An ALMA Survey of Submillimeter Galaxies in the Extended Chandra Deep Field South: The AGN ...

39 Pages·2013·1.63 MB·English

PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
AN ALMA SURVEY OF SUBMILLIMETER GALAXIES IN THE EXTENDED CHANDRA DEEP
FIELD-SOUTH: THE AGN FRACTION AND X-RAY PROPERTIESOF SUBMILLIMETER GALAXIES
S. X. Wang (王雪凇)1, W. N. Brandt1,2, B. Luo1,2, I. Smail3, D. M. Alexander3, A. L. R. Danielson3, J. A. Hodge4,
A. Karim3,5, B. D. Lehmer6,7, J. M. Simpson3, A. M. Swinbank3, F. Walter4, J. L. Wardlow8, Y. Q. Xue9,
S. C. Chapman10,11, K. E. K. Coppin12, H. Dannerbauer13, C. De Breuck14, K. M. Menten15, and P. van der
Werf16
3 ABSTRACT
1
The large gas and dust reservoirs of submm galaxies (SMGs) could potentially provide ample fuel
0
totriggeranActiveGalacticNucleus(AGN),butpreviousstudiesoftheAGNfractioninSMGshave
2
been controversial largely due to the inhomogeneity and limited angular resolution of the available

t submillimeter surveys. Here we set improved constraints on the AGN fraction and X-ray properties
c
O of the SMGs with ALMA and Chandra observations in the Extended Chandra Deep Field-South
(E-CDF-S). This study is the first among similar works to have unambiguously identified the X-ray

3 counterparts of SMGs; this is accomplished using the fully submm-identified, statistically reliable
2 SMG catalog with 99 SMGs from the ALMA LABOCA E-CDF-S Submillimeter Survey (ALESS).
Wefound10X-raysourcesassociatedwithSMGs(medianredshiftz =2.3),ofwhich8wereidentified

] as AGNs using severaltechniques that enable cross-checking. The other2 X-raydetectedSMGs have
O
levels of X-ray emission that can be plausibly explained by their star-formation activity. 6 of the 8
C SMG-AGNsaremoderately/highlyabsorbed,withN >1023cm−2. AnanalysisoftheAGNfraction,
H
. taking into account the spatial variation of X-ray sensitivity, yields an AGN fraction of 17+16% for
h −6
p AGNs with rest-frame 0.5–8 keV absorption-corrected luminosity > 7.8 1042 erg s−1; we provide
×
- estimatedAGN fractionsas a function ofX-rayflux andluminosity. ALMA’s highangularresolution
o also enables direct X-ray stacking at the precise positions of SMGs for the first time, and we found 4
r potential SMG-AGNs in our stacking sample.
t
s
a
[ 1. INTRODUCTION Over the past 15 yr, submillimeter (submm) and
millimeter surveys have discovered a population of
1
far-infrared (FIR) luminous, dust-enshrouded galax-
v
4 ies at z > 1 (e.g., Smail et al. 1997; Ivison et al.
6 1Department of Astronomy & Astrophysics, 525 Davey 1998, 2000; Coppin et al. 2006; Weiß et al. 2009;
3 Lab, The Pennsylvania State University, University Park, PA Austermann et al.2010). Multiwavelengthfollow-upob-
16802, USA; Send correspondence to [email protected] and
6 servations of these submm galaxies (e.g., Valiante et al.
[email protected]
. 2InstituteforGravitationandtheCosmos,ThePennsylvania 2007; Pope et al. 2008; Men´endez-Delmestre et al. 2007,
0
State University,UniversityPark,PA16802, USA 2009) have revealed that they are among the most lu-
1 3InstituteforComputationalCosmology,DurhamUniversity, minous objects in the Universe (e.g., Ivison et al. 2002;
3 SouthRoad,Durham,DH13LE,UK
1 4Max-Planck Institute for Astronomy, K¨onigstuhl 17, D- Chapman et al.2002;Kov´acs et al.2006),andthatthey
: 69117Heidelberg,Germany contribute significantly to the total cosmic star forma-
v 5Argelander-Institute of Astronomy, Bonn University, Auf tionaroundz 2(e.g.,Hughes et al.1998;Barger et al.
i demHu¨gel71,D-53121Bonn,Germany 1998; P´erez-G∼onza´lezet al. 2005; Aretxaga et al. 2007;
X 6The Johns Hopkins University, Homewood Campus, Balti-
Hopkins et al. 2010). These submm galaxies (SMGs)
more,MD21218,USA
ar 7NASA Goddard Space Flight Center, Code 662, Greenbelt, typically have infrared (IR) luminosities of 1012 L⊙
MD20771, USA orevengreater,andtheirstarformationrates∼(SFR)are
nia8,DIrevpianret,mCeAnt9o2f69P7h,yUsiScsA& Astronomy, Universityof Califor- estimatedto be 100–1000M⊙yr−1 (e.g., Kov´acs et al.
∼
9Key Laboratory for Research in Galaxies and Cosmology, 2006; Coppin et al. 2008; Magnelli et al. 2012). They
Center for Astrophysics, Department of Astronomy, University are massive galaxies with stellar mass M∗ 1011 M⊙
of Science and Technology of China, Chinese Academy of ∼
or greater (e.g., Borys et al. 2005; Xue et al. 2010;
Sciences, Hefei,Anhui230026, China
10Institute of Astronomy, Universityof Cambridge, Mading- Hainline et al.2011)andwithlargereservoirsofcoldgas
ley11RDoaedp,arCtmamenbtriodfgPehCyBsi3cs0aHnAd,AUtKmosphericScience,Dalhousie (&1010 M⊙; e.g., Bothwell et al. 2013).
Most commonly found around z 2–3, the
University,CoburgRoadHalifax,B3H4R2,Canada ∼
12Centre for Astrophysics, Science & Technology Research volume density of SMGs is 1000 times larger
∼
Institute, UniversityofHertfordshire,HatfieldAL109AB,UK (e.g., Chapman et al. 2003, 2005; Wardlow et al. 2011)
13Universit¨at Wien, Institute fu¨r Astrophysik, than that of the local ultraluminous infrared galax-
Tu¨rkenschanzstraße 17,1180Wien,Austria
14European Southern Observatory, Karl-Schwarzschild ies (ULIRGs), which are relatively rare in the local
Straße2,D-85748Garching,Germany universe (e.g., Sanders & Mirabel 1996; Lonsdale et al.
15Max-Planck-Institut fu¨r Radioastronomie, Auf dem Hu¨gel 2006). Also qualified as ULIRGs (LIR > 1012 L⊙;
69,D-53121Bonn,Germany
Sanders & Mirabel 1996), SMGs are often considered as
16LeidenObservatory,LeidenUniversity,POBox9513,NL-
2300RALeiden,TheNetherlands the “distantcousins”oflocalULIRGs,typically exhibit-
2
ingsimilarlyhighSFRandIRluminosity. However,they All focusing on X-ray AGNs, Alexander et al.
also differ in some important ways. The more strongly (2005a,b), Laird et al. (2010), Georgantopouloset al.
star-forming SMGs are not simply the “scaled up” ver- (2011), and Johnson et al. (2013) reported AGN frac-
sions of local ULIRGs — for example, it appears that tions among SMGs that are consistent with each other
the star formation in SMGs occurs on a larger scale within their 1σ error bars. The pioneering work by
within the galaxy instead of being concentrated at the Alexander et al. (2005a,b) studied the submm sources
corelikeforthelocalULIRGs(e.g.,Chapman et al.2004; discoveredbySCUBA(Holland et al.1999)intheChan-
Coppin et al. 2012). dra Deep Field North (CDF-N), which were matched
Believed to be the progenitors of large local ellip- to radio counterparts and spectroscopically identified
tical galaxies (e.g., Lilly et al. 1999; Smail et al. 2004; (Chapman et al. 2005). They estimated the X-ray AGN
Chapman et al.2005)andofteninvolvedinmergers(e.g., fraction among SMGs to be > 38+12%. Laird et al.
−10
Tacconi et al. 2008; Engel et al. 2010; Magnelli et al.
(2010), also using submm sources in the CDF-N but
2012), SMGs present a unique opportunity for study- with Spitzer IR counterparts identified by Pope et al.
ing the co-evolution of galaxies and their central su-
(2006), reported an X-ray AGN fraction of 29% 7%
permassive black holes (SMBHs; M > 106M⊙). The (or 20% if being conservative about AGN cla±ssifi-
cosmic star formation rate and active galactic nucleus
cation). Georgantopoulos et al. (2011) studied the
(AGN) activity both peak aroundz 2 (Connolly et al. submm sources in the Extended Chandra Deep Field
∼
1997; Merloni 2004; Hopkins et al. 2007; Cucciati et al.
