The multiple merger assembly of a hyperluminous obscured quasar at redshift 4.6

See allHide authors and affiliations

Science  30 Nov 2018:
Vol. 362, Issue 6418, pp. 1034-1036
DOI: 10.1126/science.aap7605

Mergers drive a powerful dusty quasar

Massive galaxies in the early Universe host supermassive black holes at their centers. When material falls toward the black hole, it releases energy and is observed as a quasar. Astronomers found a population of powerful distant quasars that are obscured by dust, but it has been unclear how they are formed. Díaz-Santos et al. observed the dust-obscured quasar WISE J224607.56-052634.9 at submillimeter wavelengths, finding three small companion galaxies connected to the quasar by bridges of gas and dust. They inferred that galaxy mergers can provide both the raw material to power a quasar and large quantities of dust to obscure it.

Science, this issue p. 1034


Galaxy mergers and gas accretion from the cosmic web drove the growth of galaxies and their central black holes at early epochs. We report spectroscopic imaging of a multiple merger event in the most luminous known galaxy, WISE J224607.56−052634.9 (W2246−0526), a dust-obscured quasar at redshift 4.6, 1.3 billion years after the Big Bang. Far-infrared dust continuum observations show three galaxy companions around W2246−0526 with disturbed morphologies, connected by streams of dust likely produced by the dynamical interaction. The detection of tidal dusty bridges shows that W2246−0526 is accreting its neighbors, suggesting that merger activity may be a dominant mechanism through which the most luminous galaxies simultaneously obscure and feed their central supermassive black holes.

Structure formation in the early Universe proceeded through the hierarchical assembly of dark matter haloes and the galaxies they host, with the densest structures collapsing first (1). During the periods of most intense accretion and growth, galaxies and their central supermassive black holes (SMBHs) are expected to be obscured by interstellar gas and dust (2). The obscuring material absorbs the ultraviolet and optical light from stars and the active galactic nucleus (AGN) powered by the SMBH and re-emits it in the infrared (2).

The Wide-field Infrared Survey Explorer (WISE) (3) space telescope studied the formation and evolution of galaxies in the high-redshift Universe. A population of hyperluminous obscured quasars were found at redshifts ≳2, with bolometric luminosities Lbol ≳ 1013 L (4, 5), where Lbol is the luminosity integrated across the entire electromagnetic spectrum, and L is the luminosity of the Sun. Known as hot, dust-obscured galaxies (hot DOGs), these systems are mainly powered by accretion onto their central SMBHs, which may be radiating close to the limit allowed by their own gravity (6). The dominant mechanism supplying the material necessary to sustain such high luminosities remains unknown. Rapid growth of galaxies and SMBHs can be accomplished via galaxy mergers, which would effectively funnel low angular momentum gas into the central AGN (7, 8). If hot DOGs form in over-dense environments, merger-driven instabilities could deliver large amounts of gas and dust to the galaxy. However, there is only indirect, statistical evidence that this is the case. Combined observations of 10 hot DOGs at 850 μm reveal more than twice as many sources within 1.5 arc min as in random fields (9), and shallow near-infrared (NIR) images show that the number density of red sources within 1 arc min of hot DOGs is, on average, higher than that of field galaxies (10). In submillimeter observations of 10 hot DOGs, the cumulative number counts of companion sources also support dense environments around hot DOGs (11). However, the morphological evidence for dynamical interactions in individual systems remains elusive, and observations obtained with the Hubble Space Telescope (HST) of a sample of hot DOGs at redshift z ~ 2 have yielded ambiguous results (12, 13).

With Lbol = 3.5 × 1014 L (14, 15), the hot DOG WISE J224607.56−052634.9 (hereafter W2246−0526) is the most luminous galaxy known. Previous observations with the Atacama Large Millimeter/Sub-millimeter Array (ALMA) of the ionized carbon ([C ii]) emission line at 158 μm (16) have shown W2246−0526 is located at a redshift of 4.601, which is equivalent to ~1.3 billion years after the Big Bang, assuming standard cosmological parameters (15). The line profile shows a uniform, large-velocity dispersion, with a full-width at half-maximum (FWHM) of ~500 to 600 km s−1 across the whole galaxy where emission is detected (~2.5 kpc) (16). This suggests a highly turbulent interstellar medium (ISM), likely resulting from the energy and momentum injected by the central SMBH into the surrounding gas.

