Unveiling the Nature of Submillimeter Galaxy SXDF850.6

Abstract

We present an 880 micron Submillimeter Array (SMA) detection of the submillimeter galaxy SXDF850.6. SXDF850.6 is a bright source (S(850 micron) = 8 mJy) detected in the SCUBA Half Degree Extragalactic Survey (SHADES), and has multiple possible radio counterparts in its deep radio image obtained at the VLA. Our new SMA detection finds that the submm emission coincides with the brightest radio emission that is found ~8" north of the coordinates determined from SCUBA. Despite the lack of detectable counterparts in deep UV/optical images, we find a source at the SMA position in near-infrared and longer wavelength images. We perform SED model fits to UV-optical-IR photometry (u, B, V, R, i', z', J, H, K, 3.6 micron, 4.5 micron, 5.8 micron, and 8.0 micron) and to submm-radio photometry (850 micron, 880 micron, 1100 micron, and 21 cm) independently, and we find both are well described by starburst templates at a redshift of z ~= 2.2 (+/- 0.3). The best-fit parameters from the UV-optical-IR SED fit are a redshift of z = 1.87 (+0.15/-0.07), a stellar mass of M_star = 2.5 +2.2/-0.3 x 10^11 M_sun, an extinction of A_V = 3.0 (+0.3/-1.0) mag, and an age of 720 (+1880/-210) Myr. The submm-radio SED fit provides a consistent redshift of z ~ 1.8-2.5, an IR luminosity of L_IR = (7-26) x 10^12 L_sun, and a star formation rate of 1300-4500 M_sun/yr. These results suggest that SXDF850.6 is a mature system already having a massive amount of old stellar population constructed before its submm bright phase and is experiencing a dusty starburst, possibly induced by major mergers. Comment: 7 pages, 5 figures, Accepted for publication in Astrophysical Journal

Full text

Page 1

arXiv:1001.3690v1 [astro-ph.CO] 20 Jan 2010
Accepted for Publication in the Astrophysical Journal
Preprint typeset using LATEX style emulateapj v. 10/10/03
UNVEILING THE NATURE OF SUBMILLIMETER GALAXY SXDF 850.6
B. Hatsukade1, D. Iono2, T. Yoshikawa3M. Akiyama3, J.S. Dunlop4, R.J. Ivison5,4, A. B. Peck6, S. Ikarashi1,
A. Biggs7, H. Ezawa8, H. Hanami9, P. Ho10, D. H. Hughes11, R. Kawabe2, K. Kohno1,12, S. Matsushita10,
K. Nakanishi2, N. Padilla13, G. Petitpas14, Y. Tamura2, J. Wagg11,15,16, D. J. Wilner16, G. W. Wilson17,
T. Yamada3, and M. S. Yun17
Accepted for Publication in the Astrophysical Journal
ABSTRACT
We present an 880 µm Submillimeter Array (SMA) detection of the submillimeter galaxy SXDF 850.6.
SXDF 850.6 is a bright source (S850 µm= 8 mJy) detected in the SCUBA Half Degree Extragalactic
Survey (SHADES), and has multiple possible radio counterparts in its deep radio image obtained at
the VLA. Our new SMA detection finds that the submm emission coincides with the brightest radio
emission that is found ∼8′′north of the coordinates determined from SCUBA. Despite the lack of
detectable counterparts in deep UV/optical images, we find a source at the SMA position in near-
infrared and longer wavelength images. We perform SED model fits to UV–optical–IR photometry
(u,B,V,R,i′,z′,J,H,K, 3.6 µm, 4.5 µm, 5.8 µm, and 8.0 µm) and to submm–radio photometry
(850 µm, 880 µm, 1100 µm, and 21 cm) independently, and we find both are well described by
starburst templates at a redshift of z ≃ 2.2 ± 0.3. The best-fit parameters from the UV–optical–IR
SED fit are a redshift of z = 1.87+0.15
of AV = 3.0+0.3
redshift of z ∼ 1.8–2.5, an IR luminosity of LIR= (7–26) ×1012L⊙, and a star formation rate of 1300–
4500 M⊙yr−1. These results suggest that SXDF 850.6 is a mature system already having a massive
amount of old stellar population constructed before its submm bright phase and is experiencing a
dusty starburst, possibly induced by major mergers.
