THE STRIKINGLY UNIFORM, HIGHLY TURBULENT INTERSTELLAR MEDIUM OF THE MOST LUMINOUS GALAXY IN THE UNIVERSE

Studying the most luminous galaxies in the universe is a key step to understanding how the most massive galaxies assemble and evolve (e.g., Hopkins et al. 2008)—all the way from their first appearance, about 500 million years after the Big Bang (Zheng et al. 2012), until today. Among the different populations of luminous galaxies currently known, Hot dust-obscured galaxies (Hot DOGs; Eisenhardt et al. 2012), discovered by NASA's Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010), are the most extreme in terms of their luminosities and unusual hot dust temperatures (Wu et al. 2012; Tsai et al. 2015). The infrared (IR) emission from Hot DOGs is dominated by obscured accretion onto a central supermassive black hole (SMBH), in most cases without significant contribution from star formation (Jones et al. 2014; Wu et al. 2014; Assef et al. 2015). Detailed analyses show that most Hot DOGs' rest-frame UV-near-IR spectral energy distributions (SEDs) are consistent with typically star forming, not strongly obscured host galaxies. The large contrast between the underlying galaxy and the hyper-luminous active galactic nucleus (AGN) implies that either the SMBH is much more massive than expected for the host's stellar mass or is radiating well above its Eddington limit (Assef et al. 2015; Tsai et al. 2015). Such luminous AGNs, hosted by otherwise normal galaxies, are likely to be at a key stage of their evolution—when feedback from the AGN may have started quenching star formation—just about to become regular quasars and decay into dead elliptical galaxies (Sanders et al. 1988). W2246–0526 is the most extreme of these remarkable systems. At a redshift of z = 4.601  ±  0.001, it is the most luminous galaxy known in the universe (Tsai et al. 2015).

Spatially resolving the dust emission and kinematics of the interstellar medium (ISM) provides unique insight into how extreme AGN feedback may be operating. In normal star-forming galaxies and starbursts, [C ii] emission is most likely produced in photodissociation regions (PDRs; see, e.g., Stacey et al. 2010; Carilli & Walter 2013; Díaz-Santos et al. 2013) associated with a source of intense far-UV (FUV) flux. On the other hand, it is not expected to be significantly powered by direct AGN photoionization, which would progressively increase the ionization state of carbon beyond ${{\rm{C}}}^{+}$ (Langer & Pineda 2015). Furthermore, without considerable dust shielding (e.g., Díaz-Santos et al. 2010; Esquej et al. 2014), polycyclic aromatic hydrocarbon (PAH) molecules can be photodissociated within the central kiloparsecs of a powerful AGN (Voit 1992). However, galaxies both with and without an AGN have roughly constant [C ii] /PAH ratios, independent of dust luminosity and temperature (Helou et al. 2001; L. Armus et al. 2016, in preparation), suggesting that AGNs do not heat C⁺ directly. It is possible, however, that turbulent heating related to the injection of mechanical energy from shocks produced by AGN jets or supernovae winds may substantially power [C ii] emission (Appleton et al. 2013; Guillard et al. 2015).

In this Letter, we present spatially resolved [C ii] observations obtained with ALMA of the exceptional ISM in W2246–0526, the most luminous galaxy known in the universe. Throughout the paper, we have used a cosmology with the following parameters: ${{\rm{\Omega }}}_{{\rm{M}}}$ = 0.3, ${{\rm{\Omega }}}_{{\rm{\Lambda }}}$ = 0.7, and H₀ = 70 km s⁻¹ Mpc⁻¹.

As part of an ALMA cycle-2 campaign (ID: 2013.1.00576.S; PI: R. Assef) aimed at studying the ionized gas in the most luminous Hot DOGs, we were awarded 8.1h in bands 7 and 8 to study the redshifted 157.7 μm (${}^{2}{P}_{3/2}\;\to {\;}^{2}{P}_{1/2}$) fine-structure transition of ionized carbon, [C ii], and the underlying dust continuum emission in a sample of Hot DOGs spanning a range of redshifts and luminosities. Observations of W2246–0526, the first galaxy of the sample observed, were carried out in band 7 using four spectral windows in the largest bandwidth mode (1.875 GHz). The [C ii] line was placed at the center of the reference spectral window (spw0), assuming the redshift derived from optical lines (z = 4.593; Wu et al. 2012); the other three spectral windows probing the continuum blueward of the line. The on-source integration time was 956 s in array configuration C34–4. Flux, phase, and bandpass calibrators were also obtained, for a total time of 3330 s.