The [C ii] observations had also shown two nearby companion galaxies to W2246−0526 (16). We detected a third companion in a blind search for emission line sources in the data cube, which we confirmed via its Lyman-α emission line in an optical spectrum obtained with the Keck telescope (15). The star-formation rates (SFRs) of the companions based on their [C ii] luminosities are at least 7 to 27 solar mass (M) year−1 (15). Other recent studies have also identified companion galaxies close to unobscured high-redshift quasars. ALMA [C ii] observations of luminous quasars at z ~ 4.8 and >6 (17, 18) show that a large fraction of them are accompanied by actively star-forming galaxies at projected distances <100 kpc and within radial velocities ≲600 km s−1. However, there were no direct morphological signatures that show dynamical interaction between the companion galaxies and central source.

We present deep ALMA observations of the dust continuum emission at rest frame 212 μm in W2246−0526 at an angular resolution of ~0.5 arc sec, which is equivalent to ~3.3 kpc at that distance (15). The dust continuum map shown in Fig. 1 reveals bridges of material connecting the central galaxy to the companions, which we denote C1, C2, and C3. The detection of dust indicates that the gas associated with these structures has been already enriched with elements heavier than hydrogen or helium and is thus not primordial. C2 has a stream of dust extending like a tidal tail all the way to W2246–0526, over at least 35 kpc. One of the densest regions in this structure, denoted K1 and located ~1.5 arc sec northwest of C2 (Fig. 1), has a counterpart identified in rest-frame near ultraviolet (UV) emission (Fig. 2). The UV image was obtained with the HST by using the NIR F160W filter and was previously analyzed in (16). The detection of the UV counterpart suggests that at least a fraction of the dust in the tidal tail could be heated through in situ star formation (15). Low-surface-brightness dust emission south of C3 is coincident with strong UV emission seen in the HST image as well (denoted U1) (Fig. 2), although this source could be at a different redshift because it is not detected in [C ii] line emission. Two more sources with unknown redshifts (U2 and U3) are also identified 6 arc sec to the northwest and 8.5 arc sec to the southwest of W2246−0526, respectively, in the HST image.

Fig. 1 ALMA 212 μm dust continuum emission map of the W2246−0526 merger system.

The color bar shows the flux density on a logarithmic scale in units of millijansky (1 mJy = 10−26 erg s−1 cm−2 Hz−1). North is up, and east is to the left. The angular resolution (beam size FWHM) of the observations is 0.54 by 0.46 arc sec, or ~3.6 by 3.1 kpc at the redshift of W2246−0526, and it is shown by the ellipse at the bottom left corner. The offsets in the right ascension and declination axes are given in arc seconds relative to the center of W2246−0526, whose coordinates are: 22 hours 46 min 07.55 s, −05° 26′ 35.0″. The relative velocities of three companion galaxies (labeled as C1, C2, and C3) and the redshift of W2246−0526 are measured via the [C ii] emission line (16) and suggest that W2246−0526 and its companions are gravitationally bound. A stream of dusty material resembling a tidal tail connects W2246−0526 with C2, and bridges join the central galaxy with C1 and C3. Three sources with unknown redshifts and the knot ~ 1.5 arc sec northwest of C2 are labeled as U1, U2, U3, and K1, respectively. Solid contours represent levels of [2.5, 3, 4, 6, 9, 15, 30, 50] × σ, where σ is the measured root mean square (RMS) of the background. Dotted contours indicate [–2.5, –3] × σ negative flux. An equivalent map with lower-significance contours is shown in fig. S2.

Fig. 2 Rest-frame near UV (~2860 Å) continuum image of W2246−0526 with overlaid contours of the 212 μm dust-continuum map.

The color bar shows the near-UV flux density on a logarithmic scale. The emission at the top left corner is from a low-redshift foreground galaxy. The companion galaxies C1, C2, and C3 are detected in both the dust continuum and near-UV emission, as are the sources with unknown redshifts (U1, U2, and U3) and the tidal tail knot (K1).

The detection of streams of dust emission on such large physical scales suggests that W2246–0526 is in the process of accreting its neighbors—or at least stripping a large fraction of their gas—and provides evidence that (i) merger activity is taking place and (ii) the entire system may be undergoing a morphological transformation. Gas and dust accretion triggered by galaxy mergers can provide strong, yet probably intermittent, influx of material toward the nuclei of high-redshift hyperluminous galaxies, simultaneously feeding and obscuring their SMBHs.