Subject headings: galaxies: formation, galaxies: starburst, cosmology: observations, galaxies: high
redshift, submillimeter
−0.07, a stellar mass of M⋆ = 2.5+2.2
−210Myr. The submm–radio SED fit provides a consistent
−0.3× 1011M⊙, an extinction
−1.0mag, and an age of 720+1880
1. INTRODUCTION
1Institute of Astronomy, the University of Tokyo, 2-21-1 Osawa,
Mitaka, Tokyo 181-0015, Japan
Electronic address: hatsukade@ioa.s.u-tokyo.ac.jp
2Nobeyama Radio Observatory, Minamimaki, Minamisaku,
Nagano 384-1805, Japan
3Astronomical Institute, Tohoku University, Aramaki, Aoba-ku,
Sendai, Miyagi 980-8578, Japan
4Scottish Universities Physics Alliance, Institute for Astronomy,
School of Physics and Astronomy, University of Edinburgh, Royal
Observatory, Edinburgh EH9 3HJ, UK
5UK Astronomy Technology Centre, Science and Technology
Research Council, Royal Observatory, Blackford Hill, Edinburgh
EH9 3HJ, UK
6Joint ALMA Observatory, Avenida El Golf 40, Piso 18, Las
Condes 7550108 Santiago, Chile
7European Southern Observatory, Karl-Schwarzschild-Straße 2,
D-85748 Garching, Germany
8National Astronomical Observatory of Japan, 2-21-1 Osawa,
Mitaka, Tokyo 181-8588, Japan
9Physics Section, Faculty of Humanities and Social Sciences,
Iwate University, Morioka 020-8550, Japan
10Academia Sinica Institute of Astronomy and Astrophysics,
P.O. Box 23-141, Taipei 10617, Taiwan
11Instituto Nacional de Astrofisica,
(INAOE), Aptdo. Postal 51 y 216, 72000 Puebla, Pue., Mexico
12Research Center for the Early Universe, University of Tokyo,
7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan
13Departamento de Astronom´ ıa y Astrof´ ısica, Pontificia Univer-
sidad Cat´ olica de Chile, Vicu˜ na Mackenna 4860, Santiago, Chile
14Harvard-Smithsonian Center for Astrophysics, Submillimeter
Array, 645 North A‘ohoku Place, Hilo, HI 96720, USA
15European Southern Observatory, Alonso de C´ ordova 3107,
Vitacura, Casilla 19001, Santiago 19, Chile
16Harvard-Smithsonian Center for Astrophysics, 60 Garden
Street, Cambridge, MA 02138, USA
17University of Massachusetts, Department of Astronomy,
Amherst, MA01003, USA
´Optica y Electr´ onica
Millimeter/submillimeter
tionized observational cosmology by
substantialnewpopulation
dusty starburst galaxies at high redshifts (SMGs)
(e.g., Smail, Ivison & Blain 1997; Barger et al. 1998;
Hughes et al.1998;Scott et al.
2006; Greve et al. 2004; Laurent et al. 2005; Scott et al.
2008;Perera et al.2008;
Tamura et al. 2009, and see also Blain et al. 2002 for
a review). The energy source of mm/submm emission
is primarily from intense star formation activity, with
star formation rates (SFRs) of 100 M⊙yr−1to several
1000 M⊙ yr−1, and possibly partially from an active
galactic nucleus (AGN) (e.g., Alexander et al. 2005).
While spectroscopic observations of radio-identified
SMGs find a median redshift of 2.2 (Chapman et al.
2005), several SMGs have now been found up to and
beyond z = 4 (e.g., Capak et al. 2008; Coppin et al.
