The Astrophysical Journal, 768:13 (9pp), 2013 May 1 doi:10.1088/0004-637X/768/1/13
C
2013. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
FAR-INFRARED PROPERTIES OF TYPE 1 QUASARS
D. J. Hanish
1
, H. I. Teplitz
1
, P. Capak
1
, V. Desai
1
, L. Armus
1
, C. Brinkworth
1
, T. Brooke
1
, J. Colbert
1
,
D. Fadda
1
, D. Frayer
2
, M. Huynh
3
, M. Lacy
4
, E. Murphy
5
, A. Noriega-Crespo
1
, R. Paladini
1
,
C. Scarlata
6
,S.Shenoy
7
, and the SAFIRES Team
1
Spitzer Science Center, California Institute of Technology, MC 220-6, 1200 E California Blvd.,
Pasadena, CA 91125, USA; hanish@ipac.caltech.edu
2
National Radio Astronomy Observatory, P.O. Box 2, Green Bank, WV 24944, USA
3
International Centre for Radio Astronomy Research, M468, University of Western Australia, Crawley, WA 6009, Australia
4
National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA
5
The Observatories of the Carnegie Institution for Science, Pasadena, CA 91101, USA
6
Minnesota Institute for Astrophysics, School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455, USA
7
Space Science Division, NASA Ames Research Center, M/S 245-6, Moffett Field, CA 94035, USA
Received 2012 October 5; accepted 2013 March 2; published 2013 April 9
ABSTRACT
We use the Spitzer Space Telescope Enhanced Imaging Products and the Spitzer Archival Far-InfraRed Extragalactic
Survey to study the spectral energy distributions (SEDs) of spectroscopically confirmed type 1 quasars selected
from the Sloan Digital Sky Survey (SDSS). By combining the Spitzer and SDSS data with the Two Micron All
Sky Survey, we are able to construct a statistically robust rest-frame 0.1–100 μm type 1 quasar template. We find
that the quasar population is well-described by a single power-law SED at wavelengths less than 20 μm, in good
agreement with previous work. However, at longer wavelengths, we find a significant excess in infrared luminosity
above an extrapolated power-law, along with significant object-to-object dispersion in the SED. The mean excess
reaches a maximum of 0.8 dex at rest-frame wavelengths near 100 μm.
Key words: galaxies: active – infrared: galaxies – quasars: general – surveys
Online-only material: color figure, machine-readable table
1. INTRODUCTION
Quasars, or quasi-stellar objects (QSOs), are thought to be
super-massive black holes (M>10
6
M
) actively accreting
material (e.g., Elvis et al. 1994). The energy output by these
objects may control the rate of star formation in galaxies
(Hopkins et al. 2006), so understanding and characterizing
these objects is key to galaxy evolution studies. These luminous
objects are characterized by broad emission lines in the rest-
frame optical and ultraviolet and are believed to represent the
peak accretion period for most black holes. During this period,
black-hole accretion is expected to dominate the bolometric
output of the galaxy (Hopkins et al. 2006), and so the s pectral
energy distribution (SED) template should represent the energy
output of an actively accreting black hole.
The SED of quasars is particularly important because it
traces how much energy is output by the black hole at each
wavelength, and is needed to separate the galaxy and quasar
light when determining key properties such as the star formation
rate or stellar mass of a galaxy (Fritz et al. 2006). Unfortunately,
spectral energy distributions in the mid- and far-infrared (FIR)
wavelengths are poorly understood. To date, infrared quasar
SEDs have been limited mainly by small samples and survey
biases.
In the absence of detailed spectrophotometry, SEDs can be
generated by averaging broadband photometric measurements,
making appropriate extrapolations when data are absent (e.g.,
Elvis et al. 1994; Bruzual & Charlot 2003; Richards et al.
2006; Shang et al. 2011;Wuetal.2011). In the mid-infrared
(3–100 μm), the most comprehensive study to date is the work
of Polletta et al. (2007), which used an X-ray selected sample
of 136 active galactic nuclei (AGNs) detected at 24 μmby
the Spitzer Multiband Imaging Photometer (MIPS; Rieke et al.