We performed additional observations with the Karl G. Jansky Very Large Array (VLA) of the J = 2 → 1 transition line of carbon monoxide (CO) in W2246−0526, where J is the rotational quantum number (15). The luminosity of low-J transitions of the CO molecule is regularly used as a proxy for the cold molecular gas content of galaxies. In Fig. 3, we compare the CO(2→1) line map with contours of the 212 μm dust continuum. The CO(2→1) emission is marginally resolved (the beam FWHM is 2.47 by 2.01 arc sec, which is equivalent to ~16 by 13 kpc), with tentative low-surface-brightness regions extending toward the companion C3 and in the direction of the tidal tail. The FWHM of the CO line in the central beam is ~600 km s−1, which is similar to that of the [C ii] line (fig. S3) (16), suggesting that the cold molecular gas phase of the ISM traced by CO in W2246–0526 is also very turbulent and probably affected by the strong feedback from the AGN on scales of at least a few kiloparsecs.

Fig. 3 CO(2→1) emission line map of W2246−0526 with overlaid contours of the 212 μm dust-continuum map.

The color bar shows the line flux per beam on a logarithmic scale. The angular resolution (beam size FWHM) of the observations is 2.47 by 2.01 arc sec, or ~16 by 13 kpc at the redshift of W2246−0526, and it is illustrated by the dashed ellipse. The dashed contours represent CO levels of [2.5, 3, 4, 6, 9] × σ, where σ is the measured RMS of the background. The CO(2→1) emission is only marginally resolved, slightly extending toward the companion C3, northeast of W2246–0526, and southeast in the direction of the tidal tail.

The dust mass in W2246−0526 alone (within the central 1 arc sec ~ 7 kpc) is in the range of 5.6 × 108 to 17 × 108 M (assuming a dust temperature of 100 to 50 K) (15). This is similar to the total dust mass content of dusty ultraluminous infrared galaxies (ULIRGs) in the nearby Universe, which span a range between ~108 and 109 M (although W2246−0526 is 100 times more luminous) (19). The rest of the system—including the companion galaxies C1, C2, and C3 and the extended emission—contains at least as much dust as W2246−0526 alone (table S1). The three companions contribute ~25% of the dust mass outside W2246−0526 and ~13% of the entire merger system. U2 and U3 are not included in this calculation because they may not be part of the system. The tidal tail contains almost as much dust as the sum of that of all three companion galaxies. Assuming that the dust and gas are well mixed and a standard gas-to-dust ratio (δGDR) that is typical of local, solar-metallicity galaxies (20), δGDR = 100, W2246−0526 harbors a total gas mass (Mgas) reservoir of ~0.6 × 1011 to 1.7 × 1011 M, with the entire system containing Mgas ~ 1.2 × 1011 to 3.6 × 1011 M. The total molecular gas mass estimated from the CO(2→1) line is 1.5 (± 0.8) × 1011 M (15), which is in agreement with the estimate from the dust.

On the basis of the [C ii] kinematics, we calculated the dynamical mass (Mdyn) of W2246−0526 to be ~0.8 (± 0.4) × 1011 M, which is within a factor of ~2 of the baryonic mass of the galaxy (15), which is similar to observations of some compact galaxies at z ~ 2 (21). The dynamical mass favors the lower bound of the dust-based gas estimate (15), which is as expected if most of the dust within the central core of W2246−0526 (within a few kiloparsecs) is being heated to temperatures >100 K because of its closer proximity to the central AGN.