2009; Daddi et al. 2009).There is a hypothesis that
SMGs are progenitors of present-day massive ellipticals
(e.g., Lilly et al. 1999; Smail et al. 2004), however, little
is known about their evolution process.
While multi-wavelength analysis is essential in order
to understand the nature of SMGs, the coarse angular
resolution of single dish telescopes prevents a precise
determination of the exact optical/NIR counterparts.
One of the most successful ways to pinpoint the loca-
tion of the submm emission is to obtain high resolution,
deep radio imaging (e.g, Ivison et al. 1998, 2000, 2002;
Smail et al. 2000; Barger et al. 2000).
technique reveals robust radio counterparts of ∼50%–
80% of submm sources (e.g., Ivison et al. 2005, 2007;
surveyshave
uncovering a
mm/submm-bright
revolu-
of
2002;Coppin et al.
Austermann et al.2009;
Although this

Page 2

2Hatsukade et al.
Wagg et al. 2009), sometimes multiple radio counterpart
candidates are found for a source.
The most accurate means of achieving high precision
astrometry on the submillimeter emission is clearly to
observe with high angular resolution at the wavelength
of the original detection. In this respect, the Submil-
limeter Array (SMA; Ho et al. 2004) has proved to be
a powerful instrument (Iono et al. 2006a,b; Wang et al.
2007; Younger et al. 2007, 2008; Ivison et al. 2008).
Here we present the results from the SMA observa-
tions toward an 8 mJy submm source, SXDF 850.6, de-
tected in the SCUBA Half Degree Extragalactic Survey
(SHADES). SHADES has observed a large area of the
sky (720 arcmin2) with high sensitivity (1σ ∼ 2 mJy)
with the purpose of obtaining a statistically significant
unbiased sample of submillimeter sources (Mortier et al.
2005; Coppin et al. 2006).
SHADES are divided between two fields, the Lockman
Hole and the Subaru/XMM-Newton Deep Field (SXDF).
SXDF 850.6 is a source in the SXDF with multiple
optical, IR, and radio counterparts (Ivison et al. 2007;
Clements et al. 2008), but no established submm source
identification. The strongest radio emission has no con-
firmed optical counterpart, but the two secondary radio
peaks both have apparent optical associations.
§ 2 outlines the observational and calibration details,
and the results are presented in § 3.
wavelength data are described. The results of SED model
fitting using the photometry are presented in § 5. In § 6,
we discuss the nature of SXDF 850.6. A summary is pre-
sented in § 7. Throughout the paper, magnitudes are in
the AB system, and we adopt a cosmology with H0= 70
km s−1Mpc−1, ΩM= 0.3, and ΩΛ= 0.7.
The regions covered by
In § 4, multi-
2. OBSERVATIONS AND DATA REDUCTION
SXDF 850.6 was observed on September 21 and Oc-
tober 7, 2004, and on October 9 and 14, 2005, using
a compact configuration with 7 – 8 antennas of the
SMA. The phase center was positioned at the submm
source centroid which is α(J2000) = 02h17m29.80sand
δ(J2000) = −05◦03′26′′.00. The unprojected baseline
length ranged from 23 m to 139m. The SMA correlator
was equipped with 2 GHz total bandwidth in each side-
band, yielding a total of 4 GHz bandwidth for contin-
uum observations. A continuum channel was generated
by vector averaging all of the channels after calibration.
The SIS receivers were tuned to 345 GHz for the USB,
yielding 335 GHz for the LSB. Interferometric pointing
was checked at the beginning of the track and the point-
ing offsets were usually within ∼ ±5′′(15% of the pri-
mary beam) for all antennas. We used an integration
time of 30 seconds.
