Quasar quartet embedded in giant nebula reveals rare massive structure in distant universe

See allHide authors and affiliations

Science  15 May 2015:
Vol. 348, Issue 6236, pp. 779-783
DOI: 10.1126/science.aaa5397

In a cluster of protogalaxies far, far away

Astronomers constantly scour the sky for astronomical objects that can provide insight and constrain their models and simulations of galaxy evolution. Hennawi et al. surveyed the ancient sky at an epoch when the universe was half its age for nebulae: large clouds of ionized hydrogen. They stumbled across a system containing four active galactic nuclei, or quasars; objects that are thought to be the progenitors of galaxies. Finding a nebula with a rare quadruple quasar system embedded within it allows detailed spectroscopic and motional studies that may help to refine current models of galaxy and galaxy cluster formation.

Science, this issue p. 779


All galaxies once passed through a hyperluminous quasar phase powered by accretion onto a supermassive black hole. But because these episodes are brief, quasars are rare objects typically separated by cosmological distances. In a survey for Lyman-α emission at redshift z ≈ 2, we discovered a physical association of four quasars embedded in a giant nebula. Located within a substantial overdensity of galaxies, this system is probably the progenitor of a massive galaxy cluster. The chance probability of finding a quadruple quasar is estimated to be ∼10−7, implying a physical connection between Lyman-α nebulae and the locations of rare protoclusters. Our findings imply that the most massive structures in the distant universe have a tremendous supply (≃1011 solar masses) of cool dense (volume density ≃ 1 cm−3) gas, which is in conflict with current cosmological simulations.

Cosmologists do not fully understand the origin of supermassive black holes (SMBHs) at the centers of galaxies and how they relate to the evolution of the underlying dark matter, which forms the backbone for structure in the universe. In the current paradigm, SMBHs grew in every massive galaxy during a luminous quasar phase, making distant quasars the progenitors of the dormant SMBHs found at the centers of nearby galaxies. Tight correlations between the masses of these local SMBHs and both their host galaxy (1) and dark-matter halo masses (2) support this picture, further suggesting that the most luminous quasars at high redshift (z) should reside in the most massive galaxies. It has also been proposed that quasar activity is triggered by the frequent mergers that are a generic consequence of hierarchical structure formation (3, 4). Indeed, an excess in the number of small-separation binary quasars (5, 6), as well as the mere existence of a handful of quasar triples (7, 8), support this hypothesis. If quasars are triggered by mergers, then they should preferentially occur in rare peaks of the density field, where massive galaxies are abundant and the frequency of mergers is highest (9).

Following these arguments, one might expect that at the peak of their activity, z ∼ 2 to 3, quasars should act as signposts for protoclusters, the progenitors of local galaxy clusters and the most massive structures at that epoch. However, quasar clustering measurements (10, 11) indicate that quasar environments at z ∼ 2 to 3 are not extreme: These quasars are hosted by dark-matter halos with masses Mhalo ∼ 1012.5 M (where M is the mass of the Sun), too small to be the progenitors of local clusters (12). But the relationship between quasar activity and protoclusters remains unclear, owing to the extreme challenge of identifying the latter at high redshift. Indeed, the total comoving volume of even the largest surveys for distant galaxies at z ∼ 2 to 3 is only ∼107 Mpc3, which would barely contain a single rich cluster locally.

Protoclusters have been discovered around a rare population of active galactic nuclei (AGNs) powering large-scale radio jets, known as high-redshift radio galaxies (HzRGs) (13). The HzRGs routinely exhibit giant ∼100 kpc nebulae of luminous Lyman-α (Lyα) emission LLyα ∼ 1044 erg s−1. Nebulae of comparable size and luminosity have also been observed in a distinct population of objects known as Lyα blobs (LABs) (14, 15). The LABs are also frequently associated with AGN activity (1618), although lacking powerful radio jets, and appear to reside in overdense protocluster environments (15, 19, 20). Thus, among the handful of protoclusters (21) known, most appear to share two common characteristics: the presence of an active SMBH and a giant Lyα nebula.

We have recently completed a spectroscopic search for extended Lyα emission around a sample of 29 luminous quasars at z ∼ 2, each the foreground (f/g) member of a projected quasar pair (22). Analysis of spectra from the background (b/g) members in such pairs reveals that quasars exhibit frequent absorption from a cool, metal-enriched, and often optically thick medium on scales of tens to hundreds of kiloparsecs (2228). The ultraviolet radiation emitted by the luminous f/g quasar can, like a flashlight, illuminate this nearby neutral hydrogen and power a large-scale Lyα-emission nebula, motivating our search. By construction, our survey selects for exactly the two criteria that seem to strongly correlate with protoclusters: an active SMBH and the presence of a large-scale Lyα emission nebula.

