Luminous buried AGNs as a function of galaxy infrared luminosity revealed through Spitzer low-resolution infrared spectroscopy

1.2 MB·English

by  Masatoshi Imanishi

AstrophysicalJournal
PreprinttypesetusingLATEXstyleemulateapjv.11/26/04
LUMINOUS BURIED AGNS AS A FUNCTION OF GALAXY INFRARED LUMINOSITY REVEALED
THROUGH SPITZER LOW-RESOLUTION INFRARED SPECTROSCOPY
Masatoshi Imanishi1
NationalAstronomicalObservatory,2-21-1,Osawa,Mitaka,Tokyo181-8588, Japan
Astrophysical Journal
ABSTRACT
9 We present the results of Spitzer IRS infrared 5–35 µm low-resolution spectroscopic energy diag-
0 nostics of ultraluminous infrared galaxies (ULIRGs) at z > 0.15, classified optically as non-Seyferts.
0 Based on the equivalent widths of polycyclic aromatic hydrocarbon emission and the optical depths
2 of silicate dust absorption features, we searched for signatures of intrinsically luminous, but optically
elusive,buriedAGNsintheseopticallynon-SeyfertULIRGs. Wethencombinedtheresultswiththose
n
ofnon-SeyfertULIRGsatz <0.15andnon-SeyfertgalaxieswithinfraredluminositiesL <1012L .
a IR ⊙
J We found that the energetic importance of buried AGNs clearly increases with galaxy infrared lumi-
nosity,becoming suddenly discernible inULIRGs with L > 1012L . For ULIRGs with buriedAGN
5 IR ⊙
signatures,asignificantfractionofinfraredluminositiescanbeaccountedforbydetectedburiedAGN
] and modestly-obscured (AV < 20 mag) starburst activity. The implied masses of spheroidal stellar
A components in galaxies for which buried AGNs become important roughly correspond to the value
G separating red massive and blue, less-massivegalaxies in the local universe. Our results may support
the widely-proposed AGN-feedback scenario as the origin of galaxy downsizing phenomena, where
.
h galaxies with currently larger stellar masses previously had higher AGN energetic contributions and
p star-formation-originating infrared luminosities, and have finished their major star-formation more
- quickly, due to stronger AGN feedback.
o
r Subject headings: galaxies: active—galaxies: ISM—galaxies: nuclei—galaxies: Seyfert—galaxies:
t starburst — infrared: galaxies
s
a
[
1. INTRODUCTION propertiesofdust-obscuredenergysourcesdifferbetween
1 LIRGs with L = 1011−12L and ULIRGs with L >
Infraredskysurveyshavediscoveredalargenumberof IR ⊙ IR
8v galaxies that are bright in the infrared (LIR > 1011L⊙). 1012L⊙. First, the fraction of optical Seyferts 2 system-
atically increases with increasing infrared galaxy lumi-
5 They are called luminous infrared galaxies (LIRGs), or
nosity,andissignificantlyhigherinULIRGsthanLIRGs
5 ultraluminous infrared galaxies (ULIRGs) when the in-
0 frared luminosity exceeds LIR > 1012L⊙ (Sanders & (Veilleuxetal.1999;Goto2005). Next,inLIRGs,alarge
fraction of infrared emission comes from spatially ex-
. Mirabel 1996). The spectral energy distributions of
1 tended regions, whereas the infrared dust emission from
(U)LIRGs are dominated by infrared emission, which
0 ULIRGsisdominatedbyaspatiallycompactcomponent
means that luminous energy sources are hidden behind
9
(Soiferetal.2000,2001). Thissuggeststhatmuchofthe
0 dust. The bulk of the energetic radiation from the en-
dust in LIRGs is heated by stars distributed over spa-
: ergysourcesis absorbedby the surroundingdust. Then,
v tiallyextendedgalacticregions,whileinULIRGs,alarge
the heated dust grains emit this energy as infrared dust
i amount of dust is concentrated in the nuclear regions
X emission. Theenergysourcescanbenuclearfusioninside
andheatedbyspatially compactenergysources. Since a
rapidly formed stars (starbursts), the release of gravita-
r mass-accretingSMBHcanproducehighluminosityfrom
a tional energy produced by mass accreting supermassive
acompactarea(<1pc),anAGNisaplausiblesourcefor
black holes (SMBHs) (i.e., AGN activity), or some com-
compact dust emission in ULIRG nuclei, although very
bination. The importance of (U)LIRGs to the cosmic
compact starbursts are still a possibility.
energy output increases rapidly with increasing redshift
The higher fraction of optical Seyferts and higher nu-
(Le Floc’h et al. 2005; Perez-Gonzalez et al. 2005; Ca-
clear dust concentrationin ULIRGs than LIRGs are dif-
putietal.2007). Thus,understandingthehiddenenergy
ficult to explain using the scenario that the energetic
sources of (U)LIRGs is closely related to clarifying the
role of starbursts and AGNs are similar for LIRGs and
historyofstar formationandSMBHgrowthin the dust-
ULIRGs. InopticalSeyferts,linesofsightalongthetorus
obscured portion of the universe. Since obtaining high
axis are relatively transparent to AGN ionizing radia-
quality data of distant(z > 0.5) (U)LIRGs is not simple
tion, and narrow line regions, photoionized by the cen-
with existing observational facilities, detailed studies of
tral AGN radiation, should develop at the 10–1000 pc
nearby (z < 0.3) (U)LIRGs continue to play an impor-
scale, above a torus scale height. Since narrow-line re-
tantroleinunderstandingthepropertiesofthe(U)LIRG
gionsproduceopticalemissionlineswhosefluxratiosdif-
population in the universe.
fer from those in clouds photoionized by stars, the pres-
Observations of nearby (U)LIRGs suggest that the
Electronicaddress: [email protected] 2 We denote optical Seyferts as luminous AGNs whose accre-
1Department of Astronomy, School of Science, Graduate Uni- tion disk and broad lineregions are thought to be surrounded by
versityforAdvancedStudies,Mitaka,Tokyo181-8588 torus-shaped dusty obscuring medium, responsible for absorbing
thenuclearradiationalongsomelinesofsight.
2 Imanishi
enceofsuchluminousAGNssurroundedbytorus-shaped distinguishable based on the optical depths of dust ab-
dust is easily recognizable through optical spectroscopy sorption features at different wavelengths. In a normal
(Veilleux & Osterbrock 1987; Kewley et al. 2006). As starburst, the energy sources (stars) and gas/dust are
a larger amount of dust concentrates in the nuclear re- spatially well mixed (Puxley 1991; McLeod et al. 1993;
gions of ULIRGs, even the direction of the lowest dust Forster Schreiber et al. 2001), whereas in a buried AGN
column density can be opaque to AGN ionizing radia- theenergysource(=thecentralaccretingSMBH)isvery
tion, blocking the radiation at the inner part (<10 pc) compact and more centrally concentrated than the sur-
in almost all directions. Such buried AGNs lack well- rounding gas and dust (Soifer et al. 2000; Imanishi &
developed narrow-line regions and so are classified opti- Maloney 2003; Siebenmorgen et al. 2004) (see also Fig-
cally as non-Seyferts (Imanishi et al. 2007,2008). Thus, ures 1a and 1b of Imanishi et al. 2007). The difference
withincreasingdustconcentrationaroundULIRGnuclei, in geometry is reflected in two features in the observed
it is expected that the fraction of buried AGNs (= opti- low-resolutioninfraredspectra. First,while the absolute
calnon-Seyferts)increases,while thatofopticalSeyferts opticaldepthsofdustabsorptionfeaturesinthe3–10µm
decreases. Thisis contrarytoobservations. The increas- rangecannotexceedacertainthresholdinanormalstar-
ingfractionofdetectableopticalSeyfertsinULIRGscan burstwithmixeddust/sourcegeometry,theycanbearbi-
be explained if the energetic importance of AGNs in- trarilylargeinaburiedAGN(Imanishi&Maloney2003;
trinsically increases in ULIRGs. Specifically, even if the Imanishietal.2006,2007;Levensonetal.2007). Second,
narrow-line regions under-develop, if the intrinsic AGN aburiedAGNshowsastrongdusttemperaturegradient,
luminosities increase,then AGN-signatureopticaldetec- in which inner dust, closer to the central energy source,
tion relative to stellar emission becomes easier, increas- hasahighertemperaturethanouterdust;anormalstar-
ing the fraction of optical Seyferts. If this is the case, a burst does not. The presence of this dust temperature
large fraction of ULIRGs should contain luminous, but gradient can be investigated observationally by compar-
opticallyelusive,buriedAGNs,andtheburiedAGNfrac- ing the optical depths of dust absorptionfeatures at dif-
tion should increase substantially in ULIRGs, compared ferent infrared wavelengths (Dudley & Wynn-Williams
to LIRGs. Since such luminous buried AGNs can have 1997;Imanishi 2000; Imanishi et al. 2006, 2007).
strong feedback to the surrounding dust and gas, it is Since the PAH emission and dust absorption features
very important to understand buried AGNs in ULIRGs, are spectrally very broad, low-resolution (R = 50–100)
not only to unveil the true nature of the ULIRG pop- infraredspectroscopy is adequate. We can thus examine
ulation, but also to observationally constrain the AGN- fainter sources than high-resolution (R > 500) infrared
starburst connections in galaxies. spectroscopy(Farrahet al.2007),in termsofsensitivity.
Low-resolution infrared spectroscopy is an effective Imanishi et al. (2007) performed infrared 5–35 µm low-
tool for studying buried AGNs that lack well-developed resolutionspectroscopicinvestigationsofULIRGsatz <
narrow-line regions (or narrow-line regions are obscured 0.15, classified optically as non-Seyferts, using Spitzer
by foreground dust), for the following reasons. First, IRS, and found buried AGN signatures in a significant
emissionfrompolycyclicaromatichydrocarbons(PAHs), fraction of the observed ULIRGs. However, the infrared
seen in infrared spectra at λ = 3–25 µm in the rest- luminosities of most of the observed ULIRGs are in a
rest
frame,canbeusedtodistinguishbetweenaburiedAGN narrow range of L = 1012−12.3L , hampering analysis
IR ⊙
and a normal starburst (Genzel et al. 1998; Imanishi of buried AGNs as a function of galaxy infrared lumi-
& Dudley 2000). In a normal starburst with moderate nosity. It is known that the fraction of optical Seyferts
metallicity (>0.3 solar), consisting of UV-emitting HII increases with ULIRGs with L ≥ 1012.3L , compared
IR ⊙
regions, molecular gas and dust, and photo-dissociation to those with L = 1012−12.3L (Veilleux et al. 1999).
IR ⊙
regions (PDRs), PAHs are excited by far-UV photons A similar analysis of buried AGNs in ULIRGs with L
IR
from stars, and strong PAH emission is produced in ≥ 1012.3L will help elucidate the nature of the ULIRG
⊙
PDRs (Sellgren 1981; Wu et al. 2006). When stellar en- population.
ergysourcesanddustarespatiallywellmixed,thefluxof In this paper, we present a systematic, uniform analy-
both PAH emission and the nearby continuum are simi- sis of Spitzer IRS infrared 5–35 µm low-resolution spec-
larlyattenuated,sothattheequivalentwidthofthePAH tra of ULIRGs at z > 0.15, optically classified as non-
emissionisinsensitivetodustextinction. Thus,anormal Seyferts. ByextendingourstudytoULIRGsatz >0.15,
starburstwithPDRsshouldalwaysshowlargeequivalent manyULIRGswithL ≥1012.3L willbeincluded,en-
IR ⊙
widthPAHemission,regardlessoftheamountofdustex- abling meaningful comparison of the buried AGN frac-
tinction. In a pure buried AGN, PAHs are destroyed by tion between ULIRGs with L = 1012−12.3L and L
IR ⊙ IR
strongX-rayradiationfromtheAGN(Voit1992;Sieben- ≥ 1012.3L . Throughout this paper, H = 75 km s−1
⊙ 0
morgen et al. 2004); thus, no PAH emission is seen. In- Mpc−1, Ω = 0.3, and Ω = 0.7 are adopted, to be
M Λ
stead, a PAH-free continuum from hot, submicron-sized consistent with our previously published papers.
dust grains heated by the AGN is observed. In a star-
burst/AGN composite galaxy, PAH emission is seen if 2. TARGETS
starburstsoccuratlocationsthataresufficientlyshielded We selected our targets from the IRAS 1 Jy sample
from the AGN X-ray radiation. However, the PAH- (Kim&Sanders1998). This1Jysamplelists48ULIRGs
equivalentwidthwillbe smallerthaninapurestarburst at z > 0.15. Based on the optical spectral classifica-
becauseofthedilutionbyPAH-freecontinuumproduced tions by Veilleux et al. (1999 Table 2), 33 ULIRGs are
by the AGN. Thus, we can, in principle, disentangle a classifiedoptically as non-Seyferts (i.e., LINERs, HII re-
buried AGN from a normal starburst based on infrared gions, and unclassified), and the remaining 15 ULIRGs
spectral shapes. are classified optically as Seyferts. Since our primary
Second, a buried AGN and a normal starburst are scientific goal is to study buried AGNs, these 33 opti-
Buried AGNs as a function of galaxy IR luminosity 3
cal non-Seyfert ULIRGs are our main targets. Of the conclusions. In fact, in all cases, the Spitzer IRS 25 µm
33 ULIRGs, 15 and 12 ULIRGs are classified optically flux agrees within 20% to, or is smaller than, the IRAS
as LINERs and HII-regions, respectively, and 6 ULIRGs 25 µm data. For ULIRGs with IRAS non-detection at
are optically unclassified. We analyzedthe spectra of 10 25 µm, the measured Spizter IRS 25 µm flux is always
LINER,6 HII-region,and4 unclassifiedULIRGs. These smaller than the IRAS upper limits.
20 (= 10 + 6 + 4) ULIRGs cover >60% of 33 optically For a fractionof the observedULIRGs, slight flux dis-
non-Seyfert ULIRGs at z > 0.15, and should be unbi- crepancies between SL1 and LL2 were discernible, rang-
ased in terms of their dominant energy sources. Table ingfrom30%to60%. Whenthediscrepancywaspresent,
1 summarizes the basic information and IRAS-basedin- the SL1 flux (3′.′7 wide slit) was always smaller than the
frared properties of the observed optically non-Seyfert LL2 flux (10′.′5). In these cases, we adjusted the smaller
ULIRGs. Althoughonly8outof48non-SeyfertULIRGs SL1(andSL2)fluxtomatchthelargerLL2flux. Appro-
atz < 0.15studied by Imanishietal.(2007)haveL ≥ priate spectral binning with 2 or 4 pixels was applied to
IR
1012.3L , 17 out of 20 observed non-Seyfert ULIRGs at reducethescatterofdatapointsatSL2(5.2–7.7µm)for
⊙
z > 0.15 show L ≥ 1012.3L (Table 1), substantially some faint ULIRGs, and at λ ∼ 10 µm for ULIRGs
IR ⊙ obs
increasing the number of ULIRGs with L ≥ 1012.3L . that display very strong 9.7 µm silicate dust absorption
IR ⊙
features.
3. OBSERVATIONSANDDATAANALYSIS
4. RESULTS
Observations of all 20 ULIRGs were performed using
the Infrared Spectrograph(IRS) (Houck et al. 2004)on- Figure 1 presents the infrared 5–35 µm low-resolution
board the Spitzer Space Telescope (Werner et al. 2004). spectra of the observed 20 ULIRGs. For most of the
All four modules, Short-Low 2 (SL2; 5.2–7.7 µm) and sources,full5–35µmspectraareshownhereforthefirst
1 (SL1; 7.4–14.5 µm), and Long-Low 2 (LL2; 14.0–21.3 time.
µm) and 1 (LL1; 19.5–38.0µm) were used to obtain full The spectra in Figure 1 are suitable for displaying the
5–35 µm low-resolution (R ∼ 100) spectra. Table 2 de- propertiesofoverall5–35µmspectralshapesandthe9.7
tailstheobservinglog. Theslitwidthwas3′.′6or2pixels µm and 18 µm silicate dust absorption features. How-
for SL2 (1′.′8 pixel−1) and 3′.′7 or ∼2 pixels for SL1 (1′.′8 ever,they arenotveryusefulforPAHemissionfeatures.
pixel−1). For LL2 and LL1, the slit widths were 10′.′5 Figure 2 presents enlarged spectra at λobs = 5.2–14.5
and 10′.′7, respectively, corresponding to ∼2 pixels for µm to better exhibit the properties ofthe PAH emission
both LL2 (5′.′1 pixel−1) and LL1 (5′.′1 pixel−1). features.
Thelatestpipeline-processeddataproductsatthetime
4.1. PAH emission
of our data analysis were used. Frames taken at po-
sition A were subtracted from those taken at position The majority of ULIRGs in Figure 2 show clearly de-
B to remove background emission, mostly the zodiacal tectable PAH emission features at λrest = 6.2 µm, 7.7
light. Spectra were then extracted in a standard man- µm, and 11.3 µm. To estimate the strengths of these
ner. Apertureswith4–5pixelswereemployedforSLand PAH emission features, we adopted a linear continuum,
LL data, depending on the spatial extent of individual following Imanishi et al. (2007) who made systematic
sources. Then, spectra extracted for the A and B posi- and detailed analysis of Spitzer IRS infrared 5–35 µm
tions were summed. Wavelength calibration was made low-resolution spectra of optically non-Seyfert ULIRGs
based on the files of the Spitzer pipeline processed data, at z < 0.15. For the 6.2 µm, 7.7 µm, and 11.3 µm PAH
named “b0 wavsamp.tbl” and “b2 wavsamp.tbl” for SL emission features, data at λrest = 6.1 µm and 6.45 µm,
andLL,respectively. Thesedataarebelievedtobeaccu- 7.3 and 8.1 µm, and 11.0 µm and 11.6 µm, were used,
rate within 0.1 µm. A small level of error in wavelength respectively,todeterminelinearcontinuumlevels,shown
calibration will not affect our main conclusions. Since as solid lines in Figure 2. PAH emission features, above
emissionfromallULIRGsisdominatedbyspatiallycom- the adopted continuum levels, were fitted with Gaus-
pactsourcesattheobservedwavelength,fluxcalibration sian profiles. The observed rest-frame equivalent widths
was performed using the Spitzer pipeline processed files (EWPAH) and luminosities of the 6.2 µm, 7.7 µm, and
“b0 fluxcon.tbl”(SL)and“b2 fluxcon.tbl”(LL).ForSL1 11.3 µm PAH emission features, based on our adopted
spectra,dataatλ >14.5µmintheobservedframeare continuum levels, are summarized in Table 3. The un-
obs
invalid (Infrared Spectrograph Data Handbook Version certaintiescomingfromthefittingsareunlikelytoexceed
1.0) and were discarded. For LL1 spectra, we used only 30%.
dataatλ <35µmbecausethe datapointscatterwas As noted by Imanishi et al. (2007), we estimate the
obs
large at λ > 35 µm and we did not need data at λ strengths of the 7.7 µm PAH emission feature in such a
obs obs
>35µmforourscientificdiscussions. Nodefringingwas way that the uncertainties caused by the strong, broad
attempted. 9.7 µm silicate dust absorption feature are minimized.
For flux calibration, we adopted the values of the Since our continuum definition is significantly different
pipeline processed data. We made no attempt to re- from those employed in previous papers (e.g., Genzel et
calibrate our spectra using IRAS measurements at 12 al. 1998),readersmustbecarefulwhenourvalueiscom-
µm and 25 µm, because only upper limits are provided paredwithotherestimatesintheliterature. Forthisrea-
for IRAS 12 µm and/or 25 µm photometry in many of son,althoughthe7.7µmPAHemissionstrengthisshown
theobservedULIRGs(Table1). Hence,theabsoluteflux for reference in this paper, it will not play an important
calibration is dependent on the accuracy of the pipeline partin ourdiscussions,whichwillbe basedprimarilyon
processed data, which is taken to be <20% for SL and the 6.2 µm and 11.3 µm PAH emission strengths.
LL (Infrared Spectrograph Data Handbook). This level
4.2. Silicate absorption
of flux uncertainty will not significantly affect our main
4 Imanishi
To estimate the strengths of silicate dust absorption L /L ratios in ULIRGs. The L /L ra-
11.3PAH IR 11.3PAH IR
features, we used τ′ and τ′ , defined by Imanishi et tios are (0.09–1.2) × 10−3, or 6–86 % of the value of
9.7 18
al. (2007). The τ′ value is the optical depth of the 1.4 × 10−3 for modestly-obscured starburst galaxies. In
9.7
9.7 µm silicate absorption feature against a power-law the majority of observedULIRGs, the ratios are<0.7 ×
continuum determined from data points at λ = 7.1 10−3(<50%of1.4×10−3). TheobservedL /L
rest 11.3PAH IR
µm and 14.2 µm. The τ′ value is the optical depth of ratiossuggestthatdetectedmodestly-obscuredstarburst
18
the 18 µm silicate absorption feature against a power- activitycanaccountfor5–86%(<50%inmostcases)of
law continuum determined from data points at λ = the infrared luminosities of ULIRGs. When we combine
rest
14.2µmand24µm. Thesecontinuaareshownasdotted the L /L and L /L ratios, we can con-
6.2PAH IR 11.3PAH IR
lines in Figure 1. Since these continuum levels were de- clude that the detected modestly-obscured starburst ac-
termined using data points just outside the 9.7 µm and tivity is energetically significant (say, 10–50 %), but not
18 µm features, close to the absorption peaks, the mea- dominant(say,>70–80%)intheobservedULIRGs. The
suredopticaldepths werewell-definedas dips relativeto remaining energy sources in these ULIRGs must there-
the nearby continuum emission. The τ′ values for all fore be (1) highly-obscured(A >> 20 mag)starbursts,
9.7 V
ULIRGsareshowninTable4(column2). Theτ′ values wherethefluxesofPAHemissionaresubstantiallyatten-
18
are also shown in Table 4 (column 3) for ULIRGs where uated by dust extinction, and/or (2) buried AGNs that
the 18 µm silicate feature is clearly seen in absorption. produce strong infrared radiation, but virtually no PAH
emission.
4.3. Ice and CO absorption
A significant fraction of ULIRGs display dips at the 5.2. ULIRG candidates with luminous buried AGNs
shorterwavelengthside ofthe 6.2µmPAHemissionfea- 5.2.1. Low equivalent widths of PAH emission features
ture. We ascribe the dips to the 6.0 µm H O ice ab-
2
WhetherthedominantenergysourcesofULIRGnuclei
sorption feature (bending mode). Figure 3 presents en-
are highly-obscured normal starbursts or buried AGNs
largedspectraatλ =5.2–9µmforULIRGsthatshow
obs
canbedeterminedusingtheequivalentwidthofthePAH
this absorption feature clearly. The spectrum of IRAS
emission. Since the PAH equivalent width (EW )
12018+1941isalsoshownasanexampleofnon-detection PAH
mustalwaysbe largeina normalstarburst(with PDRs)
of this feature. For ULIRGs with clearly detectable 6.0
regardless of the amount of dust extinction, a small
µmH Oiceabsorptionfeatures,observedopticaldepths
2
EW value suggests contribution from a PAH-free
(τ ) are summarized in Table 5. PAH
6.0 continuum-emittingenergysource,namelyanAGN(§1).