The raw SMA data was calibrated using the MIR
package (Scoville et al. 1993). Passband calibration was
done using bright QSOs and planets observed during the
track. Antenna based phase calibration was done using
J0238+166 (1.02 Jy), J0423-013 (1.67 Jy) and J0132-
169 (0.84 Jy). The flux levels of all sources were nor-
malized using the the quasar flux estimates derived from
the primary flux standard Uranus. Imaging was carried
out in MIRIAD (Sault et al. 1995). Maximum sensitiv-
ity was achieved by adopting natural weighting, which
gave a final synthesized beam size of 2.′′32×2.′′19 (P.A. =
79.1◦) and an RMS noise of 1.2 mJy. Because the source
-5:03:18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
38.0
Right Ascension (J2000)
Declination (J2000)
2h17m30.4s 30.0s 29.6s 29.2s
5
4
3
2
1
0
-1
-2
Signal-to-Noise Ratio
Fig.1.—
The SMA contours overlaid on the VLA 21 cm
image. The contours represent −3, −2, 2, 3, 4, and 5σ (where
1σ = 1.2 mJy). The synthesized beam (2.′′32 × 2.′′19) is shown
in lower left. The dark circle represents the approximate size of
the SCUBA/JCMT beam (FWHM ∼ 14′′). The dark cross, white
crosses, and red circles are the AzTEC position (Ikarashi et al. in
prep.), three radio counterpart candidates in Ivison et al. (2007),
and MIPS 24 µm sources in the SWIRE catalog, respectively.
is close to zero in declination, the uv coverage is under-
sampled in the north-south direction and the resultant
synthesized beam has multiple sidelobes at the 30% level
along the north-south direction about 14′′away from the
synthesized beam. The final map was made after adding
the data from all four days and two sidebands, and cor-
rected for the attenuation by the 35′′primary beam of
the SMA. The astrometric accuracy is likely dominated
by the low S/N of the image, and we assess ∼0.′′4.
3. RESULTS
The synthesized map is shown in Figure 1. A source
was detected at about 6 sigma significance 8.′′2 north of
the SCUBA coordinates.The source appears slightly
elongated in the northeast-southwest direction but this
is likely caused by the low S/N, and the source is entirely
unresolved with the SMA beam. Based on a point source
fit to the visibilities, the derived flux is 6.9 ± 1.2 mJy,
and the coordinates are α(J2000) = 02h17m29.79sand
δ(J2000) = −05◦03′18.′′65. The derived flux is consis-
tent with the SCUBA 850 µm flux of 8.15 ± 2.2 mJy
(Coppin et al. 2006) within the uncertainties of both
measurements.
4. MULTI-WAVELENGTH DATA
Multi-wavelength images around SXDF 850.6 are
shown in Figure 2. While an obvious optical counterpart
is not seen in the images from u to z′bands, the NIR
images represented in JHK and longer wavelength im-
ages show a presence of a counterpart. The photometry
of the counterpart and upper limits at the SMA position
are presented in Table 1.
4.1. Radio and Millimeter
Ivison et al. (2007) identify three radio counterpart
candidates for SXDF 850.6 in their deep VLA 21 cm

Page 3

Nature of SXDF 850.63
uBV Ri’z’ JHK
IRAC 24um 21cm
Fig. 2.— Multi-wavelength images of SXDF 850.6 with the SMA contours (2, 3, 4, and 5σ). The size of each image is 10′′× 10′′and
north is up. From left to right: rgb image of MOSAIC II/u, SuprimeCam/B, and V ; rgb image of SuprimeCam/R, i′, and z′; rgb image
of WFCAM/J, H, and K; rgb image of IRAC/ch1 (blue), ch2 (green), ch3 and 4 (red); MIPS 24 µm; VLA 21 cm.
map based on the probability analysis method of
Downes et al. (1986). All three radio sources have corre-
sponding MIPS 24 µm sources (Clements et al. 2008), as
shown in Figure 1. The SMA observation reveals that the
correct radio counterpart is the most distant source from
the SCUBA centroid, and the brightest in 24 µm emis-
sion. This demonstrates the effectiveness of the SMA
in identifying SMG counterparts, especially in situations
with multiple counterpart candidates.