2004). However, this work adopted a power-law extrapolation
beyond 3 μm due to a lack of robust mid-infrared data. Later
work by Hatziminaoglou et al. (2008) attempted to extend these
SEDs beyond 20 μm using the Spitzer Wide-area InfraRed
Extragalactic (SWIRE; Lonsdale et al. 2003) survey, but reached
inconclusive results due to the small number of high signal-
to-noise detections at 70 and 160 μm. More recent works
(Hatziminaoglou et al. 2010; Bonfield et al. 2011; Dai et al.
2012; Sajina et al. 2012) have extended quasar SEDs out
to observed wavelengths of 500 μm, using small samples of
Herschel data.
In this paper, we will focus on the 20–100 μm SEDs of type 1
quasars. The goal of this work is to characterize this template
SED and the intrinsic scatter around it which may indicate
residual galaxy light and/or obscuration of the black hole.
To probe the 3–100 μm range, we use the Spitzer Enhanced
Imaging Products (SEIP; P. Capak et al., in preparation) and
the Spitzer Archival FIR Extragalactic Survey (SAFIRES; D. J.
Hanish et al., in preparation), which provide science quality
mosaics and photometry for most cryogenic Infrared Array
Camera (IRAC; Fazio et al. 2004) and MIPS data in the
Spitzer Heritage Archive. The Sloan Digital Sky Survey (SDSS;
Abazajian et al. 2009) Data Release 7 provides a robust SED
within the observed 0.3–1 μm range, and the Two Micron All
Sky Survey (2MASS; Skrutskie et al. 2006), is used to fill in the
1–3 μm wavelength range.
The paper is organized as follows: Section 2 describes the
data used by the SAFIRES template creation process, Section 3
explains the methods used to generate our SED template,
Section 4 presents the templates created through this process,
and Section 5 presents our conclusions. We assume a ΛCDM
cosmology (Ω
0
= 0.3, Ω
Λ
= 0.7) with a Hubble constant of
H
0
= 70 km s
−1
Mpc
−1
.
1
The Astrophysical Journal, 768:13 (9pp), 2013 May 1 Hanish et al.
2. DATA
2.1. Parent Sample
The process used to create an accurate quasar SED requires
photometric data across a wide range of wavelengths. We
combine 15 wavelength bands ranging from the visible bands
of the SDSS to the far-infrared MIPS bands of the Spitzer Space
Telescope. For inclusion in this work, sources were required to
meet five criteria.
1. Inclusion in the SDSS Quasar Catalog, Data Release 7.
The optical limits associated with this catalog restricts this
sample to type 1 quasars, as type 2 quasars lack the optical
luminosity to be identified by the SDSS. This is examined
in more detail in Reyes et al. (2008)
2. A location in one of two large, well-studied extragalactic
(|b| 20.
◦
0) fields with deep Spitzer coverage: the Spitzer
Deep Wide-Field Survey (SDWFS; Ashby et al. 2009)or
the Lockman Hole (Lockman et al. 1986).
3. A S/N 10 detection exceeding the minimum depth of
coverage in at least one of the seven Spitzer bands. Only
point sources were considered; sources showing extended
emission were removed from the sample.
4. A r edshift of z 3.0, provided by the SDSS Quasar
Catalog.
5. An optical/near-infrared power-law SED slope of α
−1.2, removing a small number of severely reddened type
1 QSOs from the sample.
While the full SDSS Quasar Catalog (Schneider et al. 2010)
includes 105,873 objects, the first 3 criteria result in a sample of
328 type 1 QSOs possessing measurements in optical as well as
mid-infrared wavelengths. We further refine this sample of 328
sources with 2 additional criteria, to remove small numbers of
outlying sources which would have interfered with our ability to
reliably generate a characteristic type 1 template. High-redshift
sources tend to suffer from severe blending issues in the higher
MIPS bands, and so 17 of these 328 quasars are removed by our
z 3.0 criterion. Ten additional quasars were removed from the
template generation algorithms by the fifth and final criterion,
as they appeared to be highly dust-reddened type 1 QSOs as
described by Glikman et al. (2012). While these 10 quasars are
still classified as type 1 QSOs, their SEDs are not typical of the
general type 1 population. As a result of our last two criteria,
a total of 27 type 1 QSOs are removed from our SED template
generation.
