Mitigating Bias in CMB B-modes from Foreground Cleaning Using a Moment Expansion
One of the primary challenges facing upcoming CMB polarization experiments aiming to measure the inflationary B-mode signal is the removal of polarized foregrounds. The thermal dust foreground is often modeled as a single modified blackbody, however overly simplistic foreground models can bias measurements of the tensor-to-scalar ratio r. As CMB polarization experiments become increasingly sensitive, thermal dust emission models must account for greater complexity in the dust foreground while making minimal assumptions about the underlying distribution of dust properties within a beam. We use Planck dust temperature data to estimate the typical variation in dust properties along the line of sight and examine the impact of these variations on the bias in r if a single modified blackbody model is assumed. We then assess the ability of the moment method to capture the effects of spatial averaging and to reduce bias in the tensor-to-scalar ratio for different possible toy models of dust emission. We find that the expected bias due to temperature variations along the line of sight is significant compared to the target sensitivities of future CMB experiments, and that the use of the moment method could reduce bias as well as shed light into the distribution of dust physical parameters.