South (E-CDF-S) detected by the LABOCA E-CDF-S
2012),andtheyappeartoberelatedassuggestedbythe
Submm Survey (LESS; Weiß et al. 2009), which were
observed correlations between the properties of central
matched to 2 Ms CDF-S (Luo et al. 2008) and 250 ks
SMBHs and their host galaxies (e.g., the M-σ and the E-CDF-S (Lehmer et al. 2005) sources and also Spitzer
M-Lrelation; Ferrarese & Merritt2000;Gebhardt et al.
MIPS sources (Magnelli et al. 2009), and they found
2000; Ha¨ring & Rix 2004; Gu¨ltekin et al. 2009). More-
an X-ray AGN fraction of 18 7% among the SMGs.
over,simulations of galaxy evolution and SMBH growth ±
Johnson et al. (2013) performed a direct matching be-
showthat mergerevents cantriggerboth star-formation
tween submm sources (detected at 1.1 mm by AzTEC;
activity and the onset of powerful AGN, with the peak
Wilson et al. 2008) and X-ray sources instead of first
of the AGN activity (possibly a quasar phase) coming
matching SMGs to IR or radio counterparts, and they
shortly after the peak epoch of star formation (e.g.,
found that, for SMGs in the CDF-S and CDF-N, the
Hopkins et al. 2008; Narayananet al. 2010). Observa-
AGN fraction is about 28%.
tionally, recent studies suggest that luminous AGNs are
Though previous studies were thorough with their
moreprevalentinmassivegalaxies(e.g.,Xue et al.2010;
statistical analyses on the reliability of counterpart
Mullaney et al. 2012) and star-forming galaxies (e.g.,
matching and used supplementary IR or radio cata-
Rafferty et al. 2011; Santini et al. 2012; Rosario et al.
logs, they were largely limited by the uncertainties
2013; Chen et al. 2013),andaveryhighfractionoflocal
in finding the true X-ray counterparts of the SMGs.
ULIRGs exhibit AGN activity as indicated by line-ratio
The submm source catalogs used in Alexander et al.
diagnostics (see the review by Alonso-Herrero 2013 and
(2005a,b), Laird et al. (2010), Georgantopouloset al.
references therein).
(2011), and Johnson et al. (2013) are all from single-
AGN activity in SMGs has been identified in
dishsubmmsurveys,whichhaveatypicalangularresolu-
previous studies through mid-IR spectroscopy tionof 10′′–20′′(e.g.,Chapman et al.2005;Weiß et al.
(e.g., Valiante et al. 2007; Pope et al. 2008; ∼
2009). This poses great challenges for matching submm
Men´endez-Delmestre et al. 2007, 2009; Coppin et al.
sources to the IR/radio/X-ray sources, especially when
2010) or X-ray (e.g., Alexander et al. 2005a,b;
multiple multiwavelength counterparts are found within
Pope et al. 2006; Laird et al. 2010; Lutz et al.
the largesearchapertures. Furthermore,alargefraction
2010; Georgantopoulos et al. 2011; Gilli et al. 2011;
of the single-dish detected submm sources are actually
Hill & Shanks 2011; Bielby et al. 2012; Johnson et al.
found to resolve into multiple sources, either physically
2013) observations. For moderate-to-high X-ray lu-
unrelated or due to the clustering of SMGs, when ob-
minosity AGNs, the X-ray emission is arguably the
served with higher angular resolution instruments such
best AGN indicator as the hard X-rays (rest-frame
astheSubmmArray(SMA)andtheAtacamaLargeMil-
energies of 2–30 keV) can penetrate through obscu-
limeter/submm Array (ALMA; e.g., Wang et al. 2011;
ration (NH .1024 cm−2) and also suffer less from Barger et al. 2012; Hodge et al. 2013).
host-galaxy contamination. However, for less X-ray In this paper, we presentthe X-rayproperties and the
luminous sources, the contribution from high mass AGN fraction of the SMGs in the E-CDF-S detected by
X-ray binaries (HMXBs) in the host galaxies cannot theALMALABOCAE-CDF-SSubmmSurvey(ALESS;
be neglected, especially for extreme starburst galaxies Hodge et al. 2013; Karim et al. 2013). The ALESS is an
like SMGs (e.g., Alexander et al. 2005b). The stud- ALMA Cycle 0 surveyat870µm to followup 122of the
ies of Alexander et al. (2005a,b), Pope et al. (2006), original 126 submm sources detected by LESS, which is
Laird et al. (2010), Georgantopouloset al. (2011), and the largest and the most homogeneous 870 µm survey
Johnson et al. (2013) have all found that SMGs have a to date (Weiß et al. 2009). With the exquisite angular
high X-ray detection rate, and a significant fraction of resolution and great sensitivity of ALMA ( 1.5′′ and
the X-ray detected SMGs are AGN-dominated in the 3 deeper than LESS; Hodge et al. 2013), A∼LESS pro-
X-ray band (though the exact fraction is under debate) v×ides the first fully submm-identified sample of SMGs
while some are consistent with the X-ray emission being based on a large, contiguous, and well-defined survey
powered purely by the starburst. (LESS), and this enables robust counterpart matching
3
at other wavelengths. Pairing with the powerful ALESS ing (likelihood-ratio matching) and summarizes the re-
catalog, we use the deep Chandra data in the E-CDF-S sults.
region(Lehmer et al.2005;L05),includingthemostsen-
sitive X-ray survey to date, the 4 Ms CDF-S survey 2.1. The Submm Catalog
(Xue et al.2011;X11). CombiningthepowerofChandra
We use the ALESS SMG catalog presented in
andALMA,wehaveunambiguouslyidentifiedthe X-ray
Hodge et al. (2013) (see also Karim et al. 2013) based
counterpartsbymatchingtheX-raysourcesdirectlyonto
on ALMA follow-up observations on the submm sources
the submm positions, which is the first among similar
detected by LESS (Weiß et al. 2009). The main-source
studies.
cataloginHodge et al.(2013)contains99SMGsthatare
TheAGNfractionsinSMGspresentedinthisworkare
within the primary beam of ALMA, with low axialratio
intheformofcumulativefractionsasafunctionofX-ray
(<2),lowRMS(<0.6mJy)andhighS/N(>3.5). This
flux/luminosity (i.e., the fraction of SMGs hosting AGN
catalogisthefirstfullysubmm-identified,statisticallyre-
withX-rayflux/luminositylargerthanorequaltoagiven
liable catalog of SMGs (Hodge et al. 2013; Karim et al.
value). Here we define an AGN as an accreting SMBH
2013).
with any level of X-ray luminosity. Identification of an
Figure 1 shows the positions of the 99 ALESS main-
AGNinsideanSMGdoesnotmeantheAGNisthemain
catalog SMGs and the combined X-ray exposure maps
power source of the SMG or contributes significantly to for both the Chandra 4 Ms CDF-S and 250 ks E-CDF-S
the galaxy’s energy budget. Though some SMGs are ingrayscale. 91ofthese99SMGsliewithintheChandra
quasar powered, much evidence has shown that in the
250ksE-CDF-Sregionand44inthe4MsCDF-Sregion.
majority of SMGs, star formation is the dominant en-
We identified 10 SMGs with X-ray counterparts (large
ergy source (e.g., Chapman et al. 2004; Alexander et al.
red dots), and below we detail the X-ray catalog used
2005b; Pope et al. 2006). SMGs with AGN signatures
and our matching method.
(e.g., in the X-ray or IR bands) are ULIRG-AGN com-
posites in terms of their spectral energy distributions 2.2. The X-ray Catalog
(SEDs). Since our cumulative AGN fraction is calcu-
The X-ray catalog used for matching to the ALESS
lated as a function of X-ray flux/luminosity, we focus
SMGsconsistsoftwocatalogs: onederivedfromthe4Ms
on the AGNs that dominate in the X-ray band because
CDF-S data, and the other from the 250 ks E-CDF-S
we can measure their X-ray luminosity reliably without
data. The CDF-S catalog includes (1) 776 CDF-S 4 Ms
disentanglingthe contributionfromhost-galaxystarfor-
main and supplementary catalog sources (X11); and (2)
mation.
