MultiWavelength Surveys of Obscured AGN

1

Multi-Wavelength Surveys of Obscured AGN Eleni T.
Chatzichristou Institute of Astronomy
Astrophysics, National Observatory of Athens,
15236 Penteli, Greece
Abstract Several key goals require
measuring the number of all AGN in the Universe,
and the evolution of the ratio of obscured to
un-obscured AGN with redshift. This reflects the
structure of AGN and thus the development in the
heart of all galaxies. Hard X-rays can penetrate
most obscuring dust columns to reveal the AGN
that remains hidden in all other wavelengths.
Mid-IR surveys probe the thermal dust emission,
that is, the continuum light from the central
source after it is reprocessed by dust, and this
emission dominates the bolometric luminosities of
dusty high-redshift galaxies. Thus, combining
deep mid-IR and hard X-ray surveys can provide us
with accurate demographics of AGN especially at
high redshifts. Multi-wavelength surveys aim to
address these science goals by exploiting the
unprecedented combination of great observatories
such as HST, Chandra and Spitzer to survey the
distant universe to the faintest flux limits
across the broadest range of wavelengths. Here we
present and discuss some of the results coming
from multi-wavelength surveys placing particular
focus on the systematic study of obscured
AGN. Introduction AGN are traditionally
classified by optical spectroscopy and UV
excess/color selection techniques. These usually
miss the highly obscured or weak AGN, being
biased towards the highly luminous, unobscured
AGN (e.g. Moran et al. 2002, ApJ 579, 71
Maiolino et al. 2003, MNRAS 344, L59). The
obscured AGN can only be detected in hard X-rays
(gt2 keV) which can penetrate most obscuring
columns to reveal the AGN, and in the mid- and
far-IR wavelengths, where the majority of their
bolometric luminosity is observed. Many deep
multi-wavelength surveys such as GOODS
(Giavalisco et al. 2004, ApJ 600, L93) are
motivated by such a quest for a previously
unknown population of obscured AGN. They
furthermore give a unique opportunity to test
ideas about obscuration of AGN What are the
mid-IR properties of X-ray selected AGN and
vice-versa? What is the interplay between IR and
X-ray backgrounds?
Lack of correspondence between IR and
X-rays? Eventhough mid-IR colour criteria
identify most of the X-ray selected AGN when both
the X-ray and IR wavelengths are dominated by the
central source, they are less successful when the
host galaxy dominates the IR emission (e.g.
Barmby et al (2006, ApJ 642, 126). The inverse is
also true among the CDFS AGN selected by IR
power-law criteria only 50 are detected in at
least one of the Chandra bands, and these
represent only 30 of the hard X-ray sources with
mid-IR counterparts (Alonso-Herrero et al. 2004,
ApJS 154, 155 and 2006). Furthermore, it becomes
more and more evident that the mid-IR properties
do not directly correspond to X-ray hardness, and
inversely, X-ray selected AGN have a variety of
optical/IR spectral types (e.g. Franceschini et
al. 2005, AJ 129, 2074). What is the reason for
this lack of correspondence between IR and
X-rays? Variations of the gas/dust ratio? A
broad range of intrinsic AGN properties?
X-Ray Surveys They have identified a population
of high-z, heavily obscured luminous AGN (Stern
et al. 2002, ApJ 568, 71 Norman et al. 2002, ApJ
571, 218), as well as a population of apparently
normal galaxies with optical spectra dominated by
starlight but hosting significant X-ray emission
implying the presence of an AGN (Barger et al.
2001, AJ 121, 662 Hornschmeier et al. 2001, ApJ
554, 742 Stern et al. 2002, AJ 123, 2223). For
instance, 43 of X-ray selected AGN in the CDFS
are not classified as active by optical
spectroscopy (Szokoly et al. 2004, ApJS 155,
271). X-ray AGN selection has the advantage of
being reasonably efficient and reliable, but may
miss some obscured sources (Peterson et al. 2006,
AJ 131, 133), while IR and radio observations can
identify the AGN missed in X-rays (Donley et al.
2005 ApJ 634, 169 Alonso-Herrero et al. 2006,
ApJ 640, 167) and can help distinguish between
different activity types for X-ray selected AGN.
