seems to be related to the comparatively low black hole masses found
in these AGN rather than to the possibly high Eddington ratios (Ai et al., 2011, Dewangan et al., 2008).
As seen so far, neither optical nor X-ray properties usually associated to NLS1 are
shared by all the members of this population, nor are they exclusive of this class of
sources. The NLS1 and BLS1 classes are not completely distinct, for instance VéronCetty & Véron (2006) noticed a continuity between their optical spectral properties.
We note that accepting that the broad line region is virialized and that its radius
depends on the optical continuum luminosity at 5100 Å as rBLR ∝ L∼0.5 (Kaspi et al.,
5100Å
4
2000), it is easy to find that MBH /M ≈ Lbol /LEdd (f /592.5)2 F W HMHβ
, where f is a
factor of order unity that depends on the unknown geometry and kinematics of the
BLR (v = f ×FWHM). Therefore, selecting sources with FWHM of Hβ <2000 km s−1
imposes a maximum limit on the black hole masses of ∼ 3×107 M when Lbol /LEdd = 1
and the typical value f = 0.75 are assumed1 . An upper boundary on MBH also implies
an upper limit for the bolometric luminosity, even when allowing super-Eddington
sources.
•
quasar/QSOs
Quasars were first identified in 1950s during the first systematic radio surveys. The
optical counterparts of some of these strong radio sources were stellar in appearance
rather than a diffuse galaxy, and thus, they were so-called originally quasi-stellar objects (QSO). Although the first quasar was discovered due to its radio emission, most
quasars found in optical surveys were not present in radio surveys. In fact RL quasars
represent at most a few tenths of the total QSO population. For historical reasons, some
astronomers use the term “QSO” for the radio-quiet quasar class, reserving “quasar”
for radio-loud objects. It is clear that quasars form another class of AGN lying at high
redshift with higher luminosity than Seyfert galaxies. The latter are lower luminosity
AGN with an absolute B-magnitude MB > −23 mag, with quasars defining the higher
luminosity AGN according to an optical definition. The original distinction between
Seyfert AGN -where a bright nucleus could be resolved at the center of a host galaxyand a QSO -where the host galaxy was not detected- does no longer apply, since con1
Correcting by the broad line profile (Gaussian or Lorentzian) can result in one order of magnitude
higher black hole masses
1.3. AGN taxonomy and Unified model
11
temporary high-spatial resolution optical imaging resolves the host galaxy around most
QSOs as well. In addition, the luminosity division between Seyfert galaxies and quasars
is rather arbitrary.
•
LINERs
There is also a family of AGN whose emission does not dominate over the host
galaxy, the Low-Ionisation Nuclear Emission-line Galaxies (LINERs). They were first
identified by Heckman (1980). The optical spectrum of a LINERs is similar to that of
a Seyfert 2 as both types of AGN show low ionisation narrow emission lines. However,
the Seyfert 2 can also exhibit strong highly ionised species as well, for example the
[O II]λ3727 to [O III]λλ4959,5007 line ratio is ≈1 for LINERs whereas it is ≤ 0.5 for
Seyfert 2 galaxies.
•
Other objects
Some AGN -BL Lacs and Flat Spectrum Radio Quasars (FSRQ)- show very unusual
spectra and peculiar properties, such as featureless spectra, strong optical variability
on very short time scales, and strong and variable polarization. These objects are
collectively called blazars. BL Lacs, however, lack the strong emission lines observed
in FSRQ, suggesting a fundamental difference between the two classes in spite of their
similar peculiar properties. The unification scheme interprets both FSRQ and BL Lacs
as AGN observed at a small viewing angle so that the continuum emission is dominated
by the jet (see Figure 1.1).
RL AGN can also be classified as Fanaroff-Riley (FR) class I and class II -FRI and
FRII, respectively- according to their radio morphology. Whilst FRI are brighter at
their centers (decreasing outwards), FRII are edge-brightened, i.e. their radio surface
brightness profiles increase dramatically outwards as they reach the end of the extended
structures. The two FR classes are also broadly separated in terms of their radio
luminosity, the class I being less luminous. Their optical spectrum can show a spectrum
similar to a Seyfert of type 1 or type 2 and accordingly, they are called Broad Line
Radio Galaxies (BLRG) and Narrow Line Radio Galaxies (NLRG), respectively.
1.3.2.
Current interpretation
The first orientation-based unified scheme model was proposed by Antonucci (1993).
In the most simplified picture, there are basically two classes of AGN: RQ and RL. For
each type a range of luminosities is observed and the observational differences between
the various spectra (such as type 1 and type 2) is explained as a continuum transition
in the viewing angle from 0 to 90◦ (see Figure 1.1). A subsequent revision by Urry &
INTRODUCTION
12
Padovani (1995) explained the unification between the subgroups of RQ and RL AGN.
According to the Unified model, which uses the anisotropy owing to the dusty torus,
sources viewed through a gas and dust-free line of sight, i.e. face-on, are recognised
as type 1 AGN, while those viewed edge-on are classified as type 2 AGN. The central
engine is directly visible only for the type 1 AGN. Unified models have been supported
by the fact that X-ray absorption seems to increase when transiting from type 1 to
type 2 AGN (Bassani et al., 1999, Nandra & Pounds, 1994, Risaliti et al., 1999, Smith
& Done, 1996). The discovery of hidden type 1 nuclei at the center of type 2 AGN
when observed in polarized light (see for example Antonucci & Miller, 1985) gives more
reliability to the unified models. Hidden broad lines have also been found in type 2
AGN when observed in the IR band, less affected by reddening, and the inferred optical depths from the IR correspond to the expected optical extinction (Ward et al., 1991).
If we put aside for the time being the differences between RL versus RQ AGN, the
Unified model predicts a distinction between the various types of sources based solely
on the viewing angle. The variety of the AGN population is then assumed to be caused
by the different levels of obscuration/absorption in the line of sight. There are, however,
several optical and X-ray observational results that do not seem to fit into this simple
scheme. There are type 2 AGN which do not harbour hidden type 1 nuclei even when
observed in polarized light (see for example Bianchi et al., 2008, 2012b, Tran, 2001,
2003). The proposed explanation is that the power of the central engine is not large
enough to sufficiently illuminate the BLR, i.e. weak BLR emission or small-mass black
hole. X-ray unabsorbed type 2 AGN have also been found, like NGC 3147. Several
explanations have been considered to explain such discrepancy: the source might be
Compton-thick for which only the scattered light is visible; or the central engine might
be obscured in the optical by an ionised dusty absorber, with little effect on X-rays; or
perhaps the BLR is too weak or even does not exist. Similarly, there are also X-ray
absorbed type 1 AGN which are occasionally found in X-ray surveys (see for sample
Carrera et al., 2004, Scott et al., 2012, and references therein). A possible explanation
for such AGN could be that since an important fraction of the X-ray photoelectric absorption may arise fairly close to the nucleus, where dust cannot survive, the intrinsic
column density derived from X-rays would be larger than the one derived from optical
reddening (see Eq.1.6). The simplest version of the unified schemes has faced different
challenges in the last years, among others: the anticorrelation between the fraction of
absorbed sources and luminosity observed in hard X-ray AGN surveys (e.g., Simpson,
2005); the differences found in the luminosity functions of different AGN types (e.g.,
Bianchi et al., 2012a); how the NLS1 class can be understood in the framework of the
simple unified scheme.
1.4. The AGN spectral energy distribution
13
All these examples highlight the importance of adding to the simplest model scheme
the dependence on the AGN intrinsic properties such as the SMBH mass, its spin, the
accretion rate in terms of the Eddington ratio, the bolometric luminosity or even the
host-galaxy morphological type.
1.3.3.
X-ray obscuration
The unification scheme for AGN assumes that the differences between Seyfert galaxies of type 1 and type 2 results from the amount of absorbing material close to the central engine, encountered by the line of sight. Thus, a similar distinction between type 1
and type 2 should be evident in X-rays, based on the intrinsic absorption measurement in the X-ray band (E < 10 keV). X-ray absorption is measured as an equivalent
column density of hydrogen NH in the line of sight, in atoms per cm2 . As a dividing
line, a hydrogen column density of NH = 1022 cm−2 is often used to separate X-ray
unabsorbed objects (very often type 1), on one hand, and X-ray absorbed objects (very
often type 2), on the other. Most, but not all, Seyfert 1 have an intrinsic absorption below this threshold, while most, but not all, Seyfert galaxies with an intrinsic absorption
above this value are optically classified as Seyfert 2. Recent Chandra and XMM-Newton
studies have revealed that the widely supported correlation between optical obscuration and X-ray absorption fails for about 10-20% of the X-ray selected AGN (Corral
et al., 2011, Mateos et al., 2005, 2010, Page et al., 2003, Perola et al., 2004, Tozzi et al.,
2006). In principle, high gas-to-dust ratios in the X-ray absorbing gas (perhaps due to
dust sublimation close to the central X-ray source, Granato et al., 1997), or large dust
grains could give rise to high levels of X-ray absorption (Maiolino et al., 2001), without
much optical reddening, that could explain this lack of one-to-one connection.
1.4.
The AGN spectral energy distribution
AGN are prolific emitters of radiation that can span over the entire electromagnetic
spectrum from radio to gamma rays. In what follows, the underlying emission seen in
each of the available observational windows will be discussed briefly. Figure 1.2 shows
an schematic representation of the Spectral Energy Distribution (SED) of different
types of AGN, and a possible physical source for each emission component.
1.4.1.
Radio continuum
The radio emission from AGN consists of compact and extended components both
of which produces (often self-absorbed) synchrotron emission, which is usually well
INTRODUCTION
14
(i) Average SED of different types of AGN
(ii) Sketch of the contributions to the average SED from the AGN’s
components
Figure 1.2: Schematic representation of the spectral energy distribution of AGN.
(I) The average SED see in different types of AGN. (ii) The RQ spectrum can be
divided into three major components: the infrared bump from reprocessing of the UV
emission by the dust in a range of distance from the central engine; the BBB, which is
directly related to the accretion disk; and the X-ray region, which can be interpreted
as the reprocessed disk emission. Image credit: Koratkar & Blaes (1999) and James
Manners’s PhD thesis, respectively.
1.4. The AGN spectral energy distribution
15
fitted by a power law.
1.4.2.
Infrared emission
The AGN continuum shows a broad IR bump: from the radio continuum, there
is a sharp increase in the sub-millimeter band (known as the sub-mm break), with
emission peaking at around ∼60µm and decreasing again down to ∼1µm. Whilst
in RQ sources this sharp increase towards 60µm extends over ∼5-6 decades, for RL
sources it only takes ∼2 decades (see Figure 1.2). Depending on AGN type, the IR
emission may consist of thermal and/or non-thermal components. In RL objects, the
same synchrotron emission process producing the radio continuum is the predominant
source of IR radiation. In Seyfert galaxies and other low-luminosity AGN subtypes the
situation is more complex considering the presence of multiple thermal components:
• thermal radiation from dust and molecular gas associated with the obscuring torus
heated by the optical/UV/X-ray radiation of the central engine (dashed-red line
in Figure 1.2, Efstathiou & Rowan-Robinson, 1995, Granato et al., 2004);
• thermal dust continuum associated with star formation is believed to be the origin
of the far-IR/sub-mm emission, λ ∼100µm (dotted-purple line in Figure 1.2);
• additional line emission emanating from molecular, atomic, and ionic species originated in the NLR which can be seen in high-resolution spectra superimposed on
the IR continuum.
The IR continuum of galaxies dominated by AGN emission typically can be reproduced by a power law (fν ∝ ν α where α is the spectral index) although with a variety
of slopes (from α = −0.05 up to α = −3). Figure 1.3 shows the mean 1-10 µm SED for
Seyferts of type 1 and type 2, with the latter having steeper SEDs. A near-IR excess
is responsible for flattening the SED of type 1 nucleus, which probably comes from the
contribution of hot dust in the directly-illuminated faces of the clouds in the torus, as
well as from the direct AGN emission.
1.4.3.
Optical/UV emission
The main feature of the optical/UV spectra of AGN is the Big Blue Bump (BBB)
that dominates the SED at wavelengths shorter than ∼4000Å and peaks around ∼1000Å.
This strong and broad feature is attributed to some kind of approximately thermal emission in the range around 104 -105 K, usually the emission from a heated accretion disk
(dot-dot-dashed blue line in Figure 1.2).
INTRODUCTION
16
(i) Seyfert 1
(ii) Seyfert 2
Figure 1.3: IR SEDs of Seyfert 1 (left) and Seyfert 2 (right) normalized at 8.74 µm.
Image credit: Ramos Almeida et al. (2011).
Superimposed to this continuum, the optical spectrum shows strong broad/narrow
permitted/forbidden emission lines.
Among the most prominent emission lines in
the 3000-9000Å range (Hα, Hβ, Hγ, [N II]λλ6548,6584, Li λ6507, [O I]λλ6300,6363,
[O III]λλ4959,5007, He IIλ4686, etc.), we have used in this thesis Hα, Hβ, the [O III]
and [N II] doublets. There are also many other important and strong emission lines in
both UV and IR wavebands which we have not used because they are not covered by
the Sloan Digital Sky Survey (SDSS) spectra that we used.
These lines have complex profiles in AGN, often including a broad base and a narrow
core component, and sometimes also an intermediate component. These are revealed
when fitted by multiple Gaussian profiles (Hu et al., 2008, Mei et al., 2009, Zhu et al.,
2009). The Full Width at Half Maximum (FWHM) has been used as an estimate of
the line width, despite the fact that it remains controversial which estimated measure
of the line width best correlates with the mass of the SMBH. The observed FWHM
for the broad emission lines of AGN have typical values of ∼5000 km s−1 , reaching in
some objects widths of ∼30000 km s−1 , while the FWHM of the narrow emission line
profiles are of a few hundred km s−1 .
Optical reddening
There is no doubt of the presence of dust in AGN, nor that it modifies the optical/UV spectrum which often appears flatter (redder) than the predicted optical slope
of an accretion disk spectrum. Dust grains between the emitting region and the observer scatter and absorb photons in a wavelength-dependent way, according to the size
and composition of the grains. It is observed that the dust grains in the InterStellar
1.4. The AGN spectral energy distribution
17
Medium (ISM) scatter more blue optical light than red, resulting in an absorbed optical
spectrum that is reddened.
While Galactic reddening has been measured accurately, the intrinsic reddening
for external galaxies (especially the galaxies hosting AGN) is more difficult to quantify. The most widely used method to correct for total reddening along the line of
sight is based on the relative strengths of the lower-order Balmer lines, Hα and Hβ,
the so-called Balmer decrements. Assuming case-B recombination and optically thin
photoionised plasma (Osterbrock, 1989), the intrinsic Balmer decrement corresponding
to the Hα to Hβ ratio is Rint = 2.87. This value is found in H II regions, which have
typical densities ∼ 104 cm−3 and temperatures ∼ 104 K. However, given the different
conditions within the NLR and BLR different values are expected: 3.1 for the NLR as
a result of the presence of a partly ionised region in which the collisional excitation of
Hα becomes important (Gaskell & Ferland, 1984); for the BLR, where the density is
so high that collisional, optical-depth and radiative-transfer effects become important
(Baldwin, 1997), the Hα/Hβ ratio can be 10 or higher, and cannot be used as an indicator of reddening in BLR, as it is in AGN NLR or H II regions.
Throughout this thesis, the narrow [O III] emission line flux, which is used as an
isotropic indicator of nuclear activity, is corrected for reddening as follows
c
f[O
III]
= f[O III]
R
Rint
2.94
(1.5)
where R is the measured Balmer decrement and Rint is the intrinsic Balmer decrement
which is assumed to be equal to 3.1 in the NLR. Assuming a standard gas-to-dust ratio,
the total Hydrogen (HI+H2 ) column density, NH , can be derived. Adopting galactic
gas-to-dust ratios, Bohlin et al. (1978) presents a relation between absorbing column
density and dust reddening as follows
NH [cm
−2
22
] = 1.04 × 10
log
R
Rint
(1.6)
which can be then compared to the one obtained by X-ray spectroscopy.
1.4.4.
X-ray emission
The X-ray band is one of the most useful for AGN observations. This is not only
because the AGN X-ray emission contributes a significant fraction (∼2-5%) to the total emitted energy, but because X-rays arise from very close to the SMBH and there
is little contamination from the host galaxy at these wavelengths. Given a black hole
INTRODUCTION
18
of mass 108 M , the gravitational radius rg is 1.5×1013 cm, so the light crossing time
is ∼ rg /c ∼
= 8.3 minutes; X-ray variability has been observed in AGN down to scales
of hours or less. This suggests that the X-ray emission originates from the innermost
region, close to the central SMBH. Despite that a complete understanding of the X-ray
variability -a hallmark of the AGN phenomenon- is still to come, it is believed that
the mass accretion rate onto the SMBH is likely a fundamental parameter driving the
AGN variability.
The analysis of the X-ray spectrum is often divided into the soft and hard bands, 0.52 keV and 2-10 keV, respectively. The X-ray spectrum in the hard band often exhibits
the shape of a power law with the following form:
FE = F0 × E −Γ
[photons/s/cm2 /keV]
(1.7)
where FE is the photon flux per unit energy, E is the photon energy and Γ is called
the “X-ray photon index”. However, it is more common to show the spectrum in terms
of the energy flux log EF (E) versus log E, since EF (E)(∼ νF (ν)) can be directly
compared with estimates at other wavebands SED of the X-ray spectrum. The X-ray
photon index is predicted to be only weakly dependent on the temperature and optical
depth of the corona (e.g., Haardt & Maraschi, 1991).
To understand the origin of the X-ray emission, it is necessary to invoke the commonly accepted disk-corona model. Inverse Compton scattering of the soft optical/UV
disk photons by the relativistic electrons in the corona gives rise to an up-scatter to
X-ray energies: this is the primary X-ray emission, which can be broadly described as
a power law with a typical slope of Γ ∼1.9 (Mateos et al., 2005) from energies of ∼1
keV up to an eventual cut-off somewhere beyond 100 keV (see Figure 1.4).
The primary X-ray emission may undergo reprocessing by some mechanisms by
which these X-rays would be prevented from escaping unchanged from the source and
reaching the observer:
• When the primary X-ray power-law continuum illuminates cold material (T ≤
106 K, neutral and/or ionised in both the torus and/or in the host galaxy or even
the accretion disk itself), the incident photon can be scattered, can disappear
by Auger effect, or can be reprocessed into a fluorescent line photon. Thus,
the opacity of the material for both continuum and line radiation is the result
of photoelectric absorption and electron scattering. The competition within the
surrounding accretion flow between multiple electron (down-) scattering of high-
1.4. The AGN spectral energy distribution
19
energy photons and photoelectric absorption of low-energy photons leads to a
“Compton reflection hump” in the 5-50 keV range, and a series of emission lines
at lower energies (see Figure 1.4, top panels): low energy X-ray photons are
more likely to be absorbed than scattered, whereas high energy photons are more
likely to be Compton back-scattered. Compton electron recoil reduces the backscattered above energies of several tens of keV (see also Figure 1.4, bottom left
panel). Due to the high abundance and fluorescence yield of iron, a significant
fraction of the incident photons with energies above the iron K-shell edge at 7.1
keV give rise to iron K-shell fluorescent photons with an energy of 6.4-6.9 keV
depending on the Fe ionisation state, regularly observed in AGN X-ray spectra
with enough signal-to-noise ratio (S/N).
• When an incident power law is absorbed due to the photoelectric absorption by
neutral material, the resulting spectrum contains a number of absorption edges
corresponding to the different ionisation potentials for each species within the
absorbing material. The effective cross-section of this process depends on the
abundances in the absorbing material. The Hydrogen-equivalent column density,
NH , is used to account for the contribution of the various atomic species for a
given abundance pattern. Column densities over 1020 cm−2 are needed to produce an observable effect in X-rays at redshift zero. For higher redshifts, as the
emission and, consequently, the intrinsic absorption, are shifted towards lower
energies and eventually below the X-ray observational window, larger and larger
column densities would be needed in order to be detectable.
Those sources showing column densities greater than 1/σT ≈ 1.5 × 1024 cm−2 (σT
being the Thomson scattering cross-section) are named Compton-thick, while
AGN showing absorption below that limit are named Compton-thin. If the column density reaches values above ∼ 1024 cm−2 , all the primary X-ray emission
below 10 keV is absorbed and only reflected and scattered emission can be observed, which usually amounts only to a few percent of the direct emission, the
so-called Compton reflection emission component. The X-ray emission is thus
largely suppressed making these sources extremely hard to detect in the X-ray
observational window (see Figure 1.4, bottom right panel). The potential to observe X-rays from Compton-thick AGN does remain by virtue of the X-ray reflection and scattering component (see for example Matt et al., 2000). In Chapter 5,
we investigate whether the revised mid-IR-based IRAC criteria (see Section 1.5)
of Donley et al. (2012) are effective at selecting Compton-thin Seyfert galaxies
approaching the Compton-thick limit.
INTRODUCTION
Figure 1.4: Top panels: X-ray reflection from a cold illuminated slab (A) and from
ionised material (B) as a function of the ionisation parameter ξ (defined as L/nr2 , with
L the bolometric luminosity, n the density and r the distance between the ionizing
source and the absorbing gas). In the former, the dotted and solid lines represent the
incident power law and the reflected spectrum, respectively. Bottom-left panel, (C):
Schematic representation of a type 1 AGN spectrum in the X-ray (black line) waveband
along with the most relevant spectral components: X-ray reflected component (green
plus red line), primary X-ray emission plus photoelectric absorption by ionised material
(pink line) and soft X-ray excess (cyan line). Bottom-right panel, (D): X-ray power
law, with an X-ray photon index equal to 2, absorbed by a range of column densities
(from 1022 cm−2 up to 1025 cm−2 , being the numbers the value of NH in log units).
Image credit: George & Fabian (1991)
20
1.4. The AGN spectral energy distribution
21
If the absorbing material is partially ionised (known as warm absorber), the absorption signature depends not only on the abundances and the column density
but also on the degree of ionisation of the material. The warm absorber will
normally be only partly ionised, so the absorption could preferentially occur at
certain energies: the material could be transparent at low and high energies but
not at intermediate energies (see pink solid line in Figure 1.4, bottom left panel),
due to primarily to the O VII and O VIII edges. If the material is very highly
ionised, there is no detectable photoelectric absorption.
• Another component observed in AGN is the so-called soft X-ray excess, defined
as the excess emission below 2 keV above the extrapolated hard (above 2 keV)
power-law emission. The origin of this excess flux is still under debate among the
AGN community. In the past, the soft excess had been often associated with the
high-energy tail of the thermal emission of the disk. It can be argued that the
temperature of the disk should scale with disk parameters, however it stays almost
constant regardless of the mass and luminosity of the AGN (Crummy et al., 2006,
Czerny et al., 2003, Gierliński & Done, 2004). Three competing physical models
have been brought forward in order to explain the soft X-ray excess observed in
some AGN: i) an additional Comptonisation component (Dewangan et al., 2007),
ii) an ionised reflection from the disk (Crummy et al., 2006), iii) or an ionised
absorption (Done & Nayakshin, 2007). Done et al. (2012) have also suggested
that the accretion disk emission may also make a significant contribution in the
soft X-ray region in some low-black-hole mass, high mass accretion rate sources,
such as NLS1.
Seyfert 1 are characterized by a single underlying power-law continuum over the
0.5-10 keV regime. In the local Universe, an average value of the photon index Γ ∼ 1.9
is observed (see e.g. Nandra & Pounds, 1994). In high-quality spectra, warm absorption
signatures are usually found in ∼50% of Seyfert galaxies. The soft-excess component is
also needed in ∼30% of them. An example of the typical spectral components observed
in Seyfert viewed face-on is shown in Figure 1.4.
In Seyfert 2 the measured X-ray photon index results to be flatter than for type 1
objects. The significant amounts of neutral absorption in their X-ray spectra make
it difficult to measure the intrinsic X-ray photon index. The soft X-ray emission is
more and more suppressed as the column density increases (see Figure 1.4, bottom
right panel). In the Compton-thick regime the high energy photons that survive the
photoelectric absorption get scattered, causing the suppression of the transmitted continuum. As the amount of absorbing material increases, the spectrum can show a
flattened (log EFE vs. log E) power-law slope (Γ ∼1) at the hard X-ray band, con-
INTRODUCTION
22
sistent with a Compton reflection component, coming either from neutral or ionised
reflection. Despite the strong photoelectric absorption in Seyfert 2 galaxies, some level
of soft X-ray excess may be detected below 3 keV. The origin of these soft X-ray excess
could be due to either a strong star-forming component or scattered emission from
the nucleus. In order to observe such soft X-ray emission it is clearly a requirement
that the scattering medium should extend beyond the bounds of the obscuration of the
molecular torus, making the soft X-ray emission a peculiar property of Seyfert 1 class.
1.5.
The search for obscured AGN
The AGN phenomenon exhibits a wide range of observational features, including
emission line profiles, variability timescales, blue optical continuum and infrared excess,
as well as strong X-ray emission. Thus, having a fully efficient, reliable and complete
AGN selection technique is very difficult; and it becomes a challenge when dealing with
Seyfert 2 galaxies, which are the focus of an important part of this thesis. The wide
parameter space available for the AGN paradigm makes that different selection techniques tend to find different classes of objects.
In what follows, the most used techniques to select AGN are briefly discussed,
highlighting the methods to construct samples of Seyfert 2 galaxies.
1.5.1.
In the radio band
The earliest survey to select AGN was in the radio band. Almost all luminous
extragalactic radio sources are AGN, the only contaminant at lower luminosities being
star-forming regions. Although a mere detection in the radio for most luminous objects
indicates the presence of an AGN, and about 70 per cent of radio-selected AGN are
optically passive (Urry & Padovani, 1995), radio surveys are very incomplete: less than
10-20 per cent of AGN have powerful radio-emitting jets (White et al., 2000).
1.5.2.
In the infrared band
Near-IR (∼1-5 µm) excess appeared in early work to offer a potential advantage
over other searching techniques. One of the best known samples of infrared selection
AGN are the 2MASS red AGN (Cutri et al., 2001) in which the design of the survey
was to look for extreme sources that are missed in optical surveys. These sources have
a colour cut of J-Ks >2, i.e. very red near-IR continuum that can either be explained
by emission from very hot dust or by strong reddening of the near-IR continuum. However, the principal bias of the colour selection J-Ks >2 is that it excludes most of the
1.5. The search for obscured AGN
23
known AGN (Barkhouse & Hall, 2001).
The re-emission of the hiding dust which is heated by the strong radiation field of
the AGN, should be seen in the mid-IR (∼5-25 µm) as continuum emission (see e.g.
Haas et al., 2003, Meisenheimer et al., 2001). Thus, the advantage of techniques using
isotropic properties in the mid-IR is that type 1 sources, as well as reddened type 1
and type 2 AGN (i.e. heavily obscured type 1 objects according to the unified model),
should be found. Thus, IR-colour based surveys can potentially trace the elusive obscured accretion missed by hard X-ray surveys (Daddi et al., 2007, Fiore et al., 2008,
2009, Georgantopoulos et al., 2008, Severgnini et al., 2012). For example, it has been
claimed that objects showing excess emission at ∼24µm over that expected from star
formation might host heavily obscured or even a Compton-thick AGN (Fiore et al.,
2008, 2009).
Moreover, AGN searches in the mid-IR have been conducted in the last few years
(Alonso-Herrero et al., 2006, Barmby et al., 2006, Donley et al., 2012, 2008, 2007, Lacy
et al., 2007, 2004, Martı́nez-Sansigre et al., 2007, 2006, Park et al., 2010, Stern et al.,
2005). At these wavelengths AGN in general are dominated by a phenomenological
power-law continuum emission from hot dust heated by the intense radiation field. The
major issue in the mid-IR band however is to distinguish AGN from starburst galaxies
that also have considerable mid-IR and far-IR emission, but usually with cooler colours
and different spectral shapes (Lacy et al., 2004, Laurent et al., 2000). The mid-IR
band has the added benefit of being less affected by dust extinction and sensitive to the
highest redshift sources. The Spitzer Space Telescope and Wide-field Infrared Survey
Explorer (WISE) promise a sensitive survey for both type 1 and type 2 luminous AGN.
On the one hand, Spitzer/IRAC identification of AGN typically require all four bands
of that instrument, with data out to 8 µm, to differentiate AGN from high-redshift
(z.1.3) massive galaxies, since both have red observed colours from 3 to 5 µm (Donley
et al., 2012, 2007, Lacy et al., 2004, Stern et al., 2005). This is because, when the AGN
is sufficiently luminous compared to its host galaxy, the superposition of blackbody
emission from the AGN-heated dust fills in the dip in the galaxy SED and produces a
red, power-law thermal continuum across the IRAC bands (see Fig. 1 in Donley et al.,
2012). On the other hand, WISE suffers less pronouncedly from such contamination
and can identify AGN with just the three shortest (3.6, 4.16, 12.0 µm), most sensitive
channels (Mateos et al., 2012, Stern et al., 2012). Both IR-colour selections have similar
scopes.
For this thesis, we have used the revised IRAC selection criteria given by Donley
INTRODUCTION
24
et al. (2012) to select Seyfert galaxies in order to explore the efficiency of the IR powerlaw selection in finding absorbed X-ray sources (see Chapter 5). The Spitzer power-law
selection chooses sources whose IRAC SED follows a power law over a wide range of
slopes. A problem commonly encountered when studying AGN properties based on
IR observations is the significant contribution from the host galaxy to the near- and
mid-IR bands (Alonso-Herrero et al., 1996, Franceschini et al., 2005, Kotilainen et al.,
1992). However, this should not be a problem because we have used ultra-deep X-ray
observations (> 3Ms), the majority of the detected AGN being high-luminous objects
where the host galaxy contribution should be swamped by the nuclear activity, as we
have previously adduced.
1.5.3.
In the optical/UV band
The optical colour selection technique was first applied to AGN in the 1960’s, based
on the inference that quasars often have a larger UV excess than the hottest stars.
The optical colours of an AGN deviate from the starlight from their host galaxy as
a function of the power-law index, redshift, and the fraction of the total light in the
non-thermal continuum. Thus, it is possible to define a colour region that efficiently
selects objects with unusual colours. This kind of techniques is essentially trained on
known objects, and the region of the quasars in the colour space so obtained is chosen
to reject the vast majority of stars and thus have a highest efficiency in quasar selection, but it pays no attention to completeness (e.g. Richards et al., 2002). One of the
problems of the optical-colour selection is that the central engine must be either bright
enough to outshine the stars in the photometric extraction region or have sufficiently
different colours to be recognizable, thus stellar dilution being the main problem to find
Seyfert galaxies by a colour selection. If the relation between the host galaxy and the
nucleus properties can be modeled to remove the starlight, the selection effects could
be quantifiable. However, this relation is only known at low redshifts. Indeed, the
reddening curve in AGN is not well-known, so it is not all clear how to de-redden the
AGN spectrum to derive reliable optical colours.
Since practically all AGN vary on all timescales from hours to decades, variability
can be used as a selection criterion for AGN (e.g. Sarajedini et al., 2003). Given that
variability selection makes no prior assumption about spectral properties, it should be
a powerful method for detecting any kind of AGN, such as LLAGN in which the host
galaxy emission is dominating, as well as AGN with unusual spectral properties. However its completeness correction becomes difficult because the AGN selection function
is unknown.
1.5. The search for obscured AGN
25
As an alternative, there is the optical emission line diagnostic technique (see some
examples in Figure 1.5) which selects AGN on the strength of the UV/optical emission
lines. Emission from an AGN central engine does not resemble that from an ensemble
of normal stars. The signature of emission lines in galaxies conveys information about
the ionising source such as its power and nature, as well as the geometry, physical conditions and chemical composition of the gas among others. The presence of emission lines
Figure 1.5: Spectroscopic diagnostic diagrams: in the left panel, log ([O III]/Hβ)
versus log ([N II]/Hα), in the middle panel log ([O III]/Hβ) versus log ([O I]/Hα),
and in the right panel log ([O III]/Hβ) versus log ([S II]/Hα). The dashed lines
separating starforming galaxies, LINERs, and Seyfert galaxies are derived from Kewley
et al. (2001). In the left panel the region between the dot-dashed and dashed lines are
populated by composite galaxies, whose spectra contain significant contributions from
both AGN and star formation. Finally, the red box is the region covered principally
by LINERs.
in the spectrum of a galaxy means it is forming stars or it harbors an AGN. We note
that only spirals and some irregular galaxies form stars, while elliptical galaxies have
no ionized gas clouds. Thus, the majority of the galaxies with emission lines should
be able to be classified depending on their physical nature, like starforming or hosts
of AGN. The way AGN are usually identified among the large samples of galaxies in
massive optical spectroscopic surveys is almost invariably through their emission lines.
Baldwin et al. (1981) were amongst the first to introduce robust emission line diagnostic
ratios able to distinguish between starforming process and AGN activity. Line diagnostics based on emission line ratios were later used by Veilleux & Osterbrock (1987) to
derive a semi-empirical classification scheme to distinguish between some class of AGN
INTRODUCTION
26
and starforming galaxies. Kewley et al. (2001) used these same diagrams to derive a
purely theoretical classification scheme. Kauffmann et al. (2003) and Stasińska et al.
(2006) later extended this scheme. The underlying idea is that the emission lines in
normal starforming galaxies are powered by massive stars, so there is an upper limit on
the intensity ratio of the collisionally excited lines with respect to the recombination
lines (such as Hα or Hβ). Unlike this mechanism, photons from AGN extend to higher
energies and therefore, they induce more heating, implying that optical collisionally
excited lines will be brighter with respect to recombination lines than in the case of
ionisation by massive stars only. Thus, it is possible to construct emission line ratios
that efficiently select objects with unusual ratios. Figure 1.5 shows the most standard
optical diagnostic diagrams showing the classification scheme by both Kewley et al.
(2001) and Kauffmann et al. (2003).
In principle, this kind of methods has no false detections, however its completeness
is really difficult to evaluate, since the S/N depends on the widths of the lines. The
main limitations are,
• The observed lines depend on the redshifts of the objects, making the effective
band smaller by (1+z).
• The absorption lines due to stars should be removed to obtain high S/N spectra
of the nucleus of every object in a complete sample. This process requires extreme
care and good data.
• Emission lines can be hidden, diluted or masked by stellar light from the galaxy
(see for example Caccianiga et al., 2007, Page et al., 2006, Severgnini et al.,
2003, Trump et al., 2009). This is particularly important in low-luminosity AGN,
where the observed star formation and AGN components may be comparable.
Therefore, the evidence for AGN activity in optical spectra could be weakened in
a significant number of objects.
• In other cases the regions producing the narrow and broad emission line features
in AGN, may be obscured by dust in the host galaxy and/or in the nuclear regions
(see Civano et al., 2007, Comastri et al., 2002, Iwasawa et al., 1993, Rigby et al.,
2006). Such objects would either be optically classified as star-forming galaxies
or HII regions, on the basis of emission line diagnostics.
Among all available emission line ratios, for this thesis work we have used [O III]/Hβ
versus [N II]/Hα (Baldwin et al., 1981, often referred to as the BPT diagram) because it
is one that most clearly distinguishes between AGN and star-forming galaxies (Stasińska
et al., 2006). In the BPT diagram (see Figure 1.5) the AGN occupy the region above
1.5. The search for obscured AGN
27
and to the right of the borderline, whilst starforming (or HII regions) are found to
the bottom left of the parameter space (Kauffmann et al., 2003, Kewley et al., 2001).
The reason to use these emission-line ratios is that the lines used to compute each
ratio are very close together in wavelength, and consequently the line ratios are largely
insensitive to dust reddening. The [O III]λ5007 (hereafter [O III]) emission is assumed
to be an unbiased and orientation-independent measure of the ionizing flux from the
AGN; in contrast with the X-ray emission coming from the compact nucleus that may be
strongly attenuated in the plane of the dusty torus. The [O III] emission line luminosity
is therefore often used as an isotropic indicator of AGN activity, this diagnostic diagram
being much more suitable for our purpose.
1.5.4.