116 additional sources from a WAVDETECT catalog with
The paper is structured as follows: we first describe
a false-positive probability threshold of < 10−5 (higher
our X-ray counterpart matching for the SMGs in Sec-
thanusedforselectingthemainandsupplementarycata-
tion2,andthenpresentouranalysesoftheirX-rayprop-
logs). The E-CDF-S catalog includes: (1) 795 E-CDF-S
ertiesandalsosomerelevantmultiwavelengthproperties
250 ks main and supplementary catalog sources (L05);
in Section 3. We have used several approaches to dis-
and(2)290additionalsourcesfromaWAVDETECTcatalog
tinguish the X-ray AGNs from the SMGs that are star
with a false-positive probability threshold of 10−5. The
formation dominated in the X-ray (Section 4). Then we
WAVDETECT catalogs were used by X11 and L05 as mas-
calculate the AGN fractionamong the SMGs for various
tercatalogs,fromwhichtheyfurtherselectedsourcesand
X-ray flux/luminosity limits (Section 5). Stacking anal-
derivedthe published 4 Ms CDF-S and 250 ks E-CDF-S
yses with the X-ray undetected SMGs are described in
catalogs, respectively. Despite their relatively lower sig-
Section 6. In Section 7 we compare with previous stud-
nificance,theadditionalsourcesfromtheWAVDETECTcat-
ies, discuss our results and outlines the possible future
alogsarelikely to be realX-raysourcesif identifiedwith
work.
submm counterparts, given the low density of SMGs on
Throughout the paper, we assume a ΛCDM cosmol-
ogy with H = 70.4 km s−1 Mpc−1, Ω = 0.27, and the sky and the excellent available positions. This has
0 m
enabled us to recover genuine X-ray counterparts to the
Ω =0.73(Komatsu et al.2011). Whenevergalaxystel-
Λ
SMGs down to a lower X-ray flux limit.
lar mass and SFR are involved, we assume a Salpeter
Duplicate sources that are in both the CDF-S and
initial mass function (IMF), and we have converted the
E-CDF-S X-ray catalogs were removed. For sources in
quantitiesquotedfromotherworkstobeconsistentwith
themainandsupplementarycatalogsofbothfields, X11
the Salpeter IMF whenever necessary. We use the con-
has noted all duplicate sources in their published 4 Ms
versionfactorofM (SalpeterIMF)=1.8 M (Kroupa
⋆ ⋆
× catalog; for the additional WAVDETECTsources, duplicate
or Chabrier IMF). We adopt a Galactic column density
of N = 8.8 1019 cm−2 for the line of sight to the sourceswereidentifiedbyperformingclosest-counterpart
H × matching between the two catalogs with a search radius
E-CDF-Sregion(e.g.,Stark et al.1992),andallreported
of 1.5′′. In total, our X-ray catalog contains 892 sources
X-ray quantities are corrected for Galactic extinction.
inthe4MsCDF-Sregion(with116fromtheWAVDETECT
2. MATCHINGX-RAYSOURCESANDSUBMMSOURCES lower-significancecatalog),and762sourcesinthe250ks
E-CDF-S region but not in the CDF-S (with 255 from
We first aim to find secure X-ray counterparts for the
the WAVDETECTlower-significancecatalog).
ALESS SMGs. Section 2.1 describes briefly the ALESS
submm catalog from Hodge et al. (2013). Section 2.2
2.3. Source Matching
describes the X-ray catalogs used for finding the X-ray
counterparts for the ALESS SMGs, which include ad- Weadoptedalikelihood-ratiomatchingmethodtofind
ditional sources beyond the L05 and X11 catalogs. Sec- secure X-ray counterparts for the ALESS SMGs (e.g.,
tion2.3containsourmethodologyforcounterpartmatch- Ciliegi et al. 2003; Luo et al. 2010). This method takes
4
into account the positional uncertainties for both cat- sificationprocess(Section4)andotherfollowingsections.
alogs, as well as the expected flux distribution of the We first detail the origin of the redshifts for the SMGs
counterparts. Briefly, we computed the likelihood ra- in Section 3.1. We then describe our analyses on the
tios,definedas the ratiobetweenthe probabilitiesofthe X-ray properties and present the results in Section 3.2:
SMG being the true counterpart and being just a back- firstforthemoredirectlyobservedquantities,Γ (effec-
eff
groundsource,forallSMGswithin5′′ofanX-raysource. tive photon index) and L0.5−8 keV (rest-frame apparent
Then we iterate to find a likelihood-ratio cut that max- luminosity),andthenforthederivedrest-frameintrinsic
imizes the sum of the matching completeness and relia- properties,Γ (intrinsicphotonindex), N (absorption
int H
bility (see Luo et al. 2010 for details). We found secure columndensity)andL0.5−8 keV,corr(absorptioncorrected
X-ray counterparts for 10 ALESS SMGs with a false- luminosity). In Section 3.3, we describe the origins of
matchprobabilityof3%(i.e.,anexpectednumberoffalse some selected multiwavelength properties that are rele-
matches of 0.3). The same 10 X-ray SMGs were recov- vant for this work.
eredwhenasimpleclosest-counterpartmatchingmethod
with a matching radius of 1.5′′ was adopted. 3.1. Redshifts
Figure 2 shows the histogramfor the positionaloffsets
ExceptforALESS45.1,allX-raydetectedSMGshave
betweenthe10SMGsandtheirX-raycounterparts. The
spectroscopic redshifts either from the redshift follow-
reddashedline is the estimatednumber offalse matches
up survey zLESS (Danielson et al. in prep.) or from the
as a function of the adopted matching radius for the
literature. The origins of the spectroscopic reshifts are
closest-counterpart matching method. The number of
listedin a footnote ofTable 2. ALESS45.1 has a photo-
false matches for a certain matching radius r was esti-
s metric redshift (photo-z) from Simpson et al. (in prep.),
matedbymanuallyshiftingtheX-raycatalogsinRAand
which is based on optical-NIR (with photometric data
Decby 10–60′′in10′′ incrementsandre-matchingwith
± fromMUSYC U, B, V, R, I, z, J, H, K, and VIMOS
the SMGs within r . Then the number of false matches
s U, HAWK-I J, TENIS J, K , and IRAC 3.6–8.0 µm)
for r is just the average number of matches for these s
s SED fitting using the code Hyperz (Bolzonella et al.
shifted catalogs. As shown in Figure 2, the number of
2000). The photo-z estimate for ALESS 45.1, z =
false matches is much smaller than the actual number
of X-ray matched SMGs at all distances 6 1.5′′ and is 2.34+−00..2667, is consistent with that from Xue et al. (2012)
only 0.3 at 1.5′′. The inset plot of Figure 2 shows the derived from optical-NIR SED fitting using ZEBRA
(Feldmann et al. 2006). The median redshift for the
histogram of offset/σ , where σ is the quadrature
pos pos X-ray detected SMGs is z =2.3.
sum of the positional error of each SMG and that of its
WheneverreshiftsareneededfortheX-rayundetected
matched X-ray source (i.e., qσs2ubmm+σX2−ray). There SMGs in the E-CDF-S, we adopt the photo-z values
is no SMG andX-ray sourcepair whose positionaloffset from Simpson et al. (in prep.). For the 91 SMGs in the
exceeds2σ . X-rayandsubmmthumbnailimageswith E-CDF-S, 77 have detections in > 3 wavebands and
pos
illustrated positional error bars are in Figure 3. thus have SED fits and photo-z estimates, with a me-
AsdiscussedinSection1,whenidentifyingX-raycoun- dian redshift of z = 2.3. For the remaining 14 SMGs
terparts for SMGs, previous studies had to invoke large with detections only in 0–3 wavebands, their redshifts
search radii and/or cross-identification with radio/IR aredrawnfromthelikelyredshiftdistributionsestimated
counterparts, which suffer from larger uncertainties and from simulations by Simpson et al. (in prep.). The me-
incompleteness (e.g., see Section 5.5 of Hodge et al. dianredshiftforsourceswithdetectionsin0/1waveband
2013). The X-ray counterparts of SMGs in our study (2/3 wavebands) is z 3.5 (z 4.5). Simpson et al.
∼ ∼
are of high robustness, and our estimated false-match (in prep.) estimated a median redshift of z = 2.5 0.2
±
probability is more reliable and realistic. Our matching fortheircompletesampleof96SMGs(3ofthe99ALESS
procedure does not require the assumption that sources SMGs only have IRAC coverageand are not included in
detectedinotherbandssuchasradioorIRareverylikely their sample).
tobephysicallyassociatedwithSMGs,whichisoftenas-
sumed by previousstudies as their searchradiifor coun- 3.2. X-ray Properties
terpart matching are large. Moreover, our matching re- We present the X-ray properties of the 10 X-ray de-
sultsarerobustagainsttheclustering/blendingofSMGs tected SMGs in this section. Our goal is to derive ba-
thanks to the fully-identified ALESS SMG catalog. sicquantitiesthatdescribetheir spectralcharacteristics,
The basic properties of the 10 X-ray detected SMGs suchastheintrinsicpower-lawphotonindexΓ andthe
int
are listed in Table 1. 8 of them are in the 4 Ms CDF-S intrinsic absorption column density (neutral Hydrogen
region, and 9 have spectroscopic redshifts. As shown in equivalent) N for each source, with the hope that they
H
Figure 4, their submm flux distribution (shaded blue) will help us understand the origin of the X-ray emission
does not appear to differ from the distribution for all (see the classification of the sources in Section 4). Their
SMGs (black solid line). We performed a Kolmogorov- X-ray spectral properties are summarized in Table 2.
