The physical and chemical conditions of the interstellar medium (ISM) in galaxies can be revealed with forbidden atomic emission lines from the warm-phase ISM, such as ionized hydrogen (H
ii) regions and photodissociation regions (PDRs). A far-infrared (FIR) forbidden emission line, the [C
ii] 158-μm line predominantly coming from PDRs, has already been detected in many high-
z objects (
1,
2). Recent observations with the Atacama Large Millimeter/submillimeter Array (ALMA) have revealed the [C
ii] line emission from young star-forming galaxies emitting a strong hydrogen Lyα line, so-called Lyα emitters (LAEs), at redshift
z ~ 5 to 6 (
3,
4). However, ALMA observations have also shown that LAEs at
z > 6 have at least an order of magnitude lower luminosity of the [C
ii] line than that expected from their star formation rate (SFR) (
4–
7), suggesting unusual ISM conditions in these high-
z LAEs (
8).
Herschel observations of nearby dwarf galaxies, on the other hand, have revealed that a forbidden oxygen line, [O
iii] 88 μm, is much stronger than the [C
ii] line in these chemically unevolved galaxies (
9–
11). The Infrared Space Observatory and the Japanese infrared astronomical satellite AKARI have detected the [O
iii] line from the Large Magellanic Cloud and from many nearby galaxies (
12,
13). However, the [O
iii] line has rarely been discussed in a high-
z context, because of the lack of instruments suitable to observe the redshifted line. Only a few detections from gravitationally lensed dusty starburst galaxies with active galactic nuclei at
z ~ 3 to 4 have been reported (
14,
15) prior to ALMA. On the other hand, simulations predict that ALMA will be able to detect the [O
iii] line from star-forming galaxies with reasonable integration time even at
z > 8 (
16).
To examine the [O
iii] 88-μm line in high-
z LAEs, we performed ALMA observations of an LAE at
z = 7.2, SXDF-NB1006-2, discovered with the Subaru Telescope (
17). We have also obtained ALMA data of the [C
ii] 158-μm line of this galaxy. [The observations and the data reduction are described in (
18)]. The [O
iii] line is detected with a significance of 5.3σ (
Fig. 1A), and the obtained line flux is 6.2
× 10
–21 W m
–2; the corresponding luminosity is 3.8
× 10
35 W (
Table 1). The [C
ii] line is not detected at the position of the [O
iii] emission line, and we take the 3σ upper limit for the [C
ii] line flux as <5.3
× 10
–22 W m
–2. However, we note a marginal signal (3.5σ) that displays a spatial offset (≈ 0.4′′ ≈ 2 kpc in the proper distance) from the [O
iii] emission (fig. S4). The continuum is not detected in either of the ALMA bands, resulting in a 3σ upper limit of the total IR luminosity of <2.9
× 10
37 W when assuming a dust temperature of 40 K and an emissivity index of 1.5.
The spatial distribution of the ALMA [O
iii] emission overlaps with that of the Subaru Lyα emission (
Fig. 1A), as expected because both emission lines are produced in the same ionized gas. On the other hand, the Lyα emission is well resolved (the image resolution is 0.4′′) and spatially more extended than the [O
iii] line. This is because Lyα photons suffer from resonant scattering by neutral hydrogen atoms in the gas surrounding the galaxy. The systemic redshift of the galaxy is estimated at
z = 7.2120 ± 0.0003 from the [O
iii] emission line at an observed wavelength of 725.603 μm. The Lyα line is located at Δ
vLyα = +1.1 (±0.3)
× 10
2 km s
–1 relative to the systemic redshift (
Fig. 1C and fig. S6). This velocity offset, caused by scattering of neutral hydrogen, is relatively small by comparison to those observed in galaxies at
z ~ 2 to 3 (Δ
vLyα ~ 300 km s
–1), given the ultraviolet (UV) absolute magnitude of this galaxy (
MUV =
−21.53 magAB) (
19–
21). The observed small Δ
vLyα of SXDF-NB1006-2 may indicate an H
i column density of
NH i < 10
20 cm
–2 (
21,
22). SXDF-NB1006-2 is in the reionization era where only the intergalactic medium (IGM) with a high hydrogen neutral fraction may cause an observation of Δ
vLyα ≈ +100 km s
–1 (
23), implying an even smaller H
i column density in the ISM of this galaxy.
We performed spectral energy distribution (SED) modeling to derive physical quantities such as the SFR of SXDF-NB1006-2 (Table 1). In addition to broadband photometric data from the United Kingdom Infra-Red Telescope (UKIRT) J, H, and K bands, Spitzer 3.6-μm and 4.5-μm bands, and Subaru narrowband photometry NB1006 (table S3), we have also used the [O
iii] line flux and the IR luminosity as constraints (
Fig. 2) (
18). The extremely blue rest-frame UV color of this galaxy [slope β < −2.6 (3σ) estimated from
J −
H, where the flux density
Fλ ∝ λ
β] gives an age of ~1 million years for the ongoing star formation episode. The nondetection of the dust IR emission suggests little dust and hence negligible attenuation. The observed strong [O
iii] line flux favors an oxygen abundance of 5% to 100% that of the Sun, but rejects 2% and 200% of the solar abundance at >95% confidence. The obtained oxygen abundance is similar to those estimated in galaxies at
z ~ 6 to 7, for which UV C
iii] and C
iv emission lines were detected (
24,
25). Because ~1 million years is insufficient to produce the inferred oxygen abundance, the galaxy must have had previous star formation episodes. Therefore, the derived stellar mass of ~300 million solar masses is regarded as a lower limit. We obtain a ~50% escape fraction of hydrogen-ionizing photons to the IGM in the best-fit model. Such a high escape fraction, although still uncertain, may imply a low H
i column density of ~10
17 cm
–2 (
26) or porous structure in the ISM of the galaxy.