These fivecriteria result in a sample of 301 optically identified
quasars with measurements in at least one Spitzer infrared band.
As shown in Figure 1, these quasars show a fairly even distribu-
tion of redshifts between 0.0 and 3.0, resulting i n measurements
across a smooth distribution of rest-frame wavelengths extend-
ing from the Lyman Break to beyond 100 μm in the far-infrared
(FIR). This consistency is important for our template generation
process, as a more restrictive redshift cutoff would result in gaps
in the distribution of rest-frame wavelengths. As the distribution
of redshifts in this sample of 301 quasars matches closely that
of the SDSS sample as a whole, we feel confident that our selec-
tion methods do not inherently bias towards or against quasars
at certain redshifts.
2.2. Optical and Near-infrared: SDSS and 2MASS
To be included in this work, a quasar requires detection
in the SDSS, Data Release 7 (Abazajian et al. 2009). The
SDSS Quasar Catalog (Schneider et al. 2010) identifies 105,873
Figure 1. The redshifts for the objects included in the SDSS and SEIP/
SAFIRES samples. The full SDSS catalog is shown in black, while the two lower
histograms display all quasars located within the area observed by SAFIRES
(dark gray), and those with coverage in either 70 μm or 160 μm (light gray).
Redshift bin sizes are 0.01 for the SDSS sample and 0.1 for SEIP/SAFIRES.
distinct quasars within the full SDSS area, covering 9380 deg
2
of sky. Each potential quasar is confirmed through spectral
examination by that work; no explicit emission line equivalent
width limit was applied to rule out other types of objects, and
only a conservative emission line FWHM limit (>1000 km s
−1
)
was applied. However, Schneider et al. (2010) use catalogs
provided by previous SDSS works to explicitly remove type 2
quasars ( Reyes et al. 2008), Seyfert galaxies (Hao et al. 2005),
and BL Lac objects (Plotkin et al. 2010). The result is a sample
believed to consist only of type 1 quasars, an assumption we do
not dispute.
Each SDSS quasar possesses measured luminosities in all
five optical SDSS wavelength bands (u
, g
, r
, i
, z
), with
central wavelengths ranging from 350 to 850 nm. The sample
ranges from 0 <z<6.5 and all quasars are brighter than
m
i
19.1, which effectively removes type 2 QSOs from our
potential sample. As the SDSS Quasar Catalog only includes
spectroscopically confirmed quasars, it is unknown how many
additional type 1 quasars could also fall into the SDSS-Spitzer
coverage region. The sky-derived flux density limits within
optical bands are consistently several orders of magnitude below
the typical values for type 1 quasars, and so none of these
quasars would lack SDSS data due to insufficient flux density.
The completeness of the SDSS sample is controlled entirely by
the ability to identify quasars’ spectra.
Each SDSS quasar is also compared to data from the 2MASS
(Skrutskie et al. 2006), using both the 2MASS All-Sky Point
Source Catalog (PSC) and the 2MASS Survey Point Source
Reject Table. While this survey covers the entire sky, its
sensitivity is far lower than that of the SDSS, preventing us from
acquiring near-infrared fluxes for most of the SDSS quasars.
As the available 2MASS catalogs list only point sources, no
extended objects could be included in this sample; this criterion
greatly reduces the potential for source confusion, and more
closely matches the process used for our mid- and far-infrared
bands.
2.3. Mid-infrared: Spitzer Enhanced Imaging Products (SEIP)
The initial release of the SEIP (P. Capak et al., in preparation)
improves upon existing data sets in two significant ways: the
2
The Astrophysical Journal, 768:13 (9pp), 2013 May 1 Hanish et al.