Smirnov (K-S) test with these two distributions and the The X-ray spectral analyses were done using spectra
resultsuggeststhatthey sharethe sameparentdistribu- within the energy range 0.5 to 8 keV, following L05 and
tion, with p=0.39. X11. The spectraforsourceswithinthe 4 MsCDF-S re-
gion were extracted by X11 using ACIS Extract (AE;
3. PROPERTIESOFX-RAYDETECTEDSMGS Broos et al. 2010). The details of the AE run can
Inthissection,wedetailouranalysesandresultsonthe be found in X11. The spectra for the two sources
X-rayproperties of the X-ray detected SMGs, and other that are only in the E-CDF-S region, i.e. counter-
multiwavelengthpropertiesthatweuseintheAGNclas- parts for ALESS 66.1 and ALESS 67.1, were extracted
5
and combined for different epochs using the CIAO (ver- Galactic absorption, zwabs represents the rest-frame in-
sion 4.4.1; Fruscione et al. 2006) tools specextractand trinsic absorption(N being one ofits parameters),and
H
combine spectra. Their raw data were downloaded zpowisapower-lawmodel(withindexΓ )inthesource
int
fromtheChandraDataArchiveandwerereprocessedus- rest-frame.
ingtheCIAOtoolchandra repro. Thesourceextraction Among the 10 X-ray detected SMGs, 5 have full-
radii for them are twice the 90% encircled-energy aper- band net counts over 100 and therefore are qualified
ture radii at their off-axis angles, and the background for spectral fitting. We fitted the spectra of these 5
counts are estimated using 48 round regions around the sourceswithoutbinning,andweadoptedtheCashstatis-
source with similar or larger sizes to ensure good statis- tic (Cash 1979; cstat in XSPEC) for finding the best-
tical measurements of the background counts. fit parameters, which is well suited for fitting low-count
X-ray sources and does not require any spectral bin-
3.2.1. Hardness Ratio, Effective Photon Index Γeff, and ning (Nousek & Shue 1989). Figure 5 shows the spec-
Rest-Frame 0.5–8.0 keV Apparent Luminosity
tra of the 5 sources with full-band net counts >100,
L
0.5−8 keV ALESS11.1,57.1,66.1,84.1,and114.2,with their best-
We first derive two simple and direct spectral char- fit wabs*zwabs*zpowmodels, and the inset figures show
acteristics: the hardness ratio, defined as the ratio of the 68.3%, 90% and 99% confidence contours for Γ
int
the photon count rates in the hard band (2–8 keV) vs. N . For ALESS 66.1, the plotted best-fit model is
H
and the soft band (0.5–2 keV), and the effective pho- wabs*zpow,since its spectral fitting indicates no signifi-
ton index, Γ , for a power-law model with Galactic ab- cantevidenceforabsorption,asillustratedbyitsΓ -N
eff int H
sorption. The hardness ratios were derived using the contours. ALESS114.2doesnothavehighphotoncounts
Bayesian Estimation of Hardness Ratios (BEHR) pack- (126netcounts inthe fullband)andexhibitshighback-
age by Park et al. (2006). This package computes the groundduetoitslargeoff-axisangleintheCDF-S(>9′).
Bayesian posterior distribution for the hardness ratios Fixing its intrinsic photon index Γ at 1.8 (following
int
without requiring detections in both energy bands, and X11;for typicalAGNs) givesN 2.4+5.9 1023 cm−2.
itisespeciallyusefulforcaseswithlowphotoncounts(5 H ≈ −1.2×
The best-fit Γ and N values (and 90%CI errorbars)
of the X-ray counterparts have less than 100 net photon int H
for these 5 sources are listed in Table 2 (90% CI upper
counts in the 0.5–8.0 keV full band). The median of the
limit for the N of ALESS 66.1).
H
posteriordistributionis takenas the best-estimate value
We have also fitted the 4 obscured sources with
forthehardnessratio,andtheerrorbarsreportedinTa-
>100 full-band net counts (ALESS 11.1, 57.1, 84.1
ble 2 are the 68.3% (“1σ”) posterior confidence interval
and 114.2) with a model including an Fe Kα line,
(CI). When the hardness ratio(or its inverse)has a pos-
(zpow*zwabs+zgau)*wabs. We fixedthe rest-frameline
terior median of essentially zero (< 0.01), we adopt the
energy at 6.4 keV and width at 0.1 keV and only fitted
upper(orlower)limitvaluedefinedbythe90%posterior
for the normalization (line strength). We then calcu-
CI.
lated the equivalent width (XSPEC command eqw) and
The effective photon index Γ is then derived from
eff its 90% CI (using Markov chain Monte Carlo with the
the hardness ratio following the methods described in
chain command). We evaluated if the model including
L05 and X11. The error bars on Γ are estimated by
eff the Fe Kαline is statisticallya better modelbycomput-
converting all hardness ratios in the Bayesian posterior
ing the Bayesian Information Criterion (BIC) and com-
distribution into corresponding Γ values then taking
eff pareditwiththeBICofthemodelwithouttheFeKαline
the68.3%CI,aslistedinTable2. AsΓ valueswerede-
eff (wabs*zwabs*zpow). Briefly,BIC=C+p lnn,whereC
rivedfromhardnessratiosandarelessdirectlyrelatedto ·
istheCashstatistic,pisthenumberoffreeparametersin
theobservedquantities,theyarehardertoconstrainand
the model, andn is the number of data points in the fit.
therefore, following L05 and X11, for sources having low
The model with a smaller BIC value is the statistically
counts (see L05 and X11 for definitions) in the soft (or
preferred model (see Section 3.7.3 of Feigelson & Babu
hard)band,weadoptedthe90%CIupper(orlower)lim-
2012). ForALESS11.1,57.1,and114.2,themodelwith-
itsforΓ . Forsourceswithlowcountsinbothbands,we
eff out the Fe Kα line is favored, and they have rest-frame
fixedΓ to1.4(followingX11). UsingΓ ,redshift,and
eff eff equivalent widths consistent with 0 keV within 90% CI.
theobservedfull-bandfluxf0.5−8 keV aslistedinTable1, The 90% CI upper limits on the equivalent widths for
wederivedthe rest-frame0.5–8.0 keVapparentluminos-
ALESS11.1,57.1,and114.2are0.15keV,0.67keV,and
ity (with no intrinsic absorption correction; denoted as
0.52 keV, respectively. For ALESS 84.1, however, the
L0.5−8 keVthroughoutthispaper),foreachsourcefollow- model with the Fe Kα line is slightly favored (BIC val-
ingtheequationL0.5−8 keV =4πd2Lf0.5−8 keV(1+z)Γeff−2 ues being 494 vs. 496 for the model without the line),
(e.g., X11). and the best-fit rest-frame equivalent width is 1.17 keV,
witha90%CIof0.23–2.15keV.Sincethemodelwiththe
3.2.2. Intrinsic Photon Index Γ , Intrinsic Absorption
Column Density N , andinRtest-Frame 0.5–8.0 keV FeKαlineisonlyslightlyfavoredforonesource,ALESS
Absorption-CorrecHted Luminosity L 84.1, for simplicity and comparison purposes, we report
0.5−8 keV,corr
thespectralanalysisresultsusingthemodelwithoutthe
Wethenestimatedtheintrinsicphotonindex,Γ ,the
int Fe Kα line component for all sources.
intrinsic absorption column density, N , and the rest-
H For the 5 sources with full-band net counts fewer than
frame 0.5–8.0 keV absorption-corrected luminosity (de-
100, we estimated their N values by running simula-
H
noted as L0.5−8 keV,corr throughout this paper), for each tions in XSPEC using the wabs*zwabs*zpow model with
source. We used XSPEC (Arnaud 1996) for spectral fit-
fixed Γ = 1.8 and varying N until it reproduced the
int H
ting and modeling. The basic model we adopted was
observedhardnessratio(X11). Forthese5sources,spec-
wabs*zwabs*zpow in XSPEC, where wabs represents the
6
tral fittings does not provide more constraints on the The SEDs were fitted using Spitzer, Herschel, ALMA,
X-ray properties than the simple method adopted here. and VLA photometry at 3.6 µm, 4.5 µm, 5.8 µm,
An illustration of this method is in Figure 6 (similar to 8.0 µm, 24 µm, 250 µm, 350 µm, 500 µm, 870 µm and
Figure3inAlexander et al.2005b). TheN valuesesti- 1.4 GHz. The SED templates include the star-forming
H
matedthiswayarelistedinTable2andaredistinguished galaxytemplatesfromChary & Elbaz(2001)andthatof
from the ones derived from spectral fitting by having no SMMJ2135 0102 (the Eyelash galaxy; Swinbank et al.
−
error bars. For ALESS 45.1 and 67.1, as their hardness 2010).
ratiosweregivenas90%upperlimitsdue tolackofpho- The stellar masses for the SMGs are from
tonsinthehardband,theirN valuesaretherefore90% Simpson et al. (in prep.). Briefly, their stellar mass
H
upper limits as well. estimates are derived from the absolute H-band pho-
With the best-fit or estimated Γ and N tometry based on the optical-NIR SED fitting and
int H
values for each source, we then estimated the a mass-to-light ratio based on the best-fit star for-
rest-frame 0.5–8.0 keV absorption-corrected luminosity, mation history (either burst or constant) and stellar
L0.5−8 keV,corr, following Section 4.4 of X11, by first de- population synthesis models from Bruzual & Charlot
rivingtheintrinsicfull-bandfluxf0.5−8 keV,corr usingthe (2003). Typical error bars for M⋆ are about a factor of
wabs*zwabs*zpow model with Γ , N , and redshift, 2 (around a factor of 3–5 if taking into account model
int H
∼
and then calculating L0.5−8 keV,corr using the equation uncertainties).
L0.5−8 keV,corr = 4πd2Lf0.5−8 keV,corr(1 +z)Γint−2. This Although Simpson et al. (in prep.) only used galaxy
is corrected for both Galactic and intrinsic absorption. templates in their SED fitting, AGN contamination
Again, L0.5−8 keV,corr estimates are90%upper limits for is probably not a concern here when estimating stel-
ALESS 45.1 and 67.1 just as for their hardness ratios lar mass. Our X-ray detected SMGs have a median
and NH values. As noted by X11, L0.5−8 keV,corr val- logL0.5−8 keV = 43.0 and all but ALESS 66.1 have
ues estimated for the lowercountsourcestypically agree logL0.5−8 keV 643.7 (Table 2), which is the upper-limit
within 30% compared with those from direct spec- cut chosen by Xue et al. (2010) to minimize potential
tral fitt∼ing, but could potentially be subject to larger AGNcontaminationintheoptical-NIRbands. InSection
uncertainties since spectral components such as reflec- 4.6.3 of Xue et al. (2010), they studied 188 AGNs with
tion and scattering can play an important role for heav- 41.96logL0.5−8 keV 643.7andexaminedtheAGNcon-
ily obscured sources. This could also be true for the 5 tributiontotheirbest-fitSEDtemplates,thecorrelation
sources with spectral fits, but the precision should be oftheirrest-frameabsolutemagnitudes/colorsandX-ray
sufficient for the purposes of our study. For example, luminosities, and their fractions of optical-NIR emission
ALESS 73.1 is the known heavily obscured source re- coming from the core regions versus from the extended
ported by Coppin et al. (2010) and Gilli et al. (2011), regions. They concluded that the AGN contamination
who estimated L2−10keV 2.5 1044 erg s−1 — a bit is minimal and does not affect the optical-NIR colors or
largerthan but in agreem≈entwit×h our estimate within a themassestimatesinasignificantway. Wenotethat,as
factor of two. showninFigure9,wedonotseeanycorrelationbetween
f0.5−8 keV and the IRAC 3.6 µm magnitude/flux of the
3.3. Multiwavelength Properties 9 SMGs with logL0.5−8 keV 6 43.7, consistent with the
findings of Xue et al. (2010).
For the classificationof AGNs amongSMGs described
Also, Simpson et al. (in prep.) noted that only 3
in the next section and also for the purpose of discus-
(ALESS57.1,66.1,75.1)outof77SMGs haveχ2 >10
sion,weneed the rest-frame1.4 GHz monochromaticlu- red
due to 8 µm excesses indicative of AGN activity. For
minosity (L ), the rest-frame 8–1000 µm IR lumi-
1.4GHz ALESS 57.1, the 8 µm excess feature is consistent with
nosity (L ) and the 40–120µm FIR luminosity (L ),
IR FIR the fact that it is an obscured AGN. Because of its low
the stellar masses (M ), as well as the SFR. As we used
differentmethodstod⋆eriveSFRsindifferentAGNclassi- L0.5−8 keV value and non-power-law spectral shape, we
do not consider that ALESS 57.1 is dominated by AGN
ficationschemes, the SFR estimates are described in the
in the optical-NIR, and we take the stellar mass esti-
relevantparagraphsinSection4.2. Themultiwavelength
mate as reliable but caution the reader with this caveat.
properties are listed in Table 3.
Since ALESS 66.1 is a known optical quasar with high
The rest-frame 1.4 GHz monochromatic luminosity,
L , is calculated following Alexander et al. (2003): L0.5−8 keV andithastheworstSEDfitamongallsources
1.4GHz inSimpson et al.(in prep.),we takeits estimatedstellar
L =4πd2f 10−36(1+z)α−1, (1) massaslessreliableandlabelitdifferentlyintherelevant
1.4GHz L 1.4GHz
plots involving M .
⋆
whereL isinWHz−1,the observed1.4GHzradio
fluxf 1.4GHiszinµJy,andtheradiospectralindexisα= 4. CLASSIFICATIONSFORTHEX-RAYDETECTEDSMGS
1.4GHz
0.8 (following A05). The radio counterparts and radio Inthissection,weclassifythe10X-raydetectedSMGs
fluxes of the X-ray detected SMGs are from the catalog to assess if their X-ray emission reveals the existence of
of Biggs et al. (2011) (based on Miller et al. 2008 VLA AGNs or if they are dominated by star formation in the
maps), identified using a closest-counterpart matching X-rayregime. Todoso,weexploittheirX-rayproperties
method with r =1′′ (chosen to have a false match rate calculated in Section 3.2 as well as other characteristics
s
<1% and also verified by visual examination; 39 SMGs derived from their multiwavelength data (Section 3.3).
are matched with radio sources). We employedseveralindependent classificationmethods
Therest-frameIR(8–1000µm)luminosityL andFIR and cross-checked between them. These methods and
IR
(40–120 µm) luminosity were derived based on NIR- the derivation of the relevant multiwavelength proper-
through-radioSED fitting by Swinbank et al. (in prep.). ties used for each method are described in each of the
7
subsections. Table 4 is a summary of the classification not the case for ALESS 73.1). We note here that the 6
methods we adopted, and Table 3 lists multiwavelength sources classified as AGNs under this criterion all have
properties of the X-ray detected SMGs and the classifi- L0.5−8 keV valueslargerthan3 1042ergs−1. Therefore,
cationresults. Someofthesemethods arecloselyrelated if we are conservative and requ×ire L0.5−8 keV >3 1042
(Method IIIa, IIIb, and IV), but we have employed all ergs−1 (ratherthanL0.5−8 keV,corr >3 1042ergs×−1)as
×
to enable cross-checkbetween the results. we are dealing with high-redshift star-forming galaxies,
the conclusion would still be the same.
4.1. Method I & II. Γ and X-ray Luminosity
eff
Classification Method I. Γeff: Following 4.2. Method III. L0.5−8 keV vs. SFR
Alexander et al. (2005b) (A05 hereafter) and X11,
Thegeneralideaofthismethodistocomparetherest-
we classify sources with Γ < 1.0 as AGNs (see
eff frame 0.5–8.0 keV apparent luminosity, L0.5−8 keV, with
Figure 7). This hard signature of the X-ray spectrum
the predicted amount as expected from the level of star
is a feature of absorbed AGNs, as spectra having
formation, LX,SF. If L0.5−8 keV of an SMG is 5 or
Γ < 1.0 are empirically hard to explain with just the ×
eff more than that expected from its star formation (i.e.,
star forming component in a galaxy, which typically L0.5−8 keV > 5 LX,SF), it is classified as an AGN host
has Γ 1.5 or even softer (e.g., Teng et al. 2005; ×
eff (similar to the criterion adopted by A05 and X11). As
∼
Lehmer et al. 2008). ALESS 17.1, 57.1, 84.1, and 114.2
detailedbelow,weestimatedL withtwoapproaches,
X,SF
are classified as (obscured) AGNs under this criterion.
and they give consistent classification results.