The [O
iii]/far-UV luminosity ratio of SXDF-NB1006-2 is similar to those of nearby dwarf galaxies with an oxygen abundance of 10% to 60% that of the Sun (
Fig. 3A), which suggests that the oxygen abundance estimated from the SED modeling is reasonable and chemical enrichment in this young galaxy has already proceeded. On the other hand, the dust IR continuum and the [C
ii] line of SXDF-NB1006-2 are very weak relative to those of the nearby dwarf galaxies (
Fig. 3, B and C). The galaxies at
z ~ 3 to 4, from which the [O
iii] line was detected previously, are IR luminous dusty ones (
14,
15). Their [O
iii]/IR and [O
iii]/[C
ii] luminosity ratios are similar to those of nearby spiral galaxies (
10) and are at least one order of magnitude smaller than those of SXDF-NB1006-2. The high [O
iii]/IR ratio of SXDF-NB1006-2, despite a degree of chemical enrichment (or so-called metallicity) similar to that of the nearby dwarf galaxies, indicates a very small mass fraction of dust in elements heavier than helium (or dust-to-metal mass ratio) in SXDF-NB1006-2. The dust deficiency of this galaxy is in contrast to the discovery of a dusty galaxy at
z ≈ 7.5 (
27), suggesting a diversity of the dust content in the reionization epoch. Because the [C
ii] line predominantly arises in gas where hydrogen is neutral, the nondetection of the [C
ii] line in SXDF-NB1006-2 suggests that this young galaxy has little H
i gas.
We also compared the observed properties of SXDF-NB1006-2 with the galaxies at
z = 7.2 in a cosmological hydrodynamic simulation of galaxy formation and evolution (
18). This simulation yielded several galaxies with a UV luminosity similar to that of SXDF-NB1006-2 (fig. S10). Relative to these simulated galaxies, SXDF-NB1006-2 has the highest [O
iii] line luminosity, a similar oxygen abundance, and a lower dust IR luminosity by at least a factor of 2 to 3. This indicates a factor of >2 to 3 smaller dust-to-metal mass ratio within the ISM of SXDF-NB1006-2 relative to that in the simulated galaxies, where we assumed the dust-to-metal mass ratio of 50% as in the Milky Way ISM (
28). Therefore, the dust-to-metal ratio of SXDF-NB1006-2 is implied to be <20%. The dust-to-metal ratio is determined by two processes: (i) dust growth by accretion of atoms and molecules onto the existing grains in cold dense clouds, and (ii) dust destruction by supernova (SN) shock waves consequent upon star formation (
28). The dust-poor nature of SXDF-NB1006-2 may be explained by rapid dust destruction due to its high SN rate or by slow accretion growth due to a lack of cold dense clouds in the ISM.
In the context of cosmic reionization studies, the most uncertain parameter is the product of the escape fraction of ionizing photons and the emission efficiency of these photons,
fescξ
ion (
29). From the SED modeling, we have obtained log
10[
fescξ
ion(Hz erg
–1)] = 25.44
−0.84+0.46 for SXDF-NB1006-2 (
18). This ionizing photon emission efficiency is strong enough to reach (or even exceed) the cosmic ionizing photon emissivity at
z ~ 7 estimated from various observational constraints on reionization (
29) by an accumulation of galaxies that have already been detected (
MUV < −17), although this does not rule out contributions of fainter, currently undetected galaxies to the ionizing emissivity. The ISM properties of SXDF-NB1006-2, with little dust and H
i gas, may make this galaxy a prototypical example of a source of cosmic reionization.
Acknowledgments
This paper makes use of the following ALMA data: ADS/JAO.ALMA# 2013.1.01010.S and 2012.1.00374.S, which are available at
https://almascience.nao.ac.jp/alma-data/archive. ALMA is a partnership of the European Southern Observatory (ESO) (representing its member states), NSF (USA), and NINS (Japan), together with NRC (Canada), NSC and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, the National Radio Astronomy Observatory/Associated Universities Inc., and the National Astronomical Observatory of Japan (NAOJ). Based in part on data collected at Subaru Telescope, which is operated by NAOJ; data are available at
http://smoka.nao.ac.jp/ under project codes S08B-019, S08B-051, and S09B-055, and also at the W.M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California, and NASA; data are available at
www2.keck.hawaii.edu/koa/public/koa.php under the project code S331D. When some of the data reported here were acquired, UKIRT was operated by the Joint Astronomy Centre on behalf of the Science and Technology Facilities Council of the UK; UKIDSS data are available at
http://wsa.roe.ac.uk//dr10plus_release.html. Based in part on archival data obtained with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA; data are available at
www.cfa.harvard.edu/SEDS/data.html. Supported by JSPS KAKENHI grants 26287034 (A.K.I. and K.M.), 26247022 (I.S.), 25287050 (N.Y.), 24740112 (T.O.), 15H02073 (Y.T.), and 15K17616 (B.H.), and by a grant-in-aid for JSPS Fellows (H.U.), by a grant-in-aid for the Global COE Program “The Next Generation of Physics, Spun from Universality and Emergence” from MEXT of Japan (K.O.), the Kavli Institute Fellowship (at Kavli Institute for Cosmology, University of Cambridge) supported by the Kavli Foundation (K.O.), and the Swedish Research Council (project 2011-5349) and the Wenner-Gren Foundations (E.Z.).