Table 1
SAFIRES Data Distributions
Number of SDSS 2MASS IRAC MIPS
Quasars (All Bands) (All Bands) 3.6 μm4.5μm5.8μm8.0μm24μm70μm 160 μm
Covered 301
a
301 219 221 215 221 224 164 172
Direct 301
b
59
b
169 173 162 171 151 11 0
Indirect N/AN/A 46433236708995
Nondetections N/AN/A 4 5 21 14 3 64 77
Notes.
a
Includes any SDSS-identified quasars within the areas observed by the Spitzer Space Telescope within each band. No attempt is made to quantify the number
of quasars within this area lacking SDSS detections.
b
All objects in the provided catalogs possessed magnitudes for all bands on an instrument; these were treated as direct detections.
use of larger, deeper Spitzer fields, by combining contiguous
data from multiple observation requests (AORs), and the use
of additional indirect detection mechanisms to supplement the
S/N 10 direct measurements. This mid-infrared archival data
is now available, and contains sources in the four Spitzer IRAC
bands (at 3.6, 4.5, 5.8, and 8.0 μm) as well as the 24 μm band of
MIPS. The full SEIP will contain sources from approximately
1500 deg
2
of sky, with the subset released at the present time
spanning around 50% of the final amount. This subset contains
over 20 million detected sources of all types, with an estimated
20–30 million additional Galactic sources eventually to be
added. The initial release of the SEIP includes nine large, deep
extragalactic fields; for this work, we select quasars from two
of the largest contiguous fields, the Lockman Hole (Lockman
et al. 1986) and the field of the SDWFS (Ashby et al. 2009).
Objects are included in the SEIP Source List if they were
directly detected at 10σ in any bands, and detections in
multiple bands are treated as a single object if the detections in
multiple bands have no more than 1 arcsec separation between
IRAC bands, or 3 arcsec when comparing to the MIPS 24 μm
band. While 301 SDSS-located type 1 quasars fall within the
full area observed by Spitzer, only 222–229 quasars were located
within the area of adequate coverage depth within each IRAC
band. While their observed areas were equivalent, each IRAC
band had a slightly different total due to depth of coverage
differences or the effects of bad pixels.
2.4. Far-infrared: SAFIRES
We refine our type 1 QSO template in far-infrared wave-
lengths using source catalogs from the SAFIRES project.
SAFIRES expands the SEIP to include measurements of point
sources observed within the MIPS 70 and 160 μm bands
throughout the cryogenic phase of the Spitzer mission. Un-
like SEIP, SAFIRES is limited to only include extragalactic
(|b| 20.
◦
0) fields. With this limitation, and the lower total area
covered by the MIPS 70 and 160 μm bands, the full SAFIRES
data set will span approximately 600 deg
2
of sky when com-
pleted.
The methods used to extract these FIR sources are extremely
similartothoseusedfortheMIPS24μm measurements, as
the software used to process the cryogenic Spitzer data was
constructed with these wavelengths included. The primary
differences between these t wo MIPS bands and the previous
five Spitzer bands involve use of the Germanium Reprocessing
Tools (GeRT) to further refine data, along with an increase in the
correlation distances to compensate for the larger point-spread
functions (PSFs) and pixel scales of the two longer-wavelength
MIPS bands. Source identification mechanisms similar to those
of the SEIP were used throughout the SAFIRES project.
2.5. Detection Methods
As explained in Section 2.1, our core sample consisted of
301 type 1 quasars possessing Spitzer observations in at least 1
infrared band. Of these, 253 possess data within at least 1 MIPS
band, necessary for the generation of our far-infrared template.
The number of sources located within the area observed by
each band with sufficient depth of coverage, and the number of
detections within those areas, are shown in Table 1.Theless
resolved bands nearly always overlap with areas observed in
more resolved bands; less than 10% of the sources observed
in MIPS 70 μm or 160 μm lack coverage in MIPS 24 μm, and
almost all of the 24 μm sources possess IRAC coverage as well.