In the X-ray band
Similarly to the radio emission, compact, luminous X-ray emission is an almost
certain indicator of the existence of AGN and does not need additional observations
at other wavelength bands to confirm it. The standard threshold employed to pick up
nuclear activity is the presence of a hard X-ray luminosity L2−10keV > 1042 erg s−1 . This
is a very conservative limit, based on the fact that very few local star-forming galaxies
have higher hard X-ray luminosities, with a few possible exceptions such NGC 3256
(Lira et al., 2002). The hard X-ray luminosity is a good indicator of AGN activity,
because the X-ray spectra of star-forming galaxies are typically softer than those of the
AGN.
Like other methods it has selection effects. The main effects (and/or contaminants)
are:
• Whilst very luminous AGN can be unambiguously identified in almost any energy
band, AGN become progressively more challenging to identify at lower luminosities when their emission may be equal to, or even less than, that from the host
galaxy. Lower luminosity X-ray sources with hard spectra, can be weaker AGN
or arise from high mass X-ray binaries, or ultra-luminous X-ray sources.
• At the present time, there are very few known AGN that are not luminous Xray sources. Such class of AGN (in particular Compton-thick sources) are X-ray
quiet. At present, this is thought to be due to the high column densities in these
objects, but work is still proceeding on this.
Hence there is a problem in using X-rays alone to distinguish between low luminosity or obscured AGN, and a population of high mass X-ray binaries in starforming galaxies. One way to distinguish between these possibilities would be to use
INTRODUCTION
28
L2−10keV ∼ 1041 erg s−1 as the empirical limit in X-ray luminosity of starburst galaxies,
but it is much less conservative than 1042 erg s−1 . While galaxies with lower X-ray
luminosities than 1042 erg s−1 may be consistent with star formation1 , higher X-ray
luminosities are almost all consistent with being AGN.
There is a significant difference between soft-band (0.5-2 keV), hard-band (2-10
keV), and ultra hard-band (5-10 keV) surveys: soft X-ray selected surveys can suffer
from the same extinction effects as UV/optical surveys, while in the harder bands,
absorption is a much smaller effect. Thus, hard X-ray selection finds, in addition to
the Seyfert 1 and quasars, a population of absorbed type 2 AGN.
1.5.5.
Selection Effects
As we have seen so far, every method can be effective to select AGN, but is biased
towards one type of AGN on account of the underlying aim for which it was designed.
Most of the techniques used to search for AGN rely on the fact that the spectra of
AGN differ in shape from those of stars and normal galaxies, especially at high luminosities. In general, the results of searches at various wavelengths demonstrate that a
considerable fraction of the AGN population is missed due to observational bias and
the techniques suffer from considerable contamination by other astronomical objects,
the latter being the major problem of the AGN surveys.
Selection effects can be grouped in two classes:
• Colour/Brightness dilution by host-galaxy starlight. The dilution reduces the
equivalent widths of the lines, as well as the visibility of the non-thermal radiation
in the spectral bands where stars are luminous (i.e. from optical to near-IR).
This effect makes it very difficult to select Seyfert galaxies at z > 0.5 from nearIR to optical wavebands. In contrast, X-rays have minimal dilution by hostgalaxy starlight, this being irrelevant for luminosities above ∼ 1042 erg s−1 in
X-ray surveys.
• Obscuration/Absorption. About 70 per cent of all AGN are expected to have
large column densities of gas and dust (& 1022 cm−2 ) involving an effective optical
reddening of E(B − V ) > 5. With such an absorption, a fraction of direct line-ofsight X-rays are effectively destroyed via the combination of Compton scattering
and photoelectric absorption, even at 10-200 keV, if it rises up to ∼ 1024 cm−2 ,
i.e. Compton-thick regime. Naturally, E(B − V ) > 5 means that UV/near-IR
1
It is possible to find low luminosity X-ray sources which are weak AGN, e.g. LINERs or heavily
obscured AGN
1.6. Motivation of this thesis
29
Table 1.2: Summary of different selection methods to find AGN, illustrating the
principal advantages, disadvantages and selection effects.
WaveBand
Underlying
Principle
radioloudness
Radio
Advantage
Disadvantage
Contaminants
detect optical-silent AGN
avoid RQ AGN
SF/HII
Selection
Effect
.......................................................................................................................................
reddening
mid-IR excess
supply photometric-z
avoid LLAGN
SF
obscuration
IR
avoid low-LX AGN
reddening
colour-colour diagram
detect high-z AGN
SF
host galaxy emission
dilution2
.......................................................................................................................................
colour-colour diagram
Optical/UV
supply photometric-z
avoid LLAGN
SF
emission line ratios
supply spectroscopic-z
avoid LLAGN
redshifted lines
SF
variability
detect any type of AGN
not Completeness correction
reddening
dilution2
.......................................................................................................................................
avoid LLAGN/low-LX AGN
reddening
X-ray
L2−10keV
little contamination
X-ray bainaries, ULX1
avoid Compton-thick
obscuration
1
2
Ultra-Luminous X-ray sources
by the host galaxy
fluxes become extinguished. As we pointed at the beginning, the correlation
between high-column density and high dust extinction is not always true (such as
NGC 4151): there are sources where their X-rays are destroyed while remaining
optically visible, as well as there are sources with high optical reddening and
hardly any sign of intrinsic absorption (e.g., NGC 4151; for a review, see Comastri
& et al., 2003). If the distribution functions of each of these cases were known,
these effects could be corrected for in AGN surveys.
Regarding obscuration by the torus, there is only one privileged waveband according to the Unified model, the IR. Since optical/UV/soft X-ray emission is
re-emitted in the IR band, Seyfert 2 galaxies have low soft X-ray luminosities,
low optical luminosities, but large IR luminosities, becoming suitable targets for
IR surveys and free of this selection effect.
Table 1.2 summarizes the selection effects, as well as the advantages and disadvantages of these different techniques to select AGN.
1.6.
Motivation of this thesis
There is strong observational evidence that active galactic nuclei play an important
role in the formation and growth of galaxies (e.g. Magorrian et al., 1998). Most supermassive black hole growth takes place during an obscured quasar phase, as suggested
by the integrated energy spectrum of the cosmic X-ray background (Fabian, 1999). To
understand the evolution of galaxies and to trace the energy output due to accretion
and its cosmological evolution, mapping the history of obscured accretion is critical.
INTRODUCTION
30
Identifying complete and reliable samples of AGN has become a necessity for extragalactic surveys, whether the goal be the selection of AGN candidates or the removal
of contaminants to concentrate on star formation. Only when armed with complete
samples of AGN will we be able to determine the role of obscured accretion in the
build-up of the SMBH mass, or accurately characterize galaxy evolution.
In currently popular galaxy/SMBH co-evolution models, the differences between
obscured and unobscured AGN are no longer and uniquely described by a geometrical unification model, but can be interpreted as different evolutionary phases of the
same object. The finding that absorption is much more common at low luminosities
(and maybe at high redshift), makes this hypothesis become more plausible. In this
generic scenario, the dependence of the obscured AGN fraction with the luminosity and
redshift could be interpreted in terms of radiative (ionisation) and mechanical (winds
and outflows) feedback from the AGN. Thus, a correct and complete identification of
unobscured, obscured and highly obscured AGN at all redshifts is crucial for a detailed
understanding of the still little explored phases of the common growth of SMBHs and
their host galaxies.
Complete AGN samples are also required to test the unified scheme at cosmological
distances. Accretion rates, orientations, and intrinsic obscurations of AGN prevent
any one selection technique from reliably identifying all of them. For instance, while
current optical, UV and X-ray surveys are capable of detecting unobscured AGN, they
miss many of the obscured AGN and nearly all of the Compton-thick AGN, thought
to dominate AGN number counts at both low and high redshift.
In this thesis we compare AGN selection methods based on optical spectroscopy
and mid-IR colour selection with their X-ray emission. While no method is perfect
(in the sense that it is both efficient and reliable), we examine how they complement
each other, and the biases of each one of them. While it is relatively straightforward
to select AGN from optical multi-colour surveys due to the prominent optical/UV continua (e.g. the SDSS survey, Richards et al., 2002) or from spectroscopic samples (e.g.
Bongiorno et al., 2007), in Chapter 3 we will see that the presence of luminous X-ray
emission L2-10 keV & 1042 erg s−1 -as a reddening-independent method- becomes necessary whenever NLS1 must be selected. On the other hand, to identify the entire AGN
population independently of dust extinction, surveys in the radio, in X-rays and in the
infrared have been carried out, but one has to keep in mind that only about 10% of
the AGN are radio-loud (Urry & Padovani, 1995) and there seem to exist many X-ray
faint AGN (Wilkes et al., 2002). In Chapter 5 we will show that the revised IRAC
1.6. Motivation of this thesis
31
colour selection criterion (objects with an IR power-law spectral shape) is certainly an
absorption-independent method that becomes a powerful tool to select Compton-thin
AGN of low X-ray luminosities and also appears very efficient and reliable at high AGN
luminosities.
Summarizing, along this work we will tackle the following issues to understand
better the AGN phenomenon and explore better the parameter space covered by these
extreme sources:
1. Compare the optical classification using the BPT diagram with the X-ray emission
of a sample of narrow emission line galaxies, to study the mismatch between the
optical-and X-ray-based classifications.
2. Characterize the nature of galaxies whose optical emission line diagnostics are
consistent with star formation, but whose X-ray properties strongly point towards
the presence of an AGN.
3. Reproduce the broadband SED emission of a sample of NLS1 to understand the
lack of a soft-excess component in the observed X-ray spectrum of two of them,
as well as the lack of Fe II optical emission in a further two NLS1.
4. Explore and compare the intrinsic absorption distribution approaching the Comptonthick limit for an X-ray and IR-selected sample which has been split into two
subsamples: sources with a power-law SED in the IR and sources with a nonpower-law SED.
5. Determine whether the revised IRAC criteria of Donley et al. (2012) (objects with
an IR power-law spectral shape), are effective at selecting X-ray type 2 AGN, i.e.
sources with an intrinsic column density NH,z > 1022 cm−2 .
INTRODUCTION
32
CHAPTER 2
Data
2.1.
X-ray data: XMM-Newton
Since the Earth’s atmosphere blacks out all cosmic X-rays, only a telescope in space
can detect and study astrophysical X-ray sources. The X-ray data used for this work
come from observations performed by XMM-Newton1 , the ESA X-ray Multi-Mirror
Mission. XMM-Newton is the second cornerstone of the Horizon 2000 science program
of the European Space Agency (ESA) and was launched by an Ariane 504 on December
10th 1999.
There are three science instruments on board of XMM-Newton: the European Photon Imaging Camera (EPIC, Kendziorra et al., 1997, Kuster et al., 1999) composed by
three CCD detectors, the Reflection Grating Spectrometer (RGS, den Herder et al.,
2001) consisting of two units and the Optical Monitor (OM, Mason et al., 2001).
The three co-aligned X-ray telescopes on board have the largest effective area
achieved so far, 0.4 m2 at 1 keV and 0.05 m2 at 5 keV. On the focal plane of each
telescope there is an EPIC detector which can perform imaging over a 30 arcmin diameter Field of View (FOV). The EPIC camera is composed of three CCD detectors
based on two different technologies: two of them are based on metal-oxide semiconductor (MOS, Turner et al., 2001) and one on p-n junction (pn Strüder et al., 2001).
EPIC cameras are sensitive to photons with energies from 0.1 to 15 keV providing
moderate energy resolution (resolving power E/δE ∼ 20 − 50). The PSF of the mirror
modules determine the angular resolution (the pixel size of the EPIC cameras is much
smaller than the PSF). The accuracies on the X-ray source position are approximately
1
http://xmmssc-www.star.le.ac.uk/Catalogue/2xMM/
DATA
34
between ∼1 and ∼3 arcsec as a result of a PSF with a FWHM of 6 arcsec (the halfenergy with, in the range of 12-15 arcsec, is more often quoted to characterize the X-ray
mirror modules on board XMM-Newton).
The OM operates in the UV and the blue region of the optical spectrum (170 to 550
nm) and is co-aligned with the X-ray telescopes to allow simultaneous UV/optical/Xray observations.
2.1.1.
Data Reduction
The data used in this dissertation have been processed using the Science Analysis
System
1
(SAS, version 6.1.0) and have been analyzed using the standard software
packages (FTOOLS included in HEAsoft 5.3.1). We reprocessed the EPIC-MOS and
EPIC-pn Observation Data Files (ODFs) to obtain new calibrated and concatenated
event lists, using the SAS tasks EMPROC and EPPROC, including the latest calibration
files at the time of reprocessing. The new event files were filtered to avoid intervals of
flaring particle background, and only events corresponding to patterns 0-12 for MOS
and 0-4 for pn were used (Ehle et al., 2005). The source spectra were extracted from
00
00
circular regions, whose radii (ranging from 12 to 30 ) were chosen in each case to
optimize the S/N and to avoid the CCD gaps. The background spectra were taken in
circular source-free regions near the object, also avoiding CCD gaps. We generated the
redistribution matrices and the ancillary files (correcting for the effective area) using
the SAS tasks RMFGEN and ARFGEN. In the case of bright sources, as some of the type-2
Seyfert studied in Chapter 3, the detector can interpret as a single event all photons
falling in the same pixel during a read-out CCD cycle. As a result, the interpreted
energy of this event is registered as the sum of the individual energies of the actual
events. This is called photon pile-up. The EPATPLOT SAS task was used to evaluate
whether this phenomenon was important. This task performs pattern statistics over an
observation and compares it with the expected pattern distribution. By using that, the
pile-up fraction for the sample of Chapter 3 was estimated and the pn X-ray spectra
of all but two of the observations were found to be free from the effects of pile-up. For
these two sources this effect was reduced excising the core of the PSF2 .
2.1.2.
Spectral Analysis
In order to fit the extracted X-ray spectra, the XSPEC software was used. XSPEC
was developed at Cambridge University as a mission-independent general analysis pro1
http://xmm.esac.esa.int/sas/current/howtousesas.shtml
See http://xmm.esac.esa.int/sas/current/documentation/threads/epatplot.shtml for more information
2
2.1. X-ray data: XMM-Newton
35
gram for X-ray spectral data (Arnaud, 1996). XSPEC takes into account theoretical
models, X-ray source data, background data, and calibration data. Since X-ray spectra
are in counts per channel, a minimum number of 20 (source plus background) counts
should be selected in order to be able to use χ2 minimisation technique which requires
Gaussian statistics. Before entering the data, the X-ray spectra are grouped into a
certain minimum number of counts per energy bin.
The input spectrum for XSPEC, expressed in counts per channel C(I), does not
represent the actual incident spectrum but the detected spectrum, i.e. the monochromatic photon flux FE (E) modified by the detector response R(I, E) integrated to all
energies
Z
C(I) =
+∞
FE (E)R(I, E)dE
(2.1)
0
Since this equation cannot be inverted in an analytical way, a spectral fit is used.
A model spectrum M (E), usually described in terms of a few parameters, is selected
and a predicted counts per channel Cp(I) is computed and compared to the observed
counts C(I). The fit statistic used to compare C(I) and Cp(I) is χ2 . In this way, the
model parameters are varied so as to minimise the χ2 value of the fit,
2
χ =
X (C(I) − Cp(I))2
σ 2 (I)
(2.2)
where σ(I) is the error for channel I which is estimated by Poisson statistics. As a
rough “rule of thumb”, when the reduced χ2 (χ2 /d.o.f., where d.o.f. is the number of
degrees of freedom) is close to 1, a good fit has been achieved. To obtain the errors
in the model parameters, their values are varied until a certain ∆χ2 is reached. In
the case of errors at 90% confidence level for one varying parameter, for example, the
tabulated value for the critical ∆χ2 is 2.71.
Throughout this work, source detection was carried out on data of the three EPIC
cameras (pn, MOS1, MOS2) and on the hard 2-10 keV X-ray energy band. As the
differences between the MOS1 and MOS2 response matrices are of just a few per
cent we created combined MOS source and background spectra and response matrices.
Merged source and background spectra were obtained by adding the individual spectra.
Backscale values (size the regions used to extract the spectra) and calibration matrices for the combined spectra were obtained weighting the input data with the exposure
times. In order to use the modified χ2 minimisation during the spectral fitting, spectra
were grouped to a minimum of 20 counts per bin unless otherwise stated. The X-ray
DATA
36
spectral fitting was carried out with a joint fitting of MOS and pn spectra using the
XSPEC software (version 12.5).
2.1.3.
Spectral models
Several spectral models have been implemented in XSPEC in order to fit X-ray
spectra taking into account the physical processes that may take place. In this section
only the models that have been used in our spectral analysis are briefly summarised.
In the case of models that include redshift effects, the model parameters refer to their
rest-frame values.
There are different kinds of models according to the way they have to be used:
additive which, after convolved with the instrument response, prescribe the number of
counts per energy bin (e.g., powerlaw, zgauss, zbbody, mekal) ; and multiplicative models, which apply an energy-dependent multiplicative factor to additive models
before convolution (e.g., zphabs, zpcfabs, zwabs, redden). At least one additive
component must be included in the model, but there is no restriction on the number
of multiplicative models.
• Black body emission: zbbody
A redshifted (z) black body is an additive model that represents the emission of a
system in thermal equilibrium with the surrounding radiation field. The parameters in this case are the electronic temperature in keV, kTe , and the normalization
K,
M (E) = K
8.0525 (E (1 + z))2 dE
E(1+z)
4
kT
e
−1
(1 + z) kTe e
(2.3)
where K is the ratio between the source luminosity in units of 1039 erg s−1 and
the factor (D10 (1 + z))2 , D10 being the distance to the source in units of 10 kpc.
• Emission, hot diffuse gas: mekal
Additive model that represents an emission spectrum from hot diffuse gas based
on the model calculations of Mewe and Kaastra with Fe L calculations by Liedahl.
The model includes line emission from several elements. For our aim in Chapter 3, we used this model in the interpolate mode, where the model spectrum is
interpolated from a pre-calculated table at different temperatures. Abundances
were selected in the same way as for the zphabs model. Model parameters are:
plasma temperature (kTe , in keV), the redshift, and the normalization, H density
and metal abundance.
2.1. X-ray data: XMM-Newton
37
• Gaussian emission line: zgauss
Simple redshifted (z) additive Gaussian profiles for emission lines,
2
(E(1+z)−EL )
1
2σ 2
√ e−
M (E) = N
(1 + z)σ 2π
(2.4)
where N is the line flux (total photons cm−2 s−1 ). EL and σ are the central
energy and the line width in keV.
• Interstellar extinction : redden
Additive model for IR/optical/UV extinction E(B − V ) Cardelli et al. (1989).
The transmission is set to unity shortward of the Lyman limit. This is incorrect
physically but does allow the model to be used in combination with an X-ray
photoelectic absorption model such as zphabs to account for intrinsic absorption.
• Partial covering absorption: zpcfabs
A redshifted (z) partial covering fraction neutral absorption is a multiplicative
model whose analytical form assumes that there is neutral material surrounding
the source that partially absorbs its emission whereas the rest of it is unabsorbed.
The parameters in this case are the equivalent hydrogen column NH (in units of
1022 atoms cm−2 ) and the covering fraction (f ),
M (E) = f e−NH σ(E[1+z]) + (1 − f )
(2.5)
Abundances were selected in the same way as for the zphabs model.
• Photoelectric absorption: phabs, zphabs
Photoelectric absorption multiplicative models coming from neutral material at
redshift 0 (phabs, i.e. Galactic absorption) or redshift z (zphabs, i.e. intrinsic
absorption) have as main parameter the hydrogen column density (NH ) in units
of 1022 atoms cm−2 ,
phabs :
zphabs :
M (E) = e−NH σ(E)
M (E) = e−NH σ(E(1+z))
(2.6)
where σ(E) is the photo-electric cross-section, not including Thomson scattering.
This cross section assumes chemical abundances and ion abundances as those in
our Galaxy. Different abundance patterns are available for selection. The table
from Anders & Grevesse (1989) was selected in our case.
zwabs is an althernative photoelectic absorption multiplicative model using the
Wisconsin cross-sections. zwabs
DATA
38
• Power law: powerlaw
The powerlaw additive model corresponds to a single power law model. The
model parameters are the X-ray photon index (Γ) and the normalization at 1 keV
(K, in units of photons keV−1 cm−2 s−1 ),
M (E) = K E −Γ
(2.7)
• Thermally Comptonized continuum: nthcomp
Additive model that represents an emission spectrum from seed photons Compton
up-scattered by hot electrons. It is based on the thermally comptonized continuum model of Zdziarski et al. (1996). Typically the physical picture is that these
seed photons are quasi -blackbody or disc blackbody in shape. Either of these
shapes can be selected by par4 (0/1, respectively), both being parameterized by
a seed photon temperature of kTbb . The energy cutoff is sharper than an exponential, and is parameterized by the electron temperature kTe . This model is
a much better description of the continuum shape from thermal comptonisation
than an exponentially cutoff power law. Details of this are given in Życki et al.
(1999).
2.2.
Optical data: SDSS DR7
The SDSS1 mapped one quarter of the entire sky and performed a redshift survey
of galaxies, quasars and stars. The survey uses a dedicated 2.5m telescope (Gunn et al.,
2006) equipped with a large format mosaic CCD camera to image the sky in five optical
bands (u, g, r, i and z), and two digital spectrographs to obtain the spectra of galaxies
and quasars, which are selected in a uniform way in the imaging survey, and classified
as point or extended source. The SDSS will produce both imaging and spectroscopic
surveys having a coverage of ∼8200 deg2 , with spectra from 3800 Å to 9200 Å, and a
wavelength resolution of 1800.
SDSS DR7 is the seventh major data release of the Sloan and provides images,
imaging catalogs, spectra, and redshifts for download. It is the final data release of
SDSS-II2 . In Chapter 3 the seventh data release of the SDSS (hereafter SDSS DR7) is
used to obtain a large sample of Narrow Emission Line Galaxies (NELG). By analyzing
1
See www.sdss.org for general information
SDSS-II is an extension of the original SDSS consisting of three sub-projects: The Legancy Survey,
SEGUE and a Supernova survey
2
2.2. Optical data: SDSS DR7
39
the SDSS spectra, we can derive the parameters of the principal optical emission lines
and underlying continuum, to compute for example, the black hole mass or the relative
strength of the Fe II multiplets.
2.2.1.
Modelling the optical emission lines
Our optical spectral modelling employs linked Hα and Hβ profile fitting and the
complete optical spectral fitting. The code used is an adaptation of the IDL (Interactive Data Language) code wrote by Jin et al. (2012).
Based on current AGN emission line models, there are thought to be stratified
regions emitting different line profiles (see Chapter 1): NLR, BLR and possibly an
Intermediate Line Region (ILR). Following previous studies, several separate Gaussian
profiles are used to represent each of these emitting regions to model the Balmer line
profiles. Each of the Hα and Hβ line profiles pose distinct difficulties for the spectral analysis. In the case of the Hβ line, the permitted Fe II emission features (which
are often strong in NLS1s) and broad He II 4683 line (blended with the Hβ line) can
affect the determination of the underlying continuum and hence the Hβ line profile.
For the Hα line, there is the problem of blending with the [N II]λλ6548,6584 doublet,
improper subtraction of which may distort the intrinsic profile of Hα. Our approach,
therefore, is to fit Hα and Hβ simultaneously using the same multi Gaussian components. The assumed similarity between the intrinsic profiles of these two Balmer lines
assists in deblending from other nearby emission lines, and should yield a more robust
de-convolution for the separate components of their profile.
Fe II emission features
Fe II emission line features are fitted using the theoretical Fe II model templates of
Verner et al. (2009) which include 830 energy levels and 344035 transitions between
2000Å and 12000Å. The predicted Fe II emission depends on physical conditions such
as the micro-turbulence velocity and the hardness of the radiation field, but we use the
template which best matches the observed spectrum of I Zw 1 (Boroson & Green, 1992,
Véron-Cetty et al., 2004), i.e. the one with NH = 1011 cm−2 , micro-turbulence velocity
of 30 km/s and Fionizing = 20.5 cm−2 s−1 . High S/N spectra show that the Fe II profile
is often complex. However, for simplicity we will assume only one velocity structure
and convolve this template with a single Lorentzian profile (Véron-Cetty et al., 2004).
We fit this to the observed Fe II emission line features between 5100Å and 5600Å
DATA
40
(no other strong emission lines lie in this wavelength range) of the de-redshifted SDSS
spectra, leaving the FWHM of the Lorentzian and the normalization of the Fe II as the
free parameters. The resulting best-fit Fe II model to this restricted wavelength range
was then extrapolated and subtracted from the entire SDSS spectrum. A major benefit
from subtracting the Fe II features is that the profiles of the [O III]λ5007 lines no longer
have apparent red-wings. This is particularly important for the NLS1s, where the Fe II
emission is often strong. After subtracting Fe II, we used 2 Gaussian components to fit
the [O III]λ5007 line.
Balmer Emission Lines
After fitting the [O III]λ5007 line, Hα and Hβ line profiles are simultaneously fitted.
A simplified picture in which Balmer lines have two principal components is considered:
narrow component (from the NLR) and a broad component (from the BLR). The broad
components are represented by Gaussian profiles, whereas the narrow component is
assumed to be similar to that of [O III]λ5007. Since we do not know whether or not the
Balmer decrements are the same in these different emitting zones, the relative strengths
of different line components were not fixed, but their FWHM and relative velocity were
both kept the same. The [O III] λ4959 line was set at 1/3 of that of [O III]λ5007 from
atomic physics. The profiles of the [N II]λλ6548,6584 line doublet were also fixed to
the [O III]λ5007 line profile. For simplicity, the [S II] doublet, [O I] doublet and Li 6708
were all fitted with a single Gaussian profile each, because they are all relatively weak
lines and do not severely blend with Balmer lines. Special care was taken to separate
the narrow component of the Balmer lines from the other components considered as
accurately as possible.
2.3.
IR data: Spitzer/IRAC
The Spitzer Space Telescope (Werner et al., 2004) is the fourth and final element in
NASA’s family of Great Observatories. Spitzer was launched on the 25th of August of
2003. The observatory carries an 85-centimeter cryogenic telescope and three cryogenically cooled science instruments capable of performing IR imaging and spectroscopy in
the 3.6 to 160 µm range.
On the focal plane of the telescope there are three science instruments. Wide field,
broadband imaging is the main purpose of the Infrared Array Camera (IRAC, Fazio
et al., 2004), working in the 3.6-8 µm range, and the Multiband Imaging Photometer
for Spitzer (MIPS Rieke et al., 2004), working in the 24-160 µm.
2.3. IR data: Spitzer/IRAC
41
The near- and mid-IR source catalogue used in Chapter 5 was built from IRAC
observations. IRAC is a four-channel camera able to perform IR imaging in four bands
(3.6, 4.5, 5.8, and 8 µm) simultaneously over a 5.2×5.2 arcmin FOV. 3.6 and 5.8 µm
channels view the same telescope field, and 4.5 and 8 µm channels view a different
adjacent field simultaneously. IRAC provides diffraction-limited imaging internally but
the image quality is limited primarily by the Spitzer telescope. The angular resolution
of the four detectors is ∼1.5 arcsec.
The spectroscopic functions of the observatory are carried out mainly by the Infrared
Spectrographs (IRS Houck et al., 2004) in the 5-40 µm range. In addition, MIPS has
a low resolution spectroscopic mode between 55-95 µm.
DATA
42
CHAPTER 3
The nature of AGN missed by optical
line spectroscopy diagnostics
3.1.
Motivation
The optical and ultraviolet emission lines of galaxies are widely used to distinguish
Star-Forming (SF) galaxies from AGN. However, this type of diagnostic has some associated uncertainties, because AGN can be of low luminosity and/or heavily obscured,
and the optical emission lines may be dominated by a stellar component (see Chapter 1,
Sect. 1.5). On the other hand, and despite its limitations, X-ray emission can be used
as a reliable tracer of luminous AGN. Several well-studied examples exist where the
optical diagnostics are indicative of a SF galaxy, but the X-ray properties reveal the
presence of an AGN (see e.g. Yan et al., 2011).
In this chapter we characterize the nature of galaxies whose optical emission line
diagnostics are consistent with star formation, but whose X-ray properties strongly
point towards the presence of an AGN. Understanding these sources is of particular
importance in assessing the completeness of AGN samples derived from large galaxy
surveys, selected solely on the basis of their optical spectral properties. To address
these issues, we have characterized the optical and X-ray properties of a large sample
of NELG from the SDSS, whose X-ray emission properties are available from the Incremental Second XMM-Newton Serendipitous Source Catalogue 2XMMi-DR3 (Watson
et al., 2009) released in April 20101 . We devote special attention to those objects optically classified as SF galaxies, but with 2-10 keV luminosities in excess of 1042 erg s−1 .
1
Available from http://xmm.esac.esa.int
AGN MISSED BY THE BPT DIAGRAM
44
This chapter is structured as follows: Sect. 3.2 describes the selected NELG sample;
the optical- and X-ray-based classification are described in Sect. 3.3; optical and X-ray
properties are discussed in Sect. 3.4 and 3.5; Sect. 3.6 presents the results of the X-ray
spectral analysis; finally, Sect. 3.7 summarizes the conclusions we obtained from this
study. This study has been published in Astronomy & Astrophysics as Castelló-Mor
et al. (2012).
3.2.
Data compilation
To construct a sample of galaxies having both high quality X-ray and optical spectra
and showing prominent narrow emission lines and a lack of broad components, i.e. a
sample of NELG, we performed a cross-correlation between the 2XMMi catalogue and
the spectroscopic SDSS DR7 catalogue. Our sample selection process consisted of five
stages:
1. The BPT diagram (Baldwin et al., 1981) is the best method among all the optical
selection techniques to identify nuclear activity within a population of narrow
emission line galaxies. As mentioned in Section 1.5, one of the most important
limitations of this method is that the observed emission lines depend on the
redshift of the objects, shrinking the effective waveband. We included objects
from the SDSS DR7 catalogue in which the four emission lines involved on the
BPT diagram (i.e. Hβ, Hα, [O III], and [N II]) were detected. This selection
criteria constrains the sources redshift to z < 0.4.
2. In order to select only NELGs, we adopted an operational Hβ line width cut-off
of FWHM≤ 1200 km s−1 to reject galaxies with broad emission lines. This value
was chosen by taking into account the strongly bimodal distribution of measured
Hα FWHM values for an emission-line galaxy sample (see Figure 3.1). This would
indicate that there is a natural separation between broad and narrow line AGN.
It is known that the different emission lines in AGN originated in the same emitting region (NLR and/or BLR) have the same turbulent velocity. We have selected
our primary sample using the Hβ emission line instead of Hα because the latter
is blended with the [N II] doublet, making the velocity width estimates (σobs ) of
Hβ more reliable than those of Hα. Assuming a Gaussian line profile to model
the observed emission line, the velocity dispersion (σobs ) allows us to compute the
FWHM of each individual emission line as
√
c[km/s]
σobs [Å]
FWHM[km/s] = 2 2 ln 2
1+z
λ[Å]
(3.1)
3.2. Data compilation
45
Figure 3.1: Hα FWHM distribution for the emission-line galaxy sample given by
Hao et al. (2005). The inserted plot is the Hα FWHM distribution for narrow-line
and broad-line AGN after removing star-forming galaxies. Image credit: (Hao et al.,
2005)
where λ is the rest-frame wavelength of the emission line.
Given the typical S/N and the instrumental resolution of the SDSS spectra, we
omitted objects that had an observed FWHM (for any of the four emission lines)
smaller than ∼ 69 km s−1 , as they are likely to be spurious detections or poorly
detected lines. The Hβ FWHM measurements for each individual source were
obtained after de-convolving with an instrumental resolution1 of 69 km s−1 .
Applying these criteria, we selected from the SDSS DR7 about 150000 nearby
NELGs of different classes: type 2 AGN with a broad range of luminosities,
galaxies whose emission is not dominated by nuclear activity and are classified
as normal star-forming galaxies (or H II regions), and type 1.9 Seyferts, which
exhibit only a weak broad component of Hα (Osterbrock, 1981).
3. To identify possible X-ray counterparts, we analysed XMM-Newton observations
covering the sky positions of these NELGs. A total of 1729 SDSS source positions have some X-ray exposure time, although in the majority of cases only
upper limits to their X-ray emissivity were found. We cross-correlated this parent sample with the 2XMMi-DR3 applying a matching radius of 3 arcsec. This
radius is chosen as a compromise between allowing for some positional error, and
minimizing the probability of spurious matches (Watson et al., 2009). Figure 3.2
shows the angular distance distribution between the SDSS source and its single
1
http://www.sdss.org/dr7/instruments/spectrographs/index.html
AGN MISSED BY THE BPT DIAGRAM
46
2XMMi counterpart where the mean value is ∼1 arcsec. We considered only Xray sources with a 0.2-12 keV EPIC detection likelihood above 3σ in at least one
XMM-Newton EPIC camera.
50
number
40
30
20
10
0
0
1
2
3
r [arcsec]
Figure 3.2: Distribution of the angular separations between the SDSS-DR7 and the
2XMMi-DR3 positions for the 211 cross-correlated objects (see point 5 in Section 3.2).
4. Our first objective is to characterize the populations of NELGs by comparing
their optical classification with their X-ray properties. As we pointed out earlier,
a standard technique to identify AGN among a sample of emission line galaxies
is the hard (2-10 keV) X-ray luminosity. We have adopted an empirical X-ray
luminosity threshold of LX > 1042 erg s−1 (see Section 1.5 for an extended explanation) as the representative dividing line for starforming activity: virtually all
objects with an X-ray luminosity higher than 1042 erg s−1 are assumed to host an
active nucleus. Whilst galaxies with lower X-ray luminosities are consistent with
SF galaxies, X-ray weak AGN (like LINERs, or heavily obscured) may actually
be below this threshold.
In order to select galaxies that might potentially host an AGN, as a fourth requirement we selected only those sources with a well-defined count rate in the
hard 2-12 keV energy range, i.e. we considered only X-ray sources with a 2-12
keV EPIC detection likelihood above 3σ in at least one camera (i.e. hard X-raydetected survey). This requirement resulted in a sample of 240 NELGs selected
in the hard X-ray band with a minimum of 30 counts in at least one detector.
5. Finally, we performed a visual inspection of the optical data in each of these 240
SDSS-2XMMi pairs to confirm that all the matches were indeed genuine. Special
care was taken to examine sources that showed some signs of Hα and/or Hβ broad
3.3. Optical classification versus X-ray emission
47
emission line signatures, as well as spurious sources. After inspection of the SDSS
spectra, we excluded 29 objects. These sources showed either strong reddening
or low S/N in the blue part of the Hβ line or weak broad Hα lines (e.g. Seyfert
1.9). In some cases, we could detect weak broad Hα and Hβ lines (e.g. Seyfert
1.8). After removing these sources, our final sample contained 211 galaxies, that
had only narrow emission lines and reliable X-ray flux detections in the 2-12 keV
band.
Our final sample is, therefore, composed of 211 NELGs with Hβ line widths ranging
from 140 km s−1 up to 1200 km s−1 . The distribution of the FWHM of the four involved
emission lines on the BPT diagram are shown in Figure 3.3. The X-ray selection criteria,
on the other hand, resulted in all spectra having a minimum of 30 counts above 2 keV
in at least one detector, i.e. PN, MOS1, or MOS2.
80
60
number
number
60
40
40
20
20
0
0
200
400
600
800
1000
0
1200
0
200
Hα FWHM [km/s]
400
600
800
1000
1200
1000
1200
Hβ FWHM [km/s]
30
60
25
number
number
20
15
10
40
20
5
0
0
200
400
600
800
[NII] FWHM [km/s]
1000
1200
0
0
200
400
600
800
[OIII] FWHM [km/s]
Figure 3.3: Distribution of Hα, Hβ, [N II], and [O III] (left-to-right, top-to-bottom,
respectively) width for the 211 NELGs.
3.3.
Optical classification versus X-ray emission
The BPT diagram has been used to infer whether the gas in a given galaxy is excited by star formation or radiation from an accretion disc around a central SMBH.
