THE SECOND-GENERATION z (REDSHIFT) AND EARLY UNIVERSE SPECTROMETER. I. FIRST-LIGHT OBSERVATION OF A HIGHLY LENSED LOCAL-ULIRG ANALOG AT HIGH-z

The submillimeter view of the early universe is rapidly evolving. Improvements in instrumentation capabilities over the past 15 yr have revealed the submillimeter band (∼200–850 μm) as a unique probe of star formation in the early universe. These studies were launched by small to modest-sized ground-based surveys, largely of one or two colors, and moderate angular resolution (∼10'') far exceeding the previous submillimeter surveys of cosmology experiments (e.g., Smail et al. 1997; Barger et al. 1999; Hughes et al. 1998; Weiß et al. 2009; Coppin et al. 2006). Herschel expanded these studies to include both larger scales and multiple colors. From the ground, we began high-redshift spectroscopic surveys of far-infrared (IR) fine-structure lines in the 350 and 450 μm telluric windows with our first-generation z (redshift) and Early Universe Spectrometer (ZEUS-1) on the Caltech Submillimeter Observatory ([C ii]—Hailey-Dunsheath et al. 2010; Stacey et al. 2010; Brisbin et al. 2013; [O iii]—Ferkinhoff et al. 2010; [N ii]—Ferkinhoff et al. 2011). The Herschel-SPIRE Fourier transform spectrometer enabled similar submillimeter spectroscopy that was unobscured by telluric absorption. With the advent of the Atacama Large Millimeter Array (ALMA), studies at wavelengths between 0.43 and 3 mm are now making great leaps forward in terms of sensitivity and spatial resolution. However, because Herschel is no longer operating and ALMA has yet to open the 350 μm (band 10) window, there are limited opportunities to detect far-IR lines from high-redshift systems at wavelengths shorter than ∼ 420 μm. For the [C ii] 158 μm line, typically the brightest FIR fine-structure line, this means ALMA can observe galaxies no closer than z ∼1.7 until band 10 is deployed.

We have recently commissioned our new instrument, the second-generation z(Redshift) and Early Universe Spectrometer (ZEUS-2) on the Atacama Pathfinder Experiment (APEX) in Chile.¹¹ When fully deployed, ZEUS-2 will have wavelength coverage between 200 and 850 μm. For its first run, reported here, ZEUS-2 only accessed the 350 and 450 μm telluric windows. With a spectral resolving power of ∼1000 and background limited detectors, ZEUS-2 is optimized for detecting broad lines from galaxies (line-widths of ∼200–300 km s⁻¹). Within these windows, ZEUS-2 can rapidly survey high-redshift galaxies, detecting far-IR lines that are spatially unresolved in a typical ZEUS-2/APEX beam (∼8'') yet suitable for follow-up at high spatial resolution with ALMA. In local systems ZEUS-2 can simultaneously map their emission in the ¹³CO 6–5, CO 7–6, [N ii] 205 μm, [C i] 370 μm, and the [C i] 605 μm lines, which provide important information about the ionized, atomic, and molecular phases of the interstellar medium (ISM). Part II of this paper (C. Ferkinhoff et al. 2013, in preparation) discusses the instrument design and performance in detail.

The star formation and ISM of high-z systems can be studied with ZEUS-2 through the FIR fine-structure lines of carbon, oxygen, nitrogen, and their various ions that are redshifted into the submillimeter atmospheric windows. These lines characterize both the physical conditions of the gas and the ambient radiation fields. They are especially useful for studying dusty galaxies in the early universe where the commonly used optical emission line tracers of the ISM (e.g., [O ii] 3726, 3729 Å and [O iii] 4959, 5007 Å) undergo significant extinction, thereby limiting their use as astrophysical probes. The FIR fine-structure lines have been extensively studied in nearby extragalactic systems with the Kuiper Airborne Observatory, the Infrared Space Observatory (ISO), and recently with Herschel (e.g., Crawford et al. 1986; Stacey et al. 1991; Lord et al. 1996; Luhman et al. 1998; Malhotra et al. 2001; Negishi et al. 2001; Brauher et al. 2008; Graciá-Carpio et al. 2011). These studies were extended to high-redshift galaxies with Herschel (cf. Ivison et al. 2010; Sturm et al. 2010; Valtchanov et al. 2011; Coppin et al. 2012; A. Verma et al. 2013, in preparation) and regularly detected from the ground with ZEUS-1, as mentioned earlier, as well as with telescopes such as the IRAM 30 m, the Plateau de Bure Interferometer (PdBI), and ALMA (e.g., Maiolino et al. 2005, 2009; Iono et al. 2006; Swinbank et al. 2012; De Breuck et al. 2011; Cox et al. 2011; DeCarli et al. 2012; Carilli et al. 2013; Wang et al. 2013). Most recently, ALMA and the PdBI have enabled the spatially resolved studies of these important cooling lines (Wagg et al. 2012; Gallerani et al. 2012).