The radio coordinates derived from reanalysis of
theVLAimage(Arumugam
are α (J2000) = 02h17m29.755s(±0.005s) and
δ (J2000) = −05◦03′18.′′40 (±0.′′08), with peak and in-
tegrated flux of 99.8 µJy and 100.0 ± 10.6 µJy respec-
tively. The close similarity of the peak and integrated
flux suggests that the radio emission is unresolved with
the 1.′′79 × 1.′′51 beam (P.A. = 6.1◦).
SXDF 850.6 is detected at 1100 µm with deboosted
flux of 3.9 ± 0.5 mJy (Ikarashi et al.
ing the AzTEC camera (Wilson et al. 2008) on the
Atacama Submillimeter Telescope Experiment (ASTE;
Ezawa et al. 2004, 2008). The AzTEC coordinates are
consistent with the SMA position within a 2σ error cir-
cle.
etal.inprep.)
in prep.)us-
4.2. Mid-IR
Spitzer/IRAC and MIPS data are taken from the
SpUDS archive. The 3σ detection limits are 0.58, 0.89,
5.7, and 5.3 µJy in IRAC bands (3.6, 4.5, 5.8, and
8.0 µm), and 36 µJy in MIPS 24 µm band. The pho-
tometry was performed using the SExtractor package
(Bertin & Arnouts 1996).
bands is conducted with 3.′′8 diameter aperture, and the
photometric zero-points of the images are determined us-
ing colors of early-type stars between K-band and IRAC
bands following Lacy et al. (2005). In order to derive
total flux, the aperture photometry are corrected by as-
suming IRAC PSFs by a factor of 1.36, 1.40, 1.65, and
1.84, respectively, The photometry in the 24 µm band
is conducted using 12′′diameter aperture. The aperture
The photometry in IRAC
magnitude is corrected by a factor of 1.70.
We find emission at the SMA position in all IRAC
bands and MIPS 24 µm band. Note that the 24 µm flux
in Table 1 should be used as an upper limit since sources
around the SMA position are blended in the 24 µm image
(Figure 2).
4.3. JHK
J, H, and K band images are obtained from UKIDSS
(UKIRT InfraRed Deep Sky Surveys) Ultra Deep Survey
Third Data Release (Laurent et al. 2005). The 3σ detec-
tion limits are 24.28, 24.08, and 24.18 mag, respectively.
A faint source is detected at the SMA position in all
bands. We perform photometry using the SExtractor.
Aperture photometry with 1.′′8 diameter aperture in the
J, H, and K bands are corrected by difference between
the aperture and the total (MAG AUTO) magnitudes in
the K-band.
4.4. uBV Ri′z′
We present a u-band image of MOSAIC II on CTIO
4-m telescope (Fujishiro et al. in prep.), B,V,R,i′, and
z′images of SuprimeCam on Subaru telescope in Figure
2. B, V , R, i′, and z′band images are obtained from
Subaru/XMM-Newton Deep Survey database (SXDS;
Furusawa et al. 2008). No galaxy is detected at the SMA
position in these bands. The 3σ detection limits are
26.00, 27.94, 27.68, 27.43, 27.45, and 26.43 mag, respec-
tively.
Near the SMA position, three galaxies are seen; ∼2′′
south-east, ∼0.′′8 south-east, and ∼1′′south-west to the
SMA position. We derive photometric redshifts for the
sources and find that they are likely to be at z < 1, and
it is unlikely that they are related to the SMA source
(see § 5).
5. SED FITTING
We perform SED model fitting to the photometry data
from UV to radio to estimate photometric redshifts and

Page 4

4Hatsukade et al.
Table 1. Fluxes of SXDF 850.6
BandFluxInstrumentRef.
u
B
V
R
i′
z′
J
H
K
<0.0148 µJy
<0.0248 µJy
<0.0315 µJy
<0.0397 µJy
<0.0389 µJy
<0.0996 µJy
1.40+0.23
−0.20µJy
3.21+0.23
−0.22µJy
9.17 ± 0.23 µJy
29.0 ± 0.8 µJy
39.3 ± 1.1 µJy
54.3 ± 1.9 µJy
38.2 ± 1.4 µJy
630 ± 30 µJy
<30 mJy
<200 mJy
<81 mJy
8.15 ± 2.2 mJy
6.9 ± 1.2 mJy
3.9 ± 0.5 mJy
100.0 ± 10.6 µJy
<6.3 × 10−15erg cm−2s−1
MOSAIC II
SuprimeCam
SuprimeCam
SuprimeCam
SuprimeCam
SuprimeCam
WFCAM
WFCAM
WFCAM
IRAC
IRAC
IRAC
IRAC
MIPS
MIPS
MIPS
SCUBA
SCUBA
SMA
AzTEC
VLA
XMM-Newton
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
3
3
1
4
5
6
3.6 µm
4.5 µm
5.8 µm
8.0 µm
24 µm
70 µm
160 µm
450 µm
850 µm
880 µm
1100 µm
21 cm
0.2–12 keV
References. — (1) This work; (2) Flux limits in the SWIRE catalog
(Surace et al. in prep.); (3) Coppin et al. 2006; (4) Ikarashi et al. in
prep.; (5) Arumugam et al.in prep.
for XMM-Newton data provided by the XMM-Newton Survey Science
Centre
(6) FLIX: upper limit server
Note. — Limits are 3σ.
other physical properties (stellar mass, V -band extinc-
tion, star formation timescale, and metallicity).
photometric data are divided into two wavelength ranges:
(1) from UV to 8.0 µm, where stellar emission is domi-
nant, (2) from submm to radio, where thermal dust and
synchrotron emission is dominant.
The
5.1. UV – Optical – IR
SED fits are performed to the photometry ranging
from UV to 8.0 µm (u,B,V,R,i′,z′,J,H,K, 3.6 µm,
4.5 µm, 5.8 µm, and 8.0 µm) using the Hyperz code
(Bolzonella et al. 2000).
tral library of GALAXEV (Bruzual & Charlot 2003) for
SED templates. We adopt the Calzetti extinction law
(Calzetti et al. 2000) and the Chabrier (2003) initial
mass function with lower and upper cutoff mass of 0.1
and 100 M⊙. Two star formation histories are assumed:
(i) exponentially decaying SFR with a timescale of τ (i.e.,
SFR ∝ e−t/τ), (ii) constant SFR. We calculate χ2val-
ues for each SED template with free parameters of red-
shift (z = 0–6), stellar mass (M⋆), V -band attenuation
(AV = 0.0–6.0), star formation timescale (τ = 0.01–
5 Gyr), and metallicity (Z = 1.0,0.4,0.2 Z⊙).
best-fit results are obtained with the exponentially de-
caying SFR, z = 1.87+0.15
AV = 3.0+0.3
and Figure 3).The results suggest that SXDF 850.6
has a large population of old stars, with large amount of
dust obscuring the stellar emission.
is comparable to those of massive star-forming galax-
ies at z ∼ 2 (e.g., Shapley et al. 2005; Erb et al. 2006a)
and consistent with stellar masses recently derived for
other SMGs (∼1011–1012M⊙) (e.g., Dye et al. 2008;
We use a synthetic spec-
The
−0.07, M⋆ = 2.5+2.2
−0.3× 1011M⊙,
Myr (Table 2
−1.0mag, and age = 720+1880
−210
The stellar mass
µ
Observing Wavelength [ m]
Flux Density [ Jy]
µ
z = 1.87, AV = 3.0, age = 720 Myr
1 10
102
101
100
10-1
10-2
10-3
Fig.3.—
Best-fit SED obtained in UV–optical–IR SED fit
along with photometry data. Downward arrows represent 3σ upper
limits.
Micha? lowski et al. 2009).
5.2. Submm – Radio
Photometry at 850 µm, 880 µm, 1100 µm, and 21 cm
is used for submm-radio SED fits. The MIPS photom-
etry is also used to check for consistency. We use the
following starburst SED models: Arp 220 (Silva et al.