AGN MISSED BY THE BPT DIAGRAM
48
This diagnostic diagram has become one of the major tools for the classification and
analysis of emission line galaxies in the SDSS (York et al., 2000). To be conservative in
our analysis, we adopted the dividing line between SF and AGN galaxies presented by
Kauffmann et al. (2003, hereafter Kauf03). In that work, a refined optical classification
was obtained, based on a combination of stellar population synthesis models plus detailed self-consistent photo-ionization models, in order to create a theoretical starburst
line projected onto the BPT diagram. This is given by
log F[O III] /FHβ =
0.61
+ 1.3
log F[N II] /FHα − 0.05
(3.2)
Kewley et al. (2001) used a different separation criterion between SF and AGN galaxies
in the BPT diagram, which lies above and to the right of the Kauf03 line, defining a
region often known as the LINER/transition region where sources exhibit both AGN
and starburst activity. If we adopted the Kewley et al. (2001) borderline, then the fraction of X-ray luminous NELGs classified as SF galaxies would be significantly higher,
at least twice. In this work, we prefer to focus on the X-ray luminous NELGs that are
uncontroversially classified as SF galaxies using the BPT diagram, hence we adopt the
Kauf03 criterion. Following this criterion, we can directly obtain an optical classification for our sources:
• BPT-SF or optically-classified SF: those lying below the Kauf03’s line are starforming galaxies
• BPT-AGN or optically-classified AGN: conversely those located above the line
are AGN
Table 3.1: Summary of the classification of the NELG sample and statistics of their
principal properties: thickness parameter T , X-ray-to-optical flux ratio XO and the
FWHM of the Hβ line (T and XO defined in Section 3.4).
subsample1
BPT-SF
true-SF
missing-AGN
optical2
X-ray3
SF
SF
SF/AGN
AGN
N
38
28
T ≥ 10
[%]
0.10
0.93
XO≥ 0.1
[%]
0.00
0.96
FWHM(Hβ)
> 600km/s
0
25
weak-AGN
AGN
SF/AGN 62
0.03
0.08
3
strong-AGN
AGN
AGN
83
0.30
0.79
15
1 The sample is classified into either two (BPT-AGN, BPT-SF) or four subsamples (trueBPT-AGN
SF, missing-AGN, weak-AGN, strong-AGN) if it is based either on optical emission lines
or on both optical emission line ratios and LX , respectively;
2 SF or AGN nature according to the Kauf03’s line;
3 SF or AGN nature according to the hard X-ray luminosity;
3.3. Optical classification versus X-ray emission
49
In order to see how the optical classification compares with X-ray luminosities, we
must first obtain the intrinsic hard X-ray luminosity of each source. We used the Xray fluxes and spectroscopic redshifts1 to calculate the rest-frame 2-10 keV luminosities
(LX ) assuming an X-ray spectrum in the form of a power law with continuum spectral
slope Γ = 1.7 modified by Galactic absorption; we then corrected these luminosities for
the Galactic absorption to compute the intrinsic luminosities, but we did not correct
for any possible intrinsic absorption. Figure 3.4 shows the BPT diagram where the
dashed line is the Kauf03 separation between SF and AGN and the points change in
both shape and colour according to their value of LX to highlight the disagreements
between the optical-based and X-ray-based classifications. From the optical diagram
and according to the values of LX , the sample was split into four subsamples (see Table
3.1):
•
weak-AGN
subsample: consisting of 62 sources classified as AGN according to
the BPT diagram despite their low luminosities, not exceeding 1042 erg s−1 .
•
strong-AGN
subsample: including 83 NELGs identified as BPT-AGN that have
X-ray luminosities exceeding 1042 erg s−1 .
•
true-SF
subsample: consisting of 38 sources that were classified as SF according
to both the BPT diagram and LX criteria.
•
missing-AGN
have LX ≥
1042
subsample: involves 28 objects classified as BPT-SF that however
erg s−1 .
This classification clearly implies that there is a mismatch between the optical- and
X-ray-based classifications in the missing-AGN and possibly weak-AGN subsamples
(also found by Yan et al., 2011).
There are, on the one hand, several explanations of the low luminosity emitted by
weak-AGNs. A significant fraction of the population of AGN in the local Universe
displays a low X-ray luminosity, not exceeding 1042 erg s−1 (see Barth, 2002). In particular, low-ionization nuclear emission-line regions (LINERs) were originally defined by
Heckman (1980) as a subclass of these LLAGNs, whose optical spectra are dominated
by strong low ionization lines and much weaker higher ionization lines than classical
AGN. According to Heckman (1980), LINERs are galaxies that satisfy
[O II]λ3727 > [O III]λ5007 and
[O I]λ6300/[O III]λ5007 ≥ 1/3
1
Optical and X-ray parameters are taken from SDSS-DR7 and 2XMMi-DR3, respectively.
50
AGN MISSED BY THE BPT DIAGRAM
1.5
1
0.5
0
non-AGN
40
40
LX ≤10
LX ∈10 -1042
LX ∈1042-1044
LX ≥1044
-2
log
F[NII]
FHα
-1
AGN
0
Figure 3.4: Emission line diagnostic diagram (BPT diagram) for the 211 NELGs. The black-dashed curve separating the AGN from
the non-AGN (bottom left) zone is taken from Kauffmann et al. (2003). Symbols change in both shape and colour according to the
hard X-ray luminosity, LX . To avoid confusion, only the mean errors are reported (bottom right). The estimated errors increase with
decreasing log ([N II]/Hα).
-1
-0.5
F[OIII]
FHβ
log
3.4. Optical versus X-ray properties
51
According to these criteria, we classified 8 (13+8
−6 %) sources in our weak-AGN
subsample as pure LINERs and some additional 18 (29+11
−9 %) as weak-[O I] LINERs.
The latter fully satisfy Filippenko & Terlevich (1992)’s definition (i.e. [O II]λ6300/
Hα< 1/6).
On the other hand, the high values of the luminosity emitted by the sources in
our missing-AGN subsample suggest that these sources do contain an AGN core even
though they lie beneath Kauf03’s line, implying that optical AGN signatures are lacking
and signs of star formation, such as strong Hα and Hβ lines, are clearly visible. The
nature of this “misclassified” population is discussed throughout this chapter.
3.4.
Optical versus X-ray properties
After discussing the optical classifications of our NELG sample and the mismatches
with the X-ray luminosities for some objects, we now compare their optical and X-ray
properties in the context of three parameters in an attempt to provide clues about the
nature of the source populations within the complete sample of NELGs.
•
A Hardness Ratio Analysis:
HR
To extract the most basic X-ray spectral information, we performed a Hardness
Ratio (HR) analysis on the EPIC-pn data. We adopted the standard hardness ratio,
defined as
HR =
H −S
,
H +S
(3.3)
where S and H are the PSF and vignetting-corrected count rates in the 0.5-2 keV and
2-4.5 keV energy bands, respectively. A HR analysis is much simpler than a complete spectral analysis and is often the only X-ray spectral information available for the
faint sources in the XMM-Newton catalogue. We note that the X-ray selection criteria
resulted in a minimum count threshold of around 30, hence a proper X-ray spectral
analysis could not be performed for a number of our sources. The HR parameter is an
approximate indicator of the intrinsic X-ray spectral shape, which is also sensitive to
the level of absorption. An unabsorbed X-ray spectrum has typically lower HR than
an absorbed one. Although this correlation is relatively weak, and redshift-dependent
(Trouille et al., 2009), the vast majority of our missing-AGNs have a low HR that is
consistent with being unabsorbed as shown in Figure 3.5.
AGN MISSED BY THE BPT DIAGRAM
•
A Thickness Parameter Analysis:
52
T
An alternative method to unveil the absorption by dust in the nuclear environment
of an AGN is to measure the X-ray luminosity, and compare it with an isotropic indicator of the intrinsic power of the nuclear activity. Assuming that the unified AGN
model is correct, in absorbed sources the X-ray flux is attenuated with respect to this
isotropic indicator by an amount related to the absorbing column density. Taking the
reddening-corrected [O III]λ5007 luminosity as an isotropic indicator of the source nuclear strength, we calculated the ratio of the hard X-ray to [O III] fluxes (Bassani et al.,
1999, hereafter thickness parameter or T ). According to Bassani et al. (1999),


1 . T . 100




 0.1 . T . 10
T . 0.1






T .1
type-1 Seyfert
Compton-thin, type-2 Seyfert
Compton-thick, type-2 Seyfert
normal galaxies
The thickness parameter is defined as
c
T ≡ FXc /F[OIII]
,
(3.4)
FX being the observed 2-10 keV flux and [O III]λ5007 the emission line flux. The fluxes
c
) were corrected for Galactic absorption and optical reddening (Eq. 1.5),
(FXc , F[OIII]
respectively.
•
An X-ray-to-optical flux ratio Analysis:
XO
Finally, a useful parameter to discriminate between different classes of X-ray sources
is the X-ray-to-optical flux ratio (see Maccacaro et al., 1988, hereafter XO). For this
work, the XO parameter has been defined as the ratio between the observed X-ray
flux in the 0.5-4.5 keV energy range (corrected by Galactic absorption) and the optical
r(SDSS) band flux given by
fr [erg/s/cm2 ] = 3.36 × 10−20 10
where
dν
ν
ABr
2.5
dν c
ν λc
and λc are the band width and the central wavelength of the r filter (0.221
and 6261Å, respectively) and ABr is the AB magnitude in the rSDSS band. According
3.4. Optical versus X-ray properties
53
to Fiore et al. (2003) and Della Ceca et al. (2004),







XO & 10
0.1 . XO . 10
XO . 0.1
type-2 QSO; high-z clusters of galaxies; BL Lacs
Seyfert (both type-1 and type-2); AGN
heavily obscured AGN; Compton-thick; normal galaxies
Figure 3.5 shows the normalized distribution of these three observed parameters
(HR, T , and XO) for our subsamples. We should expect that the thickness parameter and the XO values fall within the typical range of values for BPT-SF galaxies,
i.e. T < 1 and XO< 0.1. We expect correspondingly that BPT-AGN will fall outside
of this range. We found that nearly all of the X-ray-based AGN (missing-AGN and
strong-AGN subsample, see Table 3.1) have typical AGN values for both XO and
T parameters, and the values for the true-SF and weak-AGN subsample are more
consistent with being SF galaxies, for which XO< 0.1 and T < 1. While this findings
is expected by the subsamples true-SF, strong-AGN and weak-AGN, the trend of
the missing-AGN galaxies is unexpected: despite that these sources are optically star
formers lie in the parameter space of the AGN population.
In order to explore possible correlations between these three parameters, we show
the combined information provided by XO, HR and T in as a function of the intrinsic
2-10 keV X-ray luminosity in Figure 3.6. We have used different symbols to denote
different ranges of LX : symbols denote the optical classification, filled for BPT-AGN
and empty for BPT-SF. We cannot define a range of values that isolate AGN from
the rest of the sources, either by using T or XO, i.e. a NELG classified as either a
SF galaxy or an AGN (by using either optical-based or X-ray-based criteria) does not
occupy a definite region in these parameter spaces.
Analogously, hardness ratios do not clearly cluster around different values for different subclasses of objects (see Figures 3.5 and 3.6). In general, one would expect SF
galaxies to have a X-ray spectrum dominated by a thermal component that is softer
than the typical power-law spectra exhibited by an AGN. However, the mix of BPT-SF
and BPT-AGN galaxies do not show a clear trend in their hardness ratios. Thus, we
cannot establish a clearly defined criterion to differentiate AGN from SF, in the context
of the three analysed parameters.
Nevertheless, we find that the distribution of log T as well as the distribution of
log XO display a quasi -bimodal shape for the optically-selected SF population, i.e.
54
AGN MISSED BY THE BPT DIAGRAM
0.3
0.25
0.2
0.15
0.1
0.05
0.5
0
0.4
0.3
0.2
0.1
0
-1
-0.5
HR
0
BPT-SF
BPT-AGN
weakAGN
trueSF
strongAGN
1
missingAGN
0.5
0.3
0.25
0.2
0.15
0.1
0.05
0.5
0.4
0.3
0.2
0.1
0
-2
-1
0
log T
1
2
BPT-SF
BPT-AGN
weakAGN
trueSF
strongAGN
3
missingAGN
(ii) thickness parameter
4
fraction
fraction
0.3
0.25
0.2
0.15
0.1
0.05
0.5
0
0.4
0.3
0.2
0.1
0
-4
BPT-SF
BPT-AGN
-3
-2
-1
log XO
0
weakAGN
trueSF
strongAGN
missingAGN
1
(iii) X-ray-to-optical- flux ratio
2
Figure 3.5: Normalized distribution of the hardness ratio, thickness parameter and X-ray-to-optical flux for the subsamples. Top
panels: BPT-AGN and BPT-SF subsamples (hatched and open histograms, respectively); bottom panels: true-SF, missing-AGN,
weak-AGN, and strong-AGN subsamples (solid-line open histogram, dashed-line open histogram, solid-line hatched histogram, and
dashed-line hatched histogram, respectively).
(i) hardness ratio
fraction
fraction
fraction
fraction
3.4. Optical versus X-ray properties
55
102
X/O ≡ F0.5-4.5 keV/Fr
10
1
10-1
10-2
10-3
10-4
-1
-0.5
0
0.5
1
HR
Figure 3.6: Thickness parameter versus HR and the X-ray-to-optical flux ratio
distribution as a function of HR (top and bottom panels, respectively). Different
symbols mark X-ray luminosity and the filled/empty symbols represent the optical
classification (BPT-AGN and BPT-SF, respectively). To avoid confusion, only the
mean errors are reported.
AGN MISSED BY THE BPT DIAGRAM
56
BPT-SF, opening the possibility that the emission of the missing-AGN population
has a different nature (see Figure 3.6, dashed regions).
As a further test of the disagreements between optical- and X-ray-based classifications, Figure 3.7 shows the intrinsic 2-10 keV X-ray luminosity as a function of the
Hβ FWHM. From this figure, it is evident that the bimodal nature of the BPT-SF
population is directly linked to the values of Hβ FWHM. There is an almost one-to-one
correspondence between the FWHM of their Hβ line and their X-ray luminosity for the
NELGs diagnosed as BPT-SF galaxies. Among this sample of 66 BPT-SF, we indeed
found that all those with LX < 1042 erg s−1 exhibit an Hβ FWHM.600km s−1 , while
roughly all (25/28) of the more X-ray luminous objects (missing-AGN which should
contain an AGN) have Hβ FWHMs between 600 and 1200 km s−1 . However, this division based on Hβ FWHM does not apply to the BPT-AGN: while the vast majority
of the weak-AGN galaxies (59/62) have a FWHM smaller than ∼600 km s−1 , the
values of the Hβ FWHM for the strong-AGN galaxies range from ∼200 km s−1 up
to ∼1200 km s−1 .
3.5.
An overview of the
3.5.1.
Completeness
missing-AGN
subsample
On the basis of the estimated upper limits to the 2-10 keV luminosity given by FLIX
(upper limit server for XMM-Newton data provided by the XMM-Newton Survey Science Centre)1 for objects that were identified as NELGs but lacked X-ray detections,
there are another 1207 additional sources in the SDSS-DR7 with an upper limit for the
X-ray luminosity that were classified as SF based on their position in the BPT diagram.
However, only 5% (60/1207) of them could be missing-AGN candidates, i.e. those for
which the upper limit to their 2-10 keV exceeds 1042 erg s−1 .
The missing-AGN candidates represent thus only a few percent (between 2%, i.e.
28 among 66+1207, and 7%, i.e. 28+60 among 66+1207) of the BPT-SF population,
and therefore they do not represent a major problem in terms of contamination. In
terms of the total sample of NELGs covered by X-ray observations, the missing-AGN
represent between 1.6% and 5% of the entire sample. However, the nature of the
missing-AGN subsample is poorly understood and needs to be explored further.
1
FLIX: see http://www.ledas.ac.uk/flix/flix.html
3.5. An overview of the
missing-AGN
subsample
57
LX [erg/s]
1045
1043
BPT-SF
1041
missingAGN
trueSF
BPT-AGN
weakAGN
strongAGN
1039
0
200
600
1000
Hβ FWHM [km/s]
Figure 3.7: LX as a function of Hβ FWHM. The vertical dotted line marks the
threshold of FWHM(Hβ)=600 km/s, while the horizontal dotted line corresponds to
LX = 1042 erg/s. Different symbols are related to the subsamples: true-SF, missingAGN, weak-AGN, and strong-AGN. Open symbols (squares and triangles) are
BPT-AGN sources, star and cross points are BPT-SF sources.
3.5.2.
Nature of the optical elusiveness
In Section 3.4 we have described the X-ray and optical spectral properties used to
explore the nature of the elusiveness of optical signatures in the missing-AGN subsample. Similar studies have been performed previously focusing on the nature of the
so-called elusive AGN, i.e. sources that show no signs of AGN activity in the optical regime, but display signs of AGN activity in the X-ray band, like the sources in
the missing-AGN subsample. One possibility is that they could be mildly/heavily
obscured AGN in which star formation dominates the optical emission-line ratios. Assuming that the unification model is correct, in the Compton-thick regime the X-ray
flux must be depressed with respect to the [O III] optical line flux by an amount related
to the absorbing column density (the latter corrected for extinction and used as a true
indicator of the source intrinsic luminosity, Bassani et al., 1999, Maiolino et al., 1998).
Given that the thickness parameter is in the range T ≥ 10, the hydrogen column density
would have to be NH < 1023 cm−2 and therefore obscuration by the torus is not very
likely to be the cause of the elusiveness of optical AGN signatures.
AGN MISSED BY THE BPT DIAGRAM
58
Another possibility is that the subsample contains composite objects, i.e. those
hosting both star formation and an AGN. On the basis of the relation between LX (2 −
10 keV) and LHα for SF galaxies (Kennicutt, 1998, Ranalli et al., 2003), LHα /LX
should be greater than 1, assuming Av ≤ 2. Whilst type 1 AGN have ratios lower by
two orders of magnitude, composite galaxies are expected to have intermediate ratios
(Yan et al., 2011). Figure 3.8 shows the distribution of log (LHα /LX ), where most of
the LHα /LX ratios seem to lie in-between both extremes as expected in the composite
galaxy range. This opens the possibility that the missing-AGN population could be
composite objects having both star formation and active nuclei, although largely consistent with harbouring an AGN.
15
N
10
5
0
-1.5
-1
log(LHα/LX)
-0.5
0
Figure 3.8: Distribution of the LHα /LX for the missing-AGN subsample.
The high values of XO as well as T , which are two orders of magnitude higher than
the average in SF galaxies and those typical of Seyfert 1, the low hardness ratio and
the quite high values for the Hβ FWHM make our missing-AGN sources very likely to
be NLS1, which are believed to lie in the starburst region of the BPT diagrams. The
NLS1 galaxies represent a subclass (Osterbrock & Pogge, 1985) of type 1 AGN that
manifest a distinctive ensemble of properties. As explained in the Introduction, they
are AGN with optical spectral properties similar to those of Seyfert 1 galaxies, except
for recombination lines that are only slightly broader than forbidden emission lines.
Studies of NLS1s have identified many peculiar properties that extend well beyond a
pure line-width-based distinction. Distinctive features in optical spectra of NLS1s, are
the low values of the [O III]λ5007/Hβ ratio, and the often strong blends of permitted
Fe II emission lines. Out of the 28 missing-AGNs, the Fe II multiplets were detected
in 23 objects at the >3σ level. The rest of the subsample (5/28) appears to belong
to a rare class of NLS1s that do not exhibit strong Fe II multiplets. There are three
objects, ID=203,241,338, with very narrow Balmer lines that exhibit a prominent He II
3.5. An overview of the
missing-AGN
subsample
59
broad emission line, which ensures their classification as type 1 AGN. Figure 3.9 shows
two SDSS spectrum: a typical NLS1 spectrum with a moderate Fe II optical emission
(ID=26, top panel); and a typical NLS1 spectrum without Fe II emission, but having a
prominent He II emission (ID=338). For two additional sources (ID=63,335), the SDSS
spectra were too noisy to yield reliable measurements of either Fe II multiplets or He II,
and in addition they have evidence for high reddening.
1200
Hβ
1000
SDSS J141519.50-003021.5
Hα
[NII]
[OIII]
Flux (10-17 erg/s/cm2/angstrom)
800
600
400
200
0
Fe II
SDSS J082912.68+500652.3
[OIII]
150
Hα
Hβ
[NII]
100
He II
50
0
4000
5000
6000
7000
8000
9000
Observed Wavelength (angstrom)
Figure 3.9: Example of a typical NLS1 in our sample with a moderate Fe II emission
(top panel ) and, on the other hand, a SDSS spectrum of an Fe II-lacking NLS1 (bottom
panel ). Note in the last spectrum the very narrow Balmer emission lines and also
the very weak Fe II multiplets; the prominent He II broad emission line ensures our
classification as a type 1 AGN.
AGN MISSED BY THE BPT DIAGRAM
60
The relative strength of the Fe II multiplets is usually expressed as the flux ratio of
Fe II to Hβ,
R4570 ≡
Fe IIλλ4434 − 4684
,
Hβ
(3.5)
where Fe II λλ4434 − 4684 denotes the flux of the Fe II multiplets integrated over the
wavelength range of 4434−4684 Å after subtracting the local underlying continuum and
the He II λ4686 emission line. Figure 3.10 shows the distribution of the relative strength
of the Fe II multiplets, R4570 , for the missing-AGN subsample, which is compared with
the distribution given by Zhou et al. (2006). The average of the relative strength of the
Fe II multiplets for the missing-AGN is hR4570 i = 0.88 and the 1σ dispersion is 0.5,
which is consistent with Zhou’s NLS1 sample: the probability that both distributions
(missing-AGN and the sample of NLS1 given by Zhou et al., 2006) come from the same
distribution is ∼89% according to the Kolmogorov-Smirnov test. Indeed, the average
for this kind of sources is significantly larger than the typical value of R4570 ∼ 0.4 found
in normal AGN (Bergeron & Kunth, 1984), bolstering again the idea that these two
populations (missing-AGN and true-SF) are probably of a different nature.
2
Missing-AGN subsample
Zhou’s NLS1 sample
dN
1.5
1
0.5
0
0
1
2
3
R4570
Figure 3.10: Normalized distribution of the relative strength of the Fe II multiplets,
R4570 , for the missing-AGN subsample (filled histogram) and the NLS1 sample of
Zhou et al. (2006).
3.6.
X-ray Spectral Analysis
All objects presented here were observed with XMM-Newton between 2001 June and
2007 December. The European Photon Imaging Cameras (EPIC) pn (Strüder et al.,
2001) and MOS (Turner et al., 2001, MOS1 and MOS2) were operated in full frame
3.6. X-ray Spectral Analysis
61
imaging mode during all the observations. The XMM-Newton data of some objects
were previously presented in the literature (see Table 3.2). For a fully homogeneous
analysis enabling robust conclusions, we re-analysed the XMM-Newton spectra of these
objects, in exactly the same way as for the objects whose XMM-Newton data are presented here for the first time.
The X-ray selection criteria (see Section 3.2) resulted in a minimum of 30 counts at
energies above 2 keV in at least one detector independently of the quality of the spectrum, which means that the S/N was sometimes low. Thus, the level of detail of our
spectral analysis varied for each source depending on the quality of the EPIC spectra,
ranging from a quite detailed analysis for bright sources, to only very coarse spectral
fits for the faintest. Therefore, the lowest quality spectral observations (<300 counts
adding up the three EPIC cameras) were only grouped with at least 15 counts per bin,
instead of 20. All quoted errors are for a 90% confidence interval for one parameter
(∆χ2 = 2.706).
We carried out an X-ray spectral analysis of the BPT-SF populations, consisting
of the missing-AGN and true-SF subsamples. We recall that our main aim is to
understand the nature of the missing-AGN subsample, where the sources are optically
classified as SF but have LX in excess of 1042 erg s−1 that are indicative of an AGN.
The results of the previous section (T , XO, HR, Hβ FWHM, and R4570 ) lead us to
propose that these objects are good NLS1 candidates. Thus, the missing-AGN and
true-SF subsamples were analysed as samples of different nature.
3.6.1.
missing-AGN subsample
Our aim is to estimate whether the X-ray spectra of the missing-AGN subsample
show large soft X-ray excess, an ubiquitous feature of the NLS1 class. We did not
carry out an X-ray spectral analysis for the faintest sources (<100 counts, i.e. 2XMMi
J123748.5+092323 (ID=163), 2XMM J102812.6+293222 (ID=302), and 2XMM J0148
56.9+135450 (ID=335), see Table 3.2). For three other sources (2XMM J141519.4003021 (ID=26), 2XMM J135724.5+652506 (ID=53), and 2XMM J123126.4+105111
(ID=161), see Table 3.2) the energy range used by the spectral analysis was E.3-6 keV
due to the EPIC-pn data being dominated by the background above these energies. Finally, 2XMMi J082912.8+500652 (ID=338), 2XMM J131718.6+324036 (ID=355), and
2XMM J134235.6+261534 (ID=357) were analysed using only EPIC-MOS data because
of the lack of EPIC-pn data. In Table 3.2, we give details of the X-ray observations
and the hard X-ray luminosity of each object, which was calculated using the best-fit
power-law model over the 2-10 keV keV energy band. We note that all models referred
AGN MISSED BY THE BPT DIAGRAM
62
Table 3.2: Description of the XMM-Newton observations and its SDSS counterpart.
We point out that we only present those sources that were analyzed in this work, i.e.
with an available X-ray spectrum with enough signal-to-noise ratio to make an X-ray
spectral analysis. Left to right: (1) Numeric identifier of the source; (2) SDSS object
name where the full name should be ‘SDSS J. . . ’; (3) 2XMM object name where the
full name should be ‘2XMM J. . . ’; (4) XMM-Newton’s observation number; (5) XMMNewton’s exposure time in units of ks; (6) the total counts in the EPIC monitor; (7)
spectroscopic redshiftf from the SDSS-DR3 catalogue; (8) Galactic column density
from Dickey & Lockman (1990) in units of 1020 cm−2 ; (9) the rest-frame 2 − 10 keV
luminosity derived from the best-fit model (see Section 3.6); (10) reference for its
classification: (a) Foschini et al. (2004), (b) Dewangan et al. (2008), (c) Piconcelli
et al. (2005), (d) Gallo et al. (2006), (e) Maitra & Miller (2010), (f) Grupe et al.
(2010), (g) Crummy et al. (2006) (among others).
ID
(1)
2XMMi Catalogue
SDSS DR7
SDSS . . .
(2)
2XMM . . .
Obs ID
(3)
(4)
texp
[ks]
(5)
z
NH,Gal
LX
(7)
[×1020 cm−2 ]
(8)
[×1042 erg s−1 ]
(9)
counts
(6)
Notes
(10)
missing-AGN sample
.....................................................................................................................................
26
J141519.49-003021.5
J141519.4-003021 0145480101 13
1781 ± 44
0.135
3.28
8.27
a
42
J010712.03+140844.9 J010712.0+140844 0305920101 16
4502 ± 69
0.077
3.41
2.81
b
53
J135724.52+652505.9 J135724.5+652506 0305920501 1.7
931 ± 31
0.106
1.38
6.85
b
63
J111031.61+022043.2 iJ111031.6+022043 0504101801
8
344 ± 19
0.080
3.73
2.30
65
J114008.71+030711.4 J114008.7+030710 0305920201 34 23860 ± 156 0.081
1.93
3.91
b
71
J124635.24+022208.7 J124635.3+022209 0051760101
4
31746 ± 179 0.048
1.85
12.51
c
100
J221918.53+120753.1 J221918.5+120753 0103861201
8
19684 ± 141 0.081
5.03
10.03
d
104
J092247.02+512038.0 J092246.9+512037 0300910301 6.3 12029 ± 111 0.160
1.32
18.90
111
J094240.92+480017.3 J094240.9+480017 0201470101 14
424 ± 24
0.197
1.20
5.87
125
J133141.03-015212.4
J133141.0-015212 0112240301 24
2548 ± 52
0.145
2.39
11.90
129
J081053.75+280610.9 J081053.8+280611 0152530101 16
986 ± 33
0.285
2.93
18.17
161
J123126.44+105111.3 J123126.4+105111 0145800101 7.7
269 ± 18
0.304
2.14
6.86
163? J123748.49+092323.1 iJ123748.5+092323 0504100601
0.125
1.48
171
J093922.90+370943.9 J093922.9+370942 0411980301
4
2857 ± 54
0.186
1.22
27.89
191
J155909.62+350147.4 J155909.6+350147 0112600801 11 80531 ± 285 0.031
2.11
8.22
195
J103438.59+393828.2 J103438.6+393828 0109070101 28
21857±214 0.043
1.31
c,e
203
J124013.82+473354.7 J124013.8+473355 0148740501 5.7
804 ± 29
0.117
1.32
1.03
204
J124058.45+473302.0 J124058.3+473302 0148740501
5
192 ± 15
0.367
1.33
35.88
214
J112405.15+061248.8 J112405.1+061248 0103863201
5
786 ± 29
0.272
4.61
36.76
241
J075216.55+500251.3 J075216.4+500251 0151270201 7.7
620 ± 26
0.263
5.17
57.45
275
J145108.76+270926.9 J145108.7+270926 0152660101 18 91984 ± 305 0.065
2.70
20.35
f
302? J102812.67+293222.8 J102812.6+293222 0301650401 8.4
76 ± 13
0.287
1.91
3.57
318
J122230.71+155547.9 J122230.7+155547 0106860201 8.6
238 ± 17
0.367
1.99
16.00
329
J140621.89+222346.5 J140621.8+222347 0051760201 3.1
1911 ± 44
0.098
2.05
3.24
c,g
335? J014856.95+135451.8 J014856.9+135450 0094383401
0.220
4.90
338† J082912.67+500652.3 iJ082912.8+500652 0303550901 2.2
442 ± 22
0.043
4.07
2.51
b
355† J131718.58+324035.6 J131718.6+324036 0135940201 10
161 ± 13
0.061
1.17
2.27
†
357
J134235.66+261534.0 J134235.6+261534 0108460101 26
867 ± 30
0.064
1.03
2.59
analyzed BPT-SF galaxies1
.....................................................................................................................................
8
J140919.94+262220.1 J140920.0+262219 0092850501 35
163 ± 20
0.059
1.40
6.0
56
J095848.66+025243.2 J095848.6+025243 0203362101 54
259 ± 32
0.079
1.83
36.0
79
J093402.02+551427.8 J093401.9+551428 0112520101 27
2178 ± 56
0.002
2.46
0.2
154
J162636.40+350242.0 iJ162636.5+350242 0505011201 14
127 ± 12
0.034
1.44
4.2
164
J080629.80+241955.6 J080629.7+241956 0203280201
6
156 ± 21
0.042
3.80
45.0
233
J122254.57+154916.4 J122254.6+154916 0106860201 10
473 ± 32
0.005
2.01
0.8
246
J085735.33+274605.1 J085735.4+274607 0210280101 68
256 ± 18
0.007
2.51
0.2
251
J123520.04+393109.1 J123519.9+393110 0204400101 26
97 ± 8
0.021
1.31
0.5
?
These sources could not be analysed owing to the unreliable quality of the data statistics
(number of EPIC-pn counts less than 50).
†
Were analysed using only EPIC-MOS data because of the lack of EPIC-pn data.
1
Only 8 of the 38 true-SF galaxies have been analysed, because of the poor X-ray spectral
quality of the remaining 30 (see Section 3.6.2).
3.6. X-ray Spectral Analysis
63
to in this Chapter include a multiplicative Galactic absorption component fixed at the
NH value given in that Table.
The general best-fit model of the X-ray spectrum emitted by a NLS1 has typically
four components: an underlying absorbed steep power-law, a soft X-ray excess, and a
reflection component that might also include a broad feature near the Fe line complex
at 5-7 keV. Several explanations have been proposed for the origin of the observed soft
excess, such as a relativistically blurred photo-ionized disc reflection (Crummy et al.,
2006, Ross et al., 2002), an intrinsic thermal component, or ionized absorption arising
in a wind from the inner disc (Gierliński & Done, 2004, 2006). Any discussion about
the origin of the soft X-ray excess in such class of objects is beyond the scope of this
work, and for simplicity, we only used two different two-component continua: a partial
covering and a thermal black-body model as proxies to each explanation respectively
(the quality of the X-ray data did not allow a more sophisticated analysis in the majority
of cases). In the case of the first two-component continua, we modeled the soft excess
emission due to reprocessing of the primary X-rays as a partial-covering neutral material
(hereafter PCF model). This can be regarded as a simplified model of a clumpy torus,
where the torus is a smooth continuation of the broad line region, and provides a
physical explanation of the apparent mismatch between the optical classification and
the X-ray properties of these objects. In this model, X-ray absorption, dust obscuration,
and broad line emission are produced in a single continuous distribution of clouds: the
broad line region is located within the dust sublimation radius, hence the non-dust-free
clouds obscure the optical emission but not the X-rays, whereas the torus is located
outside the dust sublimation zone. The PCF model assumes that some fraction, f , of
the X-ray source is covered by a neutral absorber with a column density of NH , while
the rest is unobscured. This could be responsible for an apparent soft excess in two
different geometries, either by reflection from optically thick material out of the line of
sight (Fabian et al., 2002), or absorption by optically thin material along the line of
sight (Chevallier et al., 2006, Gierliński & Done, 2004).
On the other hand, there are some possible ways of explaining the soft excess from
the disc itself in terms of the reprocessing of the primary X-rays in the accretion disc
as reflected emission from a geometrically flat disc, with solar abundances, illuminated
by an isotropic source. Thus, the soft excess is sometimes closely fitted by a black body
that has a roughly constant temperature of 0.1-0.2 keV. If this radiation is thermal,
this temperature is much too high to be explained by the standard accretion disc model
of Shakura & Sunyaev (1973), although it could be explained by a slim accretion disc
in which the temperature is raised by photon trapping, in which case the accretion
is super-Eddington (Tanaka et al., 2005), or by the Comptonization of extreme UV
AGN MISSED BY THE BPT DIAGRAM
64
accretion disc photons (e.g. Porquet et al., 2004).
All sources were analyzed using the following approach,
• Hard X-ray spectral fitting procedure:
1. The individual data in the 2-10 keV (when available, otherwise up to the
highest available energy) energy range are fitted by a power law model
2. The residual spectra are checked for any significant additional component
that may be present in this energy range, such as an iron emission line
(modelled with zgauss at Ec ∼ 6.4 keV) and/or an Fe K-edge, among other
features1 .
• Soft X-ray component fitting procedure:
1. We added either a redshifted black body component (zbbody in XSPEC) to
the PL model (hereafter BB-PL model) or a neutral absorber at the redshift
of the X-ray source that either fully (f = 1) or partially (f < 1) covers the
source (zpcfabs in XSPEC; hereafter PCF-PL model).
2. We then compared these two models2 , BB-PL and PCF-PL. When one of the
two models gave a fit with a ∆χ2 ≡ χ2P L − χ2BB−P L/P CF −P L ≥ 10 and/or an
F-test significance that was high enough, ≥ 99%, the new model was taken
as the baseline. In the case of sources for which the χ2 for the BB-PL and
PCF-PL models were comparable and the values of each free parameter were
physically plausible for both components, we adopted as the best fit model
the one with the least uncertain model parameters.
Discussion of the Results
We found that the PL model yields an acceptable fit for almost all objects in the
sample, in terms of the minimum of the reduced chi-squared χ2ν (see Table 3.3). No
other statistically significant and physically meaningful features were found in the hard
X-ray spectra. Figure 3.11 shows that the distribution of Γ2−10 takes values above
the typical expected value (∼ 1.8) but surely well within the mutual dispersions found
for type 2 AGN. We note that the cases where the X-ray spectra appears very hard
with a power-law slope Γ ∼ 1.5, also correspond to those with larger errors, ±0.5 (see
Table 3.3).
1
To this end we used the F-test, accepting additional spectral components only when they improved
the fit with a significance ≥3σ.
2
For most of the sources the X-ray data quality is too poor to attempt more sophisticated fits.
3.6. X-ray Spectral Analysis
65
10
number
8
6
4
2
0
1
1.5
2
2.5
3
Γ 2-10 keV
Figure 3.11: Distributio in Γ2−10 for the missing-AGN subsample when we fitted
the individual data in the 2 − 10 keV energy range with a single power law modified
by absorption in our Galaxy. The vertical line marks the expected value for type-2
Seyfert galaxies.
