X-Ray Absorption Lines Probe the Missing Half of the Cosmic Baryon Population

The topsy-turvy character of the cosmos posited by the so-called concordance model is, by now, widely known: Matter constitutes less than a third of the cosmic matter–energy budget. And only about a sixth of that matter is the ordinary baryonic matter we know anything about. Quoted as an invariant fraction of the total matter–energy density in the expanding universe, the mean cosmic baryon density Ω_b is confidently presumed to lie between 4 and 5%.

Less well known is the problem of the missing baryons in the present epoch, a small but embarrassing chink in the edifice of the concordance model. The model gets its name from the reassuring agreement of the cosmological parameters derived from a variety of independent observational regimes. The prediction of Ω_b comes primarily from the abundance ratios of the lightest nuclear species, the theory of Big Bang nucleosynthesis, and the small anisotropies seen in the cosmic microwave background. Indeed, when astronomers look back to a few billion years after the Big Bang, they find almost all of the expected baryonic matter. But by the time the cosmos was 5 billion years old, about half of that matter had somehow become invisible to the usual observing techniques.

Finding the missing baryons

A recent paper by Fabrizio Nicastro (Harvard–Smithsonian Center for Astrophysics) and coworkers reports the observation of x-ray absorption lines that appear to be the first clear sighting of the missing half of the baryon population. ¹ How did the baryons disappear from sight, and why has it been so hard to find them?

In a 1999 paper entitled “Where Are the Baryons?” Princeton theorists Renyue Cen and Jeremiah Ostriker pointed out the problem, proposed an explanation, and challenged observers to verify it.²¹ They had performed a hydrodynamic computer simulation of cosmic evolution from about 2 billion years after the Big Bang to the present. Cosmologists label the time t since the Big Bang in terms of the redshift of light emitted at t, as we see it now. Roughly, t ⌣ 14 Gy/(1 + z)^3/2. The simulation covered the interval from z = 3 to 0.

It’s known that at z = 3 only about 5% of all the baryonic matter was in galaxies. Most of it was in accumulations of cool (below 10⁵ K) intergalactic hydrogen gas with enough neutral atomic H to yield detectable Lyman a absorption lines in the UV and visual spectra of light from background quasars. From the so-called Lyman α forest of such lines at many different values of z, observers could conclude that there was enough H in discrete clouds or filaments of intergalactic gas at z = 3 to account for almost all the Ω_b expected from theory. In effect, one sees the Ly α clouds silhouetted in quasar light.

In the Cen–Ostriker simulation, the intergalactic gas gets steadily hotter after z = 3. More and more of it is shock heated as it repeatedly falls into gravitational potential wells of accumulated nonbaryonic “dark” matter. As the gas gets hotter, less and less of it remains un-ionized, so that Ly α absorption lines become increasingly hard to see. In the present epoch, the Ly α forest accounts for less than 30% of the presumed baryon density. Another 12% or so is visible in galaxies and in the x-ray emission of the very hot (above 10⁷ K) and relatively dense gas trapped in galaxy clusters.

Where are the rest of the baryons, and how can they be detected? Cen and Ostriker concluded that, after z = 1, almost half of the baryons in the cosmos have been residing in a “warm–hot intergalactic medium” (WHIM), a filamentary web of very-low-density gas—mostly hydrogen. With temperature ranging from 10⁵ to 10⁷ K, the WHIM is overwhelmingly ionized and therefore leaves no obvious trace in the Ly α forest. Figure 1 shows the simulated distribution of the WHIM in the present epoch. The largest concentrations of galaxies occur at the nodes of this cosmic web.

“We worried that the WHIM might be an artifact of our simulation code, rather than a robust consequence of the concordance model,” says Ostriker. So he organized a collaboration of theorists doing concordance-model simulations with a number of different codes. ³ “To our relief,” he recalls, “they all found essentially the same WHIM.”

The WHIM isn’t seen in Ly α absorption or emission because an ionized hydrogen atom is just a naked proton; it can cause colliding electrons to emit bremsstrahlung, but it has no excitable atomic states. The best hope for studying the WHIM is through its small admixture of oxygen. At WHIM temperatures, the predominant heavy-element ionization state is the helium-like configuration O⁶⁺, which has a strong Kα absorption line at 21.6 Å in the soft-x-ray regime.

One might think that the O⁶⁺ line would provide an x-ray analog of the Ly α forest. It turns out, however, that the x-ray flux from quasars at cosmological distances is insufficient for unambiguous identification of redshifted foreground WHIM absorption lines by Chandra or XMM-Newton, the present generation of x-ray satellites. At UV wavelengths, O⁵⁺ absorption lines presumably from WHIM filaments have been identified at various red-shifts. ³ But the less-ionized O⁵⁺ state is sufficiently populated for UV detection only at the low end of the WHIM temperature range.

