Foