Avoiding baryonic feedback effects on neutrino mass measurements from CMB lensing

A measurement of the sum of neutrino masses is one of the main applications of upcoming measurements of gravitational lensing of the cosmic microwave background (CMB). This measurement can be confounded by modelling uncertainties related to so-called "baryonic effects" on the clustering of matter, arising from gas dynamics, star formation, and feedback from active galactic nuclei and supernovae. In particular, a wrong assumption about the form of baryonic effects on CMB lensing can bias a neutrino mass measurement by a significant fraction of the statistical uncertainty. In this paper, we investigate three methods for mitigating this bias: (1) restricting the use of small-scale CMB lensing information when constraining neutrino mass; (2) using an external tracer to remove the low-redshift contribution to a CMB lensing map; and (3) marginalizing over a parametric model for baryonic effects on large-scale structure. We test these methods using Fisher matrix forecasts for experiments resembling the Simons Observatory and CMB-S4, using a variety of recent hydrodynamical simulations to represent the range of possible baryonic effects, and using cosmic shear measured by the Rubin Observatory's LSST as the tracer in method (2). We find that a combination of (1) and (2), or (3) on its own, will be effective in reducing the bias induced by baryonic effects on a neutrino mass measurement to a negligible level, without a significant increase in the associated statistical uncertainty.