Illuminating the x-ray forest

Nicastro and company found a replacement for the too-meager x-ray flux of background quasars. Their observation of several clear x-ray absorption lines lets them venture an estimate of the WHIM’s contribution to the cosmic baryon density in the recent epoch. That estimate, albeit still very rough, is consistent with the 40 or 50% of baryons that theorists had labeled missing.

The Nicastro group’s trick was to avail itself of the extraordinary fleeting x-ray brilliance of Markarian 421, a relatively nearby blazar. Blazars are active galactic nuclei characterized by variable outbursts. At a distance of 4 × 10⁸ light years (z = 0.03), Mkn 421, in its outburst phase, has an apparent x-ray brightness greater than that of any other known extra-galactic source. Alerted by the all-sky monitor aboard the Rossi X-Ray Timing Explorer, the group pointed Chandra at Mkn 421 on two dates in 2002 and 2003 for a total exposure of 60 hours. Happily, the blazar’s x-ray brilliance was at historic highs on those occasions.

Figure 2 shows the group’s principal finding: O⁶⁺ Ka absorption lines at two distinct redshifts (z = 0.011 and 0.027). These redshifts are taken to indicate the recessional velocities of absorbing WHIM filaments between us and Mkn 421 in the expanding universe. The two unredshifted lines indicate absorption in local gas. The line attributed to O⁶⁺ at z = 0.033 is more problematic. A foreground absorption line’s z should be smaller than 0.03, Mkn 421’s measured redshift. But the blazar’s z is poorly known.

The problem is more general. The blazar’s historic outbursts provided absorption lines that stand out clearly from the noise. But the provenance of each line is a subtler matter. The confident identification of a redshifted x-ray line with a particular ionization state depends crucially on finding several lines that fit the same z. That’s because of the observational degeneracy between z and transition energy. In addition to the oxygen lines, the Mkn 421 exposure yielded several weaker absorption lines that can be fitted by assuming they are nitrogen Kα lines at z = 0.011 and 0.027. But only one absorption line fits 0.033.

The line of sight to Mkn 421 seems to pierce just two WHIM filaments. From the number of filaments a typical line of sight encounters per unit distance and the depths of the absorption lines at such encounters, one can estimate the WHIM’s oxygen population. But only one line of sight had been sampled, and Mkn 421 is not very far away. Therefore, the estimate by Nicastro and company suffers the statistical uncertainty of very small samples.

Furthermore, to translate the oxygen abundance into an overall baryon density, one must estimate the WHIM’s O/H abundance ratio. And that’s poorly known at present. An educated guess is that it’s about 10% of the Sun’s O/H ratio. But that could easily be wrong by a factor of two or three. Refining the O/H ratio at WHIM temperatures will require painstaking scrutiny of existing and new UV absorption data for traces of the one H atom in a million that’s neutral at 10⁶ K.

Taking the O/H ratio to be 10% of the Sun’s and setting its large uncertainty aside, Nicastro and company estimate Ω^w _b = 2.7^+3.8 _−1.9% for the WHIM’s baryonic contribution to the total cosmic mass–energy density. The best theoretical estimate of the total baryon density Ω_b is (4.5 ± 0.2)%. In the present epoch, about half of those baryons had already been accounted for in galaxies, very hot cluster gas, the Ly α forest, and the coolest fraction of the WHIM. So the estimate from the new WHIM x-ray lines, for all its uncertainty, is consistent with what cosmologists took to be missing.

How much better can one do? X-ray astronomers are unlikely to find another source that illuminates the WHIM for them as brilliantly as Mkn 421 can. Nicastro and coworkers have recently announced ⁵ the discovery of WHIM O⁶⁺ and carbon absorption lines along a second line of sight—this time to a quasar with 10 times Mkn 421’s redshift. Its much lower x-ray brightness, however, makes for rather poor signal to noise.

The large-aperture x-ray telescopes needed to achieve good photon statistics along lines of sight to typical quasars are not expected before the next decade. For a given aperture, however, the sensitivity of an x-ray grating spectrometer improves as the square of the wavelength. “That’s why we’re now concentrating our search for WHIM lines on higher redshifts, while we’re waiting for bigger telescopes,” says Nicastro.

Andrzej Soltan (Copernicus Astronomical Center, Warsaw) and coworkers recently reported complementary evidence for the WHIM in the diffuse cosmic x-ray background. ⁶ Because the WHIM’s mean density is only a handful of protons per cubic meter, its bremsstrahlung—which increases as density squared—would be too faint to see anywhere except near galaxies, where its density in expected to be highest. And that, indeed, is what Soltan and company find.