1998), the average SED of 76 SMGs with spectroscopic
redshifts (Micha? lowski et al. 2009), and 105 SED mod-
els of Chary & Elbaz (2001). We perform minimized χ2
fits with free parameters of redshifts and flux scaling fac-
tors. We find the best-fit redshifts of z = 2.1+0.5
Arp 220 (χ2= 0.3), z = 1.8+0.5
(χ2= 0.4), and z = 2.5+0.6
−0.4for an SED of Chary & Elbaz
(2001) (χ2= 0.09). The errors are 99% confidence inter-
vals. Note that there are other templates in the library
of Chary & Elbaz (2001) that fit the photometry data
with similar χ2values in the redshift range of z ∼ 1–3.
The derived redshifts are consistent with the result in the
UV-optical-IR SED fit in the previous section. The IR
luminosities are estimated to be LIR= (7–26)×1012L⊙
from the intrinsic IR luminosities of SED templates mul-
tiplied by scaling factors for the best-fit SEDs. The best-
fit SED models are shown in Figure 4 with photometry
data from UV to radio. All of the SEDs overestimate
the UV–NIR photometry, suggesting that stellar emis-
sion is heavily attenuated by dust. This is consistent
with the large extinction (AV = 3.0) derived from the
UV–optical–IR SED fit. Finally, we note that the SMG
sample in Micha? lowski et al. (2009) is biased against op-
tically faint objects. It is clearly important to take into
account a heavily obscured SMG like SXDF 850.6 (or
e.g., GOODS 850-5 in Wang et al. 2009, AzTEC1 in
Tamura et al. in prep.) to properly understand the over-
all picture of SMGs.
−0.4for
−0.3for the average SMG
6. DISCUSSION
6.1. AGN Contribution
The rest-frame 1.6 µm bump is clearly seen at 5.8 µm
(Figure 3), suggesting that the NIR emission detected
in the IRAC bands is star-formation dominated (e.g.,
Weedman et al. 2006; Farrah et al. 2008).The flatter

Page 5

Nature of SXDF 850.65
Table 2. Best-fit Results in UV–Optical–IR SED Fit
zM⋆
(M⊙)
(2)
AV
(mag)
(3)
Age
(Myr)
(4)
τZχ2
(Myr)
(5)
(Z⊙)
(6) (1)(7)
1.87+0.15
−0.07
2.5+2.2
−0.3× 1011
3.0+0.3
−1.0
720+1880
−210
20113.4
Note. — The errors are 68% confidence intervals. (1) photometric redshift; (2) stellar mass; (3) V -band attenuation; (4) age; (5) star formation
timescale; (6) metallicity; (7) χ2value
Arp 220 z = 2.1
SMG z = 1.8
CE01 z = 2.5
µ
Observing Wavelength [ m]
Flux Density [ Jy]
µ
10-1 100 101 102 103 104 105 106
105
104
103
102
101
100
10-1
10-2
Fig. 4.— Best-fit SEDs for three different models obtained in
submm–radio SED fit. The SEDs are the average SED of 76 SMGs
(Micha? lowski et al. 2009) at z = 1.8, Arp 220 (Silva et al. 1998) at
z = 2.1, and a starburst template of Chary & Elbaz (2001) at
z = 2.5. Photometry data from UV to radio are overplotted.
spectral slope in the IRAC emission compared to AGN
populations favours a star formation as a dominant heat-
ing source (Yun et al. 2008; Ivison et al. 2004). This is
also supported by the fact that SXDF 850.6 appears
close to the starburst model tracks in the S8.0/S4.5
vs. S24/S8.0color-color diagram of Ivison et al. (2004).
These facts suggest that SXDF 850.6 favours star for-
mation as a dominant heating source and the physical
quantities derived in the SED fit are less affected by the
AGN.
The 3σ upper limit on the rest-frame X-ray luminosity
derived from 0.2–12 keV flux is 1.3×1044erg s−1(assum-
ing z = 1.87 and an effective photon index of Γ = 1.8).