We used the CASA software to process and clean the data. The cleaning algorithm was run using a natural weighting for the u–v visibility plane.⁵ The angular resolution of the observations (clean beam FWHM) is 0386 × 0356, with a major-axis position angle (P.A.) of 47° (see Figure 1). The standard deviation of the observations is 0.6 mJy/beam, measured in 35 km s⁻¹ channels. The [C ii] emission line (${z}_{[{\rm{C}}\;{\rm{II}}]}=4.601\;\pm \;0.001$) was observed in a spectral window that missed part of the red wing of the line, due to a significant blueshift of the UV/optical emission lines as compared with the ISM, a difference that is common in high-redshift dusty galaxies and quasars (e.g., Kimball et al. 2015; Willott et al. 2015).

Zoom In Zoom Out Reset image size

A Gaussian fits the emission profile well and is consistent with only 7% of the total line flux lying outside of the observed window (see Figure 2, left). The total Gaussian flux measured from a spectrum extracted with a 1'' diameter circular aperture centered on W2246–0526 is 9.3  ±  0.3 Jy km s⁻¹. The observed-frame continuum flux density underlying the emission line is 7.4  ±  0.6 mJy. Asymmetric 2D Gaussian fits to the [C ii] line and dust continuum emission provide widths of 0539 × 0511 (±0018 × 0018) with P.A. = 122° and 0433 × 0407 (±0023 × 0023) with P.A. = 33°, respectively.

Zoom In Zoom Out Reset image size

Our ALMA data reveal strong, spatially resolved emission in both the [C ii] and dust continuum. W2246–0526 is also detected in Hubble Space Telescope (HST) WFC3 rest-frame UV images (F160W; see Figure 1; C. Bridge et al. 2016, in preparation), which reveal resolved emission from unobscured, young, massive stars and the AGN. Furthermore, we detect [C ii] and UV emission from two additional, less luminous galaxies located 6.6 kpc northwest and 33 kpc southeast of W2246–0526, marked NW and SE, respectively, in the top panel of Figure 1. The spectra of W2246–0526 and its two companions are presented in Figure 2. The moment maps are shown in Figure 3 and will be discussed in Section 4. The ALMA and HST morphologies support the idea that gravitational lensing is not important in Hot DOGs (Wu et al. 2014; Tsai et al. 2015). This is in strong contrast to many other well-studied dusty, IR-bright, high-redshift systems (e.g., Vieira et al. 2013).

Zoom In Zoom Out Reset image size

Accounting for the contribution of the ALMA beam, the intrinsic size (FWHM) of the resolved [C ii] emission is 2.5  ±  0.3 kpc, very similar to that of the HST UV (see Figure 1). This suggests that a significant fraction of the [C ii] emission may be associated to extended star formation in the host galaxy (see below). In contrast, the continuum size is only 1.3  ±  0.5 kpc wide, pointing to a greater contribution from the AGN. Such a difference between ionized gas and dust emission sizes is often seen in high-redshift quasars (Wang et al. 2013; Cicone et al. 2015) and star-forming galaxies (Capak et al. 2015), although it is not ubiquitous (Kimball et al. 2015). In starbursting galaxies most of the [C ii] emission arises from dense PDRs close to massive stars, with a possible contribution from low-density, warm ionized gas heated by the dissipation of turbulent energy in shocks (Appleton et al. 2013). In addition, although AGN emission is not expected to power [C ii] (Langer & Pineda 2015), AGN-driven outflows are sometimes able to periodically inject enough energy into the ISM to enhance the [C ii] emission and drastically affect the kinematics of the gas in the host galaxy (Guillard et al. 2015; see Section 4.3).