We attempted to measure the magnitude of each source within
the sample considered here, for each of the 15 wavelength
bands used in this work. For each measurement, there were
four possible outcomes:
1. “Direct” detections. These are sources identified by re-
liable, self-contained source detection algorithms, using
no information from other bands. For all Spitzer bands,
a signal-to-noise ratio (S/N) of at least 10 is required,
while the 2MASS catalogs required only a 7σ detection
(5σ if detected within all three 2MASS bands). The tools
used to extract this information vary with instrument, with
IRAC bands using Source Extractor (Bertin & Arnouts
1996) and MIPS bands using the APEX package within
MOPEX (Makovoz & Marleau 2005). The direct SDSS and
2MASS values are drawn from three provided source cata-
logs: the SDSS Quasar Catalog, the 2MASS All-Sky PSC,
and the 2MASS Survey Point Source Reject Table. Objects
in the Point Source Reject Table are only used if they pos-
sessed confidence ratings above 90%, in order to ensure
reliable SED generation.
2. “Indirect” detections. These consist of Spitzer aperture flux
measurements centered on positions provided by the SDSS
Quasar Catalog, instead of using positions measured di-
rectly in the band in question. As this process is position-
ally less accurate than the direct detections, larger apertures
are used to ensure complete measurement of objects. This
results in an increase in quoted noise levels, and so we re-
quire only S/N 3. While the SDSS Quasar Catalog uses
a similar method to add S/N 2 2MASS measurements
to their sources, we use no analogous system for our own
2MASS measurements, and there exists no similar method
for the SDSS catalog.
3
The Astrophysical Journal, 768:13 (9pp), 2013 May 1 Hanish et al.
3. Nondetections. As the SDSS Quasar Catalog formed the ba-
sis for the sample considered here, all quasars are detected
in SDSS. There are only a handful of true nondetections in
any of the four IRAC bands, and only two quasars possess
adequate 24 μm MIPS coverage depth but lack successful
detections. While the 2MASS catalogs lack measurements
for most of the SDSS sources, the majority of sources are
nondetections in these bands due to the prohibitive magni-
tude limits of the 2MASS projects; these magnitude limits
will be discussed in more detail in Section 3. Nondetec-
tions form a significant part of this sample in the MIPS
bands, for which our provided data table gives 3σ upper
limits, although we make more realistic assumptions when
generating SED templates.
4. Rejections. Sources can be rejected from this sample for
a number of technical or statistical reasons, even if there
exist image data for those regions. These reasons include
extended features (common in the MIPS bands due to
blending issues with nearby sources), cosmic rays, edge
effects, and bad pixels. The most common reason for
rejection was lack of coverage, as the spatial coverage of
the two far-IR MIPS bands were smaller than those of the
IRAC bands, where less than half of the Lockman Hole
region was observed in at least one of the two far-infrared
bands. A less common issue is that of insufficient coverage,
with individual measurements f or an object being ignored
if the coverage in that band’s composite image fell below
a nonzero band-specific threshold. While most common in
the Spitzer MIPS bands, this last effect is most easily seen
in the IRAC bands; all IRAC-covered objects possessed
data for all four bands, but variations in depth of coverage
occasionally led to the rejection of some IRAC bands for
a given object, resulting in the differing numbers of IRAC-
covered objects in Table 1.
As shown in Table 1, the relative contributions of these cat-
egories vary significantly with wavelength band. Indirect aper-
ture measurements formed only a small part of this sample in
the IRAC bands, and were in the minority for MIPS 24 μm.
Conversely, very few 70 or 160 μm sources had direct detec-
tions, with nearly all of the detections in these bands using the
indirect method.
In total, 193 out of 301 sources within the SDWFS and
Lockman Hole fields met the minimum coverage levels of our
work in at least one FIR band, and these sources are essential
for the construction of our far-infrared template. While the areas
covered by the MIPS 70 and 160 μm bands are the same, the
depth of coverage minima reduce the numbers of sources in
both bands, with 167 possessing adequate 70 μm coverage
and 176 with 160 μm, as many far-infrared fields lacked
the depth necessary for reliable measurements. The majority
of these indirect far-infrared detections had directly detected
counterparts in at least one of the five Spitzer wavelength bands
from SEIP. In total, only 26 of the 70 μm and 31 of the 160 μm
aperture detections had adequate depth of coverage in at least
one mid-infrared SEIP band while not possessing any direct
detections, and another 21 quasars had sufficient depth in one of
the two far-infrared bands but lacked coverage in MIPS 24 μm
or any of the IRAC bands.