1.1. Studies of Far-IR [C ii] and [O i] Emission

1.2. This Paper: First ZEUS-2 Detection of a Spectral Line from a High-redshift Galaxy

2.1. H-ATLAS J091043.1−000322

2.2. Observations

Using ZEUS-2 on APEX, we observed SDP11 on 2012 November 17 under very good observing conditions (0.48 mm of precipitable water vapor). Spectra were obtained in standard chopping and beam switching mode with a chopper throw of 60''. We obtained five 13.2 minute integrations for a total integration time of 66 minutes. The ZEUS-2 velocity resolution at the observed wavelength (439.6 μm) is ∼340 km s⁻¹. The ZEUS-2/APEX beam is 8'' (∼68 kpc at z = 1.78), as measured on Uranus, which fully contains SDP11. Uranus also served as our flux calibrator using a brightness temperature of 68 K at 450 μm, as reported by Hildebrand et al. (1985). Figure 1 shows the detected [C ii] 158 μm line with a flux of (6.44 ± 0.42) × 10⁻¹⁸ W m⁻².

Zoom In Zoom Out Reset image size

2.3. Lensing Model

Figure 2 shows a Hubble Wide-Field Camera 3 (WFC3) near-IR image (F110W) of SDP11, centered on the lensing galaxy SDSS J091043−000323 (hereafter SDSSJ0910) at z = 0.793 (central elliptical at position 0'', 0''). The contours show the HST/WFC3 F160W image divided by the F110W image, i.e., the 1.5 μm/1.1 μm image.¹³ On the basis of the spectral energy distribution modeling of Negrello et al. (2010), who separately fit a spectral energy distribution to SDP11 and the lensing galaxy SDSSJ0910, we expect both SDSSJ0910 and SDP11 to be brighter in the 1.5 μm band. However, the 1.5/1.1 μm ratio for SDP11 is ∼4 times larger than that of SDSSJ0910, and the divided image should show a lensed imaged of SDP11. The divided image (i.e., the contours in Figure 2) clearly shows an Einstein ring centered on SDSSJ0910. To create a lens model, we take the positions of the ring's peak values relative to the centroid of SDSSJ0910 to correspond to lensed images of SDP11 assuming it is a point source. We input the position and values of the peaks in the divided image into LENSMODEL (Keeton 2001) to model the gravitational lens system and reproduce the peaks of the Einstein ring. The squares indicate the peak positions, and the triangles indicate the predicted image locations on the basis of our model solution. While we obtain an excellent model fit to the image positions and fluxes $\left({\tilde \chi ^2 \sim 1} \right)$, to determine the magnification and reproduce the observed Einstein ring, we use the point-source-derived model of the lens but replace the point source with an extended source. Replacing with an extended source adds uncertainty to the magnification factor because we do not know how well the width of the ring in the divided image represents the true source size. With that caveat in mind, a source with a Gaussian light profile and half-light radius of 2.1 ± 1.3 kpc is able to reproduce the observed width of the Einstein ring in Figure 2 and results in a flux magnification, μ, of 18 ± 11. This is the magnification of the NIR emission in SDP11. The magnification of SDP11's submillimeter emission could be different if it is distributed differently relative to the position of the lens. Given the lack of high-resolution submillimeter images, we adopt the NIR magnification above throughout this paper with the caveat that the submillimeter magnification may be different.