We also attempted to fit the spectrum over the entire X-ray band (0.3 − 10 keV)
with an intrinsically absorbed power-law model (absorbed-PL). We found that for all
sources this simple model was rejected over the whole X-ray band with a high statistical significance. The estimated intrinsic column density was found to be less than 1022
cm−2 , which corresponds to an unabsorbed model or at least values of absorption that
fall in the low part of the column density distribution of type-2 AGN. These very low
upper limits are not quoted in Table 3.3.
We found additional evidence against starburst activity in the missing-AGN population by comparing the soft (0.3-2.5 keV) versus hard (2-10 keV) X-ray spectral indices,
which revealed an overall spectral steepening towards low energies in many cases. This
suggests that there is a soft X-ray excess that contributes mostly below ∼2 keV. For
the majority of the sources, this soft excess represents more than 20% of the X-ray
emission, which is higher than expected for starburst activity. This soft component is
defined as the excess over an extrapolation to 0.3 keV of the PL model fitted to the
hard X-ray band (see Figure 3.12, and for those sources with more than 500 counts see
Chapter 4).
Table 3.4 shows the best-fit model parameters for each source. The addition of
either a BB or a PCF component provides a good match to the observed spectra in
almost all objects with more than 100 counts, and correspondingly provides a better
fit according to the χ2 test, than the absorbed-PL model (i.e. ∆χ2 and/or PF −test in
Table 3.4). An additional BB component was required to achieve good spectral fits in
about one-third of these objects (first part of the Table 3.4); the black-body electron
temperatures are found to be in the range of 100 − 200 eV, which is slightly higher
AGN MISSED BY THE BPT DIAGRAM
66
Table 3.3: Results of the best-fitting (absorbed) power-law model parameter values
for the missing-AGN subsample. Whilst the simple power-law model is fitted in the
hard X-ray energy band (2-10 keV), the absorbed power-law model is fitted in the
0.3-10 keV energy band.
power-law model
(over 2-10 keV)
ID
Notes
26
42
53
63
65
71
100
104
111
125
129
161
171
191
195
203
204
214
241
275
318
329
338
355
357
a.2
a.3
a
a
a
a.1
a.3
absorbed power-law model
(over 0.3-10 keV)
Γ
χ2ν /d.o.f.
ID
Notes
1.50±0.72
0.73
2.23±0.25
0.32
2.75±0.43
0.37
2.20±0.53
0.45
2.27±0.14
0.19
2.90±0.02
0.02
2.38±0.16
0.15
2.24±0.40
0.51
2.71±0.25
0.24
1.71±0.41
0.61
2.31±0.77
0.74
3.17±0.34
0.32
2.19±0.56
0.63
2.11±0.06
0.06
2.10±0.02
0.03
1.68±0.50
0.66
2.03±0.73
0.66
1.64±0.51
0.50
1.28±0.63
0.59
2.75±0.07
0.07
2.71±0.34
0.29
1.74±0.97
0.90
2.20±0.60
0.93
1.57±0.30
0.52
1.83±0.38
0.40
1.020/3
0.680/20
1.306/11
0.829/7
1.156/54
2.080/112
1.182/47
0.581/9
0.676/31
0.655/6
1.270/3
1.762/8
0.811/3
1.195/217
1.021/299
0.817/4
0.220/3
0.958/4
0.810/19
0.897/213
0.793/37
1.910/3
0.802/7
0.008/1
0.556/11
26
42
53
63
65
71
100
104
111
125
129
161
171
191
195
203
204
214
241
275
318
329
338
355
357
a.2
a.3
a
b
a
a
a.1
a.3
b
Γ
0.09
2.79±0.09
0.05
2.39±0.05
0.11
2.57±0.10
0.19
1.91±0.17
0.02
2.79±0.02
–
2.95±0.03
0.02
3.48±0.04
0.04
2.72±0.25
0.24
2.58±0.08
0.07
2.31±0.12
0.11
3.17±0.34
0.32
2.95±0.08
0.09
2.67±0.01
0.01
2.10±0.02
0.03
2.24±0.12
0.12
2.45±0.35
0.31
2.70±0.14
0.14
2.37±0.18
0.17
2.85±0.02
0.02
2.77±0.39
0.35
–
2.38±0.09
0.09
1.65±0.20
0.19
2.05±0.08
0.08
χ2ν /d.o.f.
1.213/63
0.933/161
1.154/43
1.500/18
1.370/316
–
1.360/301
1.350/170
0.676/31
0.993/85
1.294/39
1.762/8
1.437/81
1.475/562
2.021/587
0.949/41
0.367/13
1.052/32
0.899/102
1.657/558
0.894/30
–
0.979/63
1.002/13
1.231/66
a The hard photon index Γ was obtained in the energy band E ≥1 keV due to low statistics in
the hard 2-10 keV energy band. The pn data is dominated by the background above: a.1 3 keV;
a.2 4 keV; a.3 6 keV.
b The soft X-ray emission is really strong invalidating the application of a simple absorbed powerlaw model.
than those found for typical AGN. Such high temperatures could be explained by the
presence of a slim accretion disc in which the temperature is raised by photon trapping
(Abramowicz et al., 1988, Mineshige et al., 2000). For another third of the missingAGN subsample (second part of the Table 3.4), the best-fit models were achieved by the
addition of a PCF component. The spectral fits for a partial covering model indicate
3.6. X-ray Spectral Analysis
67
Table 3.4: missing-AGN: Summary of the best-fitting model parameter values for
the soft X-ray excess in the missing-AGN population. We could not analyse the X-ray
spectra of the sources 15, 163, 324, 335, and 355 because of the small number of counts.
Furthermore, the sources 111, 161, 189, 198, and 302 were not analysed using any of
these models, because their hard X-ray spectra is dominated by background, reducing
the detectability of their source counts. The table is divided into three parts: an
additional black-body component was required to find the best-fit model; the addition
of partial covering and, finally cases in which the low quality of their spectrum did
not allow us to reject any of the components. kTe is the electron temperature given
in units of eV, NH , z is the intrinsic neutral hydrogen column density in units of 1020
atoms cm−2 , f is the covering factor and PF −test denotes the probability of the F-test
(if it is low, i.e. close to zero, then it is reasonable to add the extra model component).
ID
Γ
kTe
BB ≡ PHA * ( zPOW + zBB )
χ2ν /d.o.f.
∆χ2
P†F −test
Γ
NH,z
f
χ2ν /d.o.f.
∆χ2
P†F −test
PCF ≡ PHA * zPOW * zPCF
BB-PL as the best-fit model
...........................................................................................................................
0.03
65
2.52±0.04
137±66
0.987/314
91.4
∼0
2.81±0.03
49±35
0.7±0.2
1.171/314
33.6
∼0
0.03
19
0.1
160±10
1.140/110
108
71b
2.50±0.03
0.03
10
0.03
100
2.65±0.07
160±10
1.100/299
81.5
∼0
2.98±0.03
34±19
0.6±0.1
1.230/299
41.5
∼0
0.08
10
12
0.1
4
0.04
10
104
2.82±0.09
119±
1.026/168
100.4
∼
0
3.50±
21±
0.8±0.1
1.537/168
18.9
∼0
0.08
4
0.04
40
0.2
143±30
1.140/210
68
∼0
195a 2.30±0.06
0.13
21
0.38
241
1.79±0.23
112±27
0.764/100
15.3
∼0
2.62±0.23
13±63
0.7±0.2
0.810/100
10.7
∼0
0.27
23
10
0.2
2
0.02
4
0.03
2.51±0.02
116±
1.082/556
323.0
∼
0
2.91±
15±
0.57±
1.171/556 273.5
∼0
275
0.03
2
0.02
3
0.04
329b 1.65±0.33
104±33
0.844/95 214.9
0.32
PCF-PL as the best-fit model
...........................................................................................................................
0.10
26
2.47±0.11
110±24
1.169/61
5.1
0.12
2.84±0.09
34±71
0.8±0.2
1.072/61
11.1
∼0
0.27
42
19
0.2
42
2.22±0.05
155±22
0.889/159
8.9
∼0
2.41±∗∗
65±10
0.6±0.4
0.923/159
29.8
0.157
∗∗
0.04
21
51
0.5
0.17
25
0.09
35
0.3
125
2.34±0.10
102±35
0.936/84
6.5
∼0
2.65±0.08
17±12
0.6±0.2
0.880/84
10.5
∼0
0.12
171
3.04±0.10
> 2 · 103
0.901/79
12.8
∼0
3.04±0.08
11±515
0.6±0.2
0.874/79
14.9
∼0
0.05
0.2
0.01
191
2.44±0.03
106±44
1.178/560 168.4
∼0
2.73±0.01
23±77
0.6±0.1
1.041/560
245
∼0
0.03
0.1
0.15
214
1.98±0.20
144±23
0.810/30
9.3
∼0
2.81±0.15
43±86
0.9±0.1
0.681/30
13.2
∼0
0.15
27
27
0.2
BB and/or PCF extra component could not be excluded
...........................................................................................................................
0.41
53
2.41±0.26
179±50
1.070/41
5.7
0.08
2.88±0.24
0.05±0.05
0.9±∗∗
1.130/41
<4
0.25
∗∗
∗∗
0.20
41
3
0.20
∗∗
2.03±0.24
>
2
·
10
1.553/16
<
3
0.51
2.01±
32±
0.7±∗∗
1.522/16
<3
0.44
63
∗∗
∗∗
0.22
0.62
0.24
129
2.12±0.08
87±26
1.195/37
6.3
0.08
2.67±0.46
2±157
0.5±0.2
1.256/37
<5
0.22
0.16
48
2
0.3
0.23
203
2.16±0.15
<1
0.892/39
<5
0.12
2.34±0.18
30±∗∗
0.7±∗∗
0.935/39
<5
0.09
∗∗
∗∗
0.13
∗∗
0.70
∗∗
204
2.03±0.63
120±
0.277/11
<
5
0.08
2.86±
4±
0.7±∗∗
0.243/11
<5
0.01
∗∗
∗∗
∗∗
0.48
0.60
338a 2.28±0.17
197±1000
0.985/61
<5
0.45
0.11
190
357a 2.22±0.24
1710±170
1.198/64
<5
0.16
0.18
550
a
The fit is insensitive to either the NH,z or f parameters of the PCF model.
b
The soft excess is extremely strong, invalidating the use of an abosrbed PL fitting over the whole
X-ray spectra.
∗∗
Denotes a parameter that XSPEC could not calculate the error with an error bar as precise as
90 per cent confidence.
AGN MISSED BY THE BPT DIAGRAM
68
that there were dispersions in both the absorption column density NH,z = (1 − 6) × 1021
cm−2 and covering factor f = 0.6 − 0.9. The measured strength of the non-absorbed
X-ray primary emission from the neutral material is h1 − f i = 0.32 (≥ 0.2 for the great
majority) being slightly higher than those expected by type-2 AGN (≤ 0.1 − 0.2).
For the remaining third of the sources, the low quality of their X-ray spectra did not
allow us to choose between the various possible models. In summary, the X-ray spectra
of the subsample of missing-AGNs were closely fitted by a rather steep power-law,
to which a soft excess apparent at energies . 2 keV should be added, when data of
sufficiently high quality becomes available. This is totally in line with the assumption
that this population is largely dominated by NLS1s.
3.6.2. true-SF subsample
In parallel, we conducted an X-ray spectral analysis of the true-SF subsample. The
X-ray spectra of local SF galaxies in the X-ray band (E.10 keV) can be described by a
combination of warm thermal emission (with typically kTe ∼0.5-1.0 keV, Franceschini
et al., 2003) dominating at energies ≤ 1 keV, and a power-law spectrum responsible for
producing the bulk of their 2-10 keV keV flux. The latter component has various interpretations, either in terms of an extremely hot (kTe ≥ 5 keV) thermal component or a
Γ ∼ 2 power-law model for high mass X-ray binaries. Given that SF spectra often exhibit strong collisionally excited emission lines, we used a MEKAL model (Mewe et al.,
1985, 1986) to fit the thermal component. This component appears at soft X-ray energies (below 1−2 keV), and we added a PL component to fit the hard X-ray spectrum.
We were able to perform the spectral analysis of only 8 of the 38 sources, because
of the poor X-ray spectral quality of the remaining 30. The X-ray spectra were modelled by adding both components which resulted in good fits for only 3 out of 8 objects
(2XMM J140920.0+262219 or ID=8, 2XMM J093401.9+551428 or ID=79, and 2XMM
J122254.6+154916 or ID=233, see Table 3.5) with kTe ∼ 0.6 − 3 keV and hΓi = 2.2; the
resulting parameters were consistent with those expected for a SF galaxy. The thermal
component contributes significantly over the 0.3 − 10 keV range, supplying ∼ 20% of
the total flux. Hence, a hot gas starburst component was found to be present in the
spectra of these three objects. The improvement to the fits when a MEKAL component
was added to a PL in five of the remaining eight objects, was minimal with ∆χ2 < 6
for two additional degrees of freedom (see Table 3.5). Hence, their X-ray spectra were
modelled by a PL only, which resulted in an acceptable fit; the resulting spectral indices
Γ ∼ 1.7 − 2.0 were compatible with those for a SF galaxy, as expected. We therefore
restricted our analysis of the remaining 30 sources to an inspection of the hardness
3.6. X-ray Spectral Analysis
69
ratios. In Figure 3.6, we can see that the hardness ratios of all 38 sources covers a wide
range of values, and therefore we are unable to reach any firm conclusion based on
these data. However, the invariably low values of the HR for these particular sources
are consistent with them being dominated by a thermal spectrum, in full agreement
with the expectation for SF galaxies.
As a final additional test, we fitted the missing-AGN X-ray spectra as if they were
true-SF sources. We found that the soft X-ray excess is not properly described in
terms of a thermal emission in that case, i.e. the addition of a mekal component does
not significantly improve the fit (∆χ2 < 2).
Table 3.5: Results of the best fit models for the star forming galaxies with enough
number of counts to be fit (>50 counts). ID is the numeric identifier of the source; Γ is
the X-ray photon index; kTe gives the electronic temperature in units of keV; PF −test
shows the probability of the F-test (if it is low, close to zero, then it is reasonable to
add the mekal component to the power-law model); ∆χ2 give the decrease of the χ2
when the mekal component is added.
ID model
Γ
kTe
χ2ν /d.o.f. PF −test ∆χ2
0.28
0.51
<0.2
∼ 13
8
powerlaw+mekal
2.16±0.41 0.901±0.12 1.100/8
a
0.16
0.88
79 powerlaw+mekal
2.35±0.16 3.71±3.14 1.053/98
∼0
> 200
0.08
233 powerlaw+mekal
2.11±0.07
0.62±0.05
1.163/90
∼0
>
100
0.05
................................................................................
0.22
1.126/19
56 powerlawb
1.84±0.20
0.42
0.952/5
>0.9
<6
149 powerlaw
1.83±0.38
0.31
164 powerlaw
1.94±0.29
1.219/14
>0.7
<1
0.21
246 powerlawb
1.88±0.19
0.625/22
0.27
251 powerlaw
1.40±0.25
1.270/5
>0.4
<4
21
a Plus an intrinsic absorption column density: NH,z = (2.1±0.2
atoms cm−2
0.3 ) 10
b The fit is insensitive to the electronic temperature of the B model.
3.6.3.
Missing-AGN
versus BPT-AGN
We then assessed the different nature of the missing-AGN sources in terms of
spectral fitting of known type-2 Seyfert. To avoid composite objects we adopted the
Kewley et al. (2001) criterion to secure optically classified AGN. In order to perform a
proper spectral fit to a significant amount of the BPT-AGN, we only used the sources
with a minimum of 50 counts in at least one detector. This requirement resulted in the
selection of 56 BPT-AGN. From inspection of the X-ray data, we excluded 15 of these
objects that were dominated by the background above ∼ 2-3 keV. Finally, we removed
those sources that had previously been classified as LINERS in the literature. The final
AGN MISSED BY THE BPT DIAGRAM
70
sample contains 34 bona-fide type-2 BPT-AGN candidates.
The results of our analysis are as follows:
1. The spectra of 10 type-2 AGN (∼ 29%) are best-fitted with a single power-law.
The average Γ obtained as a function of the 2 − 10 keV flux is marginally softer
(hΓi ' 1.8) than the typical values of ∼ 1.9 and ∼ 2.1 found in the unabsorbed
AGN and the missing-AGN subsamples, respectively.
2. The spectra of 9 AGN (∼ 26%) could be best fit by the inclusion of intrinsic
absorption at the level of a few ×1022 cm−2 and the average Γ is also hΓi ' 1.8.
3. The addition of another PL component as a proxy for a partially covered absorber
or scattering of the AGN light, was required to achieve good spectral fits in
another 14 AGN (∼ 41%). We found that the average photon index is hΓi = 2.2
and the average intrinsic column density hNH,z i = 5 × 1023 cm−2 . The measured
amounts of X-ray absorption for the missing-AGN subsample are lower (by two
orders of magnitude), falling in the low part of the column density distribution of
type-2 AGN. The average measured strength of the non-absorbed X-ray primary
emission by the neutral material was measured to be softer (∼ 8%) than the
average of the missing-AGN subsample. Soft excess emission is not detected
with an F-test significance > 99% in the vast majority of the objects.
4. Finally, we found a significant soft X-ray excess in only 1 AGN (∼ 3%) clearly a
negligible fraction compared with the strong soft excess displayed by the missingAGN subsample.
We have seen that the X-ray emission and the optical data of the missing-AGN
population are consistent with an NLS1 nature, and clearly differ from those of the
type-2 AGN in our sample (see Table 3.6).
3.7.
Summary and Conclusions
The availability of catalogues at different wavebands with wide sky coverage, allowed
us to assemble optical (SDSS-DR7) and X-ray (2XMMi-DR3 catalogue) information for
a large sample of NELGs. A total of 1729 NELGs fall in the region covered by XMMNewton observations, with which the 2XMMi-DR3 catalogue was built. Out of these,
we find 211 X-ray detections and 1518 upper limits.
We have compared the optical classification based on the BPT diagram with their
hard X-ray luminosity, used as an indicator of the AGN activity. Among the 211 X-ray
3.7. Summary and Conclusions
71
Table 3.6: Summary of the results of the XMM-Newton spectral analysis results.
From left to right: name of the subsample, total number of best fitted sources with
the indicated model, XSPEC model definition, mean value of the X-ray photon index,
mean value of the intrinsic column density in units of atoms cm−2 , tick to denote
whether the vast majority of the sources show a soft X-ray excess component.
N model
hΓi hNH,z i soft excess
10 powerlaw
1.8
−
7
BPT-AGN?
9 powerlaw×zphabs
1.8 3 × 1022
7
14 powerlaw×zpcfabs
2.2 5 × 1022
7
1 powerlaw×zpcfabs
2.2 1 × 1021
3
.......................................................................................
8 powerlaw+zbbody
2.3
−
3
missing-AGN 6 powerlaw×zpcfabs
2.7 < 1022
3
7 powerlaw[+zbbody/×zpcfabs] 2.2
−
3
.......................................................................................
5 powerlaw
1.8
−
7
true-SF
??
3 powerlaw+mekal
2.2
3
? There is only one source with soft-excess.
?? There is only one source with intrinsic absorption, but it is below 1022 atoms cm−2
subsample
detected NELGs (all at z < 0.4 because of our selection criteria) we find that 145
galaxies are diagnosed as AGN according to the BPT diagram, having 2-10 keV X-ray
luminosities from 1042 erg s−1 to above 1044 erg s−1 . About 43% (62/145) of these AGN
exhibit low X-ray luminosities LX . 1042 erg s−1 (weak-AGN). Using other optical
spectral features, such as the [OI] and [OII] emission lines, we find that 13% of these
weak-AGN are classical as LINERs and 29% are likely to be weak-[OI] LINERs. The
soft component in this kind of sources may arise from circumnuclear star formation
(González-Martı́n et al., 2006) that could also explain their emission-line ratios.
Out of the remaining 66 X-ray detected NELGs, which are instead diagnosed as
starforming galaxies according to the BPT diagram, we find that about 42% (28/66)
of them exhibit high luminosities & 1042 erg s−1 (missing-AGN), while the remaining
58% (38/66) have lower X-ray luminosities (true-SF) in agreement with their optical
line spectroscopy diagnostic.
To investigate these disagreements between the optical line spectroscopy diagnostic
and the hard X-ray luminosity, we have calculated the thickness parameter (T ), the
X-ray-to-optical flux ratio (XO) and the hardness ratio (HR) for the whole NELG
sample. We found that it is not clear that we can distinguish between SF and AGN
galaxies using a single criterion (Kauf03 or LX ). However, the combined use of XO, T
and HR allows us to distinguish between SF galaxies and AGN. For our sample of 211
X-ray detected NELGs, we found that the distributions of both the thickness parameter
(T ) and the X-ray-to-optical flux ratio (XO) are bimodal, with the two populations
AGN MISSED BY THE BPT DIAGRAM
72
being separated by about T ∼ 1 and XO∼ 0.1, respectively. We noted dichotomies
in the SF population, i.e. between the missing-AGN and the true-SF galaxies, and
on the other hand, within the AGN population (according to the BPT diagram), i.e.
between the weak-AGN and strong-AGN subsamples.
Finally to unravel the real nature of the missing-AGN galaxies, we have performed
a comprehensive X-ray spectral analysis of the subset of those sources with enough
signal-to-noise ratio to determine whether the emission is coming from star formation
processes or from AGN activity.
The most striking result of this inquiry is that despite the missing-AGN population
representing only a small percentage (2-7%) of the overall SF population -and therefore
presenting a minor problem in terms of contamination- these 28 sources are unquestionably NLS1. The dichotomy in the starforming population is directly linked to the
values of Hβ FWHM: all the galaxies with high LX exhibit the broadest widths in
the Hβ line, from ∼ 600 to 1200 km/s, whilst the remaining sources with LX < 1042
erg s−1 , display Hβ FWHM . 600 km/s. Indeed, the missing-AGN subsample has
high values of both XO and T and low measured hardness ratio for each source. So
we conclude that these missing-AGN are NLS1 candidates whilst the rest (true-SF
population) are consistent with being SF galaxies. Strong supporting evidence for the
NLS1 nature of the missing-AGN population comes from the spectral analysis of their
X-ray emission. The X-ray spectral properties of the missing-AGN subsample appear
to be quite uniform. These spectra are well reproduced by the combination of black
body or a partial covering absorption on a steep power-law component at hard X-ray
energies. Furthermore, we have established evidence of the missing-AGN population
displaying a soft X-ray excess, whenever spectra of sufficient S/N is available to model
them. The missing-AGN population has Hβ lines FWHMs larger than 600 km/s, and
often display strong Fe II emission. Therefore, we conclude that the population of the
missing-AGN subsample is mainly constituted of NLS1s with very moderate broadline widths (600 km/s . Hβ FWHM . 1200 km/s), which have XO> 0.1, T > 1 and
most of them strong soft X-ray excess (see Chapter 4 for an extended analysis).
All this probably shows that a NLS1 core can be identified in a large fraction of
X-ray luminous galaxies but optically diagnosed as starforming galaxies, being ∼25%
of hard X-ray-selected AGN samples (28/(28+83)).
3.7. Summary and Conclusions
73
EFE
EFE
10 -4
10 -4
10 -5
2
ratio
ratio
15
10
5
1.5
1
0
0.5
1
10
1
E [keV]
10
E [keV]
(i) ID=26
(ii) ID=42
10 -3
EFE
EFE
10 -4
3
2
2.5
1.5
2
ratio
ratio
10 -4
1.5
1
0.5
1
0.5
0
1
10
1
E [keV]
EFE
(iv) ID=63
EFE
(iii) ID=53
10 -4
10 -3
2.5
3
2.5
2
ratio
ratio
10
E [keV]
2
1.5
1.5
1
1
0.5
0.5
1
10
E [keV]
(v) ID=65
1
10
E [keV]
(vi) ID=71
Figure 3.12: X-ray spectral fitting for each object within the missing-AGN subsample. The order of the objects in the figure follows that in Table 2. The individual
hard (2-10 keV) X-ray data has been fitted with a simple power-law model corrected
from Galactic absorption. The soft X-ray excess is defined as the excess over an extrapolation down up to 0.3 keV of the best power-law model fitting to the hard X-ray
band. We note that for the three sources ID=163,302,335 we have no X-ray spectrum
due to the low counts in the X-ray energy band.
AGN MISSED BY THE BPT DIAGRAM
74
10 -3
EFE
EFE
10 -3
10 -4
2.5
8
2
6
ratio
ratio
10 -43
1.5
4
2
1
0.5
1
10
1
E [keV]
10
E [keV]
(vii) ID=100
(viii) ID=104
10 -4
EFE
EFE
10 -4
10 -5
10 -5
10 -64
6
2
ratio
ratio
3
1
4
2
0
-1
1
10
1
E [keV]
10
E [keV]
(ix) ID=111
(x) ID=125
EFE
EFE
10 -4
10 -5
10 -5
-6
1015
1.4
10
ratio
ratio
1.2
1
0.8
0.6
5
0
1
10
1
E [keV]
10
E [keV]
(xi) ID=129
(xii) ID=161
EFE
EFE
10 -3
10 -4
10 -3
2.5
3
2.5
ratio
ratio
2
1.5
1
2
1.5
1
0.5
0.5
1
10
E [keV]
1
10
E [keV]
(xiii) ID=171
(xiv) ID=191
Figure 3.12: Continued.
3.7. Summary and Conclusions
75
EFE
EFE
10 -3
10 -4
5
1.5
ratio
ratio
4
3
1
2
1
0.5
1
10
1
E [keV]
10
E [keV]
(xv) ID=195
(xvi) ID=203
EFE
EFE
10 -4
10 -5
-5
102.5
15
ratio
ratio
2
1.5
10
5
1
0.5
1
10
1
E [keV]
10
E [keV]
(xvii) ID=204
(xviii) ID=214
EFE
EFE
10 -4
10 -3
10 -5
3
8
ratio
ratio
6
4
2
2.5
2
1.5
1
0.5
0
1
10
1
E [keV]
10
E [keV]
(xix) ID=241
(xx) ID=275
10 -4
EFE
EFE
10 -3
10
10 -4
-5
10 -5
2
ratio
ratio
2.5
1.5
40
20
1
0
0.5
1
10
E [keV]
1
10
E [keV]
(xxi) ID=318
(xxii) ID=329
Figure 3.12: Continued.
AGN MISSED BY THE BPT DIAGRAM
76
10 -3
EFE
EFE
10 -4
-5
102.5
-4
102.5
ratio
2
1.5
1
1.5
1
0.5
0.5
1
10
1
E [keV]
EFE
10
E [keV]
(xxiii) ID=338
(xxiv) ID=355
10 -4
2.5
ratio
ratio
2
2
1.5
1
0.5
1
10
E [keV]
(xxv) ID=357
Figure 3.12: Continued.
CHAPTER 4
Non-Standard NLS1 galaxies
4.1.
Motivation: non-standard NLS1
The current widely accepted NLS1 paradigm is that NLS1 are BLS1s with black
holes of relatively modest mass that are accreting matter at or above their Eddington
limits, as measured in the majority of the sources. In this scenario, the narrower line
widths could be explained with a larger BLR distance to the black hole in NLS1 galaxies than those of BLS1 galaxies, as a result of the effects of winds/outflows powered
by supermassive black holes accreting close to the Eddington limit. Due to the high
accretion rates, the discs in NLS1 are likely hotter than those in BLS1, and hence the
emission is peaked at higher energies. Mathur (2000) suggested that NLS1 as high-rate
accretors might be Seyfert galaxies in their early stage of evolution residing in rejuvenated, gas rich galaxies and as such may be low redshift, low luminosity analogues of
high redshift quasars. Pounds et al. (1995) claim that NLS1 could be analogous to the
strong soft states of galactic black holes at high accretion rate.
The optical classification of NLS1 is not always straightforward given the superposition of various Hβ emission line components. As claimed in the Introduction, the
observed X-ray properties, as well as the optical features are not universal among the
NLS1 population: not all NLS1 behave the same. Ai et al. (2011) suggested the existence of two kinds of NLS1 with low black hole masses: those NLS1-like with strong
Fe II, soft X-ray excess and high-Lbol /LEdd , and others more similar to BLS1, i.e., weak
Fe II and modest or non-existing soft X-ray excess. However, these two kinds of NLS1
are unable to explain the findings of Zhou et al. (2006) where about 15% of the objects
have weak Fe II emission, but high Eddington ratio. Likewise, we also analyze two targets in this Chapter whose X-ray spectra do not shown any sign of soft X-ray emission
despite being strong Fe II optical emitters.
NON-STANDARD NLS1 GALAXIES
78
In this Chapter we analyze a sample of 19 NLS1 galaxies, constructed in Chapter 3. These are hard X-ray selected AGN, whose optical spectra reveal the Balmer
Hβ emission line being narrower than 2000 km s−1 . We use optical (SDSS), ultraviolet (XMM-Newton/OM) and X-ray (XMM-Newton/EPIC) data to derive the essential
phenomenological parameters (X-ray soft excess, Fe II emission) for these sources. In
our analysis we find that 15/19 of the NLS1 display both strong Fe II emission in the
optical spectrum and soft X-ray excess, as the majority of the NLS1. However, 2/19
display strong Fe II emission but no soft excess and the remaining 2/19 do not show
any detectable Fe II emission but show soft X-ray excess.
The main question addressed in this Chapter is what is the nature of these 2+2
non-standard NLS1, vis a vis the normal NLS1 and BLS1 populations. For the purpose
of deriving physical properties (like supermassive black hole mass or Eddington ratio),
and as an improvement over previous work, we employ a new broadband SED model
(Done et al., 2012), which combines disc emission, low temperature Comptonisation
affecting the soft excess and high-temperature Comptonisation giving rise to the hard
X-ray power law. This model provides us with an energetically self-consistent and coherent way to fit the whole collection of multi-wavelength data for each source, and to
derive meaningful physical parameters.
This Chapter is organized as follows: Sect. 4.2 describes the sample and the data
compilation; the optical properties of the sample are presented in Sect. 4.3; the broadband SED model and results are described in Sect. 4.4 and Sect. 4.5; Sect. 4.6 summarizes the conclusions we obtained from this study. There is a draft manuscript containing the results of this Chapter that will be submitted for publication to Astronomy &
Astrophysics.
4.2.
Sample and Data Preparation
The NLS1 sample studied in this work was introduced in Chapter 3 as part of
the missing-AGN subsample, based on the cross-correlation of the SDSS DR7 and
XMM-Newton DR3 catalogs. These sources are hard-X-ray-selected narrow emission
line AGN, with X-ray luminosities in excess of 1042 erg s−1 , but whose BPT diagnostic
placed them in the “star-forming” region. The NLS1 nature was derived according to
the simple criteria of Goodrich (1989). To ensure sufficient X-ray spectral quality, we
only consider those NLS1 with a minimum of 500 counts in at least one of the three
XMM-Newton EPIC cameras. This additional quality threshold results in a final sam-
4.3. Optical Spectral Properties
79
ple of 19 NLS1 (∼70% of the missing-AGN subsample).
We use the same X-ray spectra as in Chapter 3 for each individual source. For 10
out of 19 NLS1 galaxies also had XMM-Optical Monitor (OM) data (in at least one
of the V, B, U, UVW1, UVW2 filters) which extends the SDSS optical coverage into
the UV. Finally, EPIC spectra and the OM photometric points are combined with the
SDSS optical continuum points using the FTOOLS task flx2xspec tool. The SDSS
optical points are obtained from the underlying continuum approximated by a power
law after removing the Fe II false continuum and all strong emission lines. The reason
for that is that the SED model that we use (see Section 4.4) does not take into account
optical/UV emission lines. Combining these data (SDSS, OM) reduces the impact of
intrinsic variability and provides a good estimate of the spectral shape in the optical,
near UV and X-ray bands. These data allow us to construct a broadband nuclear SED
for each NLS1. There is an ubiquitous data gap in the far UV region which is due to
photoelectric absorption by Galactic gas. Unfortunately, in most cases of low-redshift
AGN, their intrinsic SED also peaks in this very UV region, and so this unobservable
energy band often encompasses a large portion of the bolometric luminosity. In order
to account for this, and to estimate the bolometric luminosity, we fit the X-ray and
UV/optical continua all together using a new broadband SED model presented by Done
et al. (2012).
4.3.
Optical Spectral Properties
In order to estimate both the relative strength of the Fe II and the black hole
mass, we must model the Balmer and [O III]λ5007 emission lines, as well as the Fe II
multiplets. Whilst the latter is usually expressed as the flux ratio of Fe II to Hβ, R4570
(see Eq. 3.5), the Hβ line width is used as a proxy for the black hole mass (see Woo &
Urry, 2002, and references therein) from
MBH [M ] = 4.817 ×
λLλ (5100Å)
1044 erg/s
0.7
2
F W HMHβ
(4.1)
where Lλ (5100Å) is obtained directly from the SDSS spectra. Comparing this method
for a sample of AGN with reverberation mapping no strong biases are evident and the
rms difference is about 0.5 dex (Woo & Urry, 2002).
Estimating the FWHM of the Hβ line is not straightforward as the line profile is
often quite complex. The narrow component to this Balmer line can be estimated by
NON-STANDARD NLS1 GALAXIES
80
Figure 4.1: Example of detailed line profile fitting to the Fe II subtracted region around
the Hβ including the [O III]λ5007/4959 doublet. In our profile fitting, three Gaussian
components are used for Hβ and Hα, two components for [O III]λ5007 (the narrow component of the Balmer lines is assumed to be similar to the [O III]λ5007 profile), and one
Gaussian for all other lines. The various Gaussian profiles are shown in blue, the total
model is shown in red.
analyzing the nearby [O III] doublet. However the [O III]λ5700Å line is probably contaminated by Fe II emission. Hence we perform our own fit of the Balmer emission
line profile once the Fe II emission is subtracted. We have used a home made code to
perform the Balmer emission line fitting after removing both the Fe II emission and
the underlying continuum. A detailed description of our spectral modelling procedures
is presented in Chapter 2. Briefly, the Hα and Hβ lines were fitted using three components: the narrow component has the same FWHM as the entire [O III]λ5007 line,
i.e. including both the central and blue component in [O III]λ5007; the intermediate
and broad components are assumed to have a Gaussian shape. All other strong nearby
emission lines are included by adding more Gaussian profiles into the whole model.
Then a complete model with up to three components per line and multiple lines was
used to fit the whole SDSS spectra, including the underlying continuum approximated
by a power law and the Fe II false continuum and all strong emission lines. Our approach is to fit Hα and Hβ simultaneously using the same multi-Gaussian components.
An example of results from SDSS spectra fitting is shown in Figure 4.1.
The most relevant properties of the fitted SDSS optical spectra are presented in
Table 4.1: the monochromatic luminosity at 5100 Å, the relative strength of the Fe II
multiplets and the black hole mass. The latter was estimated using on the one hand,
the broad component only, and on the other, the narrow-component-subtracted Hβ
line component (which is contaminated by any intermediate component). Instead of
computing directly the parameters (FWHM, flux, etc.) of the narrow-component-
4.3. Optical Spectral Properties
81
subtracted Hβ, we have calculated these parameters from the resulting profile of coadding the two best fit Gaussian profiles for the broad and intermediate line components
given by our Balmer line fitting process.
Only for 2 out of 19 sources (ID=203,241) no Fe II emission could be detected (hereafter Fe II-deficient NLS1). The optical spectrum of these two sources with very narrow
Balmer lines that exhibit a prominent He II broad emission line ensures their classification as type-1 AGN. We found that the FWHM of the narrow-component-subtracted
Hβ line is amongst the highest values of the sample for those two Fe II-deficient NLS1.
The sample contains another two objects (ID=129,357) whose X-ray spectrum also
appears to be well fit by a single power law with no obvious soft X-ray excess (hereafter X-ray-flattened NLS1). They are, however, strong iron emitters with a mean
value of R4570 = 1.41 ± 0.18. The Fe II emission of these two sources is really significant compared to the R4570 ∼1 average for NLS1 (Véron-Cetty et al., 2001). Despite
not exhibiting obvious soft excess, these two sources are optically classified as totally
normal NLS1 according to Goodrich (1989).