Of the 29 quasars surveyed, only SDSSJ0841+3921 exhibited extended large-scale (≳50 kpc) Lyα emission above our characteristic sensitivity limit of 6 × 10−18 erg s−1 cm−2 arc sec−2 (2σ). We designed a custom narrow-band filter tuned to the wavelength of Lyα at the f/g quasar redshift z = 2.0412 (central wavelength = 3700 Å, full width at half maximum = 33 Å), and imaged the field with the Keck/LRIS imaging spectrometer for 3 hours on UT 12 November 2012. The combined and processed images reveal Lyα emission from a giant filamentary structure centered on the f/g quasar and extending continuously toward the b/g quasar (Fig. 1). This nebulosity has an end-to-end size of 37′′, corresponding to 310 kpc, and a total line luminosity LLyα = 2.1 × 1044 erg s−1, making it one of the largest and brightest nebulae ever discovered.

Fig. 1 Narrow- and broadband images of the field surrounding SDSSJ0841+3921.

(A) Continuum-subtracted, narrow-band image of the field around f/g quasar. The color map and the contours indicate the Lyα surface brightness (upper color bar) and the signal-to-noise ratio per arc sec2 aperture (lower color bar), respectively. This image reveals a giant Lyα nebula on the northern side of the f/g quasar and several compact bright Lyα emitters in addition to the f/g quasar. Three of these have been spectroscopically confirmed as AGNs at the same redshift. (B) Corresponding V-band continuum image of the field presented at left with the locations of the four AGNs marked. The AGNs are roughly oriented along a line coincident with the projected orientation of the Lyα nebula. We also mark the position of the b/g quasar, which is not physically associated with the quadruple AGN system but whose absorption spectrum probes the gaseous environment of the f/g quadruple AGN and protocluster (Fig. 4).

The giant nebula is only one of the exceptional properties of SDSSJ0841+3921. Our images reveal three relatively compact candidate Lyα-emitting sources with faint continuum magnitudes V ≃ 23 to 24, embedded in the Lyα filament and roughly aligned with its major axis. Followup spectroscopy reveals that the sources labeled AGN1, AGN2, and AGN3 are three AGNs at the same redshift as the f/g quasar [see the right panel of Fig. 1 and (29)], making this system the only quadruple AGN known. Adopting recent measurements of small-scale quasar clustering (30), we estimate that the probability of finding three AGNs around a quasar with such small separations is ∼10−7 (31). Why then did we discover this rare coincidence of AGNs in a survey of just 29 quasars? Did the giant nebula mark the location of a protocluster with dramatically enhanced AGN activity?

To test this hypothesis, we constructed a catalog of Lyα-emitting galaxies (LAEs) and computed the cumulative overdensity profile of LAEs around SDSSJ0841+3921, relative to the background number expected based on the LAE luminosity function (32) (Fig. 2). To perform a quantitative comparison to other giant Lyα nebulae, many of which are known to coincide with protoclusters, we measured the giant nebulae–LAE cross-correlation function for a sample of eight systems—six HzRGs and two LABs—for which published data was available in the literature (33). In Fig. 2, we compare the overdensity profile around SDSSJ0841+3921 to this giant nebulae–LAE correlation function. On average, the environment of HzRGs and LABs hosting giant Lyα nebulae (red line) is much richer than that of radio-quiet quasars (10) (blue line), confirming that they indeed reside in protoclusters. Furthermore, the clustering of LAEs around SDSSJ0841+3921 has a steeper overdensity profile and exceeds the average protocluster by a factor of ≳20 for R < 200 kpc and by ∼ 3 on scales of R ≃ 1 Mpc. In addition to the overdensity of four AGNs, the high number of LAEs surrounding SDSSJ0841+3921 make it one of the most overdense protoclusters known at z ∼ 2 to 3.

Fig. 2 Characterization of the protocluster environment around SDSSJ0841+3921.

The data points indicate the cumulative overdensity profile of LAEs δ(<R) as a function of impact parameter R from the f/g quasar in SDSSJ0841+3921, with Poisson error bars. The red curve shows the predicted overdensity profile, based on our measurement of the giant nebulae–LAE cross-correlation function determined from a sample of eight systems—six HzRGs and two LABs—for which published data were available in the literature. Assuming a power-law form for the cross-correlation ξcross = (r/r0)−γ, we measured the correlation length r0 = 29.3 ± 4.9 h−1 Mpc, for a fixed value of γ = 1.5. The gray-shaded region indicates the 1σ error on our measurements based on a bootstrap analysis, where both r0 and γ are allowed to vary. The solid blue line indicates the overdensity of Lyman break galaxies (LBGs) around radio-quiet quasars based on recent measurements (10), with the dotted blue lines the 1σ error on this measurement. On average, the environment of HzRGs and LABs hosting giant Lyα nebulae is much richer than that of radio-quiet quasars (10), confirming that they indeed reside in protoclusters. SDSSJ0841+3921 exhibits a dramatic excess of LAEs as compared to the expected overdensities around radio-quiet quasars (blue curve). Its overdensity even exceeds the average protocluster (red curve) by a factor of ≳20 for R < 200 kpc, decreasing to an excess of ∼3 on scales of R ≃1 Mpc, and exhibits a much steeper profile.

The combined presence of several bright AGNs, the Lyα emission nebula, and the b/g absorption spectrum provides an unprecedented opportunity to study the morphology and kinematics of the protocluster via multiple tracers, and we find evidence for extreme motions (34). Specifically, AGN1 is offset from the precisely determined systemic redshift (35) of the f/g quasar by +1300 ± 400 km s−1. This large velocity offset cannot be explained by Hubble expansion [the miniscule probability of finding a quadruple quasar in the absence of clustering P ∼ 10−13 (31) and the physical association between the AGN and giant nebula demand that the four AGNs reside in a real collapsed structure] and thus provides unambiguous evidence for extreme gravitational motions. In addition, our slit spectra of the giant Lyα nebula reveal extreme kinematics of diffuse gas (Fig. 3), extending over a velocity range of −800 to +2500 km s−1 from systemic. Furthermore, there is no evidence for the double-peaked velocity profiles characteristic of resonantly trapped Lyα, which could generate large velocity widths in the absence of correspondingly large gas motions. Absorption line kinematics of the metal-enriched gas, measured from the b/g quasar spectrum at an impact parameter of R = 176 kpc (Fig. 4), show strong absorption at ≈+650 km s−1, with a significant tail to velocities as large as ≃1000 km s−1. It is of course possible that the extreme gas kinematics, traced by Lyα emission and metal-line absorption, are not gravitational but rather arise from violent large-scale outflows powered by the multiple AGNs. Although we cannot completely rule out this possibility, the large velocity offset of +1300 km s−1 between the f/g quasar and the emission redshift of AGN1 can only result from gravity.

Fig. 3 Lyα spectroscopy of the giant nebula and its associated AGNs.

(A) The spectroscopic slit locations (white rectangles) for three different slit orientations are overlaid on the narrowband image of the giant nebula. The locations of the f/g quasar (brightest source), b/g quasar, AGN1, and AGN2 are also indicated. Two-dimensional spectra for slit1 (B), slit2 (C), and slit3 (D) are shown in the accompanying panels. In the upper right and lower left panels, spatial coordinates refer to the relative offset along the slit with respect to the f/g quasar. Spectra of AGN1 are present in both slit 1 (upper right) and slit 3 (lower right) at spatial offsets of 75 and 25 kpc, respectively, whereas the Lyα spectrum of AGN2 is located at a spatial offset −60 kpc in slit 1 (upper right). The b/g quasar spectra are located in both slit 2 (lower left) and slit 3 (lower right) at the same spatial offset of 176 kpc. The spectroscopic observations demonstrate the extreme kinematics of the system: AGN1 has a velocity of +1300 ± 400 km s−1 relative to the f/g quasar, and the Lyα emission in the nebula exhibits motions ranging from −800 km s−1 (at ≈100 kpc offset in slit 3, lower right) to +2500 km s−1 (at ≈100 kpc in slit 1, upper right). A 3 × 3 pixel boxcar smoothing, which corresponds to 120 km s−1 × 0.8′′, has been applied to the spectra. In each two-dimensional spectrum, the zero velocity corresponds to the systemic redshift of the f/g quasar. The color bars indicate the flux levels in units of erg s−1 cm−2 arc sec−2 Å−1.

Fig. 4 Absorption line spectrum of cool gas in SDSSJ0841+3921.