Direct measurements were correlated purely by distance, with
each source matched to the nearest qualifying source from other
catalogs. SDSS, 2MASS, and IRAC bands required a separation
of no more than 1 arcsecond, while MIPS 24 μm detections
could vary by up to 3 arcseconds from their counterparts. The
70 μm MIPS sources were considered to match those of the
best possible SDSS or SEIP counterpart within a match radius
of 6 arcseconds, while the 160 μm data required a separation
of no more than 8 arcseconds. These distances were selected
by examining the correlations for several hundred bright point
source objects within an initial sample of quasars, and selecting
the typical distance at which no apparent mismatches occurred.
As these correlations suffer greatly from confusion issues in
the higher MIPS bands, this distance does not scale linearly
with the PSF radius of each band. The conservative nature of
this distance cutoff produces a correlated source list of high
confidence, instead of a comprehensive source catalog.
However, while the correlation distances are chosen such
that mismatches were minimized, the large size of the PSF
in the 70 and 160 μm bands means that the potential for
source confusion is much higher than in a more resolved
mid-infrared band, as any aperture is likely to include flux
contributions from a number of nearby objects as well. The
result is the distinct possibility of a single SAFIRES detection
or aperture including flux contributions from multiple distinct
sources, potentially increasing our estimated SED at far-infrared
wavelengths. Figure 2 illustrates the difficulties inherent in
correlating far-infrared sources to more resolved bands. As most
70 and 160 μm “sources” are actually the combination of several
distinct objects, their center positions are unlikely to fall within
our correlation distances, shown as the black circle in each
image. This effect will be examined in more detail in Section 4.
3. SPECTRAL ENERGY DISTRIBUTIONS
By matching the Spitzer-observed sources from the SEIP and
SAFIRES projects with their counterparts in the SDSS/2MASS
quasar catalog, as detailed in Section 2, we obtain a wide variety
of data spanning up to fifteen bands with wavelengths ranging
from 3500 Å to almost 200 μm, with a robust population of
rest-frame-adjusted effective wavelengths extending from the
near-ultraviolet to around 100 μm. The measurements of each
SAFIRES quasar for each available wavelength band are given
in our online table, with a f ormat described in Table 2.
For the Lockman Hole, the areas containing sufficient depth
of coverage in MIPS 70 and 160 μm are less than half of those
for IRAC, resulting in a somewhat lower number of potential
quasars from those bands. However, in the areas where seven
Spitzer bands were available, the detection rates for all bands
remained comparable. While IRAC and MIPS 24 μm allowed
for a large number of direct, S/N 10 detections, our indirect
aperture method was sufficient to provide S/N 3 counterparts
for these objects in the MIPS 70 and 160 μm bands. Likewise,
the SDWFS field had excellent detection rates as the spatial
coverages in all bands as the bands’ respective areas overlapped
substantially more, with over 90% of the area in each region
observed in all seven Spitzer wavelength bands. Within the
12.82 deg
2
contained within SDWFS, the SDSS Quasar Catalog
lists 81 quasars, of which 71 were directly detected in IRAC and
the 24 μm MIPS band, all 71 of which successfully acquired flux
density values through our indirect method in both the 70 and
160 μm MIPS bands.
The type 1 QSOs comprising this sample varied in luminosity
by several orders of magnitude. As most of these samples had
incomplete data, with several bands lacking adequate depth or
being removed for various technical reasons, it was necessary
to normalize each object’s flux densities to a common baseline
before any sort of logic could combine the data points into
a coherent template. As we are trying to examine the shape
4
The Astrophysical Journal, 768:13 (9pp), 2013 May 1 Hanish et al.
Figure 2. Several quasars in the SDWFS field, shown in the IRAC 3.6 μm, MIPS 24 μm, MIPS 70 μm, and MIPS 160 μm bands. All images are 1 arcmin
2
, centered
on the SDSS position for each quasar. The overlay circles show the position of each quasar, and the maximum correlation distances allowed within each band.