The spectrum of IRAS 00397−1312 in Figure 3 dis-
Following Imanishi et al. (2007), we classify ULIRGs
plays a clear CO absorption feature (λ = 4.67 µm).
rest with EW < 180 nm, EW < 230 nm, and
Its optical depth is also shown in Table 5. 6.2PAH 7.7PAH
EW < 200 nm as sources displaying clear signa-
11.3PAH
5. DISCUSSION tures of luminous AGNs. Since these equivalent width
To study buried AGNs in optically non-Seyfert values are less than one-third of the typical values for
ULIRGs at z > 0.15 and to investigate the buried AGN starburst galaxies (Brandl et al. 2006), a substantial
fraction as a function of ULIRG infrared luminosity, we contribution from AGN PAH-free continuum emission
use the same criteria as applied to non-Seyfert ULIRGs is indicated. Table 6 (columns 2–4) presents detection
at z < 0.15 (Imanishi et al. 2007). or non-detection of buried AGN signatures based on
the PAH equivalent width threshold. The buried AGN
5.1. Magnitudes of detected starbursts fraction is much larger based on the small EW6.2PAH
value (14/20; 70%) than on the small EW value
The flux attenuationof continuum emission at λ > 11.3PAH
rest
(3/20; 15%). This is reasonable, because the 11.3 µm
5 µm, aside from the strong 9.7 µm silicate absorption
PAH emission feature is inside the strong 9.7 µm sil-
peak,issmall(<1mag)fordustextinctionwithA <20
V
icate dust absorption feature; thus, buried AGN con-
mag (Rieke & Lebofsky 1985; Lutz et al. 1996). Thus,
tinuum emission at λ ∼ 11.3 µm is severely atten-
observed PAH emission luminosities can roughly trace rest
uated, not strongly diluting the 11.3 µm PAH emission
the intrinsic luminosities of modestly-obscured (A <
V
from modestly-obscured starburst regions (Imanishi et
20 mag) PAH-emitting normal starbursts (with PDRs).
al. 2007).
Table 3 (columns 8 and 9) tabulates the 6.2 µm PAH to
infrared luminosity ratio, L /L , and the 11.3 µm
6.2PAH IR 5.2.2. Absolute optical depths of dust absorption features
PAHtoinfraredluminosityratio,L /L . Thera-
11.3PAH IR
tiosinnormalstarburstgalaxieswithmodestdustobscu- Based on the EW values, we can easily detect
PAH
ration (A < 20 mag) are estimated to be L /L buried AGNs with very weak starbursts. Even if strong
V 6.2PAH IR
∼ 3.4 × 10−3 (Peeters et al. 2004) and L /L ∼ starburst activity is present, weakly obscured AGNs are
11.3PAH IR
1.4 × 10−3 (Soifer et al. 2002). detectablebecauseweaklyattenuatedPAH-freecontinua
The observed L /L ratios in the observed from the AGNs can dilute the PAH emission consider-
6.2PAH IR
ULIRGs(Table3)rangefrom<0.2×10−3to1.4×10−3, ably. However, detecting deeply buried AGNs with co-
or<6%to41%ofthevalueof3.4×10−3 formodestly- existing strong starbursts is not easy. Even if the in-
obscured starburst galaxies. In the majority of the ob- trinsic luminosities of a buried AGN and surrounding
served ULIRGs, the ratios are <1 × 10−3, or <30% of less-obscured starbursts are similar, the AGN flux will
3.4 × 10−3. Taken at face value, the detected modestly- be more highly attenuated by dust extinction than the
obscuredstarburstsintheseULIRGscanaccountfor<6 starburst emission, making the observed EW values
PAH
% to 41 % (mostly <30%) of their infrared luminosi- apparently large.
ties. The same argumentcanbe applied to the observed To determine whether a deeply buried AGN is present
Buried AGNs as a function of galaxy IR luminosity 5
inadditiontostrongstarbursts,weusetheopticaldepths dust, whereas a normal starburst nucleus with mixed
of silicate dust absorption features. As described in §1 dust/source geometry does not (§1). As explained by
andin Imanishietal.(2007)in moredetail, these values Imanishi et al. (2007) in detail, the presence of a strong
canbeusedtodistinguishwhethertheenergysourcesare dust temperature gradient can be detected by compar-
spatially well mixed with dust (a normal starburst), or ing the optical depths of 9.7 µm and 18 µm silicate dust
are more centrally concentrated than the dust (a buried absorption features because the optical depths of dust
AGN). absorption features at shorter wavelengths probe dust
In a normal starburst with mixed dust/source geom- column density toward inner hotter dust than those at
etry, observed flux is dominated by foreground, less- longer wavelengths (Figure 2 of Imanishi et al. 2007).
obscured, less-attenuated emission (which shows only Following Imanishi et al. (2007), if an observed τ′ /τ′
18 9.7
weak dust absorption features), with a small contribu- ratio is substantially smaller than τ′ /τ′ = 0.3, the
18 9.7
tion from highly-obscured, highly-attenuated emission most reasonable explanation is the presence of a strong
at the background side of the emitting regions (which dust temperature gradient. This can provide additional
shows strong dust absorption features). Thus, the ob- evidence for centrally-concentrated buried AGNs previ-
served optical depths of dust absorption features can- ouslysuggestedbylowEW orlargeτ′ values(≥2).
PAH 9.7
not exceed a certain threshold, unless very unusual dust Table 4 (column 4) summarizes the τ′ /τ′ ratios for
18 9.7
composition patterns are assumed (Imanishi & Maloney ULIRGsshowingclear18µmsilicateabsorption(mostly
2003;Imanishi et al. 2006,2007). In a buried AGN with τ′ > 2). Table 6 (column 6) displays the detection (or
9.7
centrally-concentratedenergy source geometry, the fore- non-detection) of buried AGN signatures, based on this
groundscreendustmodelisapplicable,andtheobserved small τ′ /τ′ method.
18 9.7
optical depths can be arbitrarily large. Hence, detec-
5.2.4. Combination of energy diagnostic methods
tion of strong dust absorption features, whose optical
depths substantially exceed the upper limit achieved by Table 6 (column 7) summarizes the strengths of the
themixeddust/sourcegeometry,arguesforaforeground detected buried AGN signaturesin Spitzer IRS 5–35µm
screen dust geometry, as expected from a buried AGN spectrabasedonthreemethods: (1)lowPAHequivalent
(Imanishi & Maloney 2003; Imanishi et al. 2006, 2007). width; (2) large τ′ value; and (3) small τ′ /τ′ ratio.
9.7 18 9.7
Imanishi et al. (2007) obtained a maximum value of τ′ When buried AGN signatures in individual ULIRGs are
9.7
< 1.7 fora normalstarburstwith mixed dust/sourcege- consistentlyfoundusingallormostofthesemethods,the
ometry. Consideringpossibleuncertaintiesintheτ′ es- ULIRGs are classified as very strong buried AGN candi-
9.7
timate(∼10%),weclassifyULIRGswithτ′ ≥2ascan- dates, marked with open double circles. When buried
9.7
didates for harboring luminous centrally-concentrated AGN signatures are seen only in the first method, or
buried AGNs. first and second methods, then the ULIRGs are classi-
InaburiedAGN,silicatedustattheveryinnerpartof fied as strong AGN candidates (open circles). When the
theobscuringmaterial,closetothecentralenergysource, signatures are detected only in the second method, the
can be heated to high temperature, show silicate emis- ULIRGsareclassifiedaspossibleburiedAGNcandidates
sion,anddilutethesilicateabsorptionfeature. Although (open triangles),as a normalstarburstnucleus obscured
this dilution may have significant effects on the discus- byforegrounddustinanedge-onhostgalaxy(Figure1d
sion of τ′ and τ′ for weakly-obscured AGNs (Sirocky of Imanishi et al. 2007) cannot be ruled out completely
9.7 18
et al. 2008),it should not be significantin ULIRGs with in individual cases.