In order to study the underlying emission from the host galaxy in W2246–0526, we used the clean beam of the observations to subtract an unresolved central component from both the [C ii] and dust continuum images. This provides the first insight into the spatial distribution of ionized gas in the host galaxy of a hyper-luminous AGN (see Figure 4). The beam was scaled to minimize the dispersion in the central 05 region of the residual map, which required scaling factors of 79% and 89% of the peak flux in the line and continuum, respectively. Within a 2'' diameter aperture, the subtracted images retain 67% and 28% of the total luminosity in [C ii] and the continuum, respectively. Thus, most of the [C ii], and around a quarter of the dust continuum emission originates several kiloparsecs away from the AGN. The residual extended emission structure is oriented consistently with the mild velocity gradient discussed in Section 4.2 and with a surface brightness that varies by as much as a factor of three across the galaxy. Figure 4 also shows a resolved map of the rest-frame [C ii] equivalent width from the underlying host galaxy of W2246–0526. The values are very uniform, between ∼1–2 μm, similar to those found in local luminous IR galaxies (LIRGs; Díaz-Santos et al. 2013), and consistent with the host being an actively star-forming galaxy.

Zoom In Zoom Out Reset image size

4.1. The [C ii]/FIR Deficit

The [C ii]-to-far-IR${}_{[42-122\;\mu {\rm{m}}]}$ ([C ii]/FIR) emission ratio provides insight into the physical state of the ISM in galaxies. In star-forming systems, this ratio is sensitive to the average intensity of the ionizing radiation field (Kaufman et al. 1999), with pure star-forming galaxies having values an order of magnitude greater than AGN hosts (Díaz-Santos et al. 2013). The [C ii]/FIR ratio is reduced in the nuclei of galaxies, in compact starbursts, and in general in regions with greater luminosity surface densities, regardless of the presence of an AGN (Díaz-Santos et al. 2014). However, note that all galaxies harboring deeply obscured AGNs, like W2246–0526, show [C ii]/FIR ratios less than $\sim {10}^{-3}$ (Stacey et al. 2010; Wagg et al. 2010; Carilli & Walter 2013; Díaz-Santos et al. 2013).

The total [C ii] luminosity of W2246–0526 is ${L}_{[{\rm{C}}\;{\rm{II}}]}$ = 6.1(±0.4) $\times \;{10}^{9}\;{L}_{\odot }$, and a multi-wavelength fit to its integrated SED yields a conservative estimate of its IR luminosities: ${{L}_{{\rm{IR}}}}_{[8-1000\;\mu {\rm{m}}]}$ = 2.2 (±0.4)$\;\times \;{10}^{14}\;{L}_{\odot }$ and ${{L}_{{\rm{FIR}}}}_{[42-122\;\mu {\rm{m}}]}$ = 3.5(±0.4)$\;\times \;{10}^{13}\;{L}_{\odot }$.⁶ This results in an integrated [C ii]/FIR ratio of $\sim 2\;\times \;{10}^{-4}$ (see Figure 5), which is likely an upper limit to the nuclear value since the [C ii] emission is twice as extended as the dust continuum (Section 3). Nevertheless, this ratio is still only matched in the most obscured nuclei of local dusty LIRG systems, like Mrk 231, the classical nearby archetype of a dust-enshrouded AGN. Yet, the [C ii] luminosity of W2246–0526 is more than one order of magnitude greater than that of Mrk 231 and the velocity dispersion is $\gt 2$ times larger (Fischer et al. 2010), implying a much more massive and turbulent environment. Using the resolved, rest-frame 157.7 μm continuum emission (Section 2) as a measure of the galaxy size, the IR luminosity surface density of W2246–0526 is ${{\rm{\Sigma }}}_{{\rm{IR}}}=1.6\;\times \;{10}^{14}\;{L}_{\odot }$ kpc⁻², which is in good agreement with the extrapolation of the correlation between the [C ii]/FIR ratio and ${{\rm{\Sigma }}}_{{\rm{IR}}}$ found for local LIRGs and high-z sub-millimeter galaxies (SMGs; Díaz-Santos et al. 2013, 2014; Yun et al. 2015). However, note that W2246–0526 differs substantially from the objects used to establish the local relation as the AGN dominates the total luminosity over the starburst.

Zoom In Zoom Out Reset image size

4.2. The Strikingly Uniform Velocity Dispersion of W2246–0526

4.3. An Isotropic, Large-scale Blowout of the Gas

The observed kinematic properties of the [C ii] line can be compared quantitatively with the expected stability of such a luminous system in terms of the energy and momentum injected by the AGN into the ISM. For a given velocity dispersion, σ, a system with a luminosity greater than

$$\begin{eqnarray}&&{L}_{{\rm{M}}}=\left(\displaystyle \frac{4{f}_{{\rm{g}}}c}{G}\right){\sigma }^{4}\simeq 7.8\;\times \;{10}^{12}\;{f}_{{\rm{g}}\;{\rm{0.1}}}\;{\sigma }_{200}^{4}\;{L}_{\odot }\end{eqnarray} \tag{ 1 }$$

cannot be stable in terms of the radiation pressure of photons on dust grains alone (Murray et al. 2005; King et al. 2011); ${f}_{{\rm{g}}}$ is the gas fraction, with ${f}_{{\rm{g}}\;{\rm{0.1}}}$ = ${f}_{{\rm{g}}}$/0.1 and ${\sigma }_{200}$ = σ/200 km s⁻¹. ${L}_{{\rm{M}}}$ can be understood here as an Eddington limit for an isotropic, momentum-driven wind. Even for an unrealistic extreme ${f}_{{\rm{g}}}\;\approx \;$ 1, the bolometric luminosity of W2246–0526 is at least twice the derived threshold (∼1.5 × ${10}^{14}\;{L}_{\odot }$), suggesting that the AGN is likely blowing away all the gas from the galaxy. Furthermore, in this scenario, the ejection of the gas would not be bipolar (like in M82) but rather isotropic (Murray et al. 2005). Indeed, the striking uniformity of the high-velocity dispersion resolved right across the galaxy strongly suggests that turbulent gas is expanding in all directions.

A similar argument can be made in terms of the total energy that is being injected into the ISM to create an expanding isothermal bubble through an energy deposition-driven wind (Murray et al. 2005; King et al. 2011). Here the limiting luminosity is given by

$$\begin{eqnarray}{L}_{{\rm{E}}} & \sim & \left(\displaystyle \frac{100}{\xi }\right)\left(\displaystyle \frac{4{f}_{{\rm{g}}}c}{G}\right){\sigma }^{5}\\ & \simeq & 5.2\;\times \;{10}^{12}\;{f}_{{\rm{g}}\;{\rm{0.1}}}\;{\xi }_{0.1}^{-1}{\sigma }_{200}^{5}\;{L}_{\odot },\end{eqnarray} \tag{ 2 }$$

where ξ is the efficiency of energy transfer into the ISM, with ${\xi }_{0.1}$ = ξ/0.1. W2246–0526 also exceeds this stability threshold, even for high gas fractions and low efficiencies. This matches observations of high-velocity, energy-driven molecular outflows in local galaxy mergers (Sturm et al. 2010; Veilleux et al. 2013; Tombesi et al. 2015), but in the case of W2246–0526 at much larger scales, more energetic, and with a higher degree of isotropy, as suggested by the observed uniformity of its kiloparsec-scale velocity dispersion.

The [C ii] emission from W2246–0526 has a large, uniform velocity dispersion and low shear. The system is unstable in terms of both the energy and momentum injected by the accreting SMBH into the ISM. This strongly argues that W2246–0526 is blowing out its ISM isotropically, in an unprecedented, homogeneous, large-scale turbulent outflow with a greater covering factor than the winds seen in high-redshift QSOs and local ultra-luminous IR galaxies. This suggests that W2246–0256 is near to bursting out of its dusty cocoon to become a powerful optically visible QSO. Caught at a time when the universe was ramping up to its peak of star formation and SMBH accretion, our ALMA observations clearly reveal extreme conditions in the ISM of the most luminous galaxy known, where the feedback from the powerful AGN is having a strong impact on the evolution and fate of the entire galaxy (Silk & Rees 1998).

The authors would like to thank H. Jun for suggestions provided during the analysis of the data. T.D.-S. acknowledges support from ALMA-CONICYT project 31130005 and FONDECYT 1151239. R.J.A. acknowledges support from Gemini-CONICYT project 32120009 and FONDECYT 1151408. The work of C.-W.T., P.E., D.S., and C.B. was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. M.A. acknowledges partial support from FONDECYT through grant 1140099. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada) and NSC and ASIAA (Taiwan), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.