Since this value is not properly corrected for hydrogen
attenuation, we do not constrain the AGN contribution
from the X-ray luminosity.
6.2. Dust Mass and Molecular Gas Mass
Assuming the observed 880 µm flux is dominated
by thermal dust emission, the dust mass can be de-
rived as Md= SobsD2
Hughes et al. 1997), where Sobsis the observed flux den-
sity, νrest is the rest-frame frequency, κd(νrest) is the
dust mass absorption coefficient, Td is the dust tem-
perature, and B(νrest,Td) is the Planck blackbody func-
tion. We assume that the absorption coefficient varies
as κd∝ νβand β lies between 1 and 2 (e.g, Hildebrand
1983). We adopt κd(125 µm) = 2.64 ± 0.29 m2kg−1,
the average value of various studies (Dunne et al. 2003),
Td= 30–50 K, and β = 1.5, the typical values for SMGs
L/[(1+z)κd(νrest)B(νrest,Td)] (e.g.,
(e.g., Kov´ acs et al. 2006; Pope et al. 2006; Coppin et al.
2008; Micha? lowski et al. 2009). The dust mass is esti-
mated to be Md = (4–9) ×108M⊙ for SXDF 850.6
at z = 1.87.This is consistent with previous work
on SMGs (e.g., Kov´ acs et al. 2006; Coppin et al. 2008;
Micha? lowski et al. 2009).
By adopting a gas-to-dust mass ratio of 54, which is
an average value for SMGs in Kov´ acs et al. (2006), the
molecular gas mass is Mgas= (2–5) ×1010M⊙.
6.3. Star Formation Activity and Nature of SXDF 850.6
The UV–IR SED exhibits quiescent star forming ac-
tivity dominated by old stellar components. The cur-
rent SFR estimated from the SED fit is approximately
zero.This is because the large part of the stellar
mass was formed at the early phase of star formation
and SFR decreased with time following exponential de-
cay (τ = 20 Myr for the best-fit SED). On the other
hand, the IR luminosity of the best-fit submm–radio
SED provides SFRIR= 1300–4500 M⊙yr−1(Kennicutt
1998). The SFR derived from 1.4 GHz radio emission
is 1400 M⊙ yr−1following the equation of Bell (2003),
and 1100 M⊙yr−1following Yun & Carilli (2002) (with
a spectral index of −0.8 and a radio–FIR normalization
factor of 1).
To see whether the observational data allow the coexis-
tence of the old stellar SED with quiescent star formation
and the dusty starburst SED, we create synthetic SEDs
at UV–IR wavelength using the GALAXEV library. The
synthetic SEDs are composites of an SED with the same
parameters as the best-fit SED which is dominated by
old stellar population and starburst SEDs with differ-
ent parameters. We find that the composite SEDs with
plausible parameters for starburst SEDs (e.g., SFR ∼
2000 M⊙ yr−1, AV = 3.0, and age = 1 Myr) are con-
sistent with the photometry data (Figure 5). Note that
combining the SEDs increases the total stellar mass by
only ∼1%.
Combining these results allows us to infer the nature
of SXDF 850.6: it is a mature system with a large frac-
tion of old stellar components, and currently experienc-
ing a vigorous dusty starburst. The coexistence of old
stars and a current starburst in an SMG is suggested
by Wang et al. (2009) from a detailed SED analysis.
SXDF 850.6 has enough molecular gas mass as estimated
in § 6.2 to maintain intense star formation. Such sig-
nificant star formation is likely caused by major merg-
ers (e.g., Greve et al. 2005; Tacconi et al. 2006, 2008;
Narayanan et al. 2009). If the star formation continues
with SFR ∼ a few 103M⊙ yr−1, the gas consumption
time is ∼ a few 10 Myr. This scenario is well described
by hydrodynamic simulations of Narayanan et al. (2009)
in which a major merger with a ∼1013M⊙dark matter