large τ′ (= energy sources are highly dust obscured),
as stud9i.e7d in this paper. Individual ULIRGs that show 5.2.5. Absorption-corrected intrinsic luminosities of
buried AGN signatures based on large τ′ values (>2; buried AGNs
9.7
Table 4) are marked with open circles in Table 6 (col- For ULIRGs that show buried AGN signatures and
umn 5). small PAH equivalent widths, the observed fluxes are
We have two notes on this method. First, this large mostly ascribed to AGN-heated PAH-free dust contin-
τ9.7 methodissensitivetodeeplyburiedAGNsbutobvi- uum emission. For these ULIRGs, we can estimate the
ouslymissesweaklyobscuredAGNs,whicharemoreeas- absorption-corrected intrinsic dust emission luminosity
ily detected withthe abovelow EWPAH method. Hence, at ∼10 µm (νFν) heated by the AGN, based on the ob-
this largeτ9.7 methodplaysacomplementaryroleto the served fluxes at λrest ∼ 10 µm and the dust extinction
low EWPAH method for the purpose of detecting buried toward the 10 µm continuum emitting regions inferred
AGN signatures. Second, a normal starburst nucleus from τ′ (Imanishi et al. 2007). In a buried AGN with
9.7
with mixed dust/source geometry can produce a large a strong dust temperature gradient, assuming a simple
τ′ value, if it is obscured by a large amount of fore- spherical dust distribution, dust emission luminosity is
9.7
ground screen dust in an edge-on host galaxy (Figure conserved at each temperature from hot inside regions
1d of Imanishi et al. 2007). Although Imanishi et al. to cool outside regions. Namely, the intrinsic luminos-
(2007) argued that it is very unlikely that the majority ity of inner hot dust emission at 10 µm (νF ) should be
ν
of ULIRGs with τ′ > 2 correspond to this non-AGN comparabletothatofoutercooldustemissionat60µm,
9.7
case, some particular ULIRGs could so correspond. the wavelength which dominates the observed infrared
emissionof ULIRGs (Sanders et al. 1988a). Hence, from
5.2.3. Strong dust temperature gradients
AGN-originating intrinsic νF (10 µm) values, we can
ν
A buried AGN with centrally-concentrated energy quantitatively estimate the energetic contribution from
source geometry should show a strong dust tempera- buried AGNs to the infrared luminosities of ULIRGs.
ture gradient, in which inner dust, close to the cen- Following Imanishi et al. (2007), we assume that τ′
9.7
tral energy source, has a higher temperature than outer and the extinction at λ = 8 or 13 µm continuum just
rest
6 Imanishi
outside the 9.7 µm silicate feature (A ) are related 5 µm low-resolutionspectroscopy (Imanishi et al. 2008).
cont
to τ′ /A ∼ 2.3 (Rieke & Lebofsky 1985). Based Given that the fraction of ULIRGs with optical Seyfert
9.7 cont
on a foreground screen dust absorption model applica- signatures also increases with increasing galaxy infrared
ble to buried AGNs, we obtain, as seen in Table 7, luminosity(Veilleux etal.1999;Goto2005),wecancon-
absorption-corrected intrinsic AGN luminosities for se- clude that AGN activity becomes more important with
lected ULIRGs with low PAH equivalent widths. The increasing galaxy infrared luminosity.
fluxattenuationofthe8or13µmcontinuumoutsidethe The so-called galaxy downsizing phenomenon has re-
9.7µm silicate feature rangesfroma factor of1.7(IRAS cently been proposed; it was found that galaxies with
12018+1941; τ′ ∼ 1.3) to 3.1 (IRAS 04313−1649; τ′ currently larger stellar masses finished their major star-
9.7 9.7
∼ 2.8). The absorption-corrected intrinsic AGN lumi- formation in an earlier cosmic age (Cowie et al. 1996;
nosities could explain a significant fraction (15–60%) of Bundy et al. 2005). AGN feedback is suggested to
the luminosities of these ULIRGs (Table 7). be responsible for the galaxy downsizing phenomenon
(Granato et al. 2004; Bower et al. 2006; Croton et al.
5.3. Buried AGN fraction as a function of galaxy 2006). Namely, in galaxies with currently large stellar
infrared luminosity
masses, AGN feedback was stronger in the past, heat-
Optically non-Seyfert ULIRGs at z < 0.15 studied by ing or expelling gas in host galaxies and stopping star
Imanishietal.(2007)mostlydisplayL =1012−12.3L . formation on a shorter time scale. Buried AGNs can
IR ⊙
TheextensionofourSpitzerlow-resolutioninfraredspec- have particularly strong feedback because the AGNs are
troscopicenergy diagnostic to ULIRGs at z > 0.15gives surrounded by a large amount of nuclear gas and dust.
a large number ULIRGs with L ≥ 1012.3L . Specifi- In addition, galaxies with currently larger stellar masses
IR ⊙
cally, only 8 of 48 observed non-Seyfert ULIRGs at z < should have had higher star-formation-originating in-
0.15 have L ≥ 1012.3L (Table 1 of Imanishi et al. frared luminosities in the past, as more stars were
IR ⊙
2007), while 17 of 20 observed non-Seyfert ULIRGs at formed.
z > 0.15 have L ≥ 1012.3L (Table 1). We can thus Wefoundthatthe buriedAGNfractionincreaseswith
IR ⊙
investigatetheburiedAGNfraction,separatingULIRGs increasing galaxy infrared luminosity. In ULIRGs with
into two categories: those with LIR = 1012−12.3L⊙ and LIR > 1012L⊙, the importance of buried AGNs sud-
those with LIR ≥ 1012.3L⊙. denly becomes clear, compared to galaxies with LIR <
Figure 4 shows the distribution of EW6.2PAH, 1012L⊙ (Imanishi et al. 2007; this paper), suggesting
EW , and τ′ as a function of galaxy infrared lu- thatAGNfeedbackbecomessignificantinULIRGs. The
11.3PAH 9.7
minosity. In addition to ULIRGs, optically non-Seyfert absorption-corrected intrinsic luminosities of detected
galaxieswithLIR <1012L⊙ (Brandletal.2006)areplot- buried AGNs are > a few × 1045 ergs s−1 (&1012L⊙),
ted. Itisevidentthatinallplots,thefractionofgalaxies which could accountfor a significant (15–100%)fraction
which meet the requirement of buried AGNs increases ofULIRGinfraredluminosities(Table7ofthispaperand
with increasing galaxy infrared luminosity. Imanishi et al. 2007). The infrared luminosities of de-
ToinvestigatethedetectableburiedAGNfractionasa tected modestly-obscured starbursts (Table 7, columns
function ofgalaxyinfraredluminosity in more detail, we 3 and 4) are also a few × 1045 ergs s−1, or ∼1012L⊙,
use the combined method (§5.2.4). Based on the classi- which are still higher than the total infrared luminosi-
fication of non-Seyfert ULIRGs at z > 0.15 in Table 6, ties of galaxies with LIR < 1012L⊙. Thus, both stronger
13 out of 17 ULIRGs with L ≥ 1012.3L show strong AGN feedback and higher star-formation-originating in-
IR ⊙
buried AGN signatures (open double circles and open frared luminosities are suggested in ULIRGs 3 than in
circles in column 7 of Table 6), in contrastto only 1 out lower LIR galaxies.
of 3 ULIRGs with L = 1012−12.3L . For non-Seyfert Kauffmannetal.(2003)foundthatgalaxiesaboveand
IR ⊙
ULIRGs at z < 0.15, 5 out of 8 ULIRGs with LIR ≥ belowstellarmassesofM∗ =3×1010M⊙ showdistinctly
1012.3L⊙ showstrongburiedAGNsignatures,incontrast different properties. Galaxies with M∗ > 3 × 1010M⊙
to only 11 out of 40 ULIRGs with L = 1012−12.3L are dominated by red galaxies with currently low star-
IR ⊙
(Imanishi et al. 2007). When we combine these results, formationrates,while those with M∗ < 3 × 1010M⊙ are
we obtain a fraction of strong buried AGN signatures mainly blue galaxieswith ongoingactive star-formation.