Zoom In Zoom Out Reset image size

3.1. The [C ii] 158 μm to FIR Luminosity Ratio: Intense FUV Fields

SDP11 has an apparent FIR luminosity of 3.88 × 10¹³ L_☉, placing it firmly in the hyper-luminous IR galaxy class (HyLIRG, L_FIR ⩾ 10³ L_☉). Figure 3 shows a plot of the L_[CII]/L_FIR ratio, R, for SDP11 versus its apparent FIR luminosity along with samples of local galaxies—normal star-forming galaxies and local ULIRGs—and high-z galaxies including the sources from Stacey et al. (2010) and Hailey-Dunsheath et al. (2010). The ratio R ∼ (1.0 ± 0.3) × 10⁻³ for SDP11 falls between the average line ratios observed for star-formation-dominated systems (L_[CII]/L_FIR ∼ 3 × 10⁻³) and AGN-dominated systems (L_[CII]/L_FIR ∼ 4 × 10⁻⁴) at high-z. Stacey et al. (2010) and Hailey-Dunsheath et al. (2010) show that stronger UV fields may lower the value of R and for a given value of R, i.e., for a given FUV field, the FIR luminosity indicates the spatial scale of the emission. The ratio of SDP11, when compared with the Stacey et al. sample of sources with similar ratios, suggests a FUV field, G, of ∼10,000 G₀, where G₀ is the Habing field—the strength of the local interstellar radiation field corresponding to a flux of 1.6 × 10⁻³ erg cm⁻² s⁻¹. Given the value of R from SDP11, its FIR luminosity suggests a source size greater than a kilo-parsec. However, once we account for the magnification of SDP11, its intrinsic FIR luminosity becomes similar to local ULIRGS. Magnification does not affect the estimate of the FUV field (G ∼ 10,000 G_o), which is more intense than local normal galaxies or high-z star-formation-dominated systems, but the lower FIR luminosity now suggests sub-kiloparsec size emission similar to that found in nearby ULIRGs.

Zoom In Zoom Out Reset image size

3.2. Modeling the Line and Continuum Emission

Comparisons between our [C ii] line, the [O i] line, and the FIR continuum will provide tighter constraints on the ambient FUV fields and gas density in SDP11 as well as the nature of the source. Using the Herschel-PACS spectrometer on the Herschel Space Telescope, A. Verma et al. (2013, in preparation) report a [O i] 63 μm line flux of (7.5 ± 2.3) × 10⁻¹⁸ W m⁻². Typically, the [C ii] and [O i] lines are produced in PDRs associated with star-forming regions. However, it is possible that both of these lines arise within the X-ray-dominated regions produced in molecular clouds enveloping an AGN. Meijerink et al. (2006) produce grids of observed flux for various FIR fine-structure lines, including [C ii] and [O i], across typical densities and X-ray fluxes of X-ray-dominated regions. The observed [O i]/[C ii] ratio is 1.2 ± 0.5. This is consistent only with the lowest X-ray fluxes and densities (F_X ∼ 0.1–0.3 erg cm⁻² s⁻¹ and n ∼ 100–1000), so we expect that the line emission from SDP11 is likely not dominated by an X-ray-dominated region associated with an AGN. Furthermore, since the currently available photometry of SDP11 can be fit solely by a star-formation-dominated spectral energy distribution (see Lupu et al. 2012), we proceed with an analysis of the observed [C ii] and [O i] lines and far-IR continuum emission within a star-formation-dominated PDR paradigm.

Kaufman et al. (2006) produce model grids of PDR regions over a range of densities and FUV fields, which we use for our modeling. In performing the PDR modeling, we assumed ∼70% of the observed [C ii] flux arise in the PDR, with most of the remaining 30% arising from the warm ionized medium (Oberst et al. 2006). The models of Kaufman et al. assume a single face-on cloud. In reality, the PDRs in SDP11 are more complex than used in Kaufman et al., and if we assume that they are spherical, then we will detect [C ii] and FIR emission from both the front and the back of the cloud, while we only see [O i] from the front because it is optically thick. To make use of the Kaufman et al. models, we must then multiply our observed [O i] line flux by two to account for the emission that can be self-absorbed along the line of sight by intervening atomic oxygen. Both of these corrections are appropriate for analysis within a star-formation paradigm (Kaufman et al. 1999). In Figure 4, we plot the corrected [C ii]/FIR and ([C ii] + [O i])/FIR flux ratios for SDP11. This gives a PDR model solution of G = 20,000 G₀ and a gas density n = 2500 cm⁻³. In principle, CO observations are also useful in constraining the PDR model. The observed CO lines are in agreement with our [C ii], [O i], and FIR-constrained model; however, because of the low signal-to-noise detection of these lines from SDP11, their inclusion does not help to constrain it further, as is evident by Figure 4.

Zoom In Zoom Out Reset image size

Our derived FUV field and gas density are consistent with values observed in both local ULIRGs (e.g., Luhman et al. 2003) and high-z sources dominated by AGN (e.g., Stacey et al. 2010), agreeing with our conclusions above that were drawn from the [C ii] to continuum ratio alone. While the observations of the [C ii], [O i], and FIR alone are not enough to unambiguously identify the nature of the source, describing their luminosities entirely within a star-formation paradigm suggests SDP11 most likely features a compact starburst similar to those in local ULIRGs. While this does not exclude the presence of an AGN, if one is present in SDP11 it likely does not contribute significantly to our observations or conclusions.