The Mgas and stellar mass (M) in the W2246−0526 merger system imply a baryonic gas fraction fgas ~ 0.3 to 0.6 [where fgas = Mgas/(M+Mgas)], which is lower than the value expected for most galaxies at similar redshifts (22). The SFR within the central ~4 arc sec is estimated to be ~560 M year−1, with a factor of two uncertainty (15). This translates into a specific SFR (SSFR; the star formation rate divided by the stellar mass) of ~2.2 billion year−1, which is equivalent to a mass doubling-time of ~450 million years, both with an uncertainty of a factor of three. The derived SSFR is only slightly lower than that of main-sequence galaxies at redshifts of 3.5 to 5 (22, 23). However, the gas depletion time scale is only ~125 million to 700 million years (15), which is between main-sequence and starburst galaxies at approximately the same redshift (22). The free-fall time of the gas in the system is ~100 million to 170 million years (15), which is at least an order of magnitude larger than the active period of hyperluminous quasars (24) and suggests that the hot DOG phase is shorter than the dynamical time scale of the merger. On the basis of the Mgas of the tidal tail and companion galaxies, the average accretion rate of gas toward the center of W2246−0526 could be as high as dMgas/dt ~ 550 to 900 M year−1 (15), which is similar to the estimated SFR of the underlying galaxy.

We interpret these results as showing that merger-driven accretion of neighbor galaxies can be a catalytic mechanism that simultaneously (i) obscures the central SMBH in W2246−0526 under large columns of dust and gas and (ii) provides the intermittent, large-scale influx of material needed to generate its extreme luminosity and maintain star formation in the host galaxy, which would otherwise quickly deplete its gas reservoir. The energetic AGN feedback resulting from this accretion is likely responsible for maintaining the turbulence of the gas at the center of W2246−0526. Slow, nearly isotropic ISM outflows on scales of a few kiloparsecs can coexist with the accretion of material stripped from in-falling galaxies at larger scales (even if the companions themselves may only fly by), which can be funneled efficiently into the central AGN through collimated, filamentary structures (2527). If W2246−0526 is representative of the hot DOG population, our results suggest that hyperluminous obscured quasars may be interacting systems, the result of ongoing merger-driven peaks of SMBH accretion and massive galaxy assembly in the early Universe.

Supplementary Materials

Materials and Methods

Figs. S1 to S4

Table S1

References (2857)

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We thank A. Stanford and M. Baloković for helping to obtain optical spectra for W2246−0526 on November 2010 and October 2013, respectively. We thank J. González López for helpful suggestions regarding the cleaning algorithms of CASA. ALMA is a partnership of the European Southern Observatory (ESO) (representing its member states), NSF (United States), and the National Institute of Natural Sciences (Japan), together with the National Research Council (Canada) and National Science Council and Academia Sinica’s Institute of Astronomy and Astrophysics (Taiwan), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, Associated Universities Inc. (AUI)/National Radio Astronomy Observatory (NRAO) and National Astronomical Observatory of Japan. The NRAO is a facility of NSF operated under cooperative agreement by AUI. This work is also based in part on archival observations from the Spitzer Space Telescope, the Herschel Space Observatory, WISE, as well as from the NASA/European Space agency HST. Some of the data were obtained 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. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. Funding: T.D.-S. acknowledges support from ALMA-CONYCIT project 31130005 and Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) project 1151239. R.J.A. acknowledges support from FONDECYT 1151408. The work of C-W.T., J.W., P.E., and D.S. was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA, and supported by grant ADAP13-0092. M.A. acknowledges partial support from FONDECYT through grant 1140099. J.W. acknowledges support from the Ministry of Science and Technology of China through grant 2016YFA0400702 and National Natural Science Foundation of China 11673029. This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2017R1A6A3A04005158). Author contributions: T.D.-S. lead the overall project. R.J.A. and A.W.B. contributed to the interpretation of the results. M.A. helped with processing the ALMA and VLA data. D.S. led the spectroscopic identification campaign of hot DOGs. C.-W.T., P.E., and J.W. contributed to the discussion of the results. K.D., H.I, G.L., and F.L. were part of the team that acquired the optical spectra of W2246−0526. Competing interests: The authors declare that there are no competing interests. Data and materials availability: The ALMA observations are available at under project 2015.1.00883.S (principal investigator, T.D.-S.). The VLA observations can be retrieved from under project 15B-192 (principal investigator, R.J.A.). The Spitzer data can be retrieved from the Spitzer Heritage Archive (SHA) at under project 70162 (principal investigator, P.E.). The Herschel data are available at the Herschel Science Archive (HSA) under project OT1_peisenha_1 (principal investigator, P.E.). The WISE data can be accessed at with documentation at The HST data are available at under project 12930 (principal investigator, C. Bridge). The Keck data are available in the Keck Observatory Archive at by searching with the observation dates (15).

Stay Connected to Science

Navigate This Article