As we adopted conservative Γeff estimates for sources ClassificationMethodIIIaistocompareL0.5−8 keV
with relatively low counts, some of the sources appear
againstthe rest-frame1.4 GHz monochromaticluminos-
softer than indicated by Figure 6 because their Γ
eff ity,L ,fromwhichwederivedastarformationrate
1.4GHz
values are upper limits or are fixed to 1.4 (e.g., ALESS
(SFR) and L . Figure 8a illustrates this classifica-
X,SF
73.1). For the calculation of Γ and the error bars and
eff tion scheme. SMGs above the solid line (L0.5−8 keV >
upper limits, see Section 3.2.
5 L ) are classified as hosting AGNs. The calcu-
X,SF
Classification Method II. X-ray Luminosity: ×
lation of L is described in Section 3.3. L is
1.4GHz X,SF
Following the criterion adopted in, e.g., Bauer et al.
derived using a correlation between L and SFR
1.4GHz
(2004), Lehmer et al. (2008), and X11, we classify a
forpure starburstsorhigh-redshiftstar-forminggalaxies
source with rest-frame 0.5–8.0 keV absorption-corrected
(Equation 11 of Persic & Rephaeli 2007, PR7 for short,
luminosity L0.5−8 keV,corr larger than 3 1042 erg s−1 and references therein), and then converting SFR into
×
as an AGN host. This is based on studies of local
L following Lehmer et al. (2010) (L10 hereafter).
X,SF
galaxies which found that all local star-forming galax-
The equations used are
ies have lower X-ray luminosities than 3 1042 erg s−1
(e.g., Zezas et al. 2001; Ranalli et al. 200×3; L10). The SFR=L /8.93 1020; (2)
1.4GHz
×
caveat is that the SMGs are high-redshift star-forming log(L /1.21)=39.49+0.74log(SFR/1.8), (3)
X,SF
galaxies, and it is uncertain whether the criterion of
L0.5−8 keV,corr >3 1042 establishedusinglocalgalaxies where SFR is in M⊙ yr−1 and L1.4GHz is in W Hz−1.
×
and AGNs would apply to these high-redshift sources. The factor 1.8 division of the SFR is for converting
This is why we have additional classification methods the Salpeter IMF adopted in this paper to the Kroupa
(see the following sections) to ensure a reliable identifi- IMF used by L10. The factor 1.21 is for converting the
cation of AGNs. rest-frame2–10keVluminosityL2−10keV adoptedinL10
Figure 7 illustrates this method, with the y-axis be- intothe 0.5–8.0 keVluminosity L0.5−8 keV inthis paper,
ingL0.5−8 keV,corr. Filledcirclesmarkthe L0.5−8 keV,corr and it was derived with XSPEC using a simple power-law
values, while open circles are L0.5−8 keV values. Crosses modelwithΓ=1.5(e.g.,Teng et al.2005;Lehmer et al.
mark the L0.5−8 keV (or L0.5−8 keV,corr) values with 2008). Though there might be large scatter in con-
larger uncertainties due to fixed Γeff (or Γint) as a re- versions from L1.4GHz to SFR and SFR to L0.5−8 keV,
sult of having low counts. For ALESS 45.1, 67.1 and Figure 8a clearly shows that our adopted threshold of
70.1 (with crossesin open circles), Γeff values are poorly L0.5−8 keV >5 LX,SF appears to be sufficient for iden-
×
constrained due to low counts in both X-ray bands, tifying outliers in the correlation between L and
1.4GHz
and thus Γeff = 1.4 is assumed (following X11), which L0.5−8 keV. Method IIIa classifies all but ALESS 17.1,
means their L0.5−8 keV values have larger uncertainties. 45.1 and 67.1 as AGN hosts. The results are discussed
For sources with crosses in the filled circles, their Γ at the end of this subsection.
int
values were fixed at 1.8 since they did not qualify for It is notable that all of our X-ray detected SMGs are
spectral fitting due to low counts, and therefore their also radio detected, whereas among all of the 99 SMGs
L0.5−8 keV,corr values have larger uncertainties. Arrows in the ALESS main catalog, only 39 SMGs are matched
on the L0.5−8 keV,corr of ALESS 45.1 and 67.1 indicate with a radio source within 1′′ using the Biggs et al.
that these are upper limits, because their hardness ra- (2011) radio catalog. This is perhaps not surprising,
tios and N values were given as 90% CI upper limits since X-ray luminosity correlates with radio luminos-
H
(see Fig. 6 and Table 2). ity for starburst galaxies like SMGs (e.g., Schmitt et al.
AllsourcesotherthanALESS17.1,70.1,67.1,and45.1 2006; PR07) and also the K correction is similar in the
areclassifiedasAGNs undermethodII;thefoursources X-ray and radio bands compared to submm. Further-
were not classified as AGNs because their L0.5−8 keV more, since AGNs can also contribute significantly in
and/or L0.5−8 keV,corr were relatively poorly constrained the radio regime (e.g., Roy & Norris 1997; Donley et al.
(crossesin Figure7) andthe well-constrainedL0.5−8 keV 2005;Del Moro et al.2013),itisplausiblethattheX-ray
value was not above our threshold (for ALESS 17.1, but detectedSMGsaregenerallybrighterinradiobecauseof
8
the extra radio flux contribution from AGNs. If this is µm band (rest-frame J band for a z = 2 source) to re-
true, then the SFR derived based on radio fluxes are place the R band, and utilized the classifications of the
over-estimated,whichmeans that L values areover- X11 4 Ms CDF-S sources to calibrate the classification
X,SF
estimated. Since classification method IIIa would only thresholdforlog(f0.5−8 keV/f3.6µm). Thef3.6µm datafor
become more conservative due to this effect, we do not the X11 X-ray sources were compiled by X11 based on
attempttocorrectfortheAGNcontributionintheradio the Spitzer SIMPLE catalog (Damen et al. 2011), and
band. the f data for our X-ray detected SMGs are from
3.6µm
Classification Method IIIb uses less direct but Simpson et al. (in prep.).
tighter correlations to derive SFR and L (see Fig- Figure 9 illustrates this approach: all the X11 sources
X,SF
ure 8b. The SFRs were derived following Equation 3 in having IRAC 3.6 µm detections (643 sources; see de-
Kennicutt (1998), tails in Section 4.4 of X11) are plotted here with sym-
bols representing their classifications — AGNs as red
SFR=1.8 10−10 LIR/L⊙, (4) dots and galaxies as green open squares. The classifi-
× ×
cations are the same as in X11 with only one exception:
where SFR is in M⊙ yr−1 and the solar luminosity the 52 sources that have z > 1 and were classified as
L⊙ = 3.9×1033 erg s−1. We use the LIR derived by ‘AGN’ only under the log(f0.5−8 keV/fR)> 1 criterion
Swinbank et al. (in prep.) as described in Section 3.3. −
in X11 are conservatively taken as galaxies here (hence
We then derived L from SFRs following L10:
X,SF the‘AGN ’and‘Galaxy+’labelsinthelegend). Thisis
−
for the purpose of calibrating our X-ray-to–optical/NIR
L =0.67 (9.05 1028 M +1.62 1039 SFR), (5)
X,SF × × · ⋆ × · flux ratio threshold in a reliable and conservative way.
where SFR is in M⊙ yr−1 and the galaxy stellar mass, Abalsnod,iafnadstohuurcseoinsloynhlyasdaetnecutpepdeirnltimheitsofoftrotrhheafrudll-Xb-arnady
M ,isinsolarmasses(seeSection3.3). ForSMGswhich
ha⋆ve very large SFRs, the contribution from the term flux f0.5−8 keV in X11, we used its soft- or hard-band
9.05 1028 M is negligible (<1%) except for a couple flux for our calibration (i.e. a lower limit of f0.5−8 keV).
of so×urces. ·Th⋆e factor 0.67 is for converting L2−10 keV Wabeovtehewnhliocohkeadmfoarjotrhietyloogf(fth0.e5−X81k1eVs/ofu3r.c6µesm)artehrAesGhNolsd.
into L0.5−8 keV (×1.21), and for converting M⋆ as L10 For this purpose and to avoid fine-tuning, we chose
adopted the Kroupa IMF (/1.8). This correlation has a
smaller scatter than the correlation adopted in Method log(f0.5−8 keV/f3.6µm)> −1 as our classification thresh-
old (i.e., below the solid line in Figure 9), as 95% of
IIIa (see Table 4 in L10). Method IIIb gives consistent ∼
the X11 sources that satisfy this criterion are AGNs.
classification results as IIIa, i.e., all but ALESS 17.1,
Under this criterion, all but ALESS 17.1, 45.1 and
45.1 and 67.1 are classified as AGN hosts.