Having constrained the FUV field within SDP11, we estimate the size of the emission region, i.e., the size of the starburst, following the method from Stacey et al. (2010). This will provide additional clues to the nature of the source. Since the FIR emission arises from processing of the FUV radiation by dust, the ratio of the FIR luminosity to the FUV radiation field determines the size of the emitting regions. From Wolfire et al. (1990), the source diameter, D, is proportional to (λL_FIR/G)^1/3 if the mean-free path (λ) of a FUV photon is small. If instead the mean-free path of a FUV photon is large, then D∝(L_FIR/G)^1/2. The constant of proportionality of these relations is determined by assuming that the mean-free path in SDP11 is the same for M82, which has D ∼ 300 pc (Joy et al. 1987), G ∼ 1000 G₀ (Lord et al. 1996), and L_FIR ∼ 2.8 × 10¹⁰ L_☉ (the average of the values reported in Colbert et al. 1999, and Rice et al. 1988). The source diameter is then between 0.4 and 0.9 kpc if we assume that SDP11 is lensed a factor between 7 and 29, and we account for the extremes in mean-free path as described earlier. This agrees with the size of the dust emission determined in Lupu et al. (2012) who estimate the solid angle, μΩ_d, of the dust emission region in SDP11 to be 0.43 arcsec². Accounting for the magnification by gravitational lensing, this corresponds to a circular area with diameter between 0.6 and 1.36 kpc, albeit with significant caveats as described in Lupu et al. At the same time, our PDR region size is significantly smaller than the intrinsic source size obtained from our lensing model. This disagreement is not a big concern, however, and may even be expected. Because the lensing model is based on a 1.5 μm image, i.e., ∼397 nm rest frame, it is sensitive to the total stellar component in SDP11. Our PDR analysis is sensitive only to the massive and young O and B stars necessary for the ionization and excitation of [C ii]. As such, we expect the PDR-derived size to be smaller than our lensing model. This also implies that we cannot use our PDR source size to better constrain the results of our lensing model.

Our analysis suggests that SDP11 is a highly lensed analogue of a local ULIRG system, containing a compact star-formation region and concomitant high FUV fields, albeit in the early universe. Unfortunately, the true intrinsic source luminosity depends on our adopted magnification factor, which is not well constrained. A useful diagnostic plot that can help confirm our conclusion on the nature of SDP11 is the plot of L_[CII]/L_FIR versus L_CO(1–0)/L_FIR as described in Stacey et al. (2010) and shown in Figure 5. Because in the star-forming paradigm these lines and the FIR continuum arise in the same regions, these ratios should be insensitive to variations in magnification caused by slight difference in location of the source relative to the critical curve of the lens. In Figure 5, we identify typical regions occupied by nearby sources as well as FUV radiation and density contours from the PDR models of Kaufman et al. (2006). If a source falls in the lower right section of the plot with [C ii]/CO ⩽ 4100, then the observed flux to continuum ratios can be explained fully in a star-forming paradigm. Sources that fall in the upper left half of the plot may have significant [C ii] emission from non-PDR sources such as X-ray-dominated regions within AGN (Stacey et al. 2010), low-density ionized gas, or low-metallicity molecular clouds (Stacey et al. 1991).

Zoom In Zoom Out Reset image size

The lowest CO rotational transition detected from SDP11 is the CO 2–1 line from D. A. Riechers et al. (2013, in preparation). On the basis of the PDR model solution from Section 6, we can estimate the ratio of CO(2–1) to CO(1–0) and hence the strength of the CO(1–0) line. For our model-derived FUV field strength and gas density, the expected CO(2–1)/CO(1–0) line-integrated flux ratio is 7:1, agreeing with observations of both local and high-z systems (Downes & Solomon 1998; Bradford et al. 2003; Weiß et al. 2005a, 2005b; Ward et al. 2003; Israel 2009; Aravena et al. 2008). Using this estimate along with the observed [C ii] and FIR luminosity, we place SDP11 on the plot of L_[CII]/L_FIR versus L_CO(1–0)/L_FIR. The location of SDP11 clearly places it in the ULIRG region of the plot. More generally, because it falls in the allowed region of the plot, we have further support that our analysis in the PDR paradigm and PDR derived source size are correct. As such, SDP 11 does appear analogous to local ULIRGs in terms of luminosity, FUV field strength, and source size, and it may contains a starburst produced through the interaction between two Milky-Way-sized galaxies.

