Some galaxies contain compact nuclei, which emit vast amounts of energy over the entire
electromagnetic spectrum from radio through gamma-rays, thus called active galactic nuclei
(AGN). The powerful activity originates from the release of gravitational energy through accretion
of material on to a supermassive black hole with a typical mass in the range of 106−9M⊙.
Some of the gravitational energy is thermalized in the accretion disk to radiate ultraviolet and
soft X-ray emission. Some is channeled into the generation of energetic particles (or jets) and
produces X-ray through gamma-ray emission via inverse Compton scattering and radio through
X-rays via synchrotron emission. This radiation from the vicinity of the black hole may heat
circumnuclear dust in the obscuring torus of clouds, which is believed to exist on parsec scales
outside the accretion disk. The warm torus can then be a source of a strong infrared emission.
AGN tend to show large variety in their observational properties, such as luminosities, optical
line properties, radio brightness, variability, and so on. In spite of such apparent diversity, it
is widely believed that the various AGN are intrinsically the same. Just a few parameters -
inclination, accretion rate, and presence/absence of jets - may be responsible for the apparent
differences. However, details of the geometry of the nuclei, especially of the torus, are poorly
known because their angular scales are very small and difficult to resolve spatially.
The main goal of this thesis is to study the geometry of the nuclear emission region based
upon unbiased survey data gathered from multiple wavebands. For this purpose, I take two
approaches; one is to explore the correlation between hard X-ray and infrared luminosities using
unbiased samples systematically for each AGN type. The other is to analyze variability of midinfrared
emission for individual sources and for each AGN type statistically using two all-sky
survey catalogs. This is the first large-scale systematic study of mid-infrared variability in AGN,
probing timescales of several years separately for different types of AGN.
For the first approach, I used all-sky surveys conducted by Swift in the hard X-ray (> 10 keV)
band and by AKARI in the infrared band. The Swift/Burst Alert Telescope all-sky survey
provides an unbiased, flux-limited selection of hard X-ray detected AGN. The hard X-ray band
is rather insensitive to the photo-electric absorption due to the intervening clouds up to mildly
Compton-thick (NH ∼1024 cm−2) column densities. In other words, the hard X-ray flux obtained
by the survey reflects the intrinsic luminosity for all Compton-thin AGN and also for mildly
Compton-thick ones, thus providing samples largely unbiased by obscuration. In the case of
infrared observations, high angular resolution is crucial in order to properly separate AGN from
stellar emission in the host galaxy. However, this was not possible until the advent of recent
telescopes. The AKARI satellite completed an all-sky survey whose catalog was released just
before the beginning of this research. This survey is several times more sensitive than previous
ones, and was carried out at a much higher angular resolution of the order of arcseconds. Crosscorrelating
the 22-month hard X-ray survey with the AKARI all-sky survey, I studied 158 AGN
detected by both instruments. I find a strong correlation for most AGN between the infrared
(9, 18, and 90 μm) and hard X-ray (14–195 keV) luminosities, and quantified the correlation for
various complete subsamples of AGN. Partial correlation analysis confirms that the correlation
is intrinsic; that is, the correlation between the luminosities remains significant after removing
the contribution of redshift. Under the unification scheme of AGN, this result may be viewed
as supporting clumpy torus models. The good one-to-one correlation between mid-infrared
bolometric luminosities and hard X-ray ones for over four orders of magnitude indicates that
the covering factor of torus will decrease with the increase of the intrinsic luminosity. The
correlation for radio galaxies has a slope and normalization identical to that for Seyfert 1s,
where we have a direct view of the nuclear regions in both hard X-rays and infrared, implying
similar hard X-ray/infrared emission processes in both. In contrast, sources with large Comptonthick
column densities show a large deficit in the hard X-ray band, because high gas column
densities in the torus diminish their apparent luminosities even in the hard X-ray band.
On the other hand, a few radio-loud sources (radio galaxies and blazars) show systematic deviations
toward higher X-ray luminosities in the correlations as compared to radio-quiet sources.
Origin of the broadband emission of radio galaxies is a matter of considerable debate. One possible
explanation of this deviation (i.e., excess X-rays) is the contribution of jets to the hard
X-ray emission. Observations of flux variability can be useful for isolating the jet contribution,
because strong and rapid fluctuations are characteristic of beamed jet emission. Useful hard
X-ray variability data are not yet available, but such data have recently become available in
the mid-infrared band. Furthermore, mid-infrared variations can also potentially constrain the
geometry of the dusty torus by measuring the response of the torus to changes of the nucleus
emission. This leads to the second topic of my thesis to study mid-infrared variability of AGN
systematically. I combine two mid-infrared all-sky surveys, i.e., the data released by AKARI
and Wide-field Infrared Survey Explorer (WISE). WISE was launched about four years after
AKARI and accomplished all-sky surveys with high sensitivity in several mid-infrared bands
(particularly relevant for my work are the 12 and 22 μm bands). Because the bands observed
by the two telescopes are slightly different, I calculated the flux ratio of WISE and AKARI
after subtracting the contribution of band differences, for which the spectral slope calculated
for each source was used. In addition, cross-calibration errors of the two telescopes are carefully
examined. I find significant mid-infrared variations in 3 sources, 2 blazars and 1 radio galaxy, in
either or both of the 9 and 18 μm bands. Although no significant variations are detected from
the rest of the sources, low level variations may be hidden in statistical errors. Therefore, I tried
to constrain the average sample variability for different AGN types, which was actually detected
from Seyfert 1 in the 9 μm band. This is the first detection of variability from Seyfert 1 by using
two mid-infrared all-sky surveys. I quantified the amplitudes of sample variability and found
that the amplitude reaches ∼10% in 4 years for Seyfert 1 in the 9 μm band. If this variability is
explained by the torus emission only, dust distribution in the torus should be compact, although
other possibilities, such as jet contribution, cannot be excluded. Combining the results of the
hard X-ray vs. mid-infrared correlations and mid-infrared variability, geometry and structure
of the obscuring torus are discussed.