News & Views | Published:


Feeding the first quasars

Quasars, the oldest known objects in the Universe, are powered by gas falling into black holes at their centres. How black holes formed so early in time has been hard to explain, but a new model might have the answer.

The excitement that has, in recent years, accompanied the study of supermassive black holes mirrors the excitement that followed the discovery of quasars in the early 1960s. Quasars — short for 'quasi-stellar objects' — are characterized by a prodigious outpouring of energy: hundreds of times that of a regular galaxy, but produced in a region that is only a few light days or weeks across. Quasars are also among the most distant objects known to astronomers; as such, the light reaching the Earth from them paints an invaluable picture of the history of our Universe. Supermassive black holes have long been accepted as the only viable energy source for quasars, but only now are we beginning to understand the complex connection between black holes and the formation of galaxies.

Quasars are thought to reside at the centres of massive haloes of dark matter — the mysterious, unseen matter that is needed to explain many features of our Universe. Indeed, this is a crucial assumption that underlies some of the most popular models of black-hole formation. On page 341 of this issue, Barkana and Loeb1 suggest that if a dark-matter halo is pulling gas towards a quasar at its centre, a distinctive signature should be seen in the light reaching us from the quasar. There are few data to go on so far, but if this signature is confirmed, it would provide the first observational evidence that quasars are embedded in great haloes of dark matter.

In the local Universe, there is now airtight evidence for the presence of supermassive black holes in two galaxies — the Milky Way2,3 and the nearby Seyfert II galaxy, NGC 4258 (ref. 4). Compelling, although more indirect, evidence of black holes exists for at least two dozen additional galaxies5. These supermassive black holes have masses that range from a few million to a few billion times that of our Sun. Their host galaxies are sufficiently disparate in nature that it is now possible to search for connections between their large-scale properties and the masses of their central black holes. Some connections have already been identified, most notably between supermassive black-hole masses and the velocity distribution of the surrounding hot stellar component6,7. Another connection may exist between the mass of the central black hole and the mass of the surrounding dark-matter halo8.

The demographics of more distant supermassive black holes can be inferred from a census of quasars. In the past few years, the Sloan Digital Sky Survey (SDSS) has dramatically increased the number of known quasars at 'high redshift', where redshift is a measure of an object's recessional velocity due to the expansion of the Universe. The SDSS9,8,9,10,11,12 has found several quasars with redshifts larger than 5, including the current record holder at a redshift of 6.43. Assuming that these high-redshift quasars are radiating at the Eddington limit (the maximum luminosity that can be sustained by accretion), their luminosities imply central black holes with masses at least several billion times that of the Sun.

It has been pointed out13 that at a redshift of 5 we are looking back in time to when the age of the Universe (about 1 billion years) was approximately equal to the dynamical timescale of a typical galaxy — roughly speaking, the stellar orbital time, or the time it takes a galaxy to communicate with itself through its own gravitational potential. Thus, the very existence of quasars at such high redshifts is a challenge to models of structure formation14,15. Although the details vary, the basic assumption underlying virtually all models is that, at all redshifts, the history of supermassive black holes follows the evolution of galactic dark-matter haloes. In particular, the black-hole mass is assumed to scale with halo mass, and black-hole growth proceeds through gas accretion triggered by galaxy mergers.

A relation between black-hole and dark-halo mass is a feature of models that account for supermassive black holes within this 'hierarchical' framework, in which structure formation proceeds from the smaller to the larger galaxies through merging and interactions. Models that successfully reproduce the observed quasar luminosities at high redshifts15,16,17 predict a relation between black-hole and dark-halo mass in remarkable agreement with what is observed in the local Universe.

The observations that triggered Barkana and Loeb's work1 were spectroscopic studies by the SDSS collaboration of quasars at redshifts of 4.79 and 6.28. The quasar emission lines resulting from excited hydrogen atoms have characteristic 'double-horn' profiles, with distinct red and blue peaks (Fig. 1). Although gas along the line of sight to the quasar could absorb some of the emitted radiation and produce an asymmetric profile, previous attempts at modelling the SDSS quasar spectra in this way have been unsuccessful: absorbing gas that is moving away from the quasar, following the expansion of the Universe, can 'eat away' only the blue side of the emission line profile. Barkana and Loeb realized that if gas were falling in towards the quasar at the centre of the dark-matter halo, the characteristic double-horn profile would be produced.

Figure 1: Quasars, black holes and dark matter.