To further confirm the nature of SDP11, we can ask where it falls in the gas mass–star formation relation. Genzel et al. (2010) study the relations between the star-formation rate (SFR) and the molecular gas in galaxies from z ∼ 0 to ∼3. They find that quiescent star-forming galaxies at all epochs follow a similar relation between their FIR luminosity and their CO luminosity, tracers of star formation, and molecular gas, respectively. However, merging systems—both local ULIRGs and high-z mergers—lie above the sequence of quiescently star-forming galaxies, suggesting a higher star-formation efficiency in merging systems. Where is SDP11? The FIR luminosity and CO(2–1) luminosity scaled to the expected CO(1–0) luminosity as shown earlier, places SDP11 among the merging systems suggesting a merger-induced star-formation efficiency for SDP11 higher than "normal" galaxies. Over the range of allowed magnification factors, SDP11 falls among the samples of local ULIRGS (μ = 29) or high-z mergers (μ = 7) in the L_FIR–L_CO plane of Genzel et al. (2010). Both of these samples lie on a L_FIR–L_CO relation that is ∼ 4 times that of normal galaxies.

Studies over the past decade have shown ULIRGs in the early universe to be a very diverse population. Evidence for different modes powering high-z ULIRGs, including AGN, mergers, and the accretion of gas from the intergalactic medium, have all been observed. The discovery of the latter mode, however, was very surprising (cf. Biggs & Ivison 2008; Tacconi et al. 2010; Iono et al. 2009, Hailey-Dunsheath et al. 2010; Stacey et al. 2010). In local ULIRGs, star formation is triggered by mergers of massive galaxies (Sanders & Mirabel 1996), and it was long thought that only an AGN or similar major-merger event could produce of the extreme luminosities in high-z systems. We now know the star-formation process in some z ∼ 1–3 systems can also be stimulated through the accretion of cold gas (Tacconi et al. 2013; Genzel et al. 2010; Daddi et al. 2007). The large scale of these starbursts are best understood as Schmidt–Kennicutt law star formation, with star-forming efficiencies similar to local "normal" galaxies, but with starburst-like SFRs arising from the large molecular-gas reservoirs accreted from the cosmic Web. Studies revealing the accretion mode in high-z galaxies have created a paradigm shift in our thinking about galaxy evolution and formation in the early universe. Recent models have even demonstrated that cold accretion and typical Kennicutt-type star-formation efficiencies can fully account for observed black hole growth and stellar mass assembly in the early universe (Di Matteo et al. 2012; Keres et al. 2005; Genel et al. 2012; Agertz et al. 2011; Kitzbichler & White 2007).

This is not so say that the mergers are uncommon or unimportant in the early universe. For one, observational and theoretical evidence suggests a higher merger rate in the past (Lotz et al. 2011; Hopkins et al. 2010; Bertone & Conselice 2009). Furthermore, the relative importance and prevalence of merger-driven star formation versus accretion-driven star formation in the early universe is still highly contested because of observational challenges in identifying merger events at high-z as well as measuring the full extent and size of gas disks. With the advent of high spatial-resolution submillimeter/millimeter observatories (e.g., PdBI, ALMA, and the Submillimeter Array), we are now beginning to adequately address this question.

Recently, Bothwell et al. (2013) examined the CO and FIR emission of 40 luminous (L_FIR ≳ 10¹² L_☉) galaxies at z ∼ 1.2–4.1 using the PdBI and selected for in the submillimeter (i.e., submillimeter galaxies, SMG). The authors conclude that 20%–28% of their sources show signs of an ongoing merger. Yet, within the observational constraints, there is no evidence suggesting increased star-formation efficiencies as one might expect for a merger and as seen in other works (e.g., Genzel et al. 2010). Another indicator of the star-formation efficiency in a galaxy is where it falls in the SFR and stellar mass relation (SFR–M_*), which has shown a tight relation between the SFR and stellar mass in normal galaxies (i.e., accretion mode), with them falling along a "main sequence" (Brinchmann et al. 2004; Noeske et al. 2007; Elbaz et al. 2007). Merging systems lie above the main sequence (Elbaz et al. 2011). Hung et al. (2013) examine the SFR–M_* relation and morphology of the 0.2 < z < 1.5 Herschel-selected galaxies. On the basis of morphological classification Hung et al. find that the fraction classified as "irregular" (indicating a merger) increases with IR luminosity at all values of M_*. Furthermore, they find that the fraction of interacting or merging systems increases with L_IR as well as distance above the main sequence, with nearly 50% of galaxies showing evidence of a merger at L_IR > 10^11.5 L_☉. Perhaps most interesting, however, is Hung et al.'s finding that ≳ 18% of IR-luminous galaxies on the main sequence show evidence of interactions and mergers, suggesting that evolution of galaxies on the main sequence may not be explained by gas accretion and exhaustion alone.

It is clear that there is much work yet to be done in understanding the various modes of star formation and their impact in the early universe. For example, one might ask, what is the product of a major merger at early times? One possibility is that mergers, while not responsible for the bulk of star formation in the universe, are responsible for the formation of modern-day giant ellipticals. Fu et al. (2013) report on the discovery of a system, HXMM01, featuring two submillimeter galaxies in the early stages of a merger. On the basis of the gas-depletion time scales and SFRs, these authors conclude that HXMM01 may be the progenitor of a modern day elliptical. The merging galaxies in HXMM01 feature star-forming regions of ∼1.4 kpc in diameter, similar to the size of the star-forming regions we estimate for SDP11. Perhaps then SDP11 represents a later stage in the merger process than HXMM01, in which the galaxies have already coalesced.

To test the various models of galaxy formation and constrain galaxy-evolution scenarios, we need to look for signs of different modes of star formation, i.e., accretion- versus merger-driven. Yet, identifying merging systems has proven to be a challenge at high-z. Morphological determinations such as those in Hung et al. (2013) require both high sensitivity and high spatial resolutions. Until recently, this was easiest to do at visible wavelengths, which is compromised by the large dust content of many IR luminous systems at high-z. Now, however, ALMA allows us to probe the morphologies of early galaxies in their rest-frame IR emission, such as the work in Fu et al. (2013) mentioned earlier, reducing the challenges of morphological classification. Using ZEUS-2, we can perform a census of the high-z analogs to local ULIRGs by observing their fine-structure line emission. By combining ZEUS-2 and ALMA observations, we can morphologically calibrate various FIR fine-structure line ratios in high-z systems to either merger- or accretion-driven modes of star formation instead of relying on extrapolations from local systems. This will allow us to classify the mode of star formation in systems that are spatially unresolved by ALMA. Last, in addition to assessing their total numbers, we can hope to identify the numbers of high-z ULIRG analogues in their various stages of merger to provide detailed constraints on the peak of the major merger rate in the universe and the effects of mergers on galaxy properties.

We have built a new submillimeter spectrometer, ZEUS-2, and recently commissioned it on the APEX telescope detecting the [C ii] 158 μm line from SDP11 at z ∼ 1.8. Combining our ZEUS-2 observations with a new Herschel detection of the [O i] 63 μm line, archival HST images, and data from the literature we determine the following.

• 1.  
  SDP11 is strongly gravitationally lensed with magnification factor μ ∼ 7–29.
• 2.  
  An analysis of the [C ii], [O i], and FIR emission within a PDR paradigm yields FUV fields of G ∼ 20,000 G₀, gas densities of ∼2300 cm⁻³, and a source size between 0.4 and 0.9 kpc in diameter for SDP11, all of which are similar to values seen in local ULIRGs.
• 3.  
  We conclude that SDP11 is likely a high-z analog of local ULIRGs featuring a compact and intense starburst confined to sub-kiloparsec scales and induced by a merger of two or more galaxies.

This work demonstrates the science that ZEUS-2 is capable of providing. Because of its synergy with ALMA and its being the only spectrometer capable of observing between 200 and 300 μms in the post-Herschel era, we expect ZEUS-2 will be an important tool for studying galaxies in the years to come.

Part of this work is based on observations made with the NASA/ESA Hubble Space Telescope, and obtained from the Hubble Legacy Archive, which is collaboration between the Space Telescope Science Institute (STScI/NASA), the Space Telescope European Coordinating Facility (ST-ECF/ESA), and the Canadian Astronomy Data Centre (CADC/NRC/CSA). ZEUS-2 development and observations are supported by NSF grants AST-0705256, AST-0722220, AST-1105874, and AST-1109476; NASA grant NNX10AM09H; and a grant from Georgia Southern University. C.F. thanks R. Wang for sharing her list of high-z [C ii] detections from the literature and in Wang et al. (2013). Last, the authors thank the anonymous referees for their helpful comments and acknowledge the APEX staff whose excellent support helped to make this work possible.