In 1965, just two years after quasars were first identified as cosmologically distant objects, James Gunn and Bruce Peterson at Caltech called attention to a peculiar spectroscopic aspect of these high-redshift beacons. Looking at the spectrum of a quasar with a redshift z of about 2, they pointed out that it did not exhibit a trough of total absorption on the blueward side of its prominent Lyman-α hydrogen emission peak. But if, as was generally supposed, the intergalactic medium between us and the quasar had lots of neutral atomic hydrogen, there would have to be such an absorption trough.

The absence of a “Gunn–Peterson trough” in any of the quasar spectra measured in the ensuing 35 years is the primary evidence we have that the intergalactic hydrogen is not neutral. It is, in fact, overwhelmingly ionized.

As cosmological epochs are reckoned, this ionized intergalactic medium has to be a rather recent state of affairs. The cosmic microwave background teaches us that, some 400‥000 years after the Big Bang, the universe finally became cool enough for neutral hydrogen atoms to survive (see July 2001, page 16). Astrophysicists been have striving for decades to pinpoint the time of the subsequent cosmic “reionization” phase transition—perhaps a billion years later—by looking at the spectra of ever more distant, higher-redshift quasars in search of one early enough to exhibit a Gunn–Peterson trough. The reionization time is an important parameter for models of the evolution of structure in the cosmos.

For the past year or two, this task has been greatly aided by a byproduct of the prodigious Sloan Digital Sky Survey. Gunn, who has been at Princeton since 1968, is the Sloan survey’s project scientist. The survey’s principal task is to measure the redshifts—and hence the distances—of a million galaxies. But in the process, the survey has been identifying handfuls of the most distant quasars ever seen. 1  

Now we have a paper from the Sloan team reporting that their most distant quasar, with a z of 6.28, shows the first “clear detection of a complete Gunn–Peterson trough.” 2 In a related paper, 3 George Djorgovski and colleagues at Caltech, having taken a high-resolution look at the spectrum of a z = 5.73 quasar discovered by the Sloan survey last year, argue that residual dark patches in that spectrum are a signature of the final moments of the reionization transition.

Both these new papers suggest that the intergalactic medium was almost completely reionized at a redshift zR of about 6. To circumvent various uncertainties, cosmologists like to specify moments in cosmic history by redshift rather than explicit time. If we see an emission line of a distant object redshifted by z ≡ Δλ/λ0, where λ0 is the rest-frame wavelength, we know that the linear scale of the expanding cosmos has grown by a factor 1 + z since the time of emission. And, for most of cosmic history, that scale factor has been growing like t2/3, where t is the time since the Big Bang. So, if the cosmos is now about 14 billion years old, a redshift of 6 corresponds roughly to a t of about 900 million years.

The transition from the hot primordial plasma to neutral hydrogen at about t = 400‥000 years (z ≈ 1200) rendered the cosmos, for the first time, transparent to visible light (which was rapidly fading away as the temperature of the starless universe continued to fall) and to longer wavelengths. But the neutral hydrogen would not be so kind to the ultraviolet radiation from the first stars and accreting black holes that began to appear a few hundred million years later.

An expanding universe full of neutral hydrogen would have presented an opaque barrier to ultraviolet photons emitted with energies above 10.2 eV, the rest-frame Ly-α energy (λ0 = 1216 Å). The Ly-α transition is the excitation of an electron from the ground state of atomic hydrogen to the first excited state. The Ly-α emission and absorption lines are both very strong. Even a very small residuum of neutral hydrogen presents a formidable absorption barrier. Photons that start out with energies above 10.2 eV are eventually redshifted down to the Ly-α energy by the cosmic expansion, and then promptly absorbed.

At a somewhat higher energy (13.6 eV, the limit of the Lyman spectroscopic series), an ultraviolet photon can ionize a hydrogen atom. Thus the ultraviolet radiation from the first protogalaxies and accreting black holes began to produce growing bubbles of ionized hydrogen in the neutral surrounding medium. Eventually, as time passed and galaxies proliferated, the merging of all these growing ionized bubbles is thought to have wrought a rather abrupt phase transition to the almost fully ionized intergalactic medium that surrounds us today. 4 That is to say, the utraviolet radiation from any source whose redshift is smaller than zR will not be completely absorbed by neutral hydrogen as it makes its way to us through the intergalactic medium.

A useful exception to this general assertion is the so-called Lyman-α forest—a thicket of discrete absorption lines. This seemingly random jumble of black lines is due to Ly-α absorption of ultraviolet light from distant quasars by small residual pockets of neutral atomic hydrogen along the line of sight. If one such intervening pocket happens to sit at redshift z, we see the Ly-α absorption line it engenders redshifted by a factor 1 + z. This forest of discrete absorption lines in the ultraviolet continuum of a quasar spectrum is quite different from the wide, empty trough that would be produced by pervasive neutral hydrogen not confined to occasional pockets.

For two decades, the Ly-α forest has been a superb source of information about the intergalactic medium and the objects it envelops. (See November 1987, page 17.) The pockets of neutral hydrogen are still called Lyman-α clouds, but the “cloud” metaphor is now out of favor. The pockets are thought to be the fractally distributed regions of highest mass density in the gravitationally evolving structure in the intergalactic medium. The higher the local density, the more likely ionized atoms are to recombine with electrons to become neutral again. In cosmological models, the evolution of density fluctuations depends sensitively on what one assumes to be the admixture of cold and hot dark matter.

Figure 1 shows the Sloan survey’s spectra of the three highest-redshift quasars found to date, taken in the group’s follow-up measurements at the 10-m Keck II telescope on Hawaii’s Mauna Kea. The ultraviolet region around the quasars’ prominent Ly-α peaks is accessible to ground-based spectroscopy only because the high redshifts have transposed it into the visible and near infrared.

Figure 1. Spectra of the Sloan survey’s three highest-redshift quasars, in the vicinity of the redshifted α and β lines of the hydrogen Lyman excitation series and the Lyman ionization limit. In the spectrum of the z = 6.28 quasar (bottom panel) the absence of any discernible flux between 8450 Å and the prominent Ly-α peak appears to be the first observation of a Gunn–Peterson trough.

Figure 1. Spectra of the Sloan survey’s three highest-redshift quasars, in the vicinity of the redshifted α and β lines of the hydrogen Lyman excitation series and the Lyman ionization limit. In the spectrum of the z = 6.28 quasar (bottom panel) the absence of any discernible flux between 8450 Å and the prominent Ly-α peak appears to be the first observation of a Gunn–Peterson trough.

Close modal

In the spectra of the z = 5.82 and 5.99 quasars, we see the Ly-α forest crowding right up to the quasar’s Ly-α peak from its blueward (left) side. (At much lower redshifts, the forest is more obviously a collection of narrow absorption holes piercing an emission continuum. But at these high redshifts, the great number of intervening neutral-hydrogen pockets makes the forest look more like a jumble of small emission peaks.)

The z = 6.28 spectrum is different. Between 8450 Å and the precipitous beginning of the quasar’s Ly-α emission peak, the spectral measurement is consistent with zero flux. This, then, appears to be the long-sought-after Gunn–Peterson trough. Its blueward margin at 8450 Å implies that any ultraviolet photon that redshifted down to the Ly-α energy before the time corresponding to z = 5.95 would have been promptly absorbed by neutral hydrogen. Given the sensitivity of the measurement, one can only say that the flux in the trough is dimmer by a factor of at least 150 than it would have been in the absence of intergalactic Ly-α absorption. To determine what that unabsorbed flux would have been, the Sloan group extrapolates from the essentially undimmed emission continuum on the redward shoulder of the Ly-α peak.

Even a small fraction of unionized hydrogen distributed throughout the intergalactic medium at z ≈ 6 would have rendered the flux in the trough region undetectable. “Therefore,” the Sloan team cautions, “the existence of the Gunn–Peterson trough, by itself, does not indicate that the quasar is being observed prior to the reionization epoch.” 2 One has to consider how abrupt or gradual is the increase of absorption with increasing redshift. To that end, the team has looked at how the effective optical depth of the neutral hydrogen that’s absorbing the quasar light depends on the redshift at which the absorption is taking place (see figure 2). An optical depth of τ means that Ly-α absorption along the line of sight has dimmed the quasar’s ultraviolet emission by a factor of e−τ.

Figure 2. Optical depth due to Lyman-α absorption by neutral hydrogen grows as the redshift z at which the absorption took place increases. The data are from the spectra of the Sloan survey’s highest-redshift quasars. The rightmost data point, representing a spectral trough with no discernible flux, is a lower limit. With increasing z, the data pull away from the curve that shows what one expects simply from the greater abundance of Ly-α clouds at earlier times, suggesting that the intergalactic medium also had more neutral hydrogen in that epoch.

Figure 2. Optical depth due to Lyman-α absorption by neutral hydrogen grows as the redshift z at which the absorption took place increases. The data are from the spectra of the Sloan survey’s highest-redshift quasars. The rightmost data point, representing a spectral trough with no discernible flux, is a lower limit. With increasing z, the data pull away from the curve that shows what one expects simply from the greater abundance of Ly-α clouds at earlier times, suggesting that the intergalactic medium also had more neutral hydrogen in that epoch.

Close modal

Figure 2 plots the optical depth for various z absorption intervals as calculated from the observed spectral intensity of the Sloan survey’s four highest-redshift quasars in the corresponding wavelength intervals blueward of the Ly-α emission peak. 2 The optical depth τ = 5 for the highest z interval—corresponding to the reported Gunn–Peterson trough—is a lower limit, the trough showing no discernible flux.

The curve in the figure shows the gradual increase in optical depth one would get—in the absence of a reionization phase transition—simply from the expected increase in the number density of Ly-α clouds with increasing z. Above z = 5.8, the measured optical depths appear to be pulling away from the curve. So, above and beyond the greater abundance of Ly-α clouds at z ≳ 6, the intergalactic medium between the clouds in that epoch had a lot more neutral hydrogen than it would have in the later epoch.

Djorgovski and company, having availed themselves of a six-hour exposure at the Keck II telescope to produce a very high signal-to-noise spectrum of the z = 5.73 quasar, found a revealing discontinuity that had not been evident in the earlier low-resolution spectrum: “a dramatic increase in the optical depth” 3 on the blueward side of 7550 Å. They interpret this incipient spectral trough to mean that z = 5.2, the Ly-α absorption redshift corresponding to 7550 Å, is perhaps the last gasp of a rather extended reionization transition.

To map out the reionization of the cosmos in greater detail, observers will need more lines of sight to quasars at the highest redshifts. The Sloan survey expects ultimately to bag a total of about 20 quasars with z ≳ 6. If the reionization transition was far from uniform, different lines of sight may tell different stories.

1.
X.
Fan
 et al, http://arXiv.org/abs/astro-ph/0108063,
Astron. J.
(in press).
2.
R. H.
Becker
 et al, http://arXiv.org/abs/astro-ph/0108097,
Astron. J.
(in press).
3.
S. G.
Djorgovski
,
S.
Castro
,
D.
Stern
,
A.
Mahabal
, http://arXiv.org/abs/astro-ph/0108069,
Astrophys. J. Lett.
(in press).
4.
R.
Barkana
,
A.
Loeb
,
Phys. Rep.
349
,
125
(
2001
) .