67.1 are classified as AGN hosts, consistent with
To summarize the classification results under Method
Method III. We note here that the somewhat conserva-
III, all but ALESS 17.1, 45.1 and 67.1 are classified as
AGN hosts consistently under two different ways of cal- tive log(f0.5−8 keV/f3.6µm) thresholdwe chosegiveshigh
reliability(1 95%=5%mis-classificationrate)butrel-
culating SFR and LX,SF. For the case of ALESS 17.1, atively low c−ompleteness, as 26% of the X11 AGNs in
however,we argue that it probably hosts an AGN based
onitsX-rayspectralhardness(seeSection4.1andFig.7) Figure 9 actually have log(f0.5−8 keV/f3.6µm)<−1.
and the results of classification Method IIIb. The ap-
4.4. Method V. X-ray Variability
parent X-ray ‘deficit’ shown in Figure 8a is likely due
to radiocontributionfromits AGN, andpotentially also X-ray variability is one of the distinct characteristics
combined with the effect of gas and dust obscuration, of AGNs, and it is an especially powerful tool for iden-
which is consistent with its low Γ and its high hard- tifying highly obscured or low-luminosity AGNs where
eff
ness ratio and N value (see Fig. 6, 7, and Table 2). thegalaxylightmaydominateeveninthe X-rayregime.
H
The absorption-corrected luminosity L0.5−8 keV,corr for Young et al.(2012)studied92X-raygalaxiesinthe4Ms
ALESS17.1wouldputitabovethe classificationthresh- CDF-S, and found 20 X-ray variable galaxies that are
oldofMethodIIIa(thesameappliesforALESS67.1for likely to host low-luminosity AGNs. Here we follow the
both Method IIIa, b, but its L0.5−8 keV,corr is an upper method in Section 3 of Young et al. (2012) and examine
limit). For similar reasons, this classification criterion the variability of the 8 sources in the CDF-S.17
does not rule outthe possibility ofALESS 45.1and67.1 First,wedividedthe CDF-Sobservationsinto four 1
∼
hosting AGNs, though their X-ray spectral hardnesses Msepochs,whichcangivetheamountofvariabilityeach
and luminosities are consistent with them being domi- source exhibits on month–year time scales in the ob-
∼
nated by just star-formationactivity in the X-ray band. served frame. Then through Monte Carlo simulations,
wecalculatedthe probabilityP thattheir variabilityex-
4.3. Method IV. f3.6µm vs. f0.5−8 keV ceeds that expected from Poisson statistics. If P is less
than 5%, we conclude that the source is variable.
We also use X-ray–to–optical/NIR flux ratio as
We found that ALESS 84.1 is X-ray variable (P =
an AGN activity indicator. In X11, sources with
0.025), and it has a maximum-to-minimum flux ratio of
log(f0.5−8 keV/fR) > 1 are classified as AGNs 3.1overtheobserved10.8yrtimeframe. Wewereunable
−
(e.g., Maccacaro et al. 1988; Hornschemeier et al. 2001;
toconcludesimilarlyforanyoftheother7sourcesinthe
Bauer et al.2004). However,aswearedealingwithhigh-
CDF-S. Thisdoesnotruleoutthepossibilityofthembe-
redshift (z > 1) SMGs whose observed R band corre-
sponds to the rest-frame UV band, it is more appro- 17 For ALESS 66.1 and 70.1, which are only in the E-CDF-S
priate to use a redder band to trace the stellar com- region, the total exposure time and time span for the E-CDF-S
ponent. We therefore chose the IRAC Channel 1 3.6 observationsarenotsufficientlylongforsuchstudies.
9
ing truly variable sources, since these sources have large X-ray flux listed in Table 1). Therefore, simply dividing
off-axis angles (high backgroundcounts) and/or low net the numbers of SMG-AGNs and SMGs wouldbias f
AGN
source counts (see Table 2), which give us relatively low toward either larger or smaller values depending on the
statistical power when testing their variability. spatialdistributionofSMGsonaninhomogeneousX-ray
sensitivity map.
4.5. Summary of Classification Results The AGN fraction, f , we estimated here is the
AGN
flux–orluminosity–dependentcumulativefraction. That
Combining the results of all the methods described
is, the f value for a certain X-ray flux/luminosity
above,8outofthe 10X-raydetectedSMGsshowstrong AGN
represents the fraction of SMGs that host AGNs with
evidence of containing AGNs under at least one clas-
equal or greater X-ray flux/luminosity. We first de-
sification method (see Table 4). In fact, 6 of these 8
scribe how f is calculated as a function of full-band
sources are classified as AGN hosts consistently under AGN
at least 3 methods. The only exception is ALESS 17.1, observed-frameX-rayflux(f0.5−8 keV;fX forshort). Fol-
lowingSilverman et al.(2008),thecumulativeAGNfrac-
whichisidentifiedasanAGNhostonlythroughitsspec-
tionforSMGshostingAGNswithX-rayfluxlargerthan
tral hardness (Method II) because its X-ray luminosity
or equal to f is
(L0.5−8 keV andL0.5−8 keV,corr)appearstobe lowdueto X,lim
heavy intrinsic obscuration (N > 1023 cm−2; see dis-
H N
1
cussion in Section 4.2). f (f >f )= , (6)
WedidnotfindstrongAGNsignaturesintheX-rayfor AGN X X,lim XN
SMG,i
i=1
ALESS45.1or67.1usinganyofthemethods,thoughitis
notable that their L0.5−8 keV values exceedthe expected whereN isthenumberofSMG-AGNswithfX,i >fX,lim
LX,SFbyafactorof∼3(seeSection7.3formoredetails). (fX,i being the X-ray flux of the ith SMG-AGN); and
Theyaretreatedasstarburstdominatedsystemsinstead N representsthenumberofSMGswhichlieinare-
SMG,i
of ‘X-ray AGNs’ in our AGN fraction (fAGN) analysis gion with a sufficient X-ray sensitivity limit such that
in the next section. We note here that our classifica- they would have been detected if hosting AGNs with
tion does not rule out the possibility of ALESS 45.1 or f > f . Since we have 8 SMG-AGNs in the sam-
X X,i
67.1 hosting AGNs. Future panchromatic SED analysis ple, naturally, we chose the f values to be each of
X,lim
and/or spectroscopic study may reveal hidden or weak the 8 f values in turn. The error for f is calcu-
X,i AGN
AGNs, or even AGNs in the X-ray undetected SMGs. latedbyaddinginquadraturetheerrorforeach1/N
SMG,i
For example, ALESS 45.1 actually lies within the ‘Don- term (estimated following Gehrels 1986). The results of
ley wedge’ (Donley et al. 2012), which is an AGN selec- f (f )areplottedintheupper-leftpanelofFigure10.
AGN X
tionschemeusingthefourIRACbandsandisrobustfor The f values as a function of rest-frame
AGN
screening out high-redshift starburst galaxies (as com- 0.5–8.0 keV apparent luminosity, L0.5−8 keV (LX for
pared to the ‘Lacy wedge’ or ‘Stern wedge’; Lacy et al. short), or rest-frame 0.5–8.0 keV absorption-corrected
2004, 2007; Stern et al. 2005). However, such analyses luminosity, L0.5−8 keV,corr (LX,corr for short), canbe cal-
are beyond the scope of our paper as we focus on the culated likewise. An additional step arises when deter-
X-ray properties of the SMGs. See more discussion in mining whether or notthe jth X-ray undetected SMG at
Section 7.3. redshift z would have been detected if hosting an AGN
j
with L or L . This additional step is to trans-
5. THEAGNFRACTIONINSUBMMGALAXIES lateL X,iorL X,corri,intothe(hypothetical)observedflux
X,i X,corr,i
As concluded in the previous section, we classified 8 (i.e., the same metric as the X-ray sensitivity map), as-
of the 10 X-ray detected SMGs as AGN hosts (SMG- suming an AGN with L or L in the jth X-ray
X,i X,corr,i
AGNs). That is, among the 91 ALESS SMGs in the undetected SMG.
E-CDF-S region, 8 show X-ray signatures of AGNs. As For translating L into flux, we used the equation
X,i
mentioned in Section 1, this does not mean the pri- for calculating L0.5−8 keV in Section 3.2 with the red-
maryenergysourcesof8SMGsareAGNs,thoughAGNs shift z and the Γ value of the ith SMG-AGN, Γ .
j eff eff,i
do dominate in the X-ray band (see discussion in Sec- For translating L into flux, we first calculated
X,corr,i
tion 7.2.2). In this section, we calculate the fraction of its corresponding rest-frame apparent luminosity (ab-
such‘X-rayAGNs’amongthispopulationofSMGsinthe sorbed), L′ , for the hypothetical AGN with L
X,i,j X,corr,i
E-CDF-S. We refer to this fraction simply as the AGN
hosted by the jth X-ray undetected SMG. Such a con-
fraction, f , hereafter. Fora comparisonbetweenour
AGN version requires an intrinsic column density (N ) as-
f results and the previous ones, see Section 7.1.1. H,j
AGN signed to each X-ray undetected SMG. To do so, we
drew each N value randomly from the N distri-
5.1. Methodology H,j H
bution described in Section 4 of Rafferty et al. (2011),
We follow the method discussed in Section 3.1 of whichwasformulatedbasedontheresultsinTozzi et al.
Lehmer et al. (2007) and Section 5 of Silverman et al. (2006).18 We then found L′ by calculating the ratio
X,i,j
(2008) to calculate fAGN, which takes into account L /L′ through a simulation in XSPEC using the
the spatial inhomogeneity of the X-ray sensitivity limit X,corr,i X,i,j
wabs*zwabs*zpow model with z , N , and the intrin-
across the E-CDF-S. To illustrate why such a consider- j H,j
ation is necessary, suppose that, by chance, all ALESS
SMGs lay in the least sensitive region in the E-CDF-S bu1t8ionBrfioerfly2,0th<is NloHgNdHis/tr(cibmu−ti2o)n <incl2u3dewshaichlogc-ennotremrsaladroisutnrid-
(the lightest gray in Figure 1, with X-ray sensitivity logNH/(cm−2) = 23.1 with σ 1.1. It flattens out beyond
> 10−15 erg cm−2 s−1). Then only 5 SMGs would 1023cm−2 andtruncates at1024cm≃−2,anditincludes 10% ofob-
havebeenX-raydetectedandclassifiedasAGNs(seethe jectswithlowcolumndensity(NH=1020cm−2).
10
sic photon index, Γ , of the ith SMG-AGN. Then we be viewed as the ‘fractions of SMGs that have X-ray
int,i
converted L′ into the observed flux in the same way flux/luminosity above certain values’ instead of cumula-
X,i,j
as for converting L into flux as described above. The tive AGN fractions. Future discoveries of SMG-AGNs
X,i
results of f (L ) and f (L ) are shown in the with lower X-rayluminosities (probably involvingdisen-
AGN X AGN X,corr
middle- and lower-left panels in Figure 10. tangling the contributions from the star formation and
AGN)willbeabletopushthef functiontothelower
AGN
5.2. The AGN Fractions f0.5−8 keV/L0.5−8 keV ends.
The black dots in the three left-hand panels of Fig- 6. STACKINGOFTHEX-RAYUNDETECTEDSMGS
ure 10 show the cumulative AGN fractions for the
Thanks to the high-precision positions of the SMGs
E-CDF-S ALESS SMGs. The left-most point in
provided by ALESS, we are able to assess directly and
each panel is essentially the fraction of SMGs hosting
reliably the average X-ray properties of the X-ray unde-
AGNs with flux/luminosity equal to or above the cur-
tectedSMGsthroughstacking. PreviousworksonX-ray
rent faintest SMG-AGN in X-ray ever detected in the
stacking of SMGs were typically based on the positions
E-CDF-S. These f values are marked in the plots
AGN of the IR/radio counterparts of SMGs (e.g., Laird et al.
and listed in Table 5.
2010; Georgantopouloset al. 2011; Lindner et al. 2012),
The f analysis above is for all ALESS SMGs in
AGN
which, as mentioned in Section 1 and 2, suffer from the
the E-CDF-S. This sample of SMGs, however, may not
largeruncertaintiesincounterpartmatchingdue topoor
constitute a flux-limited sample of SMGs. The reason
angular resolution of single-dish submm surveys.
is that ALESS differs from a regular flux-limited sur-
ToavoidpoorX-rayPSFregionsandhighbackground,
vey since it targeted the bright submm sources discov-
weonlystackedsourceswithin7′ oftheaimpointsofthe
ered by LESS. As mentioned earlier, the LESS survey
4 Ms CDF-S or 250 ks E-CDF-S. SMGs within 2 the
is in fact a contiguous and flux-limited survey over the
×
90%encircled-energyapertureradiiofanyX-raysources
whole field of E-CDF-S. Thus, the LESS sources con-
are also not included in our stacking analysis. There are
stitute a flux-limited sample for S > 3.5–4.5 mJy
870µm
50 sources that have small enough off-axis angles and
(LESS de-boosted flux limit; Weiß et al. 2009). Conse-
are far enough from any X-ray sources, and only one
quently, the ALESS SMGs with submm flux above the
of them lies within the 4 Ms CDF-S region; it has an
LESS flux limit are actually a flux-limited sample. We
effective exposure time 15 times larger than the other
hence removed the SMGs (including SMG-AGNs) with
∼
49sourcesintheE-CDF-S. Topreventthissinglesource
S <3.5mJyinoursampleandcalculatedtheAGN
870µm from biasing our stacking results, we did not include it
fractionfor the flux-limited (>3.5mJy) SMG sample in
inthestacking. Therefore,thereare49SMGs intotalin
the E-CDF-S. The results are shown in the right-hand
our stacking sample, all in the E-CDF-S region.
panelsofFigure10,andthef valuesfortheleft-most
AGN We follow the stacking procedures detailed in Section
points are listed in Table 5.
3.1 of Luo et al. (2011). Briefly, we extracted the X-ray
As expected, since brighter AGNs in general are rarer
counts within an aperture of 1.5′′ in radius centered
(e.g., Xue et al. 2010; Aird et al. 2012), f decreases
AGN aroundthesubmmpositionofeachofthe49SMGs. The
as X-ray flux or luminosity increases for both cases with
background counts for each SMG were estimated by ex-
orwithoutthesubmm fluxcut. Also,comparingtheleft
tractingcountswithin1000randomlyplaced1.5′′-radius
panels with the rightpanels of Figure 10, the AGN frac-
apertureswithin1′ ofeachSMG,avoidingX-raysources
tions for the S > 3.5 mJy SMG sample are larger
870µm
and the central SMG, and then taking the average. The
than those for all ALESS SMGs overall. This is not sur-
total stacked counts (S) for these 49 SMGs are 33.5 in
prising given that 5 out of 8 (63%) of the SMG-AGNs
the soft band and 54.0 in the hard band. The stacked
have flux larger than 3.5 mJy, while a smaller fraction
background counts (B) are 13.6 (soft) and 40.2 (hard).
of SMGs without AGN have S > 3.5 mJy (46 out
870µm
Thus,the netcountsare20.0(soft)and13.8(hard),and
of 91, 51%). However,the f values are actually con-
AGN
sistent between the two SMG samples within the error the signal to noise ratios S/N19 ((S B)/√B) are 5.4σ
−
bars. (soft) and 2.1σ (hard), which correspond to a probabil-
A general trend of larger f for SMG groups with ity of p = 3.7 10−6 for being generated by Poisson
AGN ×
larger submm flux S was also observed when we noiseforthe softband,andp=0.021forthe hardband.
870µm
computed f as a function of S . It rises from ThesmoothedstackedimagesareshowninFigure11. A
AGN 870µm
about 15+15% for S > 1.3 mJy (faintest ALESS summaryofourstackingresultscanbefoundinTable6.
−5 870µm
We performed a robustness test by generating 1000
SMG) to about 34+37% for S > 6 mJy. However,
−17 870µm fake submm catalogs at random RA and Dec with 49
theerrorbarsonf arelargeduetothelimitedsample
AGN sourceseach(avoidingX-raysources),andstackingthem
size,especiallyathighsubmmflux,andthef values
AGN thesamewayasdescribedabove. Noneofthe1000cases
are actually consistent with each other within 1σ error
has a S/N of σ > 5.4 (< 0.1%) in the soft band, and 21
bars, including the two f values at the lowest and
AGN cases(2.1%)haveS/Nofσ >2.1inthehardband,which
highest S ends.
870µm are both consistent with our findings.
The dashed lines in each panel of Figure 10 are
We then explore the possibility of identifying a sub-
f values if all 10 X-ray detected SMGs, including
AGN
ALESS 45.1 and 67.1, are taken as AGN hosts. Note
19 Wenoteherethatthebackgroundcountsarescaledtomatch
that this would only affect the fAGN results at the the 1.5′′ source extraction aperture. Total background counts are
low flux/luminosity end as ALESS 45.1 and 67.1 are 1000 larger since they are estimated using 1000 1.5′′ apertures.
very faint in X-ray. As we do not have strong evi- Thus×the S/N can becalculated usingGaussianstatistics as(S
−
dence that ALESS 45.1 or 67.1 host AGN, this line can B)/(√B 1000/√1000)=(S B)/√B.
× −