The spectrum of the absorbing gas detected in the b/g quasar sightline at an impact parameter of 176 kpc from the f/g quasar is shown. The gas shows strong HI and low-ionization–state metal absorption, offset by ≈650 km s−1 from the f/g quasar's systemic redshift. The CII absorption in particular exhibits a significant tail to velocities as large as ≃1000 km s−1, providing evidence for extreme gas kinematics. We modeled the strong HI Lyα absorption with a Voigt profile (blue curve with gray band indicating uncertainty) and estimate a column density logNHI = 19.2 ± 0.3. The strong low and intermediate ion absorption (SiII, CII, and SiIII) and correspondingly weak high ion absorption (CIV and SiIV) indicate that the gas is not highly ionized, and our photoionization modeling (45) implies log10xHI = −1.2 ± 0.3 or log10NH = 20.4 ± 0.4. We estimate a conservative lower limit on the gas metallicity to be 1/10th of the solar value. Spectral regions contaminated by other unassociated absorption are indicated in light gray.

One can only speculate about the origin of the dramatic enhancement of AGNs in the SDSSJ0841+3921 protocluster. Perhaps the duty cycle for AGN activity is much longer in protoclusters, because of frequent dissipative interactions (5, 6), or an abundant supply of cold gas. A much larger number of massive galaxies could also be the culprit, as AGNs are known to trace massive halos at z ∼ 2. Although SDSSJ0841+3921 is the only example of a quadruple AGN with such small separations, previously studied protoclusters around HzRGs and LABs also occasionally harbor multiple AGNs (13, 36). Regardless, our discovery of a quadruple AGN and protocluster from a sample of only 29 quasars suggests a link between giant Lyα nebulae, AGN activity, and protoclusters, similar to past work on HzRGs and LABs, with the exception that SDSSJ0841+3921 was selected from a sample of normal radio-quiet quasars. From our survey and other work (37), we estimate that ≃10% of quasars exhibit comparable giant Lyα nebulae. Although clustering measurements imply that the majority of z ∼ 2 quasars reside in moderate overdensities (1012), we speculate that this same 10% trace much more massive protoclusters. SDSSJ0841+3921 clearly supports this hypothesis, as does another quasar-protocluster association (10, 38), around which a giant Lyα nebula was recently discovered (39, 40).

Given our current theoretical picture of galaxy formation in massive halos, an association between giant Lyα nebulae and protoclusters is completely unexpected. The large Lyα luminosities of these nebulae imply a substantial mass (∼1011 M) of cool (T ∼ 104 K) gas (41), whereas cosmological hydrodynamical simulations indicate that already by z ∼ 2 to 3, baryons in the massive progenitors (Mhalo ≳ 1013 M) of present-day clusters are dominated by a hot shock-heated plasma T ∼ 107 K (42, 43). These hot halos are believed to evolve into the x-ray–emitting intracluster medium observed in clusters, for which absorption-line studies indicate negligible ≲1% cool gas fractions (44). Clues about the nature of this apparent discrepancy come from our absorption line studies of the massive ≃1012.5 M halos hosting z ∼ 2 to 3 quasars. This work reveals substantial reservoirs of cool gas ≳1010 M (2228), manifest as a high covering factor ≃50% of optically thick absorption, several times larger than predicted by hydrodynamical simulations (42, 43). This conflict most likely indicates that current simulations fail to capture essential aspects of the hydrodynamics in massive halos at z ∼ 2 (27, 42), perhaps failing to resolve the formation of clumpy structure in cool gas (41).

If illuminated by the quasar, these large cool gas reservoirs in the quasar circumgalactic medium (CGM) will emit fluorescent Lyα photons, and we argue that this effect powers the nebula in SDSSJ0841+3921 (45). But according to this logic, nearly every quasar in the universe should be surrounded by a giant Lyα nebula with size comparable to its CGM (∼200 kpc). Why then are these giant nebulae not routinely observed?

This apparent contradiction can be resolved as follows. If this cool CGM gas is illuminated and highly ionized, it will fluoresce in the Lyα line with a surface brightness scaling as SBLyαNHnH, where NH is the column density of cool gas clouds that populate the quasar halo, and nH is the number density of hydrogen atoms inside these clouds. The total cool gas mass of the halos scales as McoolR2NH, where R is the radius of the halo (45). Given our best estimate for the properties of the CGM around typical quasars (nH ∼ 0.01 cm−3 and NH ∼ 1020 cm−2 or Mcool ≃ 1011 M) (22, 26, 27), we expect these nebulae to be extremely faint SBLyα ∼ 10−19 erg s−1 cm−2 arc sec−2 and beyond the sensitivity of current instruments (22). One comes to a similar conclusion based on a full radiative transfer calculation through a simulated dark-matter halo with mass Mhalo ≈ 1012.5 M (41). Thus the factor of ∼100 times larger surface brightness observed in the SDSSJ0841+3921 and other protocluster nebulae arises from either a higher nH, NH (and hence higher Mcool), or both. The cool gas properties required to produce the SDSSJ0841+3921 nebula can be directly compared to those deduced from an absorption line analysis of the b/g quasar spectrum (46).