The remaining 15 objects (hereafter standard NLS1) that satisfied the two universal
criteria of Goodrich (1989), have proved to be strong iron emitters, hR4570 i = 1.3 ± 0.3.
These standard NLS1 galaxies exhibit the most extreme properties that define the
NLS1 class: broad Hβ emission narrower than BLS1, strong Fe II emission and steep
soft X-ray spectrum.
4.3.1.
Optical reddening
We now address the computation of the Balmer decrement for these sources, as it
may contain information about the nature or environment of the central engine. A
change in the Balmer decrement may arise from changes in the physical conditions of
the partially-ionized line emitting regions (BLR, NLR, ILR). For example, a decrease
of the Balmer decrement may be due to an increased electron density, or an increased
ionization parameter. A high Balmer decrement can also be explained by a high dust
abundance. This argument was used in Zhu et al. (2009) as evidence to support the
link between the intermediate line region and the dusty torus. Therefore, the Balmer
decrements can be used as clues to infer the physical conditions of the emission line
region, as well as to obtain the optical reddening to estimate the equivalent X-ray absorption.
NON-STANDARD NLS1 GALAXIES
82
Table 4.1: Key parameters from the fit to the optical spectrum of the NLS1 sample.
ID is the object number (the same as in Chapter 3); z is the spectroscopic redshift;
λL5100Å represents the monochromatic luminosity at 5100Å in units of 1044 erg s−1 ;
F W HMHβ provides the FWHM (in units of km s−1 ) of Hβ given by the SDSS data,
the narrow component of our Balmer fit (nlc) and the narrow-component-subtracted
Balmer line (nlc-sub) profile; log (MBH /M ) represents the estimated log of the black
hole mass in solar units given by the Eq. 4.1 using as a proxy the intermediate and
broad Balmer line components of our Balmer fit (ilc and blc, respectively); R4570
is the relative strength of the Fe II multiplets where ∗ denotes those objects whose
classification as NLS1 was based on the prominent He II broad emission line and not
on the Fe II strength; Hα/Hβ is the measured Balmer decrement in the NLR.
ID
z
λL
5100Å
×1044 erg/s
F W HMHβ [km/s]
sdss nlc nlc-sub
log (MBH /M )
ilc
blc
R4570
Hα/Hβ
nlc
non-standard NLS1
• Fe II-deficient NLS1
203 0.117
0.225
241 0.263
1.746
1246
1208
204
265
1711
1286
6.53
6.91
7.86
8.04
∗
∗
4.6
6.6
• X-ray-flattened NLS1
129 0.285
1.82
357 0.064
0.219
1206
1062
317
261
1195
1119
6.85
6.15
8.49
7.68
1.23
1.59
4.9
3.8
6.51
5.95
6.02
5.80
6.30
6.57
6.49
6.64
6.71
5.92
6.09
6.78
6.77
7.06
6.65
7.44
7.02
6.99
7.04
7.41
7.79
8.26
7.66
8.48
7.32
7.41
8.21
8.10
8.48
8.35
1.03
1.20
1.25
1.59
1.20
1.55
1.83
0.77
1.52
1.56
1.51
1.16
1.16
0.88
1.41
2.4
5.1
3.8
4.4
3.7
4.6
2.2
3.7
1.9
4.2
4.1
5.4
3.5
4.9
2.1
standard NLS1
26
42
53
65
71
100
104
125
171
191
195
214
275
318
329
0.135
0.077
0.106
0.081
0.048
0.081
0.160
0.145
0.186
0.031
0.043
0.272
0.065
0.367
0.098
0.374
0.121
0.203
0.197
0.419
0.564
1.911
0.558
2.211
0.245
0.191
2.428
2.077
2.279
1.844
1164
941
780
667
760
946
1032
1190
928
831
848
673
898
1299
1085
337
231
231
241
357
276
863
211
385
324
301
274
230
662
697
1395
1088
983
776
1051
1298
769
1420
944
821
1087
993
1032
1406
942
Last column of Table 4.1 shows the observed NLR Balmer decrement. Figure 4.2
shows the Balmer decrement distribution for both emission line regions (NLR, BLR).
The mean Balmer decrements with 1 standard deviation are 3.9 ± 1.2 and 2.7 ± 0.7
(NLR and BLR, respectively). We believe that the Balmer decrement we found for
the NLR is correct, based on the fact that this component has a narrow width, and is
matched to the observed [O III] profile. Their values in the NLR are also in agreement
with those from other NLS1 in previous works. Jin et al. (2012) reported ratios of the
Hα and Hβ lines ranging from ∼2.0 to ∼7 with a mean value of 4.8 ± 1.8, which is very
4.4. Broadband SED Modelling
83
similar to what we found (see also Shuder, 1982). However, disentangling the broad
component of the narrow profile may introduce significant uncertainties. Thus, we only
used the values given by the narrow component to estimate the intrinsic absorption,
which yields in general higher values.
BLR
8
NLR
number
6
4
2
0
1
2
3
4
5
6
7
Hα/Hβ
Figure 4.2: Balmer decrement distributions of different Balmer line components:
NLR and BLR.
There are four 4 (4/19) NLS1 galaxies with a Balmer decrement below 3.1 which is
the typical value for the NLR. Although the distribution peaks around 4.0, the large
range of Balmer decrements may probably imply the presence of some dust in the NLR
and it could also explain these higher values. The remaining NLS1 galaxies (15/19) are
expected to have some intrinsic absorption according to the optical reddening.
Obtaining the Balmer decrement for the NLS1 that do not apparently exhibit a
soft excess is of special importance, as a large reddening would imply (for normal gas
to dust ratio) significant X-ray absorption that might affect the X-ray spectrum. The
measured optical reddening predicts an intrinsic absorption of 2.1 × 1021 cm−2 and
1.0 × 1021 cm−2 for the X-ray-flattened NLS1 (ID=129,357 respectively). We will come
back to this issue later.
4.4.
Broadband SED Modelling
As discussed in Section 4.1, we use an energetically self consistent model to fit the
SED of the NLS1s, in an attempt to understand their observational properties in terms
of physical parameters, especially in the case of the 2+2 non-standard NLS1 (Fe II-
NON-STANDARD NLS1 GALAXIES
84
deficient and X-ray-flattened NLS1). Standard interpretations of the broadband SED
assume that the emission is dominated by a multi-temperature accretion disc component which peaks in the UV (e.g. Gierliński et al., 1999, XSPEC model: diskpn). This
produces the seed photons for Compton up-scattering by a hot, optically thin electron
population within a corona situated above the disc, resulting in a power-law component above 2 keV (e.g. Haardt & Maraschi, 1991, Zdziarski et al., 2000, XSPEC model:
bknpl). However, this simple interpretation does not explain the soft X-ray excess
whose origin is still uncertain (Crummy et al., 2006, Gierliński & Done, 2004, Miller
et al., 2008). Previous broadband SED modelling studies have explicitly excluded data
below 1 keV making it possible to fit the data using just a disc and (broken) powerlaw continuum (Vasudevan & Fabian, 2007, 2009). In our current study we include
all of the X-ray data (0.3 − 10 keV) plus optical and UV data points, and so we
require a self-consistent model which incorporates this soft component. This is particularly important to check whether the X-ray-flattened NLS1 galaxies have at least
a marginal soft-excess component. Whatever their origin, a simple approach to fitting the soft X-ray data is to assume that there is an additional optically thick, low
temperature Compton up-scattered component (XSPEC model: compTT). However,
it is difficult to uniquely determine the parameters of three independent components
(diskpn+compTT+bknpl) especially given the gap in spectral coverage between the
UV and soft X-ray regions caused by interstellar absorption. So instead, we use the
model optxagnf (Done et al., 2012) which assumes that the Comptonised emission
should be powered ultimately by the accretion flow, so the two components (diskpn,
compTT) should be energetically coupled.
4.4.1.
The broadband model: optxagnf
optxagnf is in essence a faster version of the models recently applied to black hole
binary spectra observed close to their Eddington limit (Done & Kubota, 2006) and to
the (possibly super Eddington) Ultra-Luminous X-ray sources (e.g. Middleton & Done,
2010).
The broadband SED model optxagnf used in this Chapter, consists of the following three continuum components:
Color-temperature-corrected blackbody component. The broadband SED model assumes that the gravitational energy released in the disc at each radius is emitted
as a colour-temperature-corrected blackbody only down to a given radius, Rcor
(red line in Figure 4.3). The disc spectra are more complex than a simple sum of
4.4. Broadband SED Modelling
85
blackbodies at the effective temperature1 , because the NLS1’s disc becomes dominated by electron scattering rather than absorption in those disc regions where
the temperature is well above the Hydrogen ionization energy of 13.6 eV (Ross
et al., 1992) making the colour-temperature-corrected blackbody a more realistic
emission model (see Done et al., 2012, for an extended explanation).
Rout = 105rg
Rcor
10-1
EF E
10-2
10-3
10-4
10-5
-3
10
10-2
10-1
E [keV]
1
10
Figure 4.3: Conceptual scheme of the model geometry and its resultant spectrum
(black line), with an outer disc (red line) which emits as a colour-temperature-corrected
blackbody, and an inner disc (green line) where the emission is Compton up-scattered
by a low temperature, optically thick electron population. Some fraction of the energy
is also Compton up-scattered by a high temperature, optically thin electron population
in a corona (blue line) to produce the hard X-ray power-law tail.
Comptonisation components. Below Rcor , the emission starts to emerge as Comptonised rather than colour-temperature-corrected blackbody, and it is distributed
between the soft excess component and the high energy tail. The model also
assumes that this Compton up-scattering takes place in the disc itself, so the
luminosity in this Compton up-scattered component is completely determined
by the integral of the emissivity from Rcor to risco . The energy within Rcor is
1
This is because the absorption opacity decreases significantly as the black hole mass increases.
Then for some AGN the accretion disc may no longer be locally thermalised: the highest temperature
photons can emerge from regions deeper in the disc, and so the disc emission extends to higher energy
than for standard accretion disc spectrum, producing the effect of a colour temperature correction. The
maximum effective temperature of the accretion disc is kTe ∼ 10(ṁ/(M/108 ))1/4 eV, so only for AGN
with both low mass and high mass accretion rate such as the NLS1 (e.g. Boller et al., 1996), electron
scattering opacity will dominate, and the effect of colour-temperature correction becomes important.
NON-STANDARD NLS1 GALAXIES
86
emitted as Compton up-scattered flux, with seed photons characterised by the
colour-temperature-corrected disc temperature at the “transition” radius Rcor .
- The energy dissipated by Compton up-scattering by a low temperature,
optically-thick electron population mainly accounts for the soft X-ray excess (green line in Figure 4.3). This component is internally modeled using
the code compTT of XSPEC.
- The energy emission above 2 keV (blue line in Figure 4.3) must also ultimately be derived from mass accretion, so some fraction of the energy
dissipated from Rcor up to risco powers this high energy and optically thin
Comptonisation component. It is internally modeled using the code nthcomp of XSPEC.
Thus the optxagnf model contains three distinct spectral components, but these
are all powered by the energy released by a single accretion flow of constant mass accretion rate, Ṁ , onto the black hole of mass MBH . The full model includes all three
components which are known to contribute to AGN SED in a self-consistent way. As
such it represents an improvement on the fits in several respects, by including the soft
excess and by requiring energy conservation, as well as by including the power-law tail.
At the same time, the internal self-consistency of the model results in a reduced number
of free parameters.
4.4.2.
Fitting Process Methodology
We use optxagnf in XSPEC v12 to perform the broadband SED fitting. The two
key free parameters are the black hole mass and the mass accretion rate in terms of the
Eddington ratio. The optical/UV data constrain the mass accretion rate through the
outer disc, provided we have an estimate of the black hole mass. So the total energy
available is determined by the accretion efficiency1 .
While the outer disc emission is a colour-temperature-corrected blackbody from
Rout up to Rcor , a fraction fpl of the remaining energy emitted as the mass accretes
from Rcor to risco appears as a high energy (>2 keV) Comptonisation, characterised by a
power law of photon index Γ and electron temperature fixed at 100 keV. The remaining
energy (1 − fpl ) is emitted as a low temperature, optically thick Comptonisation of the
disc emission, parameterised by kTe and τ . We have included two sets of corrections
1
Assuming an overall efficiency of 0.057 (a stress-free emissivity for a Schwarzschild black hole) for
an inner radius of 6rg , the total luminosity of the soft excess and power-law is 0.057Ṁ c2 (1 − 6rg /Rcor )
4.4. Broadband SED Modelling
87
for attenuation to account for the line of sight Galactic absorption and for the absorption intrinsic to each source, the latter being redshifted (XSPEC model: redden,
and zwabs, respectively). redden is used in place of zwabs or zphabs in order to
use the same used model for the BLS1 control sample analysis given by Jin et al. (2012).
All sources were analyzed using the following approach:
• Some initialisations:
− The intrinsic column density NH,int is left as a free parameter, whilst the
standard dust to gas conversion formula E(B − V ) = 1.7 × 10−22 NH.Gal is
used to fix the galactic reddening through the Galactic absorption NH,Gal
given by Dickey & Lockman (1990).
− We fixed the outer and inner disc radii: Rout = 105 rg ,risco = 6 rg , the latter
assuming a non-rotating black hole.
− The input free parameter Rcor constrains the model in the unobservable
EUV region insomuch as it sets the model output of the luminosity ratio
between the standard disc emission and Comptonisation components. The
upper limit of Rcor is set to be 100 rg , which corresponds to 81% of the
released accretion disc energy, which is based on the requirement that the
seed photons should be up-scattered (for an extended explanation see Done
et al., 2012).
− We have constrained the value of MBH to be between the region limit defined by the Hβ intermediate and broad emission line components estimated
by our Balmer line analysis (within an additional 0.5 dex error, possibly
appropriate as a systematic error for this proxy).
− The initial value of the photon index is set to be that of the photon index in
the hard energy band. The upper limit of Γ is set to be 2.2, not only because
the photon index is <2.2 for the majority of type 1 AGN, but also because
otherwise the much higher S/N in the soft excess in some observed spectra
can artificially steepen the hard X-ray power law and result in unphysical
best-fit models.
− We set some parameters at some fiducial values: τ = 15, kTe =0.2 keV and
fpl = 0.3
− The upper limit of kTe is set to be 1.0 keV based on the typical values of
the temperature for the soft X-ray excess (normally around 0.2 keV).
NON-STANDARD NLS1 GALAXIES
88
• Fitting process:
1. Holding fixed all parameters except MBH , λEdd and Rcor , the data is modeled to set the Eddington ratio λEdd = Lbol /LEdd .
2. Now, while MBH , λEdd and Rcor are frozen, τ , kTe , fpl and NH are released
to model the optical and X-ray data.
3. All the parameters (MBH , λEdd , Rcor , kTe , τ , fpl , NH ) are released and the
data is again modeled to obtain a set of physically plausible values.
4. Finally, once the values of MBH , λEdd , Rcor , τ and kTe have a physically
plausible values, Γ is released to obtain the best-fit model.
4.4.3.
Problems in the SED fitting
For some AGN, their SDSS continuum data points exhibit a very different spectral slope from that of the SED model. This cannot be reconciled by adjusting the
parameters of the accretion disc model, and thus implies the presence of an additional
component at longer optical wavelengths, which is flatter than that predicted by the
accretion disc models. A relatively modest contribution from the host galaxy could
play this role. However, since this would not affect the SED at energies higher than
optical (remember that our sources have X-ray luminosities in excess of 1042 erg s−1 ),
we have chosen not to complicate the model any further, while allowing for some misfit
at optical wavelengths.
Another additional problem in our fitting procedure is the occasional discrepancy
between OM photometry and the SDSS continuum. The OM data often appear below
the extrapolation of the SDSS continuum to the OM wavelengths. This discrepancy
may arise for several reasons, including intrinsic source variability due to the significant
time difference between the acquisition of the SDSS and the OM data. Another reason
could be the contamination by emission lines within the OM wavelength ranges; or
even a contribution by extended emission from the host galaxy, as the aperture for the
00
00
OM photometry is ≥ 12 , much higher than the 3 of the filters used in the SDSS.
These three factors could work together to produce the observed discrepancy between
the SDSS and OM data.
We should treat the optical/UV points included in our SED modelling as upper
limits when interpreting the results of our modelling. Indeed, the statistical errors in
the BH masses derived directly from the XSPEC fitting are almost certainly smaller
than the systematic errors introduced by the above uncertainties. Hence in the cases
were the discrepancy between the OM and the SDSS points was really significant, we
4.5. Physical Properties of the NLS1 sample
89
have decided do not include the OM points in the analysis due to the reasons explained
above.
4.5.
Physical Properties of the NLS1 sample
We construct SEDs ranging from about 0.9 microns to 10 keV for the 19 hard-Xray-selected NLS1 with high quality optical spectra and in some cases simultaneous
optical/UV photometric data points from the XMM-Newton OM. Best-fit broadband
SED parameters are listed in Table 4.2. We calculate the bolometric luminosity by
integrating the model emission using the best-fit parameters obtained for each continuum component. We also explicitly calculate the fraction of the soft X-ray excess fSE
in the 0.3-2 keV energy range, as the ratio between the luminosity carried by the soft
Comptonisation component and the luminosity carried by both Comptonisation components. Finally, we used a BLS1 control sample given by Jin et al. (2012) to better
illustrate the physical interpretation of the NLS1. Finally, we have also measured the
optical-to-X-ray spectral index αOX and the 2-10 keV bolometric correction κ2−10 keV .
The main uncertainty in the parameters, especially the black hole mass, is dominated by systematic uncertainties introduced by the observational data, model assumptions (e.g. the assumption of a non-spinning black hole and the inclination dependence
of the disc emission) and the analysis methods involved. Therefore the statistical uncertainties returned by model fits which are often less than 10%, are not significant in
comparison, and thus are not listed in Table 4.2.
Figures 4.4 and 4.16 present the SED fitting results for the 2+2 non-standard NLS1
and standard NLS1, respectively. We found that the broadband SEDs of the two Xray-flattened NLS1 galaxies (See Figure 4.4, plot ii) are quite similar to each other, as
well as to those of the two Fe II-deficient NLS1 objects, the SED parameters of these
four non-standard NLS1 are also similar. We note that the Fe II-deficient NLS1 objects
appear to have a harder X-ray spectrum, i.e. Γ ≈ 1.7. Although the broadband SED
of the Fe II-deficient NLS1 galaxies (both ID=203 and 241, plot i in Figure 4.4) present
a significant soft Comptonised component, their X-ray spectra alone are compatible
with being flat (Γ ∼ 2.0). The best-fit SEDs for the X-ray-flattened objects present a
flat X-ray spectrum with only a strong hard Comptonised component, which confirms
the results found by fitting a single power-law X-ray spectrum: their X-ray spectra do
not show any sign of soft X-ray excess. The broadband SED for the remaining NLS1
(15/19) appear to be well fitted with a significant soft Comptonised component (see
Figure 4.16 at the end of the Chapter).
90
NON-STANDARD NLS1 GALAXIES
Table 4.2: Broadband SED fitting Parameters, and Model outputs (Lbol , fSE , κ2−10keV , αOX ).
Lbol
×1044 erg s−1
κ2−10 keV
0.296
0.228
fSE
1.53
1.48
αOX
χ2
log (Ṁ )
g s−1
1.231
1.011
reduced
λEdd
77.53
39.68
1.61
1.50
log (MBH )
M
5.59
23.50
0.0020
0.0070
τ
25.04
25.66
1.790
2.812
KTe
keV
0.13
0.19
96.58
38.09
Rcor
rg
7.57
8.14
17.22
25.26
fpl
15.1
20.8
25.53
25.69
Γ
0.323
0.166
0.16
0.12
NH,int
×1020 cm−2
10.7
15.2
8.08
8.38
NH,Gal
×1020 cm−2
0.79
0.64
18.0
34.1
ID
1.7
1.7
0.154
0.244
non-standard NLS1
• Fe II-deficient NLS1
203
1.31
0.00
241
5.24
3.86
9.5
12.0
1.55
1.42
1.45
1.43
1.39
1.43
1.61
1.49
1.58
1.43
1.42
1.59
1.58
1.68
1.93
0.98
0.98
standard NLS1
10.0 0.410 11.7
7.49
0.24
25.23
8.78
148.20 1.067
13.4 0.175 17.6
7.25
0.10
24.61
2.10
64.04 1.168
12.1 0.262 17.1
7.40
0.19
25.05
5.82
71.82 1.208
14.2 0.316 9.8
6.98
0.48
25.06
5.90
144.22 1.231
11.9 0.219 19.0
7.32
0.41
25.36
11.82
90.10 0.965
16.2 0.238 16.3
7.59
0.22
25.34
11.21
69.60 1.417
11.8 0.166 19.9
7.66
1.10
26.04
56.81
275.95 1.144
10.9 0.363 12.4
7.50
0.30
25.32
10.81
89.31 1.398
14.6 0.209 14.3
7.81
0.63
25.94
44.22
161.35 1.162
100. 0.327 17.0
6.3
3.7
25.28
9.72
73.80 2.670
98.5 0.297 11.6
7.21
0.29
25.11
6.65
79.25 1.083
11.6 0.272 17.1
8.12
0.25
25.76
29.66
82.05 1.102
11.0 0.214 12.2
7.63
1.09
26.09
62.56
303.03 1.131
10.4 0.597 6.6
7.96
0.40
25.74
28.18
200.33 1.491
8.3 0.102 54.8
7.74
0.48
25.80
32.55
1447.88 1.901
NH,Gal and NH,int : the fixed galactic and free intrinsic neutral hydrogen column
0.547
0.283
0.580
0.669
0.548
0.564
0.743
0.501
0.629
0.916
0.458
0.500
0.675
0.300
0.830
2.1
2.0
3.28
0.03
2.1 0.42
3.41
9.41
2.2 0.48
5.00
3.65
1.9 0.46
1.91
10.44
2.2 0.16
1.87
3.36
2.2 0.42
4.90
5.52
2.2 0.32
1.33
4.82
2.2 0.14
2.36
1.60
1.8 0.55
1.22
8.05
2.1 0.14
1.31
3.71
2.2 0.09
2.11
3.57
2.1 0.11
4.45
2.99
1.7 0.58
2.77
10.30
2.3 0.18
1.99
0.00
2.2 0.32
2.07
4.23
2.1 0.14
number, the same as in Chapter 3;
• X-ray-flattened NLS1
129
2.93
0.00
357
1.03
0.22
26
42∗
53
65∗
71∗
100∗
104∗
125
171
195∗
191
214
275
318∗
329
ID: object
limit of 2.2 and were fixed there; fpl : the fraction of the power-law component in the total reprocessed disc emission; Rcor : corona
densities; Γ: the power-law components’s slope in the SED fitting, (∗ ) denotes the objects whose power-law slopes hit the upper
(truncation) radius within which all disc emission is reprocessed into the Comptonisation and power-law components; KTe :
electronic temperature of the Compton up-scattering electron population; τ : optical depth of the Comptonisation component;
log(MBH ): the best-fit black hole mass; λEdd : the Eddington ratio; Lbol : bolometric luminosity integrated from 0.001 keV to
100 keV; κ2−10 keV : 2-10 keV bolometric correction; χ2 : the reduced χ2 of the broadband SED fitting; fSE : the fraction of the
soft X-ray component emitted in the 0.3-2 keV energy band; αOX : optical-to-X-ray index
EFE
-3
10
10-6
10
-5
10-4
10
-3
10-5
-3
10
10-4
10-3
10
-2
10-2
10
10-5
-3
10
10-2
1
10
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-7
-3
10
10-6
10-5
10-4
10-3
10-2
10-2
(i) Fe II-deficient NLS1 (ID=203 and 241, respectively)
1
10-4
10-3
10-2
(ii) X-ray-flattened NLS1 (ID=129 and 357, respectively)
10
E [keV]
-1
10-1
E [keV]
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-1
E [keV]
10-1
E [keV]
10
1
10
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
1
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
Figure 4.4: Broadband SED fits: X-ray data have been rebinned for each object. The red dotted line is the pure accretion disc
component peaking in the optical/UV region, the green dotted line is Comptonisation component producing the soft X-ray excess below
2 keV, the blue dotted line is the hard X-ray Comptonisation component dominating the 2-10 keV spectrum, and the black is the total
broadband SED model. EFE in units of keV2 (photons/cm2 /s/keV).
EFE
EFE
EFE
10-2
4.5. Physical Properties of the NLS1 sample
91
NON-STANDARD NLS1 GALAXIES
4.5.1.
92
Statistical Properties
We now comment on the distribution of some properties derived from the SED
fits as histograms including optical and X-ray modelling parameters for our full NLS1
sample. Each figure through this section, shows the distribution for both the 2+2 nonstandard NLS1 population (as solid histogram) and for the entire NLS1 sample (as
hatched histogram). To unravel their role in the AGN paradigm, we include a control
sample of BLS1. The BLS1-type control sample consists of 53 bona-fide type-1 AGN
with Hβ FWHM in excess of 2000 km s−1 and was taken from Jin et al. (2012). The
reason for choosing that particular sample resides in the fact that Jin et al. (2012)
analyzed these sources using the same broadband SED model that we use here. The
distribution given by the BLS1 control sample is shown as empty histogram. We also
compare our results with previous works.
Lbol : The bolometric luminosity logarithmic distribution ranges between 44.32
15
erg s−1 (ID=275). The average bolometric luminosity is hlog (Lbol )i = 45.13 ±
0.37, which is consistent with the value
number
erg s−1 (ID=42) and 45.79
in units of
10
5
of the hlog (Lbol )i = 45.19 ± 1.01 found in
Vasudevan & Fabian (2007) in a sample
of Seyfert 1 AGN. It means that there is
no clear difference between the distribution of our NLS1 sample and that of the
0
43
44
45
46
log(L ) [erg/s]
47
48
bol
Figure 4.5: optxagnf: bolometric
luminosity
sample of Vasudevan & Fabian (2007). The non-standard NLS1 galaxies do not lie at
the tails of the distribution, except for the source with ID=203, which is one of the less
luminous sources.
MBH : The best-fit black hole mass
logarithmic distribution peaks at ∼7.5
15
number
in units of M and the mean value is
hlog (MBH /M )i = 7.61 ± 0.32. Whilst
10
it is slightly lower than that found in pre5
0
vious works (Zhou et al., 2006), the distribution ranges between 7 and 8.4 (in
7
8
log(M /M )
9
10
units of M ). This means that our sam-
BH
Figure 4.6: optxagnf: black hole
mass
ple of NLS1 is clearly biased towards high
black hole masses with respect to previ-
4.5. Physical Properties of the NLS1 sample
93
ous NLS1 samples (Kaspi et al., 2000, Zhou et al., 2006). We note that the values of
the FWHM of the narrow-component-subtracted Hβ (see values in Table 4.1), which is
correlated with MBH , ranges between 769 km s−1 (ID=104) and 1711 km s−1 (ID=203).
We found that the non-standard NLS1 galaxies display the highest values of MBH , as
well as, the highest values of Hβ FWHM (as claimed in Section 4.1).
λEdd : The average value of the Ed20
dington ratio is 0.38 ± 0.28, displaying
Although it is sig-
nificantly lower than that expected for
NLS1, this would be a consequence of the
15
number
a wide dispersion.
high black hole mass values. The aver-
10
5
age Eddington ratio are compatible with
that found by Vasudevan & Fabian (2007)
0
-2
-1
whose average value is 0.47±0.44; we note
0
1
2
log(λ Edd)
Figure 4.7: optxagnf: Eddington
ratio
that their distribution has a more pronounced peak at ∼ 0.1 (remember that
their sample contains both BLS1 and NLS1 galaxies). We found that the non-standard
NLS1 lie at the lower tail of the Eddington ratio distribution: λEdd =0.13, 0.19, 0.16,
0.12 for the sources ID=203, 241, 129 and 357, respectively.
αOX : The optical-to-X-ray index is
defined between rest-frame continuum
15
αOX
f2keV
= −(1/2.605) log
f
2500Å
The distribution is peaked at marginally
higher values (∼ 1.51) than for samples of
Seyfert 1 galaxies (∼ 1.4 found by Lusso
et al., 2010). In terms of the mean values,
number
points at 2500 Å and 2 keV as
10
5
0
1
1.2
1.4
αOX
1.6
1.8
Figure 4.8: optxagnf: optical-toX-ray spectral index
it means that there is no clear difference of our NLS1 galaxies and Seyfert 1 AGN. The
values of the optical-to-X-ray ratio for the 2+2 non-standard NLS1 are not restricted
to a privileged region and their values are almost equal to the mean value for the entire
sample 1.54 ± 0.04.
NON-STANDARD NLS1 GALAXIES
94
κ2−10 keV :
20
ric correction is defined as Lbol /L2−10keV
15
number
The 2-10 keV bolomet-
(Vasudevan & Fabian, 2007, and references therein).
10
The average value is
hlog (κ2−10keV )i = 2.00 ± 0.25. The distri5
0
0.5
bution peaks at κ2−10keV ∼50 after which
it drops sharply. We find that the frac1
1.5
2
2.5
3
log(κ 2-10keV)
tion of the bolometric luminosity emitted
as hard X-rays, i.e. κ2−10 keV , is on aver-
Figure 4.9: optxagnf: X-ray bolometric correction
age significantly higher than for Seyfert 1.
It is compatible with the findings of Vasudevan & Fabian (2007) who found that the
bolometric correction is typically 15-25 for sources with Eddington ratio of ∼0.1, and
40-70 above the value.
kTe : The average temperature of the
15
Comptonised component used to describe
the soft X-ray excess is hkTe i = 0.243 ±
keV confirms the trend seen in previous
studies for this component to exhibit a
10
number
0.078 keV. The mean value close to 0.2
5
narrow range of peak energy (see for example Gierliński & Done, 2004). There
0
0
is no differences between both populations (non-standard NLS1 and standard
NLS1).
0.4
0.6
0.8
kTe [keV]
Figure 4.10: optxagnf: temperature of the soft X-ray component
τ:
25
0.2
The optical depth of the soft
Comptonised component is always opti-
20
number
cally thick, peaking at ∼15 with a mean
15
value is hτ i = 18±10. We note that there
10
are only two sources with optical depth
5
above 20 (ID=329, 357), one of them be-
0
0.5
ing an X-ray-flattened NLS1 galaxy.
1
1.5
log(τ)
Figure 4.11: optxagnf: optical
depth of the soft X-ray component
2
However, there is no significant difference
in temperature or optical depth between
the NLS1 without soft X-ray excess and
those with an apparent soft Comptonised
component.
4.5. Physical Properties of the NLS1 sample
95
Rcor : This parameter controls the relative amount of power emerging from
the accretion disc and the soft X-ray ex-
10
cess/hard tail. We note that the best-fit
8
for very large Rcor , because the assump-
number
model parameters may not be accurate
6
tion that the seed photon energy is set
4
by the disc temperature at Rcor is prob-
2
ably unphysical (with this high values,
0
the spectrum peaks at energies lower than
10−2 keV, which is physically unplausible,
see Done et al., 2012, for an extended dis-
>20
10
15
log(Rcor/r_g)
20
Figure 4.12: optxagnf: coronal radius Rcor
cussion). Thus, the derived parameters for the sources with ID=191, 195 should be
used with caution. Excluding these two sources, the average value of Rcor is (12 ± 2) rg .
There is difference in the values of Rcor between the X-ray-flattened NLS1 and the
standard NLS1s. The X-ray-flattened sources are characterised to have an Rcor below
the mean value.
fpl : This parameter gives the fraction emitted as a high-energy Comptonised component, i.e. as a power law. The best-fit value for the X-ray-flattened NLS1 galaxies is close to 1 just as expected, while the mean value of the full NLS1 sample is
hfpl i = 0.43 ± 0.27 compatible with the previous works (∼ 0.3).
NH,intr : The intrinsic absorption distribution shows that the equivalent neutral hydrogen column densities are be30
low 1022 cm−2 , all of NLS1 galaxies be-
number
ing unabsorbed X-ray sources, in agree20
ment with the distribution of Balmer
decrements.
10
Except for the source
with ID=241, the intrinsic absorption for
0
the non-standard NLS1 sources is be0
5
10
20
NH,intr [ × 10
cm-2]
Figure 4.13: optxagnf: intrinsic
absorption
low 1020 cm−2 , which means that the lack
of soft X-ray component cannot be explained by X-ray absorption. To ensure
that the X-ray-flattened NLS1 (ID=129, 357) are unabsorbed AGN, we re-fit their
SEDs assuming a soft Comptonisation component whose profile is given by the mean
NON-STANDARD NLS1 GALAXIES
96
values (once these sources are removed): i.e. kTe = 0.267 keV, τ = 17, Rcor = 18 rg ,
fpl = 0.31. We found that a standard X-ray soft Component could not be compensated
by absorption.
4.5.2.
NLS1 vs. BLS1 and the non-standard NLS1
In previous sections, we found that while our 15 standard NLS1 are fully compliant
with all usual properties of this population, both the X-ray-flattened NLS1 and the
Fe II-deficient NLS1 do not fit very well with the general NLS1 trends. Comparing our
results on the 19 NLS1, including the four non-standard NLS1, with those from the
BLS1 control sample, we clearly see that NLS1 tend to have a softer 2-10 keV spectrum,
lower 2-10 keV luminosity, lower black hole mass, higher Eddington ratio, higher αOX
index and smaller corona radius, and correspondingly a smaller coronal component
contribution. We find that the optical and X-ray properties of our non-standard NLS1
agree better with those from the BLS1 class rather than those of the standard NLS1.
We then study the values of these parameters for the non-standard NLS1 vis a vis
their distribution in BLS1 and standard NLS1, in particular: Hβ FWHM (narrowcomponent-subtracted component), λEdd and the fraction of the soft X-ray emission
(fSE ). Figure 4.14 shows our results in the Hβ/λEdd correlation which are compatible
with previous works(e.g., Jin et al., 2012, Vasudevan & Fabian, 2007), where a correlation of the narrow-component-subtracted FWHM of Hβ with λEdd for both NLS1 and
BLS1 was found. We note that this trend remains if we use, instead, the FWHM of
the broad emission line component, as well as the intermediate component. We found
that the 2+2 non-standard NLS1 are located in the border line region between the
NLS1 and the BLS1 zones (yellow and black points on Figure 4.14, respectively). We
also found that the fraction of soft X-ray excess (fSE ) increases when increasing the
Eddington ratio for the NLS1 galaxies, and it follows a consistent trend in the BLS1
region too (see Figure 4.14, where the point size is linked with fSE ).
In order to further test the nature of the Fe II-deficient and the X-ray-flattened
NLS1 galaxies, we compare the broadband SED of our sample with the mean SEDs
given by Jin et al. (2012) which has been sub-divided according to their Hβ FWHM.
Figure 4.15 shows the comparison between the broadband SED of our sources with a
mean SED given by Jin et al. (2012). First, the individual best-fit broadband SEDs
are compared with the mean NLS1 SED from Jin et al. (2012). Whilst the standard
NLS1 of our sample appears to be fully consistent with this mean SED (see top panel
of Figure 4.15), the broadband SEDs of the 2+2 non-standard NLS1 galaxies are not
in good agreement with that average SED. However, as already discussed, when the
4.6. Summary and Conclusions
97
standard NLS1
104
Hβ
NLC-subs
FWHM [km s-1]
BLS1 (Jin et al. 2012)
103
ID=203
ID=357
ID=129 ID=241
10-2
10-1
1
λ Edd = Lbol/LEdd
Figure 4.14: Correlation of the Hβ FWHM (once the narrow component is subtracted) with the Eddington ratio. The size of the symbol is linked with the fraction
of soft X-ray component emitted in the 0.3-2 keV energy band (i.e. the strength of the
soft X-ray excess FSE ), when the fSE < 8% the source is displayed as cross-point. Blue
and black symbols are the NLS1 and the BLS1 samples, respectively. Orange, olivegreen, green and pink symbols represent the non-standard NLS1 galaxies (ID=129,
357 -Fe II-deficient- and ID=203, 241 -X-ray-flattened-, respectively).
broadband SED of these 2+2 sources are compared with the mean BLS1 SED of Jin
et al. (2012, see bottom panel of Figure 4.15) we find a much better agreement. This
suggests that our 2+2 non-standard NLS1 are much closer to being BLS1 rather than
NLS1. In addition we find that an additional source with ID=214 has a SED (black line
in the Figure 4.15) similar to that of ID=241 (one of the Fe II-deficient NLS1 galaxies)
with an X-ray photon index ∼ 1.7. The former source, that we have classified as bonafide NLS1, appears to be more consistent with being BLS1, despite having a huge soft
X-ray excess (fSE ∼ 0.5), as well as being a strong iron emitter.