of 18/25 (72%) for ULIRGs with L ≥ 1012.3L and If AGN feedback is responsible for the dichotomy, then
IR ⊙
12/43(28%)forULIRGswithLIR =1012−12.3L⊙. When buriedAGNsmaybecomeimportantingalaxieswithM∗
we include sources with possible buried AGN signatures > 3 × 1010M⊙. Based on the velocity dispersion mea-
(open triangles in Table 6 of this paper and Imanishi et surementsofULIRGhostgalaxiesintheinfrared,Dasyra
al. 2007), the fraction of detectable buried AGNs signa- et al. (2006) argued that ULIRGs will have spheroidal
tures is 22/25 (88%) for ULIRGs with LIR ≥ 1012.3L⊙ stellarmasseswithseveral×1010M⊙. AssumingtheEd-
and 19/43 (44%) for ULIRGs with L = 1012−12.3L . dington luminosity for detected buried AGNs (> a few
IR ⊙
For optically non-Seyfert galaxies with L < 1012L ,
IR ⊙
although the sample size is limited, no sources are clas- 3 The summed luminosities of detected buried AGNs and
modestly-obscured starbursts are generally smaller than the ob-
sified as buried AGNs in our criteria (Figure 4). Other
servedinfraredluminositiesof ULIRGs. This discrepancymaybe
various observations also suggest that the energetic im- dueto(1)possibleunderestimationofabsorption-correctedintrin-
portance of buried AGNs clearly decreases in galaxies sicburied AGN luminosity, caused by a small starburstcontribu-
with L < 1012L , compared to ULIRGs (Soifer et al. tiontothe observed infraredflux(§5.2.7 of Imanishiet al. 2007),
IR ⊙
or by dust extinction curves in ULIRGs that differ from our as-
2000,2001). Therefore,we clearlysee the trend that the
sumption,or(2)thepresenceofhighlydust-obscured(AV >>20
detectableburiedAGNfractionincreaseswithinfraredlu- mag)starbursts,or(3) intrinsicallylow PAHtoinfraredluminos-
minosity of optically non-Seyfert galaxies (Figure 5), as ityratioinstarburst activity inULIRGs, or somecombination of
haspreviouslybeensuggestedfromAKARIinfrared2.5– factors.
Buried AGNs as a function of galaxy IR luminosity 7
× 1045 ergs s−1), SMBH masses are estimated to be a = 1012−12.3L . Given that (1) the fraction of optical
⊙
few × 107M . If we adopt the correlation between the Seyferts is also higher in ULIRGs with L ≥ 1012.3L
⊙ IR ⊙
masses of SMBH and spheroidal stars established in the than in those with L = 1012−12.3L , and (2) signa-
IR ⊙
localuniverse(Magorrianetal.1998),similarspheroidal tures of AGNs, including both buried AGNs and optical
stellar masses with > a few × 1010M are obtained. Seyferts, areweakerin galaxies with L < 1012L than
⊙ IR ⊙
TheseestimatedspheroidalstellarmassesofULIRGsare ULIRGs, we concluded that AGN importance increases
similar to the value separating red massive (= strong asgalaxyinfraredluminosityincreases. BuriedAGNsbe-
AGNfeedbacks)andblue, less-massivegalaxies(= week come clearly discernible in ULIRGs with implied stellar
AGN feedbacks). In summary, our discovery of increas- masses of M > a few × 1010M , the value that sepa-
∗ ⊙
ingburiedAGNfractionwithgalaxyinfraredluminosity rates red massive and blue, less-massive galaxies in the
may observationally support the widely-proposed AGN nearby universe. Our overall results support the AGN-
feedback scenario as the origin of the galaxy downsizing feedback scenario as the origin of the galaxy downsizing
phenomenon. phenomenon.
6. SUMMARY
We presented the results of Spitzer IRS infrared 5–
35 µm low-resolution spectroscopic energy diagnostic We thank the anonymous referee for his/her very use-
method of optically non-Seyfert ULIRGs at z > 0.15. ful comments. This work is based on observations made
The signatures of intrinsically luminous, but optically with the Spitzer Space Telescope, operated by the Jet
elusive, buried AGNs were searched for, based on the Propulsion Laboratory, California Institute of Technol-
equivalentwidthsofPAHemissionandtheopticaldepths ogy under a contractwith NASA. Support for this work
of silicate dust absorption features in these ULIRGs. was provided by NASA and also by an award issued by
Since most of the ULIRGs at z > 0.15 have L ≥ JPL/Caltech. M.I.issupportedbyGrants-in-AidforSci-
IR
1012.3L , by combining results with our previous anal- entific Research (19740109). This research made use of
⊙
ysis of ULIRGs at z < 0.15, we could investigate the the SIMBAD database, operated at CDS, Strasbourg,
detectable buried AGN fraction by separating ULIRGs France, and the NASA/IPAC Extragalactic Database
with L = 1012−12.3L and L ≥ 1012.3L . We found (NED),whichisoperatedbythe JetPropulsionLabora-
IR ⊙ IR ⊙
thatthedetectableburiedAGNfractionisclearlyhigher tory, California Institute of Technology, under contract
in ULIRGs with L ≥ 1012.3L than those with L with NASA.
IR ⊙ IR
REFERENCES
Bower, R. G., Benson, A. J., Malbon, R., Helly, J. C., Frenk, C. Lutz,D.etal.1996,A&A,315,L269
S., Baugh, C. M.,Cole, S.,& Lacey, C.G. 2006, MNRAS,370, Magorrian,J.,etal.1998,ApJ,115,2285
645 McLeod, K. K., Rieke, G. H., Rieke, M. J., & Kelly, D. M. 1993,
Brandl,B.R.,etal.2006,ApJ,653,1129 ApJ,412,111
Bundy,K.,Ellis,R.S.,&Conselice,C.J.2005,ApJ,625,621 Peeters,E.,Spoon,H.W.W.,&Tielens,A.G.G.M.2004, ApJ,
Caputi,K.I.,etal.2007,ApJ,660,97 613,986
Cowie, L. L., Songaila, A., Hu, E. M., & Cohen, J. G. 1996, AJ, Perez-Gonzalez, P.G.,etal.2005, ApJ,630,82
112,839 Puxley,P.J.1991,MNRAS,249,11p
Croton,D.J.,etal.2006,MNRAS,365,11 Rieke,G.H.,&Lebofsky,M.J.1985,ApJ,288,618
Dasyra,K.M.,etal.2006,ApJ,651,835 Sanders,D.B.,&Mirabel,I.F.1996, ARA&A,34,749
Dudley,C.C.,&Wynn-Williams,C.G.1997,ApJ,488,720 Sanders,D.B.,Soifer,B.T.,Elias,J.H.,Madore,B.F.,Matthews,
Farrah,D.etal.2007,ApJ,667,149 K.,Neugebauer,G.,&Scoville,N.Z.1988a,ApJ,325,74
Forster Schreiber, N. M., Genzel, R., Lutz, D., Kunze, D., & Sanders, D. B., Soifer, B. T., Elias, J. H., Neugebauer, G., &
Sternberg, A.2001,ApJ,552,544 Matthews,K.1988b, ApJ,328,L35
Houck,J.R.,etal.2004, ApJS,154,18 Sellgren,K.1981,ApJ,245,138
Genzel,R.etal.1998,ApJ,498,579 Siebenmorgen, R., Krugel, E., & Spoon, H. W. W. 2004, A&A,
Goto,T.2005, MNRAS,360,322 414,123
Granato, G. L.,DeZotti, G.,Silva,L.,Bressan,A.,&Danese, L. Sirocky,M.M.,Levenson,N.A.,Elitzur,M.,Spoon,H.W.W.,&
2004, ApJ,600,580 Armus,L.2008,ApJ,678,729
Imanishi,M.2000,MNRAS,319,331 Soifer,B.T.etal.2000,AJ,119,509
Imanishi,M.,&Dudley,C.C.2000,ApJ,545,701 Soifer,B.T.etal.2001,AJ,122,1213
Imanishi,M.,&Maloney,P.R.2003,ApJ,588,165 Soifer, B. T., Neugebauer, G., Matthews, K., Egami, E., &
Imanishi,M.,Dudley,C.C.,&Maloney,P.R.2006,ApJ,637,114 Weinberger,A.J.2002,AJ,124,2980
Imanishi, M., Dudley, C. C., Maiolino, R., Maloney, P. R., Veilleux,S.,Kim,D.-C.,&Sanders,D.B.1999, ApJ,522,113
Nakagawa, T.,&Risaliti,G.2007,ApJS,171,72 Veilleux,S.,&Osterbrock,D.E.1987,ApJS,63,295
Imanishi, M., Nakagawa, T., Ohyama, Y., Shirahata, M., Wada, Voit,G.M.1992, MNRAS,258,841
T.,Onaka,T.,&Oi,N.2008,PASJ,inpress(ArXiv:0808:0363) Werner,M.W.,etal.2004,ApJS,154,1
Kauffmann,G.,etal.2003, MNRAS,341,54 Wu,Y.,Charmandaris,V.,Hao,L.,Brandl,B.R.,Bernard-Salas,
Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. 2006, J.,Spoon,H.W.W.,&Houck,J.R.2006,ApJ,639,157
MNRAS,372,961
Kim,D.-C.,&Sanders,D.B.,1998,ApJS,119,41
LeFloc’hetal.2005,ApJ,632,169
Levenson, N. A., Sirocky, M. M., Hao, L., Spoon, H. W. W.,
Marshall, J. A., Elitzur, M., & Houck, J. R., 2007, ApJ, 654,
L45
8 Imanishi
TABLE 1
Observed ULIRGsatz> 0.15 andtheir IRAS-based infraredemission properties
Object Redshift f12 f25 f60 f100 logLIR f25/f60 Optical
(Jy) (Jy) (Jy) (Jy) L⊙ Class
(1) (2) (3) (4) (5) (6) (7) (8) (9)
IRAS03521+0028 0.152 <0.11 0.20 2.52 3.62 12.5 0.08(C) LINER
IRAS09463+8141 0.156 <0.07 <0.07 1.43 2.29 12.3 <0.05(C) LINER
IRAS10091+4704 0.246 <0.06 <0.08 1.18 1.55 12.6 <0.07(C) LINER
IRAS11028+3130 0.199 <0.09 0.09 1.02 1.44 12.4 0.09(C) LINER
IRAS11180+1623 0.166 <0.08 <0.19 1.19 1.60 12.2 <0.16(C) LINER
IRAS11582+3020 0.223 <0.10 <0.15 1.13 1.49 12.5 <0.14(C) LINER
IRAS12032+1707 0.217 <0.14 0.25 1.36 1.54 12.6 0.18(C) LINER
IRAS16300+1558 0.242 <0.07 0.07 1.48 1.99 12.7 0.05(C) LINER
IRAS16333+4630 0.191 <0.06 0.06 1.19 2.09 12.4 0.05(C) LINER
IRAS23129+2548 0.179 <0.08 0.08 1.81 1.64 12.4 0.04(C) LINER
IRAS00397−1312 0.261 0.14 0.33 1.83 1.90 12.9 0.18(C) HII-region
IRAS01199−2307 0.156 <0.11 <0.16 1.61 1.37 12.3 <0.1(C) HII-region
IRAS01355−1814 0.192 <0.06 0.12 1.40 1.74 12.4 0.09(C) HII-region
IRAS08201+2801 0.168 <0.09 0.15 1.17 1.43 12.3 0.13(C) HII-region
IRAS13469+5833 0.158 <0.05 0.04 1.27 1.73 12.2 0.03(C) HII-region
IRAS17068+4027 0.179 <0.08 0.12 1.33 1.41 12.3 0.09(C) HII-region
IRAS01494−1845 0.158 <0.08 <0.15 1.29 1.85 12.2 <0.12(C) unclassified
IRAS04313−1649 0.268 <0.07 0.07 1.01 1.10 12.6 0.07(C) unclassified
IRAS10035+2740 0.165 <0.14 <0.17 1.14 1.63 12.3 <0.15(C) unclassified
IRAS12018+1941 0.168 <0.11 0.37 1.76 1.78 12.5 0.21(W) unclassified
Note. — Col.(1): Objectname. Col.(2): Redshift. Col.(3)–(6): f12,f25, f60,andf100 areIRASfluxesat12µm,25µm, 60µm,and100µm,
respectively,takenfromKim&Sanders(1998). Col.(7): Decimallogarithmofinfrared(8−1000µm)luminosityinunitsofsolarluminosity(L⊙),
calculatedwith LIR = 2.1×1039× D(Mpc)2 × (13.48 × f12 + 5.16 × f25 + 2.58×f60+f100) ergs s−1 (Sanders& Mirabel1996). Since the
calculation is based on our adopted cosmology, the infrared luminositiesdiffer slightly (<10%) from the values shown in Kim & Sanders (1998,
Table1, column15). Forsourceswith upperlimitsinsome IRASbands, we canderiveupperandlower limitsfor infraredluminosity,assuming
thattheactualfluxistheIRAS-upperlimitandzerovalue,respectively. Thedifferenceintheupperandlowervaluesisusuallyverysmall,less
than 0.2 dex. We assume that the infrared luminosity is the average of these values. Col.(8): IRAS 25 µm to 60 µm flux ratio. ULIRGs with
f25/f60 <0.2and>0.2areclassifiedascoolandwarm sources(denotedas“C”and“W”),respectively(Sandersetal. 1988b). Col.(9): Optical
spectralclassificationbyVeilleuxetal.(1999).
TABLE 2
SpitzerIRS observinglog
Object PID Date Integrationtime[sec]
[UT] SL2 SL1 LL2 LL1
(1) (2) (3) (4) (5) (6) (7)
IRAS03521+0028 105 2004Feb27 240 240 180 180
IRAS09463+8141 105 2004Mar23 240 240 120 120
IRAS10091+4704 105 2004Apr19 240 240 120 120
IRAS11028+3130 30407 2007Jun9 240 240 180 180
IRAS11180+1623 30407 2007Jun8 196 196 180 180
IRAS11582+3020 105 2005Dec16 240 240 120 120
IRAS12032+1707 105 2004Jan4 240 240 120 120
IRAS16300+1558 105 2005Aug13 240 240 300 300
IRAS16333+4630 105 2004Mar4 240 240 120 120
IRAS23129+2548 105 2003Dec17 360 360 300 300
IRAS00397−1312 105 2004Jan4 240 240 180 180
IRAS01199−2307 105 2004Jul18 240 240 180 180
IRAS01355−1814 105 2005Jul10 240 240 120 120
IRAS08201+2801 30407 2007May3 196 196 180 180
IRAS13469+5833 20589 2006Apr26 240 112 480 960
IRAS17068+4027 105 2004Apr16 240 240 180 180
IRAS01494−1845 105 2005Jul14 240 240 120 120
IRAS04313−1649 105 2004Mar1 240 240 300 300
IRAS10035+2740 30407 2007Jun8 196 196 180 180
IRAS12018+1941 105 2004May15 120 120 180 180
Note. — Col.(1): Object name. Col.(2): PID number. Col.(3): Observing date in UT. Col.(4): Net on-source integration
time for SL2 spectroscopy in sec. Col.(5): Net on-source integration time for SL1 spectroscopy in sec. Col.(6): Net on-source
integration time for LL2 spectroscopy in sec Col.(7): Net on-source integration time for LL1 spectroscopy in sec.