The b/g quasar sightline pierces through SDSSJ0841+3921 at an impact parameter of R= 176 kpc, giving rise to the absorption spectrum shown in Fig. 4. Photoionization modeling of these data constrains the total hydrogen column density to be log10NH = 20.4 ± 0.4 (45), implying a substantial mass of cool gas 1.0 × 1011 M < Mcool < 6.5 × 1011 M within r = 250 kpc. Assuming that the Lyα emitting gas has the same column density as the gas absorbing the b/g sightline, reproducing the large fluorescent Lyα surface brightness requires that this gas be distributed in compact rcloud ∼ 40 pc clouds at densities characteristic of the interstellar medium nH ≃ 2 cm−3, but on ∼100-kpc scales in the protocluster.

Clues to the origin of these dense clumps of cool gas come from their high enrichment level, which we have determined from our absorption line analysis (46) to be greater than 1/10th of the solar value. At first glance, this suggests that strong tidal interactions due to merger activity or outflows due to powerful AGN feedback are responsible for dispersing dense cool gas in the protocluster. However, the large cool gas mass ∼1011 M and high velocities ∼1000 km s−1 imply an extremely large kinetic luminosity Lwind ∼ 1044.6 for an AGN-powered wind, making the feedback scenario implausible (25). An even more compelling argument against a merger or feedback origin comes from the extremely small cloud sizes rcloud ∼ 40 pc implied by our measurements. Such small clouds moving supersonically ∼1000 km s−1 through the hot T ∼ 107 K shock-heated plasma predicted to permeate the protocluster will be disrupted by hydrodynamic instabilities in ∼5 × 106 years and can thus only be transported ∼5 kpc (47). These short disruption time scales instead favor a scenario where cool dense clouds are formed in situ, perhaps via cooling and fragmentation instabilities, but are short-lived. The higher gas densities might naturally arise if hot plasma in the incipient intracluster medium pressure-confines the clouds, compressing them to high densities (48, 49). Emission line nebulae from cool dense gas have also been observed at the centers of present-day cooling flow clusters (50, 51), albeit on much smaller scales ≲50 kpc. The giant Lyα nebulae in z ∼ 2 to 3 protoclusters might be manifestations of the same phenomenon, but with much larger sizes and luminosities, reflecting different physical conditions at high redshift.

The large reservoir of cool dense gas in the protocluster SDSSJ0841+3921, as well as those implied by the giant nebulae in other protoclusters, appear to be at odds with our current theoretical picture of how clusters form. This is likely to be symptomatic of the same problem of too much cool gas in massive halos already highlighted for the quasar CGM (27, 42, 43). Progress will require more cosmological simulations of massive halos M ≳ 1013 M at z ∼ 2, as well as idealized higher-resolution studies. In parallel, a survey for extended Lyα emission around ∼100 quasars would uncover a sample of ∼10 giant Lyα nebulae, likely coincident with protoclusters, possibly also hosting multiple AGNs, and enabling continued exploration of the relationship between AGNs, cool gas, and cluster progenitors.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S10

Tables S1 to S6

References (52133)

References and Notes

  1. See supplementary text S2.
  2. See supplementary text S3.
  3. See supplementary text S4.
  4. See supplementary text S6.
  5. See supplementary text S5.
  6. See supplementary text S7.
  7. See supplementary text S8.
  8. See supplementary text S9.
  9. See supplementary text S10.
  10. Acknowledgments: We thank the staff of the W. M. Keck Observatory for their support during the installation and testing of our custom-built narrow-band filter. We are grateful to B. Venemans and M. Prescott for providing us with catalogs of LAE positions around giant nebulae in electronic format. We also thank the members of the ENIGMA group ( at the Max Planck Institute for Astronomy for helpful discussions. J.F.H. acknowledges generous support from the Alexander von Humboldt foundation in the context of the Sofja Kovalevskaja Award. The Humboldt foundation is funded by the German Federal Ministry for Education and Research. J.X.P. acknowledges support from NSF grant AST-1010004. The data presented here 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 financial support of the W. M. Keck Foundation. We acknowledge the cultural role that the summit of Mauna Kea has within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain. The data reported in this paper are available through the Keck Observatory Archive.

Stay Connected to Science

Navigate This Article