4.6.
Summary and Conclusions
In this Chapter we have presented a study of the X-ray to optical properties of
a sample of 19 NLS1 found in a hard-X-ray-selected sample of NELG. Among these
NLS1 galaxies there are two sources with non-detectable Fe II emission, and another
NON-STANDARD NLS1 GALAXIES
98
two with non-detectable soft X-ray emission. To study the nature of these non-standard
NLS1 galaxies, we assembled X-ray data from the EPIC X-ray detector on board the
XMM-Newton satellite, and optical data from the SDSS DR7. In addition we added
optical/UV data from the XMM-Newton OM instrument when available. We fit the
Hβ and Hα line profiles, with multiple-component models to deblend the narrow, intermediate and broad components by simultaneous modelling of the Fe II continuum and
other blended lines. We then used results from the Hβ fitting to constrain the black
hole mass. The FWHM of their intermediate and broad components give a lower and
upper limit for the mass, respectively. In agreement with previous work, we find that
NLS1 tend to have lower black hole masses and higher Eddington ratios than BLS1, although their bolometric luminosities are not substantially different from those of BLS1.
The optical, UV and X-ray data were fitted using a physically coherent SED model
(optxagnf in XSPEC), which assumes that the gravitational potential energy is emitted as optically thick blackbody emission at each radius down to some specific value
Rcor . Below this radius down to the last stable orbit, the remaining energy is divided
between a soft X-rays excess component and a hard X-ray tail, resulting from Comptonisation processes. This energetically constrains the model in the unobservable EUV
region. We constructed SEDs for each source and then the resulting best-fit broadband
SEDs were compared with that of a BLS1 control sample of Jin et al. (2012) to understand how these 2+2 non-standard NLS1 fit in this picture.
Fe II-deficient NLS1s (sources ID=203 and 241) exhibit a best-fit SED which match
better the mean SED of BLS1, rather than that of NLS1. Therefore their classification
as NLS1 relies only on the FWHM of their Hβ line, being smaller than the usually
adopted threshold of 2000 km s−1 (Goodrich, 1989). By the way, the Hβ FWHM values of both sources are among the largest within the entire NLS1 sample studied here.
The threshold separating the NLS1 and BLS1 is rather arbitrary and therefore we believe that these sources are really normal BLS1, perhaps with a marginally smaller
black hole mass. Both Fe II-deficient NLS1 appear to have a a flatter X-ray power-law
photon index than other NLS1, although their fitted blackbody temperature or optical depth take average NLS1 values. The location of both of these sources in the Hβ
FWHM versus Eddington ratio diagram (Figure 4.14) suggests a “borderline” nature
in between BLS1 and NLS1, but the lack of Fe II and overall SED shape confirm that
they are best classified as BLS1. We also caution that other similar sources might be
present in other NLS1 samples selected exclusively on the basis of the Hβ FWHM.
From our statistical study we found that the Fe II-deficient NLS1 galaxies have
4.6. Summary and Conclusions
99
significant difference in the X-ray photon indices (being significantly harder than for
common NLS1), whereas there is no significant difference in temperature or optical
depth. Their lack of iron emission along with a moderate soft X-ray emission, as well
as their higher Hβ FWHM (& 1200 km s−1 ) lead us to believe that they are rather BLS1
instead of NLS1, the Hβ FWHM threshold between both classes being somewhat arbitrary. It was confirmed by the correlation between the narrow-component-subtracted
FWHM of Hβ and the Eddington ratio in which they are located in the border region
midway between both classes: having the lowest Eddington ratio and highest FWHM
values compared with the standard NLS1 of our sample.
As regards the X-ray-flattened NLS1 galaxies, there are two potential interpretations for the lack of soft X-ray emission. The first one is that these sources could have
similar X-ray broadband SED to those standard NLS1 with a detected soft Comptonised
component, but this component would be suppressed by a significant amount of photoelectric absorption. However, the amount of photoelectric absorption needed (around
1023 cm−2 ) largely exceeds the estimates based on the Balmer decrement (which are
∼100 times smaller). In addition, the fit to the X-ray spectrum would be statistically
excluded at high statistical significance, as its overall shape (and not only the spectral region below 2 keV) would be grossly distorted by such a large absorbing column
density. The second possibility is that the X-ray-flattened NLS1 galaxies represent a
new population of NLS1’s with intrinsically harder-X-ray spectral shape. In this case,
the lack of soft X-rays would be due to a fundamental difference in the central black
hole and accretion disc properties. Certainly, these two sources have larger black hole
masses and lower Eddington ratios, the accretion disc spectral distribution is shifted
to lower energies and effectively it starts to flatten below 1 keV. The picture emerging
from this explanation, is that X-ray-flattened NLS1 contain relatively large black holes
(similar to those in BLS1), while they are accreting at a relatively high Eddington ratio
(similar to those of typical NLS1). Whether this corresponds to a steady state situation
for these sources, or instead the X-ray-flattened NLS1 are simply normal NLS1 in the
low (and hard) state (paralleling what happens in stellar-mass black holes) would require long term X-ray monitoring of these sources in the hope that changes in the mass
accretion rate could show up as a detectable soft excess. This would need, however, a
very long sequence of X-ray observations spanning long timescales.
NON-STANDARD NLS1 GALAXIES
100
5
E
log( const*EF )
4
ID=26
ID=275
ID=104
ID=42
ID=318
ID=65
ID=191
ID=125
ID=100
ID=71
ID=214
ID=329
ID=171
ID=53
ID=203
ID=241
ID=129
ID=357
3
2
1
0
-3
10
10-2
10-1
1
Energy [keV]
10
102
(i) mean NLS1 SED with hFWHMHβ i = 1400km s−1
5
ID=203
ID=241
ID=129
ID=214
ID=357
E
log( const*EF )
4
3
2
1
0
-3
10
10-2
10-1
1
Energy [keV]
10
102
(ii) mean BLS1 SED with hFWHMHβ i = 3600km s−1
Figure 4.15: Rest-frame best-fit broadband SED models of our NLS1 sample (green
and red lines represents the Fe II-deficient and X-ray-flattened NLS1 galaxies, respectively). The dashed region indicates a one standard deviation region on either side of
the mean SED of Jin et al. (2012). The SEDs have all been re-normalized to the mean
flux at 2 keV.
4.6. Summary and Conclusions
10-2
101
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
EFE
10-3
EFE
10-3
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-4
10-4
10-5
-3
10
10-2
10-1
E [keV]
1
10-5
-3
10
10
(i) ID=26
10-2
10-2
10-1
E [keV]
1
10
(ii) ID=42
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-2
EFE
EFE
10-3
10-3
10-4
10-5
-3
10
10-4
10-2
10-1
E [keV]
1
-3
10
10
(iii) ID=53
10
10-1
E [keV]
1
10
(iv) ID=65
10-1
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
-1
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
EFE
EFE
10-2
-2
10
10-3
-3
10
-3
10
10-2
10-1
E [keV]
1
10-4
-3
10
10
(v) ID=71
10
10-1
E [keV]
1
10
(vi) ID=100
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
-2
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
EFE
EFE
10-3
10-3
10-4
10-4
-3
10
10-2
10-1
E [keV]
(vii) ID=104
1
10
10-5
-3
10
10-2
10-1
E [keV]
1
10
(viii) ID=125
Figure 4.16: Broadband SED fitting plot: X-ray data has been rebinned for each object. The red dotted line is the pure accretion disc component peaking at optical/UV
region, the green dotted line is Comptonisation component producing soft X-ray excess below 2 keV, the blue dotted line is the hard X-ray Comptonisation component
dominating 2-10 keV spectrum, and the black line is the total broadband SED model.
EFE is in units of keV2 (photons/cm2 /s/keV).
NON-STANDARD NLS1 GALAXIES
102
10-1
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
EFE
EFE
10-2
10-3
10-3
10-4
-3
10
10-2
10-1
E [keV]
1
10-4
-3
10
10
10-2
(i) ID=171
1
10
(ii) ID=191
10-1
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-3
EFE
EFE
10-2
10-3
10-4
10-4
-3
10
10-2
10-1
E [keV]
1
10-5
-3
10
10
10-2
(iii) ID=195
1
10
10-2
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
-1
10
EFE
10
10-1
E [keV]
(iv) ID=214
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
-1
10-3
10-2
10-3
-3
10
10-2
10-1
E [keV]
1
-3
10-2
10
10
(v) ID=275
10-1
E [keV]
(vi) ID=318
10-1
best-fit model
pure AD component
soft Comptonisation component
hard Comptonisation component
data points
10-2
EFE
EFE
10-1
E [keV]
10-3
10-4
10-5
-3
10
10-2
10-1
E [keV]
1
(vii) ID=329
Figure 4.16: Continued.
10
1
10
CHAPTER 5
IR power-law
selected AGN,
potentially less biased against
obscuration
5.1.
Motivation
In heavily obscured and Compton-thick AGN the observed flux below 10 keV can
be as low as a few per cent of the intrinsic nuclear flux. In the Compton-thick regime,
the high energy photons that survive the photoelectric absorption get scattered in the
absorber, losing part of their energy. This is an important effect that can significantly
suppress the transmitted continuum (e.g. Matt, 2002). In the Unified model framework,
surveys at near- and mid-IR wavelengths are much less affected by extinction since the
obscuring dust re-emits the nuclear optical-to-X-ray radiation at infrared wavelengths.
Thus, mid-IR based surveys (or the combination of mid-IR and data at shorter wavelengths) should potentially trace the elusive obscured accretion missed by hard X-ray
surveys (Daddi et al., 2007, Donley et al., 2012, Fiore et al., 2009, Georgantopoulos
et al., 2008, Severgnini et al., 2012). For example, it has been claimed that objects
showing excess emission at ∼ 24µm over that expected from star formation might host
heavily obscured and Compton thick AGN (Fiore et al., 2008, 2009). However the exact
contribution of heavily obscured AGN to the IR-excess galaxy population remains an
open issue (Alexander et al., 2011).
In this Chapter we present a study of a sample of AGN which have been reported
to exhibit a power-law continuum in the IR waveband. Among all employed IR colourcolour selections to search for AGN in which strong re-radiation from obscuring dust
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
104
is expected, we have picked the revised IRAC selection criterion given by Donley et al.
(2012). For deep IRAC data, the AGN selection criteria previously in use might be
contaminated by star-forming galaxies, especially at high redshift. Donley et al. (2012)
redefined the AGN selection criteria for deep IRAC surveys using large samples of luminous AGN and high-redshift star-forming galaxies in COSMOS. Their revised IRAC
criterion have been designed to be both highly complete and reliable, and incorporate
the best aspects of the current AGN selection wedges and of the IR power-law selection (Alonso-Herrero et al., 2006, Donley et al., 2007, Park et al., 2010). It excludes
high-redshift star-forming galaxies which are otherwise present in other high-redshift
galaxy samples of IR-based AGN candidates. The Spitzer IRAC power-law selection
(Alonso-Herrero et al., 2006, Donley et al., 2012, 2007) chooses sources whose IRAC
SEDs follow a power law over a wide range of slopes. This is because, when the AGN
is sufficiently luminous compared to its host galaxy, the superposition of blackbody
emission from the AGN-heated dust fills in the dip in the galaxy SED and produces a
red, power-law-like thermal continuum across the IRAC bands. A problem commonly
encountered when studying AGN properties based on IR observations is the significant
contribution of the host galaxy to the near-IR and mid-IR (Alonso-Herrero et al., 1996,
Franceschini et al., 2005, Kotilainen et al., 1992). However, in high-luminosity objects
where the AGN outshines the host galaxy by a large factor in the rest-frame optical
and near- and mid-infrared, this should not be an issue.
We compare the X-ray properties of the sources detected both in X-rays and in
the four IRAC bands with and without a power-law-like continuum shape. The goal
is to explore and quantify the efficiency in finding obscured AGN (in the ultra-deep
XMM-Newton observations in the Chandra Deep Field South) and especially whether
it effectively selects type-2 AGN at some luminosity range when selecting the sources
that display a power-law SED in the IR.
This Chapter is organized as follows. The parent sample and the IR power-law
selection are presented in Sect. 5.2. The X-ray spectral analysis and our results are discussed in Sect. 5.3. A detailed study of the absorption column density distribution and
the obscured AGN fraction as a function of the intrinsic X-ray luminosity and the nearand mid-IR-continuum shape is reported in Sect. 5.4. Finally in Sect. 5.5 a summary
of the main results obtained is presented. Throughout this chapter, we have estimated
the most probable value for the fractions using a Bayesian approach and the binomial
distribution from Wall & Jenkins (2008) for which the quoted errors are the narrowest
interval that includes the mode and encompasses 90% of the probability. All the results
in this Chapter are contained in the already published paper Castelló-Mor et al. (2012).
5.2. Data compilation
5.2.
105
Data compilation
To study the X-ray properties of a sample of IR power-law AGN in the CDF-S
and the Extended CDF-S fields using ultra-deep XMM-Newton observations we took
advantage of the deep Spitzer images available in that region. The average Galactic
column density towards the CDF-S is 0.9 × 1020 cm−2 (Dickey & Lockman, 1990)
providing a relatively clean vision of the extragalactic X-ray sky even down to soft
X-ray energies.
5.2.1.
Infrared data
The near- and mid-IR source catalogue used for this work has been built from
Spitzer/IRAC observations (the selection being made at 3.6 µm and 4.5 µm) as a
compilation from a number of different Spitzer/IRAC surveys in the extended CDF-S
area, of different depths. The median 1σ sensitivity for the four IRAC channels is 0.60
µJy, 1.2 µJy, 8.0 µJy, and 9.8 µJy, respectively, for 100 seconds of integration with
low background. The total surveyed area is 664 arcmin2 . This IRAC sample is 75%
complete down to 1.6 µJy.The data are described in detail in Pérez-González et al.
(2008). A total of 23,044 IRAC sources are selected from this sample with a high
signal-to-noise ratio (S/N> 5) in each of the four bands (3.6, 4.5, 5.8, and 8.0 µm).
5.2.2.
X-ray data
The X-ray data presented here were obtained from the CDF-S XMM-Newton survey,
which surveyed the CDF-S and the extended CDF-S areas. An extensive and detailed
description of the full data set, including the data analysis and reduction as well as the
X-ray catalogue was published in Ranalli et al. (2013). Briefly, the bulk of the X-ray
observations were made between July 2008 and March 2010, and were combined with
archival data taken between July 2001 and January 2002. The total net integration
time (after removal of background flares) is ∼2.8 Ms and ∼2.5 Ms for the EPIC MOS
and pn detectors, respectively. The source catalogue used for this work contains X-ray
sources detected in the 2-10 keV band with conservative detection criteria: >8 σ significance and an exposure time >1Ms. These requirements resulted originally in 171
X-ray-detected sources. We then sub-selected the 150 objects for which spectral data
are available for at least one of the three EPIC cameras: pn, MOS1 or MOS2 (for more
information about the definition of the spectral catalogue see Comastri et al., 2013).
The exclusion of the 21 sources without X-ray spectral data, spanning a wide range of
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
106
redshifts up to z ∼3, does not introduce any additional bias. Finally, we excluded 3
further objects for which the redshift is unknown. Our final X-ray sample contained
then 147 AGN with spectroscopic and/or photometric redshifts, all with >180 counts
in the observed 0.5-9 keV energy band. We note that the sample is not statistically
complete, as a result of the primary selection criteria (X-ray spectrum are needed),
however, the aim of the work is not heavily dependent on any selection bias.
Hereafter we use “ID210=” for the identification number of X-ray sources listed
in Ranalli et al. (2013). Spectroscopic redshifts are available for 124 objects, while
only photometric redshifts for 23 further objects. The photometric and spectroscopic
redshifts adopted in this paper are taken from Ranalli et al. (2013). The redshift distribution of the sample is shown in Figure 5.1 where we can see that our sample spans
a broad range of redshifts from 0 up to z = 3.6, 1.4 being the mean redshift.
We note that a classification purely based on optical properties (type-1/type-2) has
not been possible for the majority of the sources, however as we pointed in the Introduction, correlations between optical and X-ray properties of obscured and unobscured
AGN show a very good match between optical reddening and X-ray absorption (e.g.,
Page et al., 2003).
30
total
spectroscopic (124)
25
photometric (23)
number
20
15
10
5
0
0
1
2
3
4
5
redshift
Figure 5.1: Spectroscopic and photometric redshift distribution as solid-line and
hatched histograms, respectively. The dashed-line histogram represents the distribution of the redshift of the entire sample.
5.2. Data compilation
5.2.3.
107
Cross-correlation
To identify X-ray/IR associations, we analysed the CDF-S ultra-deep XMM-Newton
observations covering the sky positions of the IRAC catalogue. We cross-correlated
both catalogues applying the method developed by Pineau et al. (2011). The method
is based on a likelihood ratio (LR) technique. For a given source, this method provides
the probability of association for each candidate counterpart, which is a function of the
positional errors, relative distance and the local density of potential counterparts. The
much higher density of IRAC sources with respect to that of XMM-Newton sources
and the large uncertainties in the XMM-Newton’s positions make this cross-correlation
exercise particularly difficult. To facilitate our task, we used the cross-correlation between CDF-S XMM-Newton and the CDF-S Chandra catalogues as presented in Ranalli
et al. (2013) as a first step to find the CDF-S XMM-Newton counterparts to our IRAC
targets. The reason for this is that the positions of the Chandra sources are much
more accurate (a fraction of an arcsec, Lehmer et al., 2005, Xue et al., 2011) than
those from XMM-Newton, while the densities of both Chandra and XMM-Newton catalogues are similar. In practice this means that the associations between Chandra and
XMM-Newton X-ray sources are highly reliable and this step does not introduce any
uncertainty in our process to associate XMM-Newton to IRAC sources. We then use
the Chandra’s positions instead of the less accurate XMM-Newton positions to search
for IRAC counterparts.
Spitzer/IRAC data cover very well over 95% of the XMM-Newton exposure area used
in this work. Only 6 X-ray sources fall in regions where there is partial or no coverage
by the Spitzer/IRAC catalogue data. For the remaining sources, we found one single
IRAC counterpart for each X-ray source, and therefore our final sample contains 147
cross-correlations with a single counterpart. Thus, all sources have an X-ray spectrum
from the CDF-S XMM-Newton observations, Spitzer/IRAC fluxes in the 4 bands and
a spectroscopic or photometric redshift. The mean value of the X-ray-Chandra-IRAC
source separation is . 1.5 arcsec (see the distribution of the distance in Figure 5.2).
We also estimate a fraction of spurious matches of randomized IRAC and X-ray sources
of < 1.5% (i.e., <2 sources in our case) from the cross-matching of IRAC and X-ray
sources using a large offset in IRAC coordinates (3 arcmin in either RA or Dec).
5.2.4.
Selection of IR power-law galaxies
Following the revised AGN selection by Donley et al. (2012) defined by the following
wedge,
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
108
40
N
30
20
10
0
0
0.5
1
1.5
2
2.5
3
source separation [arcsec]
Figure 5.2: Distribution of the angular separation between the X-ray source and
their single IRAC counterpart, where the mean value is ∼1.5 arcsec.



x




 y


fν,5.8µm
fν,8.0µm 
x = log
, y = log
y
fν,3.6µm
fν,4.5µm 


 y





f
ν,3.6µm
≥ 0.08
≥ 0.15
≥ (1.21 × x) − 0.27
≤ (1.21 × x) + 0.27
< fν,4.5µm < fν,5.8µm < fν,8.0µm
we can directly obtain an IR-based classification: IR power-law are those galaxies
lying within the revised IRAC-selection wedge which include those IRAC SEDs (fλ
monotonically-rising within the IR bands. Thus, we label as IR non-power-law
those galaxies which fall outside the Donley et al. (2012)’s wedge which also exclude all
the sources with non-monotonically-rising IRAC SEDs. The latter largely removes any
possible contamination due to low-redshift star-forming galaxies (which would have
little X-ray emission) in which their 1.6 µm stellar bump passes through the IRAC
bandpass.
Using this criterion, we have 147 X-ray-detected sources with known redshift and X-ray
spectra, from which 60 are IR power-law and 87 IR non-power-law galaxies.
Hereafter, we concentrate on the X-ray spectral properties of these 147 X-raydetected galaxies, highlighting possible differences between these two sub-samples (IR
power-law and IR non-power-law). The main aim of this work is to investigate
their intrinsic absorption distribution up to column densities near the Compton-thick
limit and specially, to infer whether obscured AGN are the major contributor to IR
5.3. X-ray spectral analysis
109
power-law-selected galaxy surveys or not, and specifically whether this criterion is effective at selecting type-2 AGN at some luminosity range. By design, we cannot address
in this work the question of the nature of the IR power-law X-ray-undetected sources.
Figure 5.3 shows the IRAC colour distribution for the entire IRAC sample. The
colour symbols are our 147 X-ray-detected sources, with the blue circles showing the
IR power-law galaxies and the red triangles the IR non-power-law objects. Filled
intr > 1022 cm−2 ) and unabsorbed (N intr ≤
and empty symbols denote absorbed (NH
H
1022 cm−2 ) sources, respectively (see Section 5.4). The symbols also change in size to
denote different ranges of the rest-frame 2-10 keV intrinsic luminosity.
1.5
Abs. | UnAbs.
LX ≤ 1043
(1043 < LX < 1044)
LX ≥ 1044
log S8.0µm/S4.5µm
1
IR power-law
0.5
0
IR non-power-law
-0.5
-0.5
0
0.5
log S5.8µm/S3.6µm
1
Figure 5.3: IRAC colour-colour diagram for the entire IRAC sample (plotted as a
surface grey map) and for the cross-correlation sample: circles and triangles represent
IR power-law and IR non-power-law galaxies according to Donley et al. (2012),
respectively. The filled and empty symbols denote absorbed and unabsorbed sources,
respectively. The symbols change in size to denote different ranges of X-ray luminosity.
The solid line shows the revised IRAC criteria by Donley et al. (2012).
5.3.
X-ray spectral analysis
We have carried out an EPIC X-ray spectral analysis for the 147 sources with
background-subtracted EPIC counts in the 0.5-9 keV band above 180. We ignore the
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
110
data below 0.5 keV to avoid uncertainties in the EPIC calibration. The observed 1.4-1.6
keV energy range was excluded from further analysis to avoid any contamination from
the Al-K fluorescence line from the internal XMM-Newton EPIC background. Finally,
the observed >9 keV energy range was also excluded because the efficiency of XMMNewton decreases rapidly at high energy and the energy bins at >9 keV are noise and
background-dominated for most sources.
The ability to obtain a reliable fit depends on the X-ray spectral quality. The distribution of the net counts in the 2-9 keV band for all the sources in our sample peaks
at ∼700 (see Figure 5.4). Despite the relatively high mean value of the net counts
in the hard energy band, there are many cases in which the spectrum is dominated
by the background. Therefore, the strategy for the X-ray spectral analysis must be
appropriate for the high background regime.
105
60
0.5-2 keV energy band
2-9 keV energy band
4
40
10
number
Background Counts
50
3
30
20
10
10
3
104
10
Source Net Counts
(i)
5
10
0
1
2
3
4
5
log( Source Net Counts )
(ii)
Figure 5.4: (i) Background net counts versus source net counts for the 2 − 9 keV
energy range. (ii) Source net count distribution for both the soft and 2 − 9 keV energy
bands as hatched and empty histograms, respectively.
As discussed in Chapter 1 the X-ray spectrum emitted by an AGN can be typically
modelled with four components: an underlying absorbed power law, a reflection component, a soft excess above the power law at energies below ∼ 1 keV, and an iron Kα
emission line. All sources were analyzed using the following scheme:
1. Underlying absorbed power-law. We started with a joint fit of MOS and
pn spectra with a power-law model1 (model A, see Table 5.1). The power-law
component is associated with the direct X-ray emission of the central engine of
the AGN, where the optical/UV photons emitted by the accretion disc are converted to X-ray photons by high energy electrons surrounding the disc through
1
All models referred to in this Chapter implicitly include a multiplicative Galactic absorption
component fixed at the Galactic NH value from Dickey & Lockman (1990) (phabs)
5.3. X-ray spectral analysis
111
inverse Compton scattering (see Chapter 1).
Additional intrinsic absorption on the power law was included where it was statistically significant. The model used for it was a redshifted neutral absorption
model by cold matter (model C, see Table 5.1), with redshift fixed to the source’s
spectroscopic or photometric value (see Table 5.5).
2. Soft X-ray Excess. When the fit with a simple absorbed/unabsorbed power
law was not a good description of the X-ray spectrum, we then checked for any
significant additional component. Despite the fact that the high background
regime does not allow us to investigate more complex spectral features often
present in AGN, we identify one possible additional spectral component: a soft
excess. The origin of this excess flux for Seyfert 2 galaxies is still a source of
controversy among the AGN community. Despite the fact that the origin of the
soft excess component in type-2 AGN is clearer than for type-1 AGN, we have
adopted a simple parametrisation for this component for simplicity, since more
detailed models do not make much sense given the quality of the X-ray spectra
in hand. Thus, the soft-excess components have been only modeled in the two
simplest ways: one as a thermal emission (model B or D, see Table 5.1) from a
collisionally ionized plasma, which is heated by shocks induced by AGN outflows
(see e.g. King, 2005) or intense star formation (see e.g. Schurch et al., 2002); and
the other one as a power-law-like emission (model E, see Table 5.1) to account for
the spectral complexity observed in some of our sources and which may not be
well fit by a simple thermal emission.
3. Reflection component. Strongly obscured AGN can show a reflection-dominated
spectrum and/or a high equivalent width iron line. These spectral features can
be used as additional criteria to signpost Compton-thick AGN (e.g., Comastri &
XMM-CDFS Team, 2013, Georgantopoulos et al., 2013). This kind of Comptonthick AGN would not necessarily be identified unambiguously as such in our
analysis. Only when the resulting photon index is much lower than the typical
values for unabsorbed AGN (.1) we then checked for a reflection-dominated spectrum: sources with such an observed flat spectrum could be in the Compton-thick
limit in which all the direct emission is suppressed and only reflected emission is
observed at energies below 10 keV.
4. Iron emission line. Finally, we checked for any Fe Kα emission line.
We measured the significance of the detection of additional components in the X-ray
spectra of our sources using the F-test statistic. The F-test measures the significance
of a decrease in χ2 when new components are added to the model. We used a signif-
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
112
icance threshold of 95% (.2σ) to accept the detection of soft excess and/or intrinsic
absorption. In three sources, the temperature of the soft X-ray excess component of
the best-fit model was far too high (>10 keV) to render this spectral component physically plausible. We consequently adopted as the best fit the one that yielded the least
uncertain model parameters leaving out the F-test criteria (provided that the values of
these parameters are physically plausible).
5.3.1.
Spectral Results
Table 5.1 summarizes the models used for the spectral fits. A total of 58 sources
appear unabsorbed (model A or B, i.e. their best-fit X-ray absorption column density is
NH < 1022 cm−2 ). That is, 89 (60+7
−6 %) of the galaxies in our sample host significantly
obscured active nuclei from which 36 are IR power-law and 53 are IR non-powerlaw. The fraction of obscured AGN is not far but still on the lower side from that
predicted by the X-ray background synthesis models (∼60-80%, Gilli et al., 2007). An
example spectral plot for each case (i.e., absorbed and unabsorbed vs IR power-law
vs IR non-power-law) is shown in Figure 5.5.
Table 5.1: Summary of the results of the spectral fitting. The XSPEC model definition is given in the column 1. The percentage of sources best fitted with the indicated
model for the entire sample, the IR power-law and the IR non-power-law subsamples, respectively.
Model†
N (f %)
A: powerlaw
45 (31±66 )
B: zbbody + powerlaw
13 (9±43 )
C: zphabs × powerlaw
69 (47±77 )
D: zbbody + (zphabs × powerlaw)
12 (8±43 )
E: powerlaw + (zphabs × powerlaw) 8
(5±43 )
† See Chapter 2 for the XSPEC model definition.
NIRpl (f %)
15 (25±10
8 )
9
(15±86 )
27 (45±10
−10 )
5
(8±75 )
4
(6±74 )
NIRnon−pl (f %)
30
(34±98 )
4
(5±43 )
42
(48±98 )
7
(8±64 )
4
(5±43 )
Table 5.2: Summary of the mean value of some X-ray properties (given by the
best-fit model). The last column (KS-test) is the probability that both distributions
(IR power-law and IR non-power-law) come from the same distribution according to the Kolmogorov-Smirnov test.
X-ray parameter
hΓi N
hlog cmH,z
−2 i
hzi F
keV
hlog erg2−10
i
−1
−2
s cm L2−10 keV
hlog erg s−1 i
entire Sample
1.72 ± 0.36
IR power-law
1.74 ± 0.36
IR non-power-law
1.71 ± 0.37
KS-test
∼ 65%
22.00 ± 1.81
22.13 ± 1.36
21.89 ± 1.34
. 10%
1.42 ± 0.81
1.81 ± 0.79
1.15 ± 0.70
...
−14.35 ± 0.63
−14.33 ± 0.91
−14.36 ± 0.32
∼ 30%
43.50 ± 0.84
43.80 ± 0.99
43.28 ± 0.63
. 3%
5.3. X-ray spectral analysis
113
Absorbed Sources
10−5
PID 130
PID 283
10−6
10−7
10−9
10−8
keV (Photons cm−2 s−1 keV−1)
5×10−6
2×10−6
10−6
5×10−7
keV (Photons cm−2 s−1 keV−1)
10−5
2×10−5
Unabsorbed Sources
0.5
1
2
Energy (keV)
5
0.5
1
1
2
Energy (keV)
(iii) model B
5
10−7
10−6
PID 161
keV (Photons cm−2 s−1 keV−1)
PID 6
0.5
5
(ii) model C
10−6
10−7
keV (Photons cm−2 s−1 keV−1)
10−5
(i) model A
2
Energy (keV)
0.5
1
2
Energy (keV)
5
(iv) model E
Figure 5.5: Example X-ray spectra of four objects whose best-fit model is a simple
power law (i), absorbed power law (ii), a simple power law with thermal soft excess (iii)
and an absorbed power law with a Thomson scattering component (iv), respectively
(ID210=130, ID210=283, ID210=6, ID210=161).
Table 5.2 displays the mean values of some model parameters of our samples, as
computed from the best-fit models: the X-ray power-law photon index, the intrinsic
absorption column density, the observed 2-10 keV fluxes, and the 2-10 keV rest-frame
luminosity corrected for both intrinsic and Galactic absorption. On the one hand, we
found that both the measured X-ray photon index and the observed X-ray flux are
very similar between both samples. There is one unabsorbed AGN (ID210=114 as
IR power-law) for which the resulting photon index is >3. This very steep slope
is probably a consequence of both the small number of counts and high background
in its X-ray spectrum, although we cannot exclude a very high accretion rate source,
e.g. a NLS1 galaxy. On the other hand, the distributions of the intrinsic absorption
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
114
and the rest-frame 2-10 keV luminosity appear to span distinct regions of values for
each subsample. This will be discussed in the following subsections. The distribution
of these X-ray parameters are shown in Figure 5.6. Whilst the solid-line hatched and
hollow histograms show the distribution for IR power-law and IR non-power-law
galaxies, respectively, the dashed-line empty histogram shows the distribution for the
entire sample.
0.6
0.6
total
IR power-law (60)
IR non-power-law (87)
0.5
0.4
fraction
fraction
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
total
IR power-law (60)
IR non-power-law (87)
0.5
0
1
2
3
0
-16
4
-15
-12
0.5
IR power-law (60)
IR non-power-law (87)
total
Tozzi’s distribution
0.4
fraction
30
number
-13
) [erg/s/cm2]
2-10 keV
40
20
10
0
-14
log(F
Γ
total
IR power-law (60)
IR non-power-law (87)
0.3
0.2
0.1
20
21
22
23
intr
H
log(N )
24
25
0
41
42
43
log(L
44
) [erg/s]
45
46
2-10 keV
Figure 5.6: Distribution of the values of some parameters from the best-fit models
to each source. Top panels: Normalized distribution of the X-ray power-law photon
index and the observed 2-10 keV flux. Bottom panels: Normalized distribution of the
intrinsic absorption and the rest-frame 2-10 keV luminosity corrected for both Galactic
and intrinsic absorption. The empty circles with error bars in the intrinsic absorption
intr
distribution show the NH
distribution for the sample given by Tozzi et al. (2006).
Solid-line shaded and empty histograms show the distribution for the IR power-law
and IR non-power-law populations, respectively. The dashed-line empty histogram
shows the distribution for the entire sample.
No correlation of the power-law slope with the detected absorption was found, as
we see in Figure 5.7 (Spearman Rank coefficient SR∼-0.03). Note that if the intrinsic
absorption is close to the Galactic value for the CDF-S field (≈ 9 × 1019 cm−2 ) we are
not able to derive any meaningful value and can only provide an upper limit for the
absorption, NH < NH,Gal . In such cases, we keep a simple power law as the best-fit
model and we plot them at NH = 1020 cm−2 in our figures. We do not see any clear
trend of the fitted photon index of our objects with the hard rest-frame X-ray intrinsic
5.3. X-ray spectral analysis
115
4
Γ
3
IR power-law (60)
IR power-law (60)
IR non-power-law (87)
IR non-power-law (87)
2
1
IR power-law (60)
IR non-power-law (87)
0
21
10
23
22
10
intr
NH
10
[cm-2]
24
10
25
41
10 10
42
10
43
10
log(L
44
10
45
10
) [erg/s]
46
10
0
2-10 keV
1
2
3
4
5
redshift
Figure 5.7: Left to right. Scatter plot of the best fit values of Γ against absorption, rest-frame 2-10 keV intrinsic luminosity and redshift for both populations: circle
and triangle points represent IR power-law and IR non-power-law populations,
respectively. Error bars are at the 90% confidence level.
(unabsorbed) luminosity in the range of luminosities 1042 −1046 erg s−1 (see Figure 5.7)
where the Spearman’s Rank correlation coefficient is compatibles with zero. Finally we
do not find evidence for a correlation of the X-ray photon index γ with the redshift,
the Spearman’s Rank correlation coefficient being almost null.
Soft X-ray Component
Among the entire sample, only 33 sources (22+6
−2 %) have a soft emission component
(B, D, and E models, see Table 5.1). In the full X-ray spectral analysis presented
in Comastri & XMM-CDFS Team (2013) the origin of this soft X-ray emission
component is studied in more detail. We note that the low fraction of sources
with a significant soft X-ray component may be due to the high redshift of our
sources (see Figure 5.1), for which the soft component is often shifted below the
X-ray spectral band used.
Compton-thick candidates
In a few cases (5/147), the resulting photon index is .1 (even when accounting
for uncertainties), much lower than the typical values for unabsorbed AGN. As
we pointed earlier, a possible explanation would be that such sources are, in
fact, Compton-thick AGN. However, this explanation does not seem to be valid
because in all such cases we find a significant amount of absorption but not in
the Compton-thick regime (NH between 1022 and a few 1023 cm−2 ); and besides,
the iron Kα emission line is not detected within sensitive limits (> 3σ level). Our
adopted explanation for the sources with an apparent very flat X-ray spectral
index is the combination of a moderate absorption with poor statistics and/or
high background. In those cases, an upper limit for the absorption is given at
90% level by freezing the power-law photon index at 1.9 (see Chapter 1).
As noted earlier, the simple power-law model gives a good fit for 58 sources while
the remaining 89 (61+6
−7 %) sources display a best fit with a X-ray absorption, i.e.