Barkana and Loeb1 show that the observed flux of light from a quasar is modified as gas falling in towards the black hole at its centre, under the gravitational influence of a surrounding halo of dark matter, absorbs some of the radiation. If the same amount of gas is falling in from all directions, the flux as a function of the gas velocity relative to the observer takes on a distinctive 'double-horn' profile. The 'red' peak arises because gas moving away from the observer towards the centre of the quasar hits the accretion shock — a shock barrier due to gas ionization — and inside this barrier the transmitted flux follows the intrinsic spectrum, as the gas no longer absorbs the quasar light. Although Barkana and Loeb have little information to work with so far, their model is in good agreement with the existing spectra of two high-redshift quasars.

As gas falls towards the quasar, an increasing volume becomes ionized. Barkana and Loeb's model predicts a sudden change in the fraction of neutral gas around 100 kiloparsecs (roughly 1018 km) from the quasar. At this radius an accretion shock sets in, producing a sharp cut-off in the absorption profile (Fig. 1) — as is seen in the spectra of the SDSS quasars (Fig. 2 on page 342). The velocity of the infalling gas, the radius of the accretion shock and the rate at which mass is being collected towards the centre all depend on the gravitational force exerted by, and hence on the mass of, the dark-matter halo.

Barkana and Loeb's calculations for the two SDSS quasars imply black holes of 108–109 solar masses. Relating these values to the dark haloes, as in hierarchical models, they find that these quasars are embedded in dark-matter haloes each of around 1012 solar masses. And, reassuringly, the mass collection rate is large enough that the quasar host galaxies could have been assembled on timescales shorter than the age of the Universe at a redshift of 6.28. But, as Barkana and Loeb admit, more observational data are needed for a rigorous test of their model.

What challenges lie ahead in the study of supermassive black holes? At present, hierarchical models cannot easily account for nuclear black holes with masses smaller than 105–106 solar masses16,17. If such black holes exist, they must be formed through a completely different mechanism — or the entire picture needs to be revised. In fact, the smallest supermassive black hole detected so far is at the centre of the Milky Way (with a mass of 3 million solar masses). Unfortunately, the dynamical signature of lower-mass black holes is subtle at best. Their detection in even the nearest galaxies poses a strong technological challenge as their 'sphere of influence' (the region of space within which the black hole dominates the gravitational potential of the surrounding stars) is smaller than the resolution limits of current instruments. Once again, the ball is in the court of the observers.


  1. 1

    Barkana, R. & Loeb, A. Nature 421, 341–343 (2003).

  2. 2

    Schödel, R. et al. Nature 419, 694–696 (2002).

  3. 3

    Ghez, A. M., Morris, M., Becklin, E. E., Tanner, A. & Kremenek, T. Nature 407, 349–351 (2000).

  4. 4

    Miyoshi, M. et al. Nature 373, 127–129 (1995).

  5. 5

    Kormendy, J. & Gebhardt, K. in Proc. XX Texas Symp. Relativistic Astrophysics Vol. 586 (eds Wheeler, J. C. & Martel, H.) 363 (Am. Inst. Phys., New York, 2001).

  6. 6

    Ferrarese, L. & Merritt, D. Astrophys. J. 539, L9–L12 (2000).

  7. 7

    Gebhardt, K. et al. Astrophys. J. 539, L13–L16 (2000).

  8. 8

    Ferrarese, L. Astrophys. J. 578, 90–97 (2002).

  9. 9

    Fan, X. et al. Astron. J. 122, 2833–2849 (2001).

  10. 10

    Fan, X. et al. Astron. J. 121, 54–65 (2001).

  11. 11

    Fan, X. et al. Astron. J. (in the press); Preprint astro-ph/0301135,

  12. 12

    Schneider, D. P. et al. Astron. J. 121, 1232–1240 (2001).

  13. 13

    Turner, E. L. Astron. J. 101, 5–17 (1991).

  14. 14

    Haiman, Z. & Loeb, A. Astrophys. J. 552, 459–463 (2001).

  15. 15

    Wyithe, S. B. & Loeb, A. Astrophys. J. 581, 886–894 (2002).

  16. 16

    Haiman, Z. & Loeb, A. Astrophys. J. 503, 505–517 (1998).

  17. 17

    Haehnelt, M. G., Natarajan, P. & Rees, M. J. Mon. Not. R. Astron. Soc. 300, 817–827 (1998).

Download references

Author information

Correspondence to Laura Ferrarese.

Rights and permissions

Reprints and Permissions

About this article


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.