Mid-IR Surveys Mid-IR selected AGN show higher
surface density relative to optically-selected
AGN of comparable mid-IR flux (Stern et al. 2005,
ApJ 631, 163 Gorjian et al. 2006, in
preparation). Furthermore, among the most IR
luminous galaxies at both intermediate and high
redshifts, the AGN fraction can be very high, at
least 40 (Alexander et al., 2005, Nature 434,
738 Yan et al. 2005, ApJ 628, 604). Deep mid-IR
imaging studies with Spitzer, as well as
radio-selection criteria have begun to detect a
population of type 2 QSOs which seem to be at
least comparable in number density to the
un-obscured type 1 population (Martinez-Sansigre
et al. 2005, Nature 436, 666 Donley et al. 2005,
ApJ 634, 169). Some of these show no hard X-ray
counterpart, indicating that they are hidden by
Compton-thick material even at high X-ray
luminosities. Treister et al (2006, ApJ 640, 603)
furthermore show that the contribution of AGN
missed in X-rays to the IR background can add 45
more flux, 30 coming from obscured AGN.
Figure 7 from Barmby et al. 2006, showing (a)
hard X-ray vs. 24 micron fluxes and (b) and (c)
full band X-ray vs. IRAC 8.0 and 3.6 micron
fluxes.
Trends with Luminosity Hard X-ray studies have
shown a decreasing fraction of type 2 AGN with
increasing X-ray luminosities. Could this
indicate that higher-luminosity AGN clean out
their environment? Or rather that the torus
opening angle is larger in more luminous objects?
If these explanations are true, one would expect
that the ratio of mid-IR/X-ray emission should be
lower in type 2 QSOs than in Seyfert 2s. Is this
what we indeed see? In fact, reality seems to be
more complex. Although some correlation between
mid-IR emission and QSO 2 luminosity can indeed
be seen (e.g. Lutz et al. 2004, AA 418,465), and
the above ratio is found to decrease with X-ray
luminosity, there seems to be no correlation with
hard X-ray or IR luminosity (Treister et al.
2006). Furthermore, it seems that in general the
mid-IR spectra of luminous QSO2s are
significantly different from those of lower
luminosity type 2 AGN, in fact they are more
similar to type 1 AGN, probably because the
contribution of QSO 2 host galaxies to the mid-IR
spectra is small (e.g. Sturm et al. (2006, ApJ
642, 81). If this is true, it could mean that the
QSO2 contribution to the peak IR background is
less important than predicted.
Figure 4 from Treister et al. 2006, showing hard
X-ray to mid-IR luminosity ratio as a function of
X-ray (left) and IR (right) luminosities.
Triangles are GOODS AGN and circles are local
AGN. Filled/open symbols denote
obscured/unobscured sources, respectively.
Figure 1 from Alonso-Herrero et al. 2006, showing
mid-IR vs. hard X-ray fluxes, for X-ray selected
sources in the CDFS. The lightly shaded area is
the extrapolation of values for local AGN
Figure 1 from Stern et al. 2005, showing the
success of Spitzer mid-IR colour criteria for
separating quasars and Seyfert 1s from galaxies.
Is there a good indicator of obscured
AGN? Obscured AGN locally seem to have higher
mid-IR/hard-X-ray flux ratios compared to
unobscured AGN (e.g. Fadda et al. 2002, AA 383,
838 Silva et al. 2004, MNRAS 355, 973
Alonso-Herrero et al. 2006). One would thus
expect to find a large population of X-ray hard,
IR-bright AGN. However, most recent studies
indicate that this flux ratio does not depend on
X-ray hardness or column density NH, and does not
differ between optically-classified broad-line
and narrow-line AGN (e.g. Lutz et al. 2004
Treister et al. 2006). There are even findings of
X-ray hard AGN with very low mid-IR/X-ray flux
ratios (Rigby et al. 2004, ApJS 154, 160). This
is not expected within the unification paradigm.
Could it be interpreted as indicating an
intrinsic difference between high-z and local
AGN, in their dust properties or absorbing
geometries? Or, that very hard sources are also
very faint IR (and maybe also X-ray) emitters?
Alternatively, one could simply suggest that,
while there must be certainly some connection
between X-ray hardness ratio and mid-IR spectral
indices, the latter is not a particularly good
indicator of the AGN type, but rather indicating
the degree to which the mid-IR is dominated by
the AGN or the host galaxy emission.
The question that is posed is then If indeed
hard X-ray and mid-IR selection techniques find
few Compton-thick AGN, what is a better way to
search? Improving statistics with larger, more
uniform samples and cross identification with
other wavelengths is certainly needed.
Furthermore it is essential to reach a better
understanding of their nature in order to be able
to uncover the elusive population of obscured
AGN. All of these issues are and will continue to
be in the focus of the Spitzer and Herschel Space
Observatories.