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
116
C, D, and E models. The intrinsic absorption distribution measured is bimodal
(see Figure 5.6), in the sense that 39+7
−6 % of sources have very small NH (below
the Galactic value) and appear separated from the distribution of the bulk of
the sources. The lowest bin in the distribution (i.e., that with NH ∼ 1020 cm−2 )
accumulates those sources for which we cannot measure any intrinsic absorption,
i.e. it includes all the sources whose best-fit calls for an unabsorbed model (model
A and B, see Table 5.1). The last bin at NH = 1024 cm−2 includes the few sources
with NH > 1024 cm−2 but their error bars make them compatible with being
Compton-thin. Thus, there are no Compton-thick sources securely detected in
our sample using the first-order spectral model adopted in this work (see Table 5.3 and further discussion below). In Figure 5.6, we also show the normalized
distribution of the intrinsic absorption for the sample by Tozzi et al. (2006). We
remark that both distributions have in general a similar shape but they differ at
large absorptions (see Figure 5.6, bottom left panel), in the Compton-thick region.
Among the heavily obscured sources, we find two Compton-thick candidates in
the sample (right-most bin in the bottom panel in Figure 5.6), in the sense that
the column density has some likelihood of exceeding 1.5 × 1024 cm−2 . However
the uncertainty in the intrinsic column densities (see Figure 5.8) of these sources
(ID210=66 IR non-power-law, and ID210=144 IR power-law), is such that
their Compton-thin nature cannot be ruled out. In Figure 5.8 the X-ray spectrum using the best-fit model (C and D model, respectively) for both sources is
displayed. From this figure, we can see that this dichotomy could not be resolved
with additional criteria such as strong iron emission line or Compton reflection
component at least up to the 3σ confidence level.
While all X-ray background synthesis models predict that Compton-thick sources
have an important role to play in filling the so far unresolved XRB above ∼ 8 keV,
such sources have been generally elusive in X-ray surveys conducted both with
Chandra and XMM-Newton. This is also evidenced in the current work, which
shows that even with the deepest exposures it is very difficult to unambiguously
find sources with column densities in excess of 1024 cm−2 . In the case of our
XMM-Newton data this is likely due to a combination of large background at
high energies together with a modest (but still significant) effective area. In the
case of Chandra data, a much more reduced background (since the better angular resolution enables a much smaller extraction region for the X-ray spectra) is
hardly compensated by a smaller effective area with respect to XMM-Newton.
Detecting significant numbers of Compton-thick sources remains a challenge with
current X-ray instruments.
5.3. X-ray spectral analysis
117
300
min = 2.555787e+02; Levels = 2.578787e+02 2.601887e+02 2.647887e+02
+
100
10−6
200
10−5
200
ratio
ID210=66 (absorbed power−law model)
Parameter: nH (1022)
normalized counts s−1 keV−1
ID210=66 (absorbed power−law model)
100
0
0.5
1
2
Energy (keV)
5
0
400
5
min = 3.033334e+02; Levels = 3.056334e+02 3.079434e+02 3.125434e+02
10−6
10−7
300
+
100
200
10−5
50
ratio
2
3
4
Parameter: PhoIndex
ID210=144 (absorbed power−law model with SE)
10−4
Parameter: nH (1022)
normalized counts s−1 keV−1
ID210=144 (absorbed power−law with SE)
10−3
1
0
−50
0.5
1
2
Energy (keV)
5
1
2
3
Parameter: PhoIndex
4
5
Figure 5.8: Left panels: best-fit model and data (top) and the ratio between both
(bottom): C (top) and D (bottom) models. Right panels: Solid contour showing the
1σ, 2σ, and 3σ confidence level likelihood contours for the X-ray photon index (Γ)
vs. the intrinsic absorption (NH ).
Comparison with previous results
• ID210=144,147
We find that our best-fit value for these two sources of the intrinsic column
density using an absorbed power-law model is in very good agreement with
the fit to the XMM-Newton spectrum of the same sources with the more
sophisticated torus model of Murphy & Yaqoob (2009) by Comastri et al.
(2011) using the same ultra-deep XMM-Newton data in the CDFS. The
uncertainties on the fitted column density do not allow us to assert that
they are Compton-thick (see Figure 5.8), but Comastri et al. (2011) use the
presence of a strong Iron line and a strong reflection component to classify
source ID210=147 as such.
• ID210=48, 66, 144, 147, 214, 222, 245, 289, 324
Similarly, Georgantopoulos et al. (2013) have conducted a search for Compton-
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
118
Table 5.3: Summary of the sources in our sample that were identified as Comptonthick candidates (CThick) in other works.
ID210
us
T06
C11
B12
I12
G13
30
heavily
heavily
44
unabsorbed CThick
heavily
48
moderately
heavily
64
heavily
heavily
66
heavily
secure-CThick
106
moderately CThick
heavily
114
heavily
heavily
144
heavily
CThick CThick heavily heavily secure-CThick
147
heavily
CThick CThick heavily
secure-CThick
155
moderately CThick
heavily
180
heavily
heavily
heavily
214
heavily
heavily
222
unabsorbed
heavily
245
heavily
heavily
heavily
289
moderately
heavily
324
unabsorbed
secure-CThick
us: this work; T06: Tozzi et al. (2006); C11: Comastri et al. (2011); B12: Brightman
& Ueda (2012); I12: Iwasawa et al. (2012); G13: Georgantopoulos et al. (2013)
thick AGN using the same XMM-Newton exposures. The spectral model that
they fitted to the data is PLCABS (Yaqoob, 1997) which is specially tailored
to Compton-Thick sources, as it takes properly into account Compton scattering and reflection up to column densities up to ∼ 5 × 1024 cm−2 . A total
of 9 candidate Compton-Thick sources were found by them. Our analysis
of the X-ray spectra of these 9 sources is coincident with theirs in terms
of best-fit spectral index and absorbing column. They classify four of these
(ID210=144, 66, 147, 324) as Compton-thick: the first two had transmissiondominated spectra with strong Iron lines (we also find them to be heavily
absorbed) and the last two because of their reflection-dominated spectra (we
also detect 66 as a strongly absorbed source, but not 324, because in our fit a
flat unabsorbed model mimics its intrinsic reflection-dominated spectrum).
• ID210=30, 64, 144, 180, 245, 114
Another work using the same XMM-Newton exposure dealing with obscured
AGN is Iwasawa et al. (2012), who label these sources as strongly absorbed
sources in agreement with our findings. Again, the last source was found
by them to be a possible Compton-thick candidate because of its reflectiondominated spectrum, while our best-fit model is a flat spectrum with a
moderate column density.
• ID210=44, 106, 155, 144, 147
Other works to characterize obscuration using deep X-ray data in the same
5.4. Intrinsic absorption
119
region of the sky have also been carried out by Tozzi et al. (2006) and Brightman & Ueda (2012). Sources ID210=44, 106, 155, 144, and 147 were identified as Compton-Thick sources by Tozzi et al. (2006) with their 1 Ms Chandra data. We disagree with the Compton-Thick character of ID210=44, 106,
and 155, which appear to be at most moderately absorbed in our analysis,
while the last two would be heavily absorbed, in agreement with Brightman
& Ueda (2012), but see the discussion above about those last two sources
using additional criteria.
IR power-law
versus
IR non-power-law
from best-fit model
On the one hand, we found that both the measured X-ray photon index and the
observed X-ray flux remain virtually confined to the mean value (see Table 5.2)
for both subsamples (IR power-law and IR non-power-law). The distribution of the intrinsic absorption and the rest-frame 2-10 keV luminosity appears
to span distinct regions, on the other hand. There are some evidences that the
distribution of the intrinsic absorption column densities seems to have a different
shape for IR power-law and IR non-power-law populations according to the
Kolmogorov-Smirnov test (. 10%) which suggests that both samples might have
different parent distributions. Finally, we also found that the rest-frame 2-10
keV luminosity distribution appears to have a different shape for IR power-law
and IR non-power-law populations according to the Kolmogorov-Smirnov test
(. 3%). The mean values for the logarithmic X-ray luminosity are 43.8 ± 0.99
and 43.28 ± 0.63 for the IR power-law and IR non-power-law subsamples,
respectively. The IR power-law population appears to select better higher luminosity AGN because as expected, the IRAC selection cannot efficiently identify
low-luminosity AGN (see Donley et al. 2008, 2007).
5.4.
Intrinsic absorption
To explore and quantify the efficiency in finding highly-obscured AGN and/or QSO
when selecting the sources as IR power-law and IR non-power-law, all individual
spectra were re-fitted by an absorbed power-law model (plus a soft-excess component
when required by the X-ray data), i.e., using models C, D or E. In addition, now, for
those sources with a small number of counts (< 500), the X-ray power-law photon index
(Γ) was fixed to 1.9 (the typical average value for unabsorbed AGN) to constrain better
the intrinsic absorption. As expected, for the sources, which had already been fitted
with models C, D, or E the results remain essentially unchanged, the only difference is
that now we assign a 1σ uncertainty interval to the intrinsic column density. Similarly,
for the sources whose best-fit model was A or B the results are essentially compatible
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
120
except for 8 sources, which are now classified as absorbed. This happens because of two
reasons: in two cases, the photon index is now fixed to 1.9, because of the low number
of counts, hence the fitted absorption increased considerably because these sources had
low values of the photon index; for the other six sources, the F-test probability of
models C,D,E versus A,B was &90% (hence the former were not significantly better than
the latter, according to our criterion in Section 5.3: the 95% confidence interval on
intr includes zero) while the bottom of the 1σ confidence interval on N intr is above
NH
H
1022 cm−2 (i.e., they are absorbed according to our classification in this Section). The
intrinsic hard X-ray luminosities derived in this Section and in Section 5.3 are very
similar, only in two cases the difference exceeds 50%, but in all cases the source remains
in the same luminosity bin (as defined below).
5.4.1.
Absorbed fraction from best-fit values
We have subdivided our IR power-law and IR non-power-law samples accordintr values, also taking the 1σ uncertainty intervals into account, in the
ing to their NH
sense that those sources whose 1σ uncertainty interval is fully above (below) 1022 cm−2
are called absorbed (unabsorbed) at the 1σ level. Those objects whose 1σ interval
crosses the 1022 cm−2 border have been labelled as unclassified. The number of sources
in each of these subdivisions in bins of 2-10 keV luminosity is listed in Table 5.4 and the
individual classification of each source is shown in Table 5.5. As a further test of the
robustness of our estimates, we also redefined the absorbed and unabsorbed samples
using a 2σ threshold around 1022 cm−2 . The number of sources in each subsample
changes very little, and our conclusions remain unaffected. In what follows, in our
conservative approach those labelled unclassified will be considered as unabsorbed.
We measured a significant intrinsic absorption in excess of 1022 cm−2 for 83 sources
(see Table 5.4), of which 41 are IR power-law and 42 are IR non-power-law AGN.
A further 27 galaxies are classified as unabsorbed, out of which 6 are IR power-law
and 21 IR non-power-law galaxies. 37 sources were unclassified at the 1σ level.
An important result from this work is that the fraction of absorbed sources (see
Table 5.4) is higher among the IR power-law galaxies (41 out of 60) than among
the IR non-power-law galaxies (42 out of 87). Bayesian error estimates return a
fraction of absorbed sources among the IR power-law galaxies of 68+9
−10 % and among
the IR non-power-law galaxies of 48+9
−8 %. A Bayesian estimate of the probability
of these two fractions coming from the same parent distribution (Stevens et al., 2005)
yields a probability of 0.0023, and therefore the fraction of absorbed sources among
IR power-law galaxies is higher than that of IR non-power-law galaxies at about
3σ confidence level.
5.4. Intrinsic absorption
121
Table 5.4: Number of sources classified as absorbed/unabsorbed (at 1σ significance
level) and unclassified as a function of the intrinsic 2-10 keV luminosity according to
intr
their intrinsic column density. The NH
is obtained assuming an absorbed power-law
model (i.e., C, D, and E model). We remark that absorbed/unabsorbed and unclassified
groups are disjoint sets.
Absorbed
Sample
IR power-law
log LX [erg/s]
< 43 43 − 44 ≥ 44
5
15
21
Unclassified
all
41
log LX [erg/s]
< 43 43 − 44 ≥ 44
0
6
7
Unabsorbed
all
< 43
13
1
log LX [erg/s]
43 − 44 ≥ 44
0
5
Total
all
6
60
IR non-power-law
5
28
9
42
10
10
4
24
10
10
1
21
87
..........................................................................................................................
Total
5.4.2.
10
43
30
83
10
16
11
37
11
10
6
27
147
Dependence on X-ray luminosity from best-fit values
Our next step was to check for a possible dependence of the column density on
luminosity, within the two subsamples. We selected three 2-10 keV X-ray luminosity
ranges, LX ≤ 1043 erg s−1 , LX ∈ 1043 −1044 erg s−1 , and LX > 1044 erg s−1 with median
redshifts of 0.65, 1.59, and 2.21, respectively (with a standard deviation of 0.11, 0.69
and 0.75, respectively). We compute the fraction of IR power-law (IR non-powerlaw) sources, which have been classified as absorbed AGN at each luminosity range,
understood as the ratio of the number of absorbed IR power-law (IR non-powerlaw) at 1σ level to the total number of IR power-law (IR non-power-law) sources
at this luminosity bin. We find that the fraction of absorbed sources appears roughly
constant with luminosity for the IR power-law AGN, while it grows from ∼ 20% to
∼ 64% for the IR non-power-law galaxies (see Figure 5.9).
5.4.3.
Dependence on X-ray luminosity from probability density functions
intr distributions
To take the full distribution of probabilities when building the NH
into account, we followed the following procedure. First, when fitting the X-ray spectrum of each individual source, we used the steppar command under the XSPEC fitting
intr ). We explored sufficiently wide
package, to find out how the χ2 varies with log(NH
intr ) in such a way that ∆χ2 = χ2 − χ2
2
ranges of log(NH
min (where χmin corresponds
intr value) reaches sufficiently high values (hence low probabilities,
to the best-fit NH
int ) = 20.0, because, as discussed
see below). We set an absolute minimum of log(NH
before, our data are not sensitive to lower values of the intrinsic absorption. Next, we
intr of each individual source by using
constructed a probability density function for NH
intr )) ∝ exp(−∆χ2 /2), and normalised it between log(N int /cm−2 ) =20 and 25.
p(log(NH
H
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
122
1.2
IR power-law
IR power-law (pdf’s)
IR non-power-law
1
IR non-power-law (pdf’s)
f abs
0.8
0.6
0.4
0.2
0
42
43
log( L
44
) [erg/s]
45
2-10 keV
Figure 5.9: Fraction of absorbed sources (at ≥ 1σ) for IR power-law and IR nonpower-law galaxies as a function of the rest-frame absorption corrected 2-10 keV
luminosity. The open squares and star symbols show the fraction of absorbed sources
taking into account the probability density function for NH of each individual source
for IR power-law and IR non-power-law, respectively. Note that these fractions
refer exclusively to our sample and not to the overall AGN population.
intr )) into a common grid of values of log(N intr ) spaced 0.2
We registered these p(log(NH
H
units. We finally sum up the resulting probabilities in three bins of X-ray luminosity
and for the entire sample, also taking both IR power-law and IR non-power-law
populations into account, normalising each of the summed probabilities to a total unit
area (see Figure 5.10).
intr for the entire sample show some difference
The normalised distributions of NH
between the two populations of IR power-law and IR non-power-law galaxies,
with the former being more heavily absorbed than the latter. This behaviour is not
evident at high or intermediate luminosities, where both populations have an overall
indistinguishable distribution. However, at the lowest luminosity regime, the distriintr distribution for the
butions appear to have quite a different shape. Whilst the NH
IR power-law sample peaks at ∼ 1023 cm−2 , the IR non-power-law sample appears to have largely unabsorbed sources. This is somewhat unexpected because of
the known incompleteness of the power-law selection at low AGN luminosities. We
interpret this as due to the different shapes of the IR AGN SEDs of type-1 and type-2,
with type-2 having steeper SEDs (see Figure 1.3). The host galaxy, on the other hand,
has a “broad” bump peaking at ∼1.6µm (rest-frame). Therefore, the combination of a
5.4. Intrinsic absorption
123
relatively low luminosity type-1 AGN and its host galaxy would result in a relatively
flat spectral shape in the IRAC bands, likely not meeting the IRAC IR power-law criterion.
As a further test of the robustness of our estimates, we also measured the fraction of
absorbed sources in both samples at each luminosity bin using the probability density
intr of each individual source. The absorbed source fractions change
function for NH
very little (see Figure 5.9) both results being compatible. Moreover, if we compute
the absorbed source fraction excluding those sources with photometric redshift (11/60
IR power-law and 12/87 IR non-power-law sources), our findings are also in good
agreement when accounting for uncertainties.
0.2
0.15
LX<1043
IR power-law
IR non-power-law
IR non-power-law
0.1
fraction
fraction
0.15
1043<LX<1044
IR power-law
0.1
0.05
0.05
0
20
21
22
23
24
0
20
25
21
22
intr
log N
H
(i) LX ≤ 1043
24
25
24
25
(ii) 1043 < LX ≤ 1044
0.15
0.15
LX>1044
IR power-law
IR power-law
IR non-power-law
IR non-power-law
0.1
0.1
fraction
fraction
23
intr
H
log N
0.05
0
20
0.05
21
22
23
intr
log N
H
(iii) LX > 1044
24
25
0
20
21
22
23
intr
H
log N
(iv) the entire sample
Figure 5.10: Normalized distribution of intrinsic absorbing column densities (in
log units) in three absorption corrected 2-10 keV luminosity bins (upper panels and
lower left panel) and for the entire sample (lower right panel). The distributions were
intr
computed using the probability density function for NH
of each individual source.
intr ) into account, our
Therefore, taking the full probability distribution of log(NH
analysis is in full agreement with the one using just the 1σ confidence levels: the fraction of absorbed sources among IR power-law AGN is higher both in the full sample
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
124
and among lower luminosity sources (log LX /erg s−1 < 43), while it is compatible
with being similar at higher luminosities. We note that, on the one hand, we are not
trying to determine the absolute fraction of absorbed sources as a function of the X-ray
luminosity, but to evaluate the effectiveness of the mid-IR selection to identify obscured
AGN by comparing the fraction of absorbed sources inside and outside the IRAC criterion. On the other hand, we do not apply any completeness correction to the absorbed
source fractions, thus these fractions of absorbed sources should not be compared with
results referring to the overall AGN population (Gilli et al., 2007, Treister & Urry, 2005,
Ueda et al., 2003). We also point out that the difference on the fraction of obscured
AGN at the lowest luminosity bin should be interpreted with caution, since there are
only 6 IR power-law sources in this low-luminosity bin.
In conclusion, although the IR power-law selection only picks up 60/147 (about 41%)
of our X-ray-selected high-spectral-quality sources at high X-ray luminosities it singles
out about 70% of our sources (33/47). Concentrating on the IR power-law sources,
the overall percentage of such sources, which are absorbed is 68% (41/60), essentially
independent of the AGN luminosity, better than the overall fraction of absorbed sources
in our entire sample (83/147 ∼56%) and significantly higher than that of IR nonpower-law sources (42/87∼48%). As Donley et al. (2012) found, the IR powerlaw selection produces an incomplete census of AGN, its completeness being a strong
function of AGN luminosity. Our overall estimate of its efficiency to find absorbed
sources (∼70%) is similar, if marginally lower, than their estimate of ∼75%.
5.5.
Summary and Conclusions
In this work we have investigated the subset of X-ray sources in the ultra-deep
XMM-Newton observation of the Chandra Deep Field South (Ranalli et al., 2013) with
a highly significant (>8σ) detection, high exposure time (>1Ms) and known (spectroscopic or photometric) redshift, totalling 147 sources. All of them turn out to have a
unique Spitzer/IRAC counterpart, and they are all detected in the four IRAC bands.
Consequently, our final sample is biased towards high signal-to-noise X-ray sources
and cannot be considered as a complete sample of X-ray selected or mid-IR-selected
sources. However, this does not affect the main goal of this work: to test the efficiency
of the IR power-law criterion in selecting absorbed X-ray sources. We have used IRAC
photometry to classify these sources into IR power-law and IR non-power-law
galaxies, according to whether or not their IR SED has a power-law-like shape and is
monotonically increasing, following Donley et al. (2012).
We have estimated the absorbing column density assuming an absorbed power-
5.5. Summary and Conclusions
125
law model. Each source was classified as absorbed or unabsorbed at 1σ level, in the
sense that those sources whose 1σ uncertainty interval is fully above 1022 NH or not
(respectively). And finally, we further explored possible mismatches in the observed
X-ray absorption distributions for these subsamples (IR power-law and IR nonpower-law) in three intrinsic rest-frame 2-10 keV luminosity bins (< 1043 erg s−1 ,
1043−44 erg s−1 , and > 1044 erg s−1 ). The goal was to investigate whether the IR
power-law criterion selects more absorbed AGN or not, and specifically whether it is
effective at selecting type-2 AGN at a given luminosity range. The main results from
our work are as follows:
• All high-X-ray-spectral-quality sources in the deepest XMM-Newton field have
counterparts in the Spitzer/IRAC bands, therefore using mid-IR data to search
for obscured sources does not leave any candidates out.
• With our absorbed-power-law X-ray spectral analysis, 21 out of 147 sources are
intr > 3 × 1023 cm−2 ) but, when taking the uncertainties into
heavily absorbed (NH
account, we cannot confirm the Compton-Thick nature of any of our sources. We
are, however, in full agreement with other papers using the same deep XMMNewton data, which use additional criteria to that end (Georgantopoulos et al.,
2013).
• At more than 3σ level, we find that the fraction of absorbed sources among the
IR power-law populations (68−9
+8 %) appears significantly higher than that for
IR non-power-law galaxies (48−10
+9 %).
• We also found that the percentage of absorbed sources appears roughly constant
(∼ 70%) with luminosity for the IR power-law AGN, while it grows from ∼ 20%
to ∼ 65% for the IR non-power-law galaxies with increasing luminosity.
• The main difference in the absorbed fraction between the IR power-law and the
IR non-power-law sources happens at the lowest X-ray luminosities (< 1043
erg/s). We understand this in terms of contrast with the host galaxy, in the sense
that type-2 AGN (in principle absorbed in X-rays) are more easily picked up by
those criteria than (unabsorbed) type-1 AGN at low luminosities.
We conclude that the Donley et al. (2012) IRAC criterion, if admittedly incomplete, favour the selection of absorbed sources among the X-ray detected AGN. This
is particularly clear at low X-ray luminosities. This prompts the question about the
nature of the IR power-law in the XMM-Newton area without X-ray detection. Since
we do not miss any X-ray-detected sources by using mid-IR data and about 2/3 of the
selected sources turn out to be absorbed, it is likely that the mid-IR power-law criterion
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
126
would pinpoint absorbed AGN among X-ray undetected sources. These sources would
still be detected in the mid-IR (from the reprocessing of the AGN radiation), but they
would have high X-ray absorbing column densities, placing them beyond the current
capabilities of our most powerful X-ray observatories pushed to their limit.
We dare to postulate that the revised-IRAC criterion is efficient in finding X-rayabsorbed sources, to such an extent we would then expect that the long-sought dominant population of absorbed AGN is abundant among IR power-law spectral shape
sources not detected in X-rays (it is only a speculation that needs extending the present
work).
127
5.5. Summary and Conclusions
z
(2)
1.622
1.591
0.526
1.050
0.662
1.156
1.368
3.198
1.936
2.583
1.843
1.633
0.624
1.598
0.512
0.619
1.167
1.508
0.717
0.298
2.298
1.218
2.571
1.049
2.561
3.341
1.185
2.005
0.494
0.839
0.737
1.887
2.561
1.271
3.417
2.819
1.170
0.878
2.732
3.001
1.571
2.034
0.667
0.665
2.562
0.679
1.097
0.738
1.806
FX
(3)
9.89
4.20
9.15
4.19
6.00
3.68
2.67
2.73
9.86
7.67
35.02
1.13
3.75
9.37
4.58
3.01
3.89
3.45
4.23
42.64
4.55
3.65
2.71
3.78
6.51
6.75
6.25
6.99
3.99
3.31
17.06
17.70
2.44
0.00
1.15
2.48
5.29
3.08
0.99
2.38
0.40
9.41
2.33
6.05
2.99
3.48
11.03
1.69
5.54
LX
(4)
195.68
64.01
8.48
16.42
10.16
22.15
34.46
239.94
204.64
176.62
637.08
27.06
8.03
129.04
4.43
4.71
24.80
22.13
7.67
11.94
103.95
26.88
184.29
17.66
369.28
327.35
112.48
173.27
4.53
11.09
44.77
311.20
88.87
0.00
176.05
122.41
71.49
8.56
133.75
119.75
67.80
231.52
5.09
9.73
109.99
29.01
72.33
4.20
15.74
Γ
(5)
0.1
2.1±0.1
1.9±0.4
0.4
0.1
1.6±0.1
0.4
1.4±0.3
0.3
1.7±0.3
0.6
1.7±0.5
1.9±1.7
1.1
1.2
1.9±0.8
0.8
1.5±0.6
0.1
1.3±0.1
0.0
1.7±0.1
1.9†
1.9†
0.1
1.8±0.1
0.1
1.9±0.1
0.2
1.9±0.2
0.1
1.8±0.1
0.3
1.1±0.3
1.6±0.1
0.1
0.1
1.5±0.1
0.3
1.5±0.2
0.3
1.8±0.3
0.1
2.2±0.1
1.6±0.4
0.3
0.1
2.0±0.1
0.6
1.4±0.4
1.7
1.9±1.5
0.1
1.9±0.1
0.6
2.3±0.4
0.2
2.0±0.2
1.7±0.3
0.3
0.1
1.6±0.1
0.4
1.7±0.4
0.4
1.3±0.4
0.4
2.2±0.4
1.7±0.6
0.5
0.9
2.1±0.8
0.2
1.5±0.1
1.2
2.4±1.0
0.3
1.7±0.3
≤ 4.0
0.1
1.8±0.1
1.9†
0.1
1.6±0.1
0.3
1.7±0.3
1.2
2.7±1.0
0.2
1.7±0.2
1.9±0.6
0.5
≤ 5.0
intr
NH
(6)
3.83
8.37±2.84
0.96
0.86±0.75
0.71
1.57±0.56
2.86
3.04±1.81
29±27
15
60
46±29
60
77±36
82±36
30
20±811
0.17
0.21±0.16
0.81
4.36±0.62
2.25
3.43±1.86
1.23
2.16±0.95
4.24±1.73
1.42
44
70±31
139
143±101
3.47
2.30±0.92
0.52
1.47±0.43
16±43
14±68
12±58
19±811
19±14
10
36
65±30
64
59±36
5.01
5.84±3.83
44
43±26
0.79
3.35±0.73
30±88
0.22
0.77±0.20
30±912
34
71±26
30±67
3.60±2.08
1.55
-
KT
(7)
0.024
0.067±0.021
0.072
0.224±0.068
0.081
0.242±0.049
0.119
0.278±0.130
0.712
0.135±0.098
0.059
0.411±0.077
-
model
(8)
A
C
B
C
C
C
C
C
C
A
B
C
C
A
A
C
A
A
A
D
C
C
A
C
B
C
C
A
D
C
C
B
C
C
C
C
C
A
C
C
E
C
C
C
C
C
E
C
A
IRpl
(9)
3
7
3
7
7
7
7
3
7
3
3
3
7
3
7
7
7
7
7
7
3
7
3
7
3
3
7
7
3
7
7
3
7
3
7
3
7
7
7
3
3
3
7
7
3
7
7
3
3
ID210
(1)
116
118
120
123
124
130
133
134
137
138
142
144
146
147
151
155
158
161
165
166
173
174
175
178
180
182
185
190
191
194
197
198
200
201
203
204
210
211
213
214
219
221
222
227
228
230
231
232
233
z
(2)
3.531
1.609
3.591
1.438
0.346
1.628
0.671
1.609
0.751
1.220
1.379
3.700
0.122
1.536
1.508
0.732
2.394
0.686
1.016
1.361
1.920
1.581
1.519
1.380
3.064
1.034
0.735
3.101
0.821
2.838
0.667
0.860
2.567
2.713
0.544
0.599
3.193
2.612
2.202
1.499
2.308
1.220
0.424
0.834
1.216
0.520
1.730
1.620
1.615
FX
(3)
4.91
8.60
3.31
2.93
1.39
13.44
2.06
16.81
1.82
3.44
3.19
0.93
8.24
5.85
7.31
3.23
4.54
7.71
4.22
1.90
4.78
13.85
5.04
1.24
5.43
4.96
1.65
3.75
6.07
3.64
1.62
5.82
10.00
3.07
78.64
4.11
1.22
3.66
3.04
15.36
1.68
2.97
2.87
8.81
2.84
2.16
2.92
2.58
5.67
LX
(4)
475.68
172.30
217.70
29.43
0.53
212.81
3.62
224.82
6.00
23.26
19.82
5592.82
0.33
58.53
74.73
6.39
141.01
17.00
17.69
36.06
100.08
180.35
57.70
19.47
376.04
22.11
4.03
204.51
14.90
219.13
2.48
32.79
455.99
164.74
87.31
20.06
115.02
122.21
35.26
124.88
63.03
57.61
1.93
30.39
21.25
2.13
33.05
34.80
75.77
Γ
(5)
0.3
1.8±0.2
0.1
2.2±0.1
0.1
1.6±0.1
0.6
1.7±0.5
0.2
1.8±0.1
0.0
1.9±0.0
1.8±0.1
0.1
0.0
1.8±0.0
1.1
2.3±0.8
0.1
1.7±0.1
0.5
1.2±0.4
≤ 4.0
0.2
1.6±0.2
0.7
1.3±0.6
0.1
1.6±0.1
1.0
1.3±1.0
0.3
1.6±0.3
0.3
1.9±0.3
1.5±0.3
0.3
1.7
2.3±1.0
0.1
1.8±0.1
0.1
1.8±0.1
0.4
1.6±0.3
2.3±0.4
0.3
0.4
1.8±0.3
0.1
1.7±0.1
0.4
2.0±0.3
0.2
1.7±0.1
0.1
1.6±0.1
0.2
1.9±0.2
1.5±0.4
0.3
0.8
2.2±0.7
0.1
1.9±0.1
0.1
1.9±0.1
0.0
1.9±0.0
2.7±0.8
0.7
0.5
2.0±0.4
0.2
1.6±0.2
0.2
1.0±0.2
0.2
1.2±0.2
0.2
1.9±0.2
0.7
2.2±1.1
1.9†
0.4
1.9±0.5
0.2
1.8±0.2
0.4
1.8±0.3
0.4
1.4±0.4
1.8±0.1
0.1
0.4
1.7±0.3
intr
NH
(6)
16
43±14
0.27
2.40±0.25
2.56
2.34±1.77
3.10
4.50±2.21
10±68
249±89
132
0.33
1.79±0.29
42
59±32
0.92
4.22±0.82
22
18±15
22±78
2.44
7.10±2.07
6.49±2.37
1.91
66
40±25
24±710
1.11±0.73
0.61
28
87±23
0.43
0.59±0.35
1.72
2.48±1.55
28±67
0.91±0.85
0.60
23±913
48±17
13
7.09
7.29±4.99
2.34
4.32±1.97
26±68
36
86±48
3±11
2.08
3.64±1.51
0.93
2.65±0.79
0.70
1.02±0.49
14±68
12±45
KT
(7)
0.049
0.067±0.038
0.469±0.469
0.267
0.069±0.069
0.052
0.140
0.208±0.093
0.146
0.362±0.098
0.155
0.039±0.009
0.902
0.902±0.902
0.212
0.466±0.392
0.087
0.116±0.055
model
(8)
E
C
A
C
A
A
A
A
D
A
C
D
C
C
C
C
C
E
D
C
A
A
C
C
C
A
C
C
A
C
C
C
B
A
A
C
C
C
A
D
B
D
C
D
C
C
C
A
D
IRpl
(9)
3
3
7
7
7
3
7
3
7
7
7
3
7
7
3
7
3
7
7
7
7
7
3
3
3
7
7
7
7
3
7
7
3
3
3
7
7
3
7
7
3
7
7
7
3
7
3
7
3
ID210
(1)
234
237
244
245
248
249
262
265
269
273
277
281
283
284
285
289
291
297
301
305
307
308
310
311
312
315
318
319
320
321
323
324
326
330
337
338
341
344
345
358
359
361
366
374
375
382
385
388
390
z
(2)
0.952
0.620
1.260
1.864
0.922
0.738
1.621
0.665
1.764
0.733
1.323
1.264
2.291
0.566
2.072
0.607
1.215
1.440
0.665
0.467
0.857
0.736
0.668
2.091
1.406
0.534
0.663
0.742
0.735
0.733
0.181
1.222
1.089
2.208
0.837
1.270
1.324
1.512
1.235
0.976
1.574
0.977
2.347
1.041
1.350
0.527
1.220
0.605
1.118
FX
(3)
3.11
4.10
4.45
1.01
0.77
7.35
8.61
3.46
4.74
4.97
6.71
8.42
4.46
4.51
3.58
11.70
3.30
7.54
2.25
28.99
5.53
13.52
10.21
9.34
7.44
2.83
6.12
80.80
3.69
5.77
2.95
2.32
2.80
5.11
25.94
11.04
10.53
18.92
6.91
68.91
34.95
20.00
3.27
13.20
4.34
46.32
7.46
4.24
6.30
LX
(4)
12.45
6.09
37.87
6.88
2.39
18.99
134.48
5.61
66.75
11.38
69.56
69.26
147.98
4.80
101.20
12.87
26.54
118.57
3.72
20.75
16.76
30.09
20.66
211.98
65.37
2.91
11.30
189.17
4.64
8.18
0.25
7.63
17.57
132.73
84.93
55.82
144.78
206.39
69.43
267.86
428.02
89.39
73.65
58.81
49.77
50.58
36.03
5.44
41.36
Γ
(5)
0.2
1.8±0.2
0.1
1.8±0.1
0.1
1.9±0.1
≤ 1.52
0.3
1.5±0.3
0.1
2.1±0.1
1.9±0.0
0.0
0.1
1.5±0.1
0.3
1.6±0.3
0.1
1.9±0.1
0.1
2.0±0.1
1.8±0.1
0.1
0.3
1.8±0.3
1.5±0.2
0.2
0.1
1.9±0.1
0.2
1.2±0.2
1.3
1.6±1.0
0.6
2.0±0.5
1.7±0.1
0.1
0.1
1.6±0.1
0.3
1.7±0.3
0.0
1.8±0.0
0.2
1.9±0.2
1.7±0.2
0.1
0.3
1.6±0.2
0.1
1.8±0.1
0.2
1.9±0.2
0.0
1.9±0.0
≤ 1.87
0.1
1.0±0.1
1.4±0.2
0.2
0.3
0.8±0.3
1.9†
0.1
1.7±0.1
0.1
1.9±0.1
1.2±0.3
0.2
0.0
2.3±0.0
0.1
1.6±0.1
0.1
2.2±0.1
0.1
1.5±0.1
0.0
1.7±0.0
0.0
1.9±0.0
1.5±0.2
0.2
0.3
1.5±0.3
0.1
2.1±0.1
0.1
1.7±0.1
0.3
1.3±0.3
1.6±0.2
0.1
0.1
1.9±0.1
intr
NH
(6)
0.53
1.04±0.45
43
27±17
1.71
2.32±1.36
0.86±0.22
0.21
18±56
2.56
4.18±1.74
50
39±26
16
30±11
0.34
0.46±0.22
0.60
1.06±0.48
1.50
6.93±1.38
6.36±1.46
1.31
3.54
7.38±3.22
9.86±18.25
6.51
5±22
0.80
1.06±0.74
2.68±1.74
1.32
1.87
9.95±1.75
12±12
13±35
3.36
3.78±1.84
1.59
2.70±1.23
-
KT
(7)
0.042
0.075±0.025
0.007
0.007±0.007
0.038
0.218±0.048
0.052
0.090±0.056
0.047
0.186±0.066
0.018
0.214±0.020
0.022
0.225±0.026
0.278
0.278±0.160
0.063
0.063±0.063
0.063
0.398±0.077
-
model
(8)
C
A
A
C
A
B
A
B
C
A
A
C
C
B
A
E
C
C
A
D
C
A
E
C
E
A
B
B
C
A
A
A
C
C
B
C
A
E
A
D
A
A
A
C
B
D
C
A
A
IRpl
(9)
7
7
7
7
7
7
3
7
3
7
7
7
3
7
7
7
3
3
7
3
7
7
3
3
3
7
7
3
7
7
7
3
3
7
3
3
3
7
7
3
3
3
3
7
3
7
7
7
7
Table 5.5: Details of the X-ray sources with IRAC counterpart found in this analysis presented in three panels. Column (1) is the
identification number of the X-ray sources listed in Ranalli et al. (2013) column (2) is the redshift of the source; column (3) gives the
observed 2 − 10 keV flux of the source in units of 10−15 erg/s/cm2 ; column (4) gives the unabsorbed rest-frame 2 − 10 keV luminosity
of the source using the best fit model in units of 1042 erg/s; column (5) is the X-ray photon index, Γ, where † indicates that the index
was frozen in the fit; column (6) is the measured NH in units of 1022 cm−2 ; column (7) gives the electron temperature for the soft
Comptonisation component in units of keV; column (8) is the best-fitting model for the spectrum as described in Table 5.1; column (9)
is a tick to denote whether the sources is classified as IR power-law or IR non-power-law galaxy.
