Anastasia Fialkov (Tel Aviv University; École Normale Supérieure, Paris) [link], Rennan Barkana (Tel Aviv University) [link], Eli Visbal (Columbia University; Harvard University)

Nature, advance online publication on Wednesday, Feb. 5'th 2014 (in print Thursday, Feb. 13'th).

More details and related images

The intensity of the emission by hydrogen is affected by the surrounding stars. In particular, massive stars emit energetic ultra-violet radiation that breaks up ("ionizes") hydrogen atoms that it hits. This creates bubbles that are nearly empty of hydrogen atoms (and thus of 21 cm emission) around the early galaxies. A second major effect comes from stellar systems called X-ray binaries, especially black-hole binaries. These are binary stars in which the more massive star ended its life with a supernova explosion that left a black-hole remnant in its place; gas from the companion star is pulled in towards the black hole, gets ripped apart in the strong gravity, and produces high-energy (X-ray) radiation. This radiation reaches large distances, and is believed to have caused the first substantial heating of the cosmic gas (after it had cooled down as a result of the cosmic expansion).

It has been firmly believed that the heating occurred very early, so that by the time of reionization, all the gas was already hot. In this case, the hydrogen gas emitted brightly in the radio everywhere, except in the ionized bubbles (where there are no hydrogen atoms). The result is a clear expectation for a 21 cm signal driven by reionization: uniform intensity initially, increasing spatial fluctuations as the ionized bubbles grow, with the emission finally disappearing as all the gas becomes ionized. We have discovered, however, that this standard picture depends sensitively on the spectrum (energy distribution) of the heating X-ray radiation. Previously it has been assumed that most of the X-rays come out at relatively low energies (below 1 keV: the energy gained by an electron moving through 1000 Volt); however, accumulating data on the radiation from X-ray binaries shows that high-energy X-rays actually dominate. We find that this substantially changes the expectations for cosmic heating.

One consequence is that the heating is substantially delayed. High-energy X-rays typically travel a long distance, over a long time, before their energy is absorbed and heats the gas. As a result, the heating should still be underway during cosmic reionization, with the gas still cold in some areas. Unlike the previous prediction, the observed pattern of 21 cm emission of hydrogen should still be affected by the distribution of gas temperature as well as by the ionized bubbles of reionization.

Map of the radio intensity, according to our new prediction, at redshift 12.1. For the new prediction, this corresponds to the heating transition (i.e., when the average gas temperature equals that of the cosmic microwave background). The color scale indicates the intensity in millikelvin units, where negative numbers correspond to absorption. This map ranges from -4.8 mK to 8.7 mK, but it is shown on a common scale with the previous prediction (i.e., the corresponding image on the right-hand side). In terms of the paper, the model shown here corresponds to atomic cooling halos, late reionization, and our standard X-ray normalization.

Map of the radio intensity, according to the previous prediction, shown at the same time (redshift 12.1) and stage of reionization (14% ionization of the cosmic gas) as for the new prediction. This map ranges from 0 mK to 27.4 mK, but it is shown on a common scale with the new prediction. The slice shows a region that in today's Universe corresponds on each side to 384 Megaparsecs (1.25 billion light years). Observed at redshift 12.1 (when the Universe was 360 million years old), it corresponds to 2.2 degrees on the sky (about four times the angular size of the moon). Note the ionized bubbles in blue (corresponding to zero intensity, since there are no hydrogen atoms there).

A key moment is when the cosmic gas is heated to the point that its average temperature equals that of the cosmic microwave background (the latter is the radiation that fills the Universe and is a remnant of an early time in which the Universe was hot, dense, and full of light). This moment should have a distinct, previously unexpected, signature in observations: the radio intensity at this moment should be quite uniform on the sky. The reason is that high-energy X-rays come from large distances, and this smooths over small-scale fluctuations.

The two images show examples of predicted radio emission maps at an early time in cosmic reionization, when 14% of the cosmic gas was in ionized bubbles. The corresponding redshift in this model is 12.1 (which means that the Universe has expanded by a factor of 13.1 from that early time - a cosmic age of 360 million years - until today's 14 billion year old Universe). The small ionized bubbles can be seen in the blue corresponding to zero on the intensity scale (They are easy to see in the right-hand image, where they are the only blue regions). The time shown here corresponds (for the new prediction) to the key heating moment noted above. Compared to the previous prediction, under the new prediction the gas is colder (and so emits less, even showing absorption - negative intensity), and the intensity map is much more uniform and especially smooth (fuzzy) on small spatial scales.

Several international teams of radio astronomers are currently competing to be the first to detect the radio waves emitted at 21 cm by hydrogen in the early Universe (acronyms of some of the experiments are LOFAR, MWA, PAPER, GMRT, EDGES, and the future SKA). If the key moment of smooth 21 cm intensity is detected, this would confirm the predicted scenario, and indirectly measure the average gas temperature at that time as well as the properties of the X-ray radiation from some of the earliest black holes.

Map of the radio intensity, according to our new prediction, at redshift 12.1. Shown on a separate intensity scale.

Map of the radio intensity, according to the previous prediction, at redshift 12.1. Shown on a separate intensity scale. This gives a different comparison, showing more clearly that under the new prediction the intensity map is smoother (fuzzier) on small spatial scales

Map of the gas temperature, according to our new prediction, at redshift 12.1. Shown on a common (logarithmic) temperature scale with the previous prediction. Range: 30.6 - 55.2 Kelvin. Ionized regions are shown with the color of the max of the scale (although ionized gas is actually above 10,000 K).

Map of the gas temperature, according to the previous prediction, at redshift 12.1. Shown on a common (logarithmic) temperature scale with the new prediction. Range: 75 - 1562 Kelvin. Ionized regions are shown with the color of the max of the scale (more easily seen in the map of the new prediction).

Map of the radio intensity, according to our new prediction, at redshift 8.7, when half the cosmic gas is ionized. Shown on a common scale with the previous prediction. Range: 0 - 14.5 mK. Ionized bubbles are in blue.

Map of the radio intensity, according to the previous prediction, at redshift 8.7, when half the cosmic gas is ionized. The gas is hotter (and thus more strongly emitting) under the old prediction, but the overall pattern of intensity fluctuations is similar in the two cases. Range: 0 - 16.8 mK. Ionized bubbles are in blue.

Map of the gas temperature, according to our new prediction, at redshift 8.7 (midpoint of reionization). Shown on a common (logarithmic) temperature scale with the previous prediction. Range: 164 - 200 Kelvin. Ionized regions are shown with the color of the max of the scale (although ionized gas is actually above 10,000 K).

Map of the gas temperature, according to the previous prediction, at redshift 8.7 (midpoint of reionization). Shown on a common (logarithmic) temperature scale with the new prediction. Range: 691 - 2566 Kelvin. Ionized regions are shown with the color of the max of the scale (more easily seen in the map of the new prediction).

Map of the fluctuation in the density of matter (mostly dark matter) at redshift 12.1 (zero indicates the cosmic mean density). Density fluctuations are the underlying driver of all the other maps. Regions of higher density have stronger gravity and form more galaxies, and thus tend to both heat and reionize earlier than regions of lower density.