Importance of high-frequency bands for removal of thermal dust in ECHO

The Indian Consortium of Cosmologists has proposed a cosmic microwave background (CMB) space mission, Exploring Cosmic History and Origin ($ECHO$). A major scientific goal of the mission is to detect the primordial $B$-mode signal of CMB polarization. The detection of the targeted signal is very challenging as it is deeply buried under the dominant astrophysical foreground emissions of the thermal dust and the Galactic synchrotron. To facilitate the adequate subtraction of thermal dust, the instrument design of $ECHO$ has included nine dust-dominated high-frequency bands over the frequency range 220--850 GHz. In this work, we closely reexamine the utility of the high-frequency $ECHO$ bands in foreground subtraction using the Needlet Internal Linear Combination component separation method. We consider three dust models: a physical dust model, a dust spectral energy distribution (SED) with a single modified black body (MBB) emission law and a multilayer dust model with frequency-frequency decorrelation. We consider eleven $ECHO$ bands in the 28--190 GHz range as our baseline configuration and investigate the changes in the level foreground and noise residuals as subsequent dust-dominated high-frequency bands are added. We find that adding the high-frequency bands leads to a consistent decrease in the level of residual foreground and noise, and the sensitivity of $r$ measurement improves. Most of the reduction in both residual levels and enhancement in the sensitivity is achieved in the 28--600 GHz frequency range. Negligible change in residual levels is seen by extending the frequency range from 600 GHz to 850 GHz. We find that extending the $ECHO$ frequency bands from 190 GHz to 340 GHz leads to a 40%--50% reduction in the foreground and noise residual levels in the recovered CMB map. Correspondingly the sensitivity of $ECHO$ toward $r$ also improves by a similar amount. Furthermore, incorporating higher frequencies up to 600 GHz yields an additional reduction of 12%--15% in the residual levels and uncertainty on $r$. However, extending observations up to the 850 GHz frequency band only leads to a marginal improvement in sensitivity, ranging from 3%--7%.