ID210
(1)
2
3
6
13
19
21
24
26
30
31
33
34
37
38
40
42
43
44
45
48
49
50
57
60
62
64
66
68
71
72
78
81
84
89
92
93
95
96
97
98
102
103
106
107
108
109
111
113
114
IR POWER-LAW AS AN INDICATOR OF OBSCURATION
128
CHAPTER 6
Conclusions and Future Work
The standard unified model for AGN has been tested in many different ways in
the past few years, through a large set of new observations. Overall, the fundamental
aspect of the model is that the non-spherically symmetric absorption medium plays a
major role in explaining the differences in the observed features among all AGN types.
In this framework, the variety of the AGN population is assumed to be caused by the
different levels of obscuration/absorption along the line of sight: (obscured/absorbed)
type 2 and (unobscured/unabsorbed) type 1 AGN are intrinsically the same class of
objects. Although this has been confirmed, and even reinforced by the most recent
observations in the local Universe, there are still open issues. There is also strong evidence that AGN play a key role along a galaxy’s lifetime. Major galaxy mergers have
been suggested to funnel gas to the nuclear region of galaxies triggering star-formation
and feeding black hole growth, although other processes might also be responsible for
the trigger of AGN activity. The main phase of black hole growth appears to occur in
heavily obscured environments where outflows -AGN feedback - regulate the growth of
the black holes and their host galaxies. Identifying complete and reliable samples of
AGN, with full knowledge of any underlying biases or contamination in the sample, has
become a necessity for galaxy surveys. One can then determine which processes (e.g.,
secular evolution, mergers, and environment) are the dominant fueling mechanisms in
the growth and evolution of super-massive black holes. However, accretion rates, orientations, and intrinsic obscurations of AGN prevent any single selection technique from
reliably and completely identifying all AGN types.
The scientific aim of this dissertation comes out mainly of this need for a complete
census of the type 2 AGN population at cosmological distances.
CONCLUSIONS AND FUTURE WORK
130
AGN missed by the BPT diagram
When both X-ray data and optical spectra are available for the same galaxy, galaxies
optically classified as starforming by optical line spectroscopy diagnostics may be found
to have high X-ray luminosities, in excess of the most luminous starforming galaxies
known in the local Universe, by over an order of magnitude (i.e., X-ray luminosities
> 1042 erg s−1 ). Although the origin of this classification discrepancy has already been
addressed in previous works, it is not fully understood. Trouille & Barger (2010) found
that the Baldwin et al. (1981)-type empirical diagnostic diagram misidentifies 20-50%
of the X-ray selected AGN that are star formers according to that diagram. They postulate that these optically misidentified hard X-ray-selected AGN have low [O III]λ5007
luminosities due to the complexity of the structure of the narrow-line region. The work
of Yan et al. (2011) also reports another case of misclassification in agreement with
previous works (e.g., Netzer & Trakhtenbrot, 2007, Trouille & Barger, 2010). All these
previous works are in good agreement with that these emission-line ratio diagnostic
diagrams lose many AGN in star-forming galaxies.
We compare the optical and hard X-ray identifications of AGN of a low redshift
(z . 0.4) sample of 211 Narrow Emission Line Galaxies (NELG) to study this discrepancy. Chapter 3 provides a quantitative estimate of the incidence of such mismatches: although this population of NELG optically-classified as starforming galaxies
with rest-frame 2-10 keV luminosity in excess of 1042 erg s−1 represents only a small
fraction (∼2-7%) of the overall starforming population and hence they do not represent
a major problem in terms of contamination, this discrepancy means ∼25% of the hard
X-ray-selected AGN.
For the first time, our work shows that these X-ray luminous (LX & 1042 erg s−1 )
star-forming galaxies largely belong to a specific class of AGN: from the high values
of Hβ FWHM (& 600 km s−1 ), the strong Fe II optical emission present in the vast
majority of them and the almost ubiquitous presence of an X-ray soft excess, we conclude that these objects are compatible with being Narrow Line Seyfert 1 (NLS1). This
finding confirms the Trouille & Barger (2010)’s postulate, as NLS1 galaxies show different physical conditions for the narrow-line region than Seyfert 1 (Xu & Komossa, 2011).
The results obtained from this work can be further refined in the future through
a comparison with other different methods for optical selection of such sources. To
strengthen the conclusions that a NLS1 could be selected from their high hard X-ray
luminosity (characteristic of AGN), but optically diagnosed as starforming galaxies,
131
this study should be repeated using other optical galaxy surveys (such as COMBO-17
or even MUSYC).
non-standard NLS1
NLS1 class is an exceptional subclass of AGN, harbouring low-mass black holes
accreting at a high rate. As such, they may hold important clues on the nature of
black hole growth and evolution, and of feeding and feedback in the course of galaxy
evolution. Studying the multi-wavelength properties of NLS1 galaxies, the links and
correlations among them, and the physics that drive them, is therefore of great interest.
One of the prime problems in understanding this kind of sources is that their defining
properties (such as soft X-ray excess, Fe II optical emission, Eddington ratio, etc.) as
an AGN class depend on the selection method which is far from being uniquely defined
(e.g. Goodrich, 1989, Laor, 2000, Netzer & Trakhtenbrot, 2007, Osterbrock & Pogge,
1985).
Complementing previous works, we have tried to unravel the role of the nonstandard NLS1 population: why not all NLS1 galaxies have Fe II optical emission,
or soft X-ray excess?
In order to find a place for the non-standard NLS1 in the AGN framework, we have
studied in Chapter 4 the correlations among the measured and derived properties of a
sample of 19 NLS1 galaxies (among which there are 4 non-standard NLS1, including
both NLS1 without Fe II optical emission and without soft X-ray emission) on the basis of a broadband SED analysis (from about 0.9 microns to 10 keV), accompanied by
a comparison sample of Broad Line Seyfert 1 (BLS1) galaxies analyzed in the same way.
From this multi-wavelength study we conclude, for the first time, that the nonstandard NLS1 galaxies whose classification was based only on the classical FWHM
cut-off of Hβ, are probably BLS1 having the lowest values of the Hβ FWHM: the
correlation between the Hβ FWHM and the Eddington ratio shows that these nonstandard NLS1 are midway between typical NLS1 and Seyfert 1 which is reinforced by
low Eddington ratios, large black hole masses, as well as hard X-ray photon indices.
In agreement with previous works, we also found that the NLS1 classification should
not be based solely on the FWHM of Hβ, but should probably include another physical parameter, such as the Eddington ratio as Netzer & Trakhtenbrot (2007) suggested.
CONCLUSIONS AND FUTURE WORK
132
The reason why there are NLS1 without soft X-ray excess or without Fe II optical
emission has not yet been established and an analysis of a large sample of non-standard
NLS1 galaxies are needed to account for it. Our hypothesis that these non-standard
NLS1 are an extension of Seyfert 1 that lie at the high-end tail of the Eddington ratio distribution could be tested by further observational studies based on larger, and,
most importantly, complete sample of Fe II-deficient NLS1, on the one hand, and NLS1
without any sign of soft X-ray excess, on the other. Moreover our broadband SED
analysis could be complemented with theoretical models of AGN and other NLS1 templates. Another extension of this work would be a spectral variability analysis -which
is believed to be directly linked to the soft X-ray emission- of those sources without
soft X-ray excess, which would provide more detailed evidence of their nature.
IR power-law SED as an indicator of obscuration
Obscured AGN growth is a key phase in super-massive black hole-growth and galaxy
co-evolution models. In what concerns the open issues about the type 2 AGN population, the exact fraction of obscured AGN is not yet fully known, although it is well
known that obscured AGN likely represent a large fraction of the total AGN population
at all luminosities. In the last few years, deep X-rays surveys provide a reliable means
of selecting AGN, in that they introduce few false positives, as the AGN light generally outshines light from even highly active starforming galaxies at X-ray wavelengths.
However, hard X-ray surveys fail to identify the most heavily absorbed AGN. According
to the standard unified models, a mid-infrared selection technique should offer much
potential for discovery of obscured AGN, since any primary AGN continuum (i.e., disc
emission) that is absorbed must ultimately come out at these wavelengths after being
reprocessed by the torus. While some previous works ensure that mid-infrared colorcolor diagrams are reliable in finding heavily obscured AGN (e.g., Donley et al., 2012,
Stern et al., 2005), others claim that only X-ray data have been able to provide the
smoking-gun of the truly Compton-thin/Compton-thick nature (e.g., Polletta et al.,
2006).
We compare mid-infrared and hard X-ray identifications (in ultra deep X-ray observations) of AGN to study whether the selection of galaxies emitting in the infrared
as a power law (i.e. the revised-IRAC selection criterion) is potentially less biased
against obscuration. The study presented in Chapter 5 shows that absorbed AGN are
significantly more likely to be detected (at about 3 sigma level) among sources with
mid-infrared colours that meet the revised IRAC selection criterion, than among those
133
do not meet this criterion. This likelihood is defined in terms of the fractions of absorbed sources among the sources that meet those criterion and among those that do
not meet them. This contrast is essentially independent of the AGN luminosity (the
main difference occurs in the lowest X-ray luminosities where there are only 6 sources).
In agreement with previous works, we can not conclude that this selection scheme
based on mid-IR colors at relatively low X-ray luminosities (LX < 1045 erg s−1 ) is much
more efficient than a selection in the X-rays: the overall fraction of absorbed sources is
compatible with that found by the revised-IRAC selection criterion; indeed, the midinfrared selection loses about 50% of the X-ray-detected absorbed AGN.
We conclude that separating the most heavily obscured AGN (Compton-thick nature) from the remaining source populations adopting the revised-IRAC selection criterion is not a trivial job. As the census of such objects is difficult to complete, especially
at high redshifts, a multi-wavelength synergistic approach is needed, requiring deep
X-ray exposure, mid-IR data and, possibly, optical spectroscopy.
The results of this work can be extended in the future through a comparison with
other infrared-selected galaxy samples (for example using WISE) in order to check the
efficiency of the low X-ray-to-infrared luminosity selection method in finding heavily
obscured AGN. Moreover, to generalize our findings we would need to characterize the
X-ray-undetected infrared-selected AGN population, searching for evidence of intrinsic
absorption in their SEDs.
CONCLUSIONS AND FUTURE WORK
134
Glossary
Absorption
Decrease in the intensity of radiation, representing energy converted into excitation or ionization of electrons in the region through which the radiation travels.
As contrasted with monochromatic scattering (in which reemission occurs in all
directions at the same frequency) the process of emission refers to radiation that
is reemitted in general in any direction and at any frequency.
Auger Effect
A radiationless quantum jump that occurs in the X-ray region. When a Kelectron is removed from an atom and an L-electron drops into the vacancy in
the K-shell, the energy released in the latter transition is invested in ejecting an
L-shell electron.
Back-scattering
Scattering of radiation (or particles) through angles greater than 90o with respect
to the original direction of motion.
Big Blue Bump
The flux densities of most Seyfert 1 AGN continua have an average slope of -1
(i.e. fν ∼ ν −1 ) extending from the visual to a far-IR turnover (around 80 µm).
Relative to this red power law, there is generally an excess of flux in the blue and
ultraviolet, which constitutes the so-called Big Blue Bump.
Compton effect
Elastic collision between an electron and a photon. Energy is transferred from
the more energetic of the two particles to the other one.
Compton scattering
The scattering of photons by free electrons in an ionized medium.
Glossary
136
Extinction
The combined effects of absorption and scattering of light, usually by interestellar
dust, which dims our view of a distant object..
FWHM
The full width of a spectral line profile between the two points where the value
is 50% of the peak value.
Hubble constant
According to Hubble’s law, discovered by Edwin Hubble in 1929, distant galaxies
are receding from us, on average, with a speed equal to the product of the Hubble
constant and the distance to the galaxy. Hubble’s “constant” is independent of
distance, but actually decreases slowly in time as the expansion is slowed by the
gravitational pull of each galaxy on all the others.
PSF
The size and shape of the actual image of a point source as a result of the combined
effects of atmosphere, optics, guiding.
Reddening
The process by which light from an astronomical object grows red as the light
travels through interestelar dust. Dust scatters blue light more than red, thus
leaving predominantly red light transmitted.
References
Abramowicz, M. A., Czerny, B., Lasota, J. P., & Szuszkiewicz, E. 1988, ApJ, 332, 646
Ai, Y. L., Yuan, W., Zhou, H. Y., Wang, T. G., & Zhang, S. H. 2011, ApJ, 727, 31
Alexander, D. M., Bauer, F. E., Brandt, W. N., et al. 2011, ApJ, 738, 44
Alonso-Herrero, A., Pérez-González, P. G., Alexander, D. M., et al. 2006, ApJ, 640,
167
Alonso-Herrero, A., Ward, M. J., & Kotilainen, J. K. 1996, MNRAS, 278, 902
Anders, E. & Grevesse, N. 1989, GeCoA, 53, 197
Antonucci, R. 1993, ARA&A, 31, 473
Antonucci, R. R. J. & Miller, J. S. 1985, ApJ, 297, 621
Arnaud, K. A. 1996, in Astronomical Society of the Pacific Conference Series, Vol. 101,
Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby & J. Barnes,
17
Baldwin, J. A. 1997, in Astronomical Society of the Pacific Conference Series, Vol. 113,
IAU Colloq. 159: Emission Lines in Active Galaxies: New Methods and Techniques,
ed. B. M. Peterson, F.-Z. Cheng, & A. S. Wilson, 80
Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5
Ballo, L., Heras, F. J. H., Barcons, X., & Carrera, F. J. 2012, A&A, 545, A66
Barkhouse, W. A. & Hall, P. B. 2001, AJ, 121, 2843
Barmby, P., Alonso-Herrero, A., Donley, J. L., et al. 2006, ApJ, 642, 126
Barth, A. J. 2002, in Astronomical Society of the Pacific Conference Series, Vol. 258,
Issues in Unification of Active Galactic Nuclei, ed. R. Maiolino, A. Marconi, & N. Nagar, 147
REFERENCES
138
Bassani, L., Dadina, M., Maiolino, R., et al. 1999, ApJS, 121, 473
Beckmann, V. & Shrader, C. R. 2012, Active Galactic Nuclei
Bergeron, J. & Kunth, D. 1984, MNRAS, 207, 263
Bianchi, S., Corral, A., Panessa, F., et al. 2008, MNRAS, 385, 195
Bianchi, S., Maiolino, R., & Risaliti, G. 2012a, Advances in Astronomy, 2012
Bianchi, S., Panessa, F., Barcons, X., et al. 2012b, MNRAS, 426, 3225
Bohlin, R. C., Savage, B. D., & Drake, J. F. 1978, ApJ, 224, 132
Boller, T., Brandt, W. N., & Fink, H. 1996, A&A, 305, 53
Bongiorno, A., Zamorani, G., Gavignaud, I., et al. 2007, A&A, 472, 443
Boroson, T. A. & Green, R. F. 1992, in American Institute of Physics Conference Series,
Vol. 254, American Institute of Physics Conference Series, ed. S. S. Holt, S. G. Neff,
& C. M. Urry, 584–587
Brandt, W. N., Mathur, S., & Elvis, M. 1997, MNRAS, 285, L25
Brightman, M. & Ueda, Y. 2012, MNRAS, 423, 702
Caccianiga, A., Severgnini, P., Della Ceca, R., et al. 2007, A&A, 470, 557
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
Carrera, F. J., Page, M. J., & Mittaz, J. P. D. 2004, A&A, 420, 163
Castelló-Mor, N., Barcons, X., Ballo, L., et al. 2012, A&A, 544, A48
Chevallier, L., Collin, S., Dumont, A.-M., et al. 2006, A&A, 449, 493
Civano, F., Mignoli, M., Comastri, A., et al. 2007, A&A, 476, 1223
Comastri, A. & et al. 2003, Memorie della Societa Astronomica Italiana Supplementi,
3, 179
Comastri, A., Mignoli, M., Ciliegi, P., et al. 2002, ApJ, 571, 771
Comastri, A., Ranalli, P., Iwasawa, K., et al. 2011, A&A, 526, L9
Comastri, A. & XMM-CDFS Team. 2013, ArXiv e-prints
Corral, A., Della Ceca, R., Caccianiga, A., et al. 2011, A&A, 530, A42
REFERENCES
139
Crummy, J., Fabian, A. C., Gallo, L., & et al. 2006, MNRAS, 365, 1067
Cutri, R. M., Nelson, B. O., Kirkpatrick, J. D., Huchra, J. P., & Smith, P. S. 2001, in
Bulletin of the American Astronomical Society, Vol. 33, AAS, 829
Czerny, B., Nikolajuk, M., Różańska, A., et al. 2003, A&A, 412, 317
Daddi, E., Alexander, D. M., Dickinson, M., et al. 2007, ApJ, 670, 173
Della Ceca, R., Maccacaro, T., Caccianiga, A., et al. 2004, A&A, 428, 383
den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001, A&A, 365, L7
Dewangan, G. C., Griffiths, R. E., Dasgupta, S., & Rao, A. R. 2007, ApJ, 671, 1284
Dewangan, G. C., Mathur, S., Griffiths, R. E., & Rao, A. R. 2008, ApJ, 689, 762
Dickey, J. M. & Lockman, F. J. 1990, ARA&A, 28, 215
Dietrich, M., Kollatschny, W., Alloin, D., et al. 1994, A&A, 284, 33
Done, C., Davis, S. W., Jin, C., Blaes, O., & Ward, M. 2012, MNRAS, 420, 1848
Done, C. & Kubota, A. 2006, MNRAS, 371, 1216
Done, C. & Nayakshin, S. 2007, MNRAS, 377, L59
Donley, J. L., Koekemoer, A. M., Brusa, M., & et al. 2012, ApJ, 748, 142
Donley, J. L., Rieke, G. H., Pérez-González, P. G., & Barro, G. 2008, ApJ, 687, 111
Donley, J. L., Rieke, G. H., Pérez-González, P. G., Rigby, J. R., & Alonso-Herrero, A.
2007, ApJ, 660, 167
Efstathiou, A. & Rowan-Robinson, M. 1995, MNRAS, 273, 649
Ehle, M., Loiseau, N., & Bailey, R. 2005, in In IAU Commission 55: Communicating
Astronomy with the Public 2005, ed. I. Robson & L. L. Christensen, 318–320
Fabian, A. C. 1999, MNRAS, 308, L39
Fabian, A. C., Ballantyne, D. R., Merloni, A., et al. 2002, MNRAS, 331, L35
Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10
Filippenko, A. V. & Terlevich, R. 1992, ApJ, 397, L79
Fiore, F., Brusa, M., Cocchia, F., et al. 2003, A&A, 409, 79
REFERENCES
140
Fiore, F., Grazian, A., Santini, P., et al. 2008, ApJ, 672, 94
Fiore, F., Puccetti, S., Brusa, M., et al. 2009, ApJ, 693, 447
Foschini, L., Braito, V., Palumbo, G. G. C., et al. 2004, A&A, 428, 51
Franceschini, A., Braito, V., Persic, M., et al. 2003, MNRAS, 343, 1181
Franceschini, A., Manners, J., Polletta, M. d. C., et al. 2005, AJ, 129, 2074
Fritz, J., Franceschini, A., & Hatziminaoglou, E. 2006, MNRAS, 366, 767
Gallo, L. C., Lehmann, I., Pietsch, W., et al. 2006, MNRAS, 365, 688
Gaskell, C. M. & Ferland, G. J. 1984, PASP, 96, 393
Georgantopoulos, I., Comastri, A., Vignali, C., et al. 2013, A&A, 555, A43
Georgantopoulos, I., Georgakakis, A., Rowan-Robinson, M., & Rovilos, E. 2008, A&A,
484, 671
George, I. M. & Fabian, A. C. 1991, MNRAS, 249, 352
George, I. M., Turner, T. J., Yaqoob, T., et al. 2000, ApJ, 531, 52
Gierliński, M. & Done, C. 2004, MNRAS, 349, L7
Gierliński, M. & Done, C. 2006, MNRAS, 371, L16
Gierliński, M., Zdziarski, A. A., Poutanen, J., et al. 1999, MNRAS, 309, 496
Gilli, R., Comastri, A., & Hasinger, G. 2007, A&A, 463, 79
González-Martı́n, O., Masegosa, J., Márquez, I., Guerrero, M. A., & Dultzin-Hacyan,
D. 2006, A&A, 460, 45
Goodrich, R. W. 1989, ApJ, 342, 224
Granato, G. L., Danese, L., & Franceschini, A. 1997, ApJ, 486, 147
Granato, G. L., De Zotti, G., Silva, L., Bressan, A., & Danese, L. 2004, ApJ, 600, 580
Grupe, D., Komossa, S., Leighly, K. M., & Page, K. L. 2010, ApJS, 187, 64
Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006, AJ, 131, 2332
Haardt, F. & Maraschi, L. 1991, ApJ, 380, L51
Haas, M., Klaas, U., Müller, S. A. H., et al. 2003, A&A, 402, 87
REFERENCES
141
Hao, L., Strauss, M. A., Tremonti, C. A., et al. 2005, AJ, 129, 1783
Heckman, T. M. 1980, A&A, 87, 152
Hoenig, S. F. 2013, ArXiv e-prints
Houck, J. R., Roellig, T. L., Van Cleve, J., et al. 2004, in Society of Photo-Optical
Instrumentation Engineers (SPIE) Conference Series, Vol. 5487, Optical, Infrared,
and Millimeter Space Telescopes, ed. J. C. Mather, 62–76
Hu, C., Wang, J.-M., Ho, L. C., et al. 2008, ApJ, 683, L115
Iwasawa, K., Gilli, R., Vignali, C., et al. 2012, A&A, 546, A84
Iwasawa, K., Koyama, K., Awaki, H., et al. 1993, ApJ, 409, 155
Jin, C., Ward, M., Done, C., & Gelbord, J. 2012, MNRAS, 420, 1825
Kaspi, S., Smith, P. S., Netzer, H., et al. 2000, ApJ, 533, 631
Kauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003, MNRAS, 346, 1055
Kendziorra, E., Bihler, E., Grubmiller, W., et al. 1997, in Society of Photo-Optical
Instrumentation Engineers (SPIE) Conference Series, Vol. 3114, EUV, X-Ray, and
Gamma-Ray Instrumentation for Astronomy VIII, ed. O. H. Siegmund & M. A.
Gummin, 155–165
Kennicutt, Jr., R. C. 1998, ARA&A, 36, 189
Kewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, C. A., & Trevena, J. 2001,
ApJ, 556, 121
Khachikian, E. Y. & Weedman, D. W. 1974, ApJ, 192, 581
King, A. 2005, ApJ, 635, L121
Koratkar, A. & Blaes, O. 1999, PASP, 111, 1
Kotilainen, J. K., Ward, M. J., Boisson, C., Depoy, D. L., & Smith, M. G. 1992,
MNRAS, 256, 149
Kuster, M., Benlloch, S., Kendziorra, E., & Briel, U. G. 1999, in Society of PhotoOptical Instrumentation Engineers (SPIE) Conference Series, Vol. 3765, EUV, XRay, and Gamma-Ray Instrumentation for Astronomy X, ed. O. H. Siegmund &
K. A. Flanagan, 673–682
Lacy, M., Petric, A. O., Sajina, A., et al. 2007, AJ, 133, 186
REFERENCES
142
Lacy, M., Storrie-Lombardi, L. J., Sajina, A., et al. 2004, ApJS, 154, 166
Laor, A. 2000, NewAR, 44, 503
Laurent, O., Mirabel, I. F., Charmandaris, V., et al. 2000, in ESA Special Publication,
Vol. 456, ISO Beyond the Peaks: The 2nd ISO Workshop on Analytical Spectroscopy,
ed. A. Salama, M. F. Kessler, K. Leech, & B. Schulz, 249
Lehmer, B. D., Brandt, W. N., Alexander, D. M., et al. 2005, ApJS, 161, 21
Leighly, K. M. 1999, ApJS, 125, 317
Lira, P., Ward, M., Zezas, A., Alonso-Herrero, A., & Ueno, S. 2002, MNRAS, 330, 259
Lusso, E., Comastri, A., Vignali, C., et al. 2010, A&A, 512, A34
Maccacaro, T., Gioia, I. M., Wolter, A., Zamorani, G., & Stocke, J. T. 1988, ApJ, 326,
680
Magorrian, J., Tremaine, S., Richstone, D., et al. 1998, AJ, 115, 2285
Maiolino, R., Marconi, A., & Oliva, E. 2001, A&A, 365, 37
Maiolino, R., Salvati, M., Bassani, L., et al. 1998, A&A, 338, 781
Maitra, D. & Miller, J. M. 2010, ApJ, 718, 551
Martı́nez-Sansigre, A., Rawlings, S., Bonfield, D. G., et al. 2007, MNRAS, 379, L6
Martı́nez-Sansigre, A., Rawlings, S., Lacy, M., et al. 2006, MNRAS, 370, 1479
Mason, K. O., Breeveld, A., Much, R., et al. 2001, A&A, 365, L36
Mateos, S., Alonso-Herrero, A., Carrera, F. J., et al. 2012, MNRAS, 426, 3271
Mateos, S., Barcons, X., Carrera, F. J., et al. 2005, A&A, 444, 79
Mateos, S., Carrera, F. J., Page, M. J., et al. 2010, A&A, 510, A35
Mathur, S. 2000, MNRAS, 314, L17
Matt, G. 2002, in Astronomical Society of the Pacific Conference Series, Vol. 258, Issues
in Unification of Active Galactic Nuclei, ed. R. Maiolino, A. Marconi, & N. Nagar,
3
Matt, G., Fabian, A. C., Guainazzi, M., et al. 2000, MNRAS, 318, 173
REFERENCES
143
Mei, L., Yuan, W.-M., & Dong, X.-B. 2009, Research in Astronomy and Astrophysics,
9, 269
Meisenheimer, K., Haas, M., Müller, S. A. H., et al. 2001, A&A, 372, 719
Mewe, R., Gronenschild, E. H. B. M., & van den Oord, G. H. J. 1985, A&AS, 62, 197
Mewe, R., Lemen, J. R., & van den Oord, G. H. J. 1986, A&AS, 65, 511
Middleton, M. & Done, C. 2010, MNRAS, 403, 9
Miller, L., Turner, T. J., & Reeves, J. N. 2008, A&A, 483, 437
Mineshige, S., Kawaguchi, T., Takeuchi, M., & Hayashida, K. 2000, PASJ, 52, 499
Nandra, K. & Pounds, K. A. 1994, MNRAS, 268, 405
Nenkova, M., Sirocky, M. M., Nikutta, R., Ivezić, Ž., & Elitzur, M. 2008, ApJ, 685,
160
Netzer, H. & Trakhtenbrot, B. 2007, ApJ, 654, 754
Osterbrock, D. E. 1981, ApJ, 249, 462
Osterbrock, D. E. 1989, Astrophysics of gaseous nebulae and active galactic nuclei
Osterbrock, D. E. & Ferland, G. J. 2006, Astrophysics of gaseous nebulae and active
galactic nuclei
Osterbrock, D. E. & Pogge, R. W. 1985, ApJ, 297, 166
Osterbrock, D. E. & Pogge, R. W. 1987, ApJ, 323, 108
Page, M. J., Loaring, N. S., Dwelly, T., et al. 2006, MNRAS, 369, 156
Page, M. J., McHardy, I. M., Gunn, K. F., et al. 2003, AN, 324, 101
Park, S. Q., Barmby, P., Willner, S. P., et al. 2010, ApJ, 717, 1181
Pérez-González, P. G., Rieke, G. H., Villar, V., et al. 2008, ApJ, 675, 234
Perola, G. C., Puccetti, S., Fiore, F., et al. 2004, A&A, 421, 491
Peterson, B. M. 1997, An Introduction to Active Galactic Nuclei
Piconcelli, E., Jimenez-Bailón, E., Guainazzi, M., et al. 2005, A&A, 432, 15
Pineau, F.-X., Motch, C., Carrera, F., et al. 2011, A&A, 527, A126
REFERENCES
144
Polletta, M. d. C., Wilkes, B. J., Siana, B., et al. 2006, ApJ, 642, 673
Porquet, D., Reeves, J. N., O’Brien, P., & Brinkmann, W. 2004, A&A, 422, 85
Pounds, K. A., Done, C., & Osborne, J. P. 1995, MNRAS, 277, L5
Ramos Almeida, C., Levenson, N. A., Alonso-Herrero, A., et al. 2011, ApJ, 731, 92
Ranalli, P., Comastri, A., & Setti, G. 2003, A&A, 399, 39
Ranalli, P., Comastri, A., Vignali, C., et al. 2013, A&A, 555, A42
Richards, G. T., Fan, X., Newberg, H. J., et al. 2002, AJ, 123, 2945
Rieke, G. H., Young, E. T., Cadien, J., et al. 2004, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 5487, Optical, Infrared, and
Millimeter Space Telescopes, ed. J. C. Mather, 50–61
Rigby, J. R., Rieke, G. H., Donley, J. L., Alonso-Herrero, A., & Pérez-González, P. G.
2006, ApJ, 645, 115
Risaliti, G., Maiolino, R., & Salvati, M. 1999, ApJ, 522, 157
Robson, I. 1996, Active galactic nuclei
Rodrı́guez-Ardila, A., Pastoriza, M. G., & Donzelli, C. J. 2000, ApJS, 126, 63
Ross, R. R., Fabian, A. C., & Ballantyne, D. R. 2002, MNRAS, 336, 315
Ross, R. R., Fabian, A. C., & Mineshige, S. 1992, MNRAS, 258, 189
Sarajedini, V. L., Gilliland, R. L., & Kasm, C. 2003, ApJ, 599, 173
Schartmann, M., Meisenheimer, K., Camenzind, M., et al. 2008, A&A, 482, 67
Schurch, N. J., Roberts, T. P., & Warwick, R. S. 2002, MNRAS, 335, 241
Scott, A. E., Stewart, G. C., & Mateos, S. 2012, MNRAS, 423, 2633
Severgnini, P., Caccianiga, A., Braito, V., et al. 2003, A&A, 406, 483
Severgnini, P., Caccianiga, A., & Della Ceca, R. 2012, A&A, 542, A46
Seyfert, C. K. 1943, ApJ, 97, 28
Shakura, N. I. & Sunyaev, R. A. 1973, A&A, 24, 337
Shuder, J. M. 1982, ApJ, 259, 48
REFERENCES
145
Simpson, C. 2005, MNRAS, 360, 565
Smith, D. A. & Done, C. 1996, MNRAS, 280, 355
Spergel, D. N., Bean, R., Doré, O., et al. 2007, ApJS, 170, 377
Spergel, D. N., Verde, L., Peiris, H. V., et al. 2003, ApJS, 148, 175
Stasińska, G., Cid Fernandes, R., Mateus, A., Sodré, L., & Asari, N. V. 2006, MNRAS,
371, 972
Stern, D., Assef, R. J., Benford, D. J., et al. 2012, ApJ, 753, 30
Stern, D., Eisenhardt, P., Gorjian, V., et al. 2005, ApJ, 631, 163
Stevens, J. A., Page, M. J., Ivison, R. J., et al. 2005, MNRAS, 360, 610
Strüder, L., Briel, U., Dennerl, K., et al. 2001, A&A, 365, L18
Tanaka, Y., Boller, T., & Gallo, L. 2005, in Growing Black Holes: Accretion in a
Cosmological Context, ed. A. Merloni, S. Nayakshin, & R. A. Sunyaev, 290–295
Tozzi, P., Gilli, R., Mainieri, V., & et al. 2006, A&A, 451, 457
Tran, H. D. 2001, ApJ, 554, L19
Tran, H. D. 2003, ApJ, 583, 632
Treister, E. & Urry, C. M. 2005, ApJ, 630, 115
Trouille, L. & Barger, A. J. 2010, ApJ, 722, 212
Trouille, L., Barger, A. J., Cowie, L. L., Yang, Y., & Mushotzky, R. F. 2009, ApJ, 703,
2160
Trump, J. R., Impey, C. D., Taniguchi, Y., et al. 2009, ApJ, 706, 797
Turner, M. J. L., Abbey, A., Arnaud, M., et al. 2001, A&A, 365, L27
Turner, T. J., George, I. M., & Netzer, H. 1999, ApJ, 526, 52
Ueda, Y., Akiyama, M., Ohta, K., & Miyaji, T. 2003, ApJ, 598, 886
Urry, C. M. & Padovani, P. 1995, PASP, 107, 803
Vasudevan, R. V. & Fabian, A. C. 2007, MNRAS, 381, 1235
Vasudevan, R. V. & Fabian, A. C. 2009, MNRAS, 392, 1124
REFERENCES
146
Veilleux, S. & Osterbrock, D. E. 1987, ApJS, 63, 295
Verner, E., Bruhweiler, F., Johansson, S., & Peterson, B. 2009, Physica Scripta, 014006
Véron-Cetty, M.-P., Joly, M., & Véron, P. 2004, A&A, 417, 515
Véron-Cetty, M.-P. & Véron, P. 2006, A&A, 455, 773
Véron-Cetty, M.-P., Véron, P., & Gonçalves, A. C. 2001, A&A, 372, 730
Wall, J. V. & Jenkins, C. R. 2008, Practical Statistics for Astronomers, p. 21
Ward, M. J., Blanco, P. R., Wilson, A. S., & Nishida, M. 1991, ApJ, 382, 115
Watson, M. G., Schröder, A. C., Fyfe, D., et al. 2009, A&A, 493, 339
Werner, M. W., Roellig, T. L., Low, F. J., & et al. 2004, ApJS, 154, 1
White, R. L., Becker, R. H., Gregg, M. D., et al. 2000, ApJS, 126, 133
Wilkes, B. J., Schmidt, G. D., Cutri, R. M., et al. 2002, ApJ, 564, L65
Woo, J.-H. & Urry, C. M. 2002, ApJ, 579, 530
Xu, D. & Komossa, S. 2011, in Narrow-Line Seyfert 1 Galaxies and their Place in the
Universe
Xue, Y. Q., Luo, B., Brandt, W. N., et al. 2011, ApJS, 195, 10
Yan, R., Ho, L. C., Newman, J. A., et al. 2011, ApJ, 728, 38
Yaqoob, T. 1997, ApJ, 479, 184
York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579
Zdziarski, A. A., Johnson, W. N., & Magdziarz, P. 1996, MNRAS, 283, 193
Zdziarski, A. A., Poutanen, J., & Johnson, W. N. 2000, ApJ, 542, 703
Zhou, H., Wang, T., Yuan, W., et al. 2006, ApJS, 166, 128
Zhu, L., Zhang, S. N., & Tang, S. 2009, ApJ, 700, 1173
Życki, P. T., Done, C., & Smith, D. A. 1999, MNRAS, 309, 561

Download Report

Muestras de galaxias activas, del infrarrojo a los Rayos X

Paperzz.com

Your Paperzz