Near the last phases of recombination, when the n = 2 to n = 1 transition was yet to be recombined, there were two plausible paths: recombination either by the 2p-1s Lyα transition or by the 2s-1s two-photon decay.

The Lyα channel is optically thick and has a large cross-section of interaction that would immediately ionise its neighbour, making this channel less efficient in bringing about recombination. For this channel to work, there needs to be a mechanism with which we can remove Lyα photons from the medium, and it turns out that the cosmological redshifting of these photons leads to recombination in this channel. The number of scatterings a typical Lyα photon would undergo is very large and depends on the medium number density, expansion rate, as well as its cross-section. The transition rate of Lyα is 7x10^8 s-1, but since the scattering rate is of the same order, the effective channel of recombination via Lyα is reduced. The scattering rate is proportional to the Sobolev optical depth of the medium, and a typical Lyα photon undergoes about 10^8 scatterings before getting redshifted out of the Lyα resonance.

On the other hand, the 2s-1s two-photon decay channel has a decay timescale of 8.2 s-1, and due to the reduced efficiency of Lyα, it is observed that this channel contributes almost equally to the Lyα channel in bringing about recombination (57 % to 43 % in favour of 2s-1s channel). This channel adds photons as a continuum since the only constraint is that the two-photon energy sum equals the energy difference between 2s-1s states, making it possible for photons to take away equal or disproportionate energies out of the n = 2 to n = 1 transition energy. This is thought to proceed via the creation of a meta-stable state between the two levels, and a probability distribution of the two photon energies exists. Hence, this channel can also potentially alter the CRR spectrum, including the low-frequency (higher transitions) part of the signal.

Plot on the left: Step-like feature in the CMB spectrum due to the accumulation of photons from the blue side of the Lyα resonance to the red - redshifting of photons. The number of photons on the blue side is reduced, and they appear on the red side, leading to the feature. Production of resonance photons by the formation of H atoms produces a step in the spectrum. This feature is present only for the Lyα line (last transition to recombine), as other transitions would end up in neighbouring transitions. This step feature might have the intensity of the order of typical recombination line features. The CRR additive distortions are added on top of this feature?   

Recombination: We need a process of removing the Lyα photons from the medium. In our universe, the Hubble expansion does that for us - it redshifts the CMB photons, favouring the recombination to begin. It also saturates the transitions whose reaction rate has gone below the Hubble rate. An interesting scenario to ponder is the question of whether recombination can occur in the absence of the expansion of the universe. The answer is yes, and this proceeds via the diffusive effects of Lyα resonance. The redshifting of Lyα brought about by the expansion of the universe is a linear process dependent on speed v. However, the diffusive process, proportional to temperature and having a dependence of v^2/c^2, broadens the Lyα resonance. In this case, the plot on the left would have symmetric red and blue shifted components, where the blue component can further interfere with Lyβ or higher energy transitions.  

ARCADE-2, a NASA balloon based experiment to make an absolute measurement of radio sky temperature and to constrain the CMB and its anisotropies at various spot radio frequencies of 3.3, 8, 10, 30 and 90 GHz made an unusual observation (Kogut et. al 2011). The observed 'sky temperature' (intensity expressed in units of Kelvins) was higher than expected, most prominently at lower frequencies less than 10 GHz and all the way down to 50 MHz, as later confirmed by the LWA observations (Dowell & Taylor, 2018). 

From sky temperature, on subtracting an estimate of our diffuse Galactic emissions, extra-Galactic emissions and the CMB monopole, one would expect to get a zero residual, as all known sources of diffuse sky emissions have been accounted for. Interestingly, ARCADE-2 observed that this residual is non-zero, and infact follows a power-law like feature with amplitude 18.4 ± 2.1 K at 0.31 GHz and a spectral index of −2.57 ± 0.05 (Seiffert et. al 2011). 

*** Stay tuned ***

*** Stay tuned ***