Table 2
SAFIRES Photometry Format
Column Name Descriptions
(1) ID Name from SDSS Quasar Catalog
(2) RA Right ascension (J2000)
(3) DEC Declination (J2000)
(4) Z Spectroscopic redshift from SDSS
(5) REGID Region ID from SEIP; see P. Capak et al. (in preparation)
(6)–(10) U_MAG, G_MAG, R_MAG, I_MAG, Z_MAG Magnitudes (AB) for quasars in the SDSS u
, g
, r
, i
,andz
bands
(11)–(13) J_MAG, H_MAG, K_MAG Magnitudes (AB) for quasars in the 2MASS J, H,andK
S
bands
(14)–(17) I1_MAG, I2_MAG, I3_MAG, I4_MAG Magnitudes (AB) for quasars in the Spitzer IRAC 3.6, 4.5, 5.6, and 8.0 μm bands
(18)–(20) M1_MAG, M2_MAG, M3_MAG Magnitudes (AB) for quasars in the Spitzer MIPS 24, 70, and 160 μm bands
(21)–(23) J_STAT, H_STAT, K_STAT Detection status for 2MASS bands: direct or nondetection
(24)–(27) I1_STAT, I2_STAT, I3_STAT, I4_STAT Detection status for IRAC bands: direct, indirect, nondetection, or low coverage
(28)–(30) M1_STAT, M2_STAT, M3_STAT Detection status for MIPS bands: direct, indirect, nondetection, or low coverage
Notes. The data table is available online for the 301 type 1 quasars used by the template generation algorithm. For nondetections possessing adequate
coverage, the quoted magnitudes are the 3σ flux density limits at the appropriate location.
(This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its form and
content.)
of the far-infrared SED, we normalize each object based on its
luminosities at rest-frame wavelengths between 0.2 and 1.0 μm,
the most well-studied part of the quasar SEDs. Nearly every
quasar (285 out of 301) possessed at least two measurements
within this region, as every quasar in our sample had a complete
set of five SDSS measurements, with many objects adding
2MASS detections as well. More than half of our sample (166
objects) possess at least four observations within that rest-frame
wavelength range, resulting in a robust normalization process.
All fluxes corresponding to rest-frame wavelengths within those
limits were compared to the Polletta QSO1 template value at the
appropriate wavelength, to derive a band-specific deviation from
the QSO1 template; each object’s fluxes were then normalized
by the mean deviation for all bands within that rest-frame
window. This process effectively normalizes each object’s SED
by its optical and near-infrared intensities without penalizing
objects lacking depth in individual bands, which is essential
given the wide variation in spatial coverage for the various
bands.
We used these normalized luminosities to generate a single
SED template, with a method based on that of several earlier
works (Richards et al. 2006; Polletta et al. 2007; Hatziminaoglou
et al. 2008). Measurements are sorted into a series of rest-frame
wavelength bins, which begin with a width of 0.2 dex. This
size was chosen to minimize the chance of an object possessing
multiple rest-frame measurements within a single bin, a very
real possibility when dealing with the optical SDSS bands. We
then use an iterative adjustment process to narrow the bins
down further while keeping a minimum number of 100 rest-
frame source measurements within each bin, with the final bins
being as small as 0.03 dex. The resulting distribution of binned
data provides greater detail to the templates in a wavelength
regime that previously used a simple power-law relationship,
while keeping the number of detections within each bin at a
reasonable value. To minimize the impact of erroneous data
points, we use the weighted medians within our wavelength bins
to generate a final SED template, shown in yellow in Figure 3(a).
4. RESULTS
4.1. SED Template
We have created an SED template using the methods de-
scribed in Section 3, for the SAFIRES-SDSS sample of 301
type 1 QSOs, 167 of which possess data in either the 70 or
160 μm wavelength bands. The results are shown in Figure 3(a).
This template, using the binned median of detections,
is shown in yellow, while the previous SED curves are
shown in blue (Polletta et al. 2007, QSO1) and green
(Richards et al. 2006). We have also confirmed the relia-
bility of this template through the use of a Kaplan–Meier
5
