By using N-body hydrodynamical cosmological simulations in which the chemistry of major metals and molecules is consistently solved for , we study the interaction of metallic fine-structure lines with the cosmic microwave background radiation ( CMB ) .
Our analysis shows that the collisional induced emissions in the OI 145 \mu m and CII 158 \mu m lines during the epoch of reionization ( z > 5 ) introduce a distortion of the CMB black-body spectrum at low frequencies ( \nu < 300 GHz ) with amplitudes up to \Delta I _ { \nu } / B _ { \nu } ( T _ { CMB } ) \sim 10 ^ { -8 } – 10 ^ { -7 } , i.e. , at the \sim 0.1 percent level of FIRAS upper limits .
Shorter wavelength fine-structure transitions ( like OI 63 \mu m , FeII 26 \mu m , SiII 35 \mu m , FeII 35 \mu m , and FeII 51 \mu m ) typically sample the reionization epoch at higher observing frequencies ( \nu \in [ 400 , 1000 ] GHz ) .
This corresponds to the Wien tail of the CMB black-body spectrum and thus the distortion level induced by those lines may be as high as \Delta I _ { \nu } / B _ { \nu } ( T _ { CMB } ) \sim 10 ^ { -4 } .
Consequently , the brightness temperature anisotropy produced by these lines should be more relevant at higher frequencies : while practically negligible at \nu = 145 GHz , signatures from CII 158 \mu m and OI 145 \mu m should amount to 1 % –5 % of the anisotropy power measured at l \sim 5000 and \nu = 220 GHz by the ACT and SPT collaborations ( after taking \Delta \nu _ { obs } / \nu _ { obs } \simeq 0.005 for the line observations ) .
More importantly , our simulations indicate that anisotropy maps from different lines ( e.g. , OI 145 \mu m and CII 158 \mu m ) at the same redshift show a very high degree ( > 0.8 ) of spatial correlation , allowing for the use of observations at different frequencies ( under different systematic and noise contributions ) to unveil the same snapshot of the reionization epoch .
Finally , our simulations also demonstrate that line-emission anisotropies extracted in narrow frequency/redshift shells are practically uncorrelated in frequency space , thus enabling standard methods for removal of foregrounds that vary smoothly in frequency , just as in HI 21 cm studies .