Buried AGNs as a function of galaxy IR luminosity 9
TABLE 3
Observed propertiesof PAH emission features
Object EW6.2PAH EW7.7PAH a EW11.3PAH L6.2PAH L7.7PAH a L11.3PAH L6.2PAH/LIR L11.3PAH/LIR
[nm] [nm] [nm] 1042 [ergss−1] 1042 [ergss−1] 1042 [ergss−1] [×10−3] [×10−3]
(1) (2) (3) (4) (5) (6) (7) (8) (9)
IRAS03521+0028 245 855 450 6.1 24.3 5.2 0.5 0.4
IRAS09463+8141 225 765 530 2.9 11.3 1.7 0.4 0.3
IRAS10091+4704 <50 480 560 <1.7 29.1 3.5 <0.2 0.2
IRAS11028+3130 160 870 460 1.9 12.1 1.7 0.2 0.2
IRAS11180+1623 165 855 340 4.1 25.6 2.7 0.6 0.4
IRAS11582+3020 70 235 325 6.4 63.1 5.9 0.5 0.5
IRAS12032+1707 70 195 315 5.2 26.1 5.9 0.3 0.4
IRAS16300+1558 85 410 400 7.4 40.6 6.0 0.4 0.3
IRAS16333+4630 275 680 485 13.1 42.6 11.6 1.4 1.2
IRAS23129+2548 110 320 330 5.9 30.8 4.1 0.6 0.4
IRAS00397−1312 25 160 35 21.4 135.5 2.9 0.6 0.09
IRAS01199−2307 130 440 215 4.3 21.8 1.9 0.6 0.3
IRAS01355−1814 170 605 195 4.2 19.1 1.7 0.4 0.2
IRAS08201+2801 105 425 280 5.5 33.8 4.2 0.8 0.6
IRAS13469+5833 235 675 460 4.0 15.7 3.4 0.7 0.6
IRAS17068+4027 110 405 185 8.5 40.1 5.0 1.0 0.6
IRAS01494−1845 365 810 485 8.0 25.0 5.0 1.3 0.8
IRAS04313−1649 <120 305 240 <3.2 15.3 2.1 <0.3 0.1
IRAS10035+2740 660 900 275 5.0 9.8 1.1 0.8 0.2
IRAS12018+1941 120 220 35 9.3 24.5 2.3 0.9 0.2
Note. — Col.(1): Objectname. Col.(2): Rest-frame equivalentwidth of the6.2 µm PAHemission. Col.(3): Rest-frame equivalentwidth of
the7.7µmPAHemission. Col.(4): Rest-frameequivalentwidthofthe 11.3µmPAHemission. Col.(5): Luminosityofthe6.2µmPAHemission
in units of 1042 ergs s−1. Col.(6): Luminosity of the 7.7 µm PAHemission in unitsof 1042 ergs s−1. Col.(7): Luminosity of the 11.3 µm PAH
emissioninunitsof1042 ergss−1. Col.(8): The6.2 µm PAHtoinfraredluminosityratioinunitsof10−3. Theratiofornormalstarburstswith
modestdustobscuration(AV <20mag)is∼3.4×10−3 (Peetersetal.2004). Col.(9): The11.3µmPAHtoinfraredluminosityratioinunitsof
10−3. Theratiofornormalstarburstswithmodestdustobscuration(AV <20mag)is∼1.4×10−3 (Soiferetal.2002).
a
Weregardfluxexcessatλrest =7.3–8.1µmaboveanadoptedcontinuumlevelas7.7µmPAHemission,toreducetheeffectsofthestrong9.7
µmsilicatedustabsorptionfeature. Ourdefinitionisdifferentfromthosepresentedinpreviouspapers.
TABLE 4
9.7 µm and18 µm silicatedust absorption featureopticaldepths
Object τ9′.7 τ1′8 τ1′8/τ9′.7
(1) (2) (3) (4)
IRAS03521+0028 1.3 0.4 0.31
IRAS09463+8141 2.0 0.6 0.30
IRAS10091+4704 2.5 1.0 0.40
IRAS11028+3130 2.5 1.1 0.44
IRAS11180+1623 2.0 0.6 0.30
IRAS11582+3020 2.7 0.8 0.30
IRAS12032+1707 2.6 0.6 0.23
IRAS16300+1558 2.6 0.9 0.35
IRAS16333+4630 1.3 ··· ···
IRAS23129+2548 2.6 0.9 0.35
IRAS00397−1312 2.7 0.6 0.22
IRAS01199−2307 2.4 0.7 0.29
IRAS01355−1814 2.4 0.7 0.29
IRAS08201+2801 2.2 0.5 0.23
IRAS13469+5833 1.7 0.5 0.29
IRAS17068+4027 1.8 0.5 0.28
IRAS01494−1845 1.6 ··· ···
IRAS04313−1649 2.8 0.7 0.25
IRAS10035+2740 2.0 0.7 0.35
IRAS12018+1941 1.3 0.3 0.23
Note. — Col.(1): Objectname. Col.(2): τ9′.7 is opticaldepthof the 9.7 µm silicatedust absorption feature,against apower-law continuum
shownasdottedlinesinFigure1. Oncethecontinuumlevelsarefixed,theuncertaintyofτ9′.7 is<5%forULIRGswithlargeτ9′.7 values(>2)and
canbe∼10%forULIRGswithsmallτ9′.7. Col.(3): τ1′8isopticaldepthofthe18µmsilicatedustabsorptionfeature,againstapower-lawcontinuum
shown asdottedlinesin Figure1. Oncethecontinuumis fixed,theuncertaintyof τ1′8 is<10% becausethe valueis estimatedonly for ULIRGs
withclearlydetectable18µmsilicateabsorptionfeatures. Col.(4): τ1′8/τ9′.7 ratioforULIRGswithclearlydetectable18µmsilicateabsorption.
10 Imanishi
TABLE 5
IceandCO absorption features
Object τ6.0 τCO
(1) (2) (3)
IRAS09463+8141 0.7 ···
IRAS11582+3020 0.3 ···
IRAS12032+1707 0.5 ···
IRAS16300+1558 0.3 ···
IRAS16333+4630 0.4 ···
IRAS23129+2548 0.4 ···
IRAS00397−1312 0.2 1.0(P),1.2(R)
IRAS01355−1814 0.3 ···
IRAS08201+2801 0.5 ···
IRAS13469+5833 0.9 ···
IRAS17068+4027 0.3 ···
Note. — Col.(1): Objectname. Col.(2): Opticaldepthofthe6.0µmH2Oiceabsorptionfeatureforclearlydetectedsources. Col.(3): Optical
depth of the 4.67 µm CO absorption feature. “P” means the P-branch of the CO absorption feature(the sub-peak at longer wavelength). “R”
meanstheR-branchoftheCOabsorptionfeature(thesub-peakatshorterwavelength).
TABLE 6
BuriedAGNsignatures
Object EW6.2PAH EW7.7PAH EW11.3PAH τ9′.7 T-gradient Total
(1) (2) (3) (4) (5) (6) (7)
IRAS03521+0028 X X X X X X
IRAS09463+8141 X X X (cid:13) X △
IRAS10091+4704 (cid:13) X X (cid:13) X (cid:13)
IRAS11028+3130 (cid:13) X X (cid:13) X (cid:13)
IRAS11180+1623 (cid:13) X X (cid:13) X (cid:13)
IRAS11582+3020 (cid:13) X X (cid:13) X (cid:13)
IRAS12032+1707 (cid:13) (cid:13) X (cid:13) (cid:13) ⊚
IRAS16300+1558 (cid:13) X X (cid:13) X (cid:13)
IRAS16333+4630 X X X X X X
IRAS23129+2548 (cid:13) X X (cid:13) X (cid:13)
IRAS00397−1312 (cid:13) (cid:13) (cid:13) (cid:13) (cid:13) ⊚
IRAS01199−2307 (cid:13) X X (cid:13) X (cid:13)
IRAS01355−1814 (cid:13) X (cid:13) (cid:13) X (cid:13)
IRAS08201+2801 (cid:13) X X (cid:13) (cid:13) ⊚
IRAS13469+5833 X X X X X X
IRAS17068+4027 (cid:13) X (cid:13) X X (cid:13)
IRAS01494−1845 X X X X X X
IRAS04313−1649 (cid:13) X X (cid:13) (cid:13) ⊚
IRAS10035+2740 X X X (cid:13) X △
IRAS12018+1941 (cid:13) (cid:13) (cid:13) X (cid:13) ⊚
Note. — Col.(1): Objectname. Col.(2): BuriedAGNsignaturesbasedonthelowequivalentwidthofthe6.2µmPAHemission(EW6.2PAH <
180nm)(§5.2.1). (cid:13): present. X:none. Col.(3): BuriedAGNsignaturesbasedonthelowequivalentwidthofthe7.7µmPAHemission(EW7.7PAH
< 230 nm) (§5.2.1). (cid:13): present. X: none. Col.(4): Buried AGN signatures based on the low equivalent width of the 11.3 µm PAH emission
(EW11.3PAH <200nm)(§5.2.1). (cid:13): present. X:none. Col.(5): BuriedAGNsignaturesbasedonthelargeτ9′.7 value(>2)(§5.2.2). (cid:13): present.
X:none. Col.(6): BuriedAGNsignaturesbasedonthesmallτ1′8/τ9′.7 ratio(§5.2.3). (cid:13): present. X:none. Col.(7): BuriedAGNsignaturesfrom
⊚
combinedmethodsinCol. (2)–(6). : verystrong. (cid:13): strong. △: possible. X:none. Pleasesee§5.2.4formoredetails.