Mitigating Complex Dust Foregrounds in Future CMB Polarization Experiments

Polarized Galactic foregrounds are one of the primary sources of systematic error in measurements of the B-mode polarization of the Cosmic Microwave Background (CMB). Experiments are becoming increasingly sensitive to complexities in the foreground frequency spectra that are not captured by standard parametric models, potentially affecting our ability to efficiently separate out these components. Employing a suite of dust models encompassing a variety of physical effects, we simulate observations of a future seven-band CMB experiment to assess the impact of these complexities on parametric component separation. We identify configurations of frequency bands that minimize the `model errors' caused by fitting simple parametric models to more complex `true' foreground spectra, which bias the inferred CMB signal. We find that: (a) fits employing a simple two parameter modified blackbody (MBB) dust model tend to produce significant bias in the recovered polarized CMB signal in the presence of physically realistic dust foregrounds; (b) generalized MBB models with three additional parameters reduce this bias in most cases, but non-negligible biases can remain, and can be hard to detect; and (c) line of sight effects, which give rise to frequency decorrelation, and the presence of iron grains are the most problematic complexities in the dust emission for recovering the true CMB signal. More sophisticated simulations will be needed to demonstrate that future CMB experiments can successfully mitigate these more physically realistic dust foregrounds.