Distinctive rings in the 21 cm signal of the epoch of reionization
1 March 2011
9 March 2011 — The epoch of reionization is the moment in the universe when the hydrogen gas, first essentially neutral, becomes ionized by the first UV and X-ray sources (stars and quasars). Astrophysicists hope to get information on the epoch of reionization from future observations of the 21 cm transition of neutral hydrogen, for which numerous projects are in development ((LOFAR, SKA, MWA..). However, it will be difficult to extract the cosmological signal from the raw data because of foreground sources, ionospheric refraction, or the instrumental contamination. The intensity of the foregrounds will be several orders of magnitude greater than that of the 21 cm signal. Any prediction about the properties of the signal is therefore useful as a diagnostic to test systematics and foreground removal procedures. Such a prediction has been precisely done by researchers of the Paris Observatory (LERMA). Using numerical simulations, they emphasized the existence of a ring-shaped signal around the first light sources.
As a result, any source emitting in the UV band is surrounded by a series of concentric shells, so-called Lyman horizons, that photons emitted in given frequency bands cannot overtake. Therefore, the Lyman-alpha flux radial profile around UV sources shows characteristic discontinuities at distances where the emitted photons reach the different Lyman-series lines (Figure 1).
Figure 1: Radial profile for the coupling coefficient for Lyman-alpha excitation, x_alpha, at z=13.42, around the first source appearing in a 140 Mpc size simulation (this size corresponds to the size measured today, subtracting expansion, we call it "comoving"). The red line is the correct profile, while the black line is the result of a simulation that neglects the contribution of higher-order Lyman-series lines. Sharp discontinuities are detected at the predicted Lyman-gamma, Lyman-delta and Lyman-epsilon horizons (arrows).
Because of the Wouthuysen-Field effect, this translates into a similar profile for the differential brightness temperature of the 21 cm signal. On maps of that signal, Lyman horizons are marked as concentric, perfectly spherical rings around the sources (Figure 2).
Figure 2: Map of the quantity Delta T_b r2 at z=13.42, where Delta T_b is the differential (deviation from the CMB) brightness temperature of the 21 cm signal, and r the distance to the source center. The color scale is logarithmic and in arbitrary units. The Lyman-epsilon, Lyman-delta and Lyman-gamma horizons are indicated with white, yellow and red arrows respectively. The map is 140 comoving Mpc on a side and the slice width is 2.8 comoving Mpc. The radiation source is located in the center of a ionized hydrogen bubble (central white spot).
Unfortunately, other nearby sources, appearing progressively in the simulation, will also contribute to the Lyman-alpha background, making it more and more anisotropic. This will progressively wipe out the ring-shaped discontinuities. The LERMA team showed that rings are visible during a redshift interval Delta z 2 after the first light source lit up. Using a stacking technique (averaging the radial profiles of all the sources in the simulation), they are able to extend this interval to Delta z 4, i.e. from z=14 to z=10 approximately.
It is interesting to determine whether that predicted signature will be detectable by the planned Square Kilometre Array, a huge radiotelescope in development which will be considerably more sensitive than any other radio instrument. In order to do so, instrumental noise has been added to the predicted signal. Analysis shows that if the number of sources is sufficient, the profile averaging procedure allows individual fluctuations to be smoothed, and then Lyman horizons are still detectable (Figure 3).
Figure 3: Gradient of the 21cm signal at z=11.05, in a 280 comoving Mpc simulation. The black line refers to the first source appearing in the simulation, without instrumental noise. The red line refers to the same source, this time with the addition of noise. Finally, the blue line shows the gradient of the profile when summing all the sources in the simulation. This averaging method is efficient, and we clearly detect the Lyman-delta and Lyman-epsilon horizons (gradient peaks, shown by arrows), which are not detected in the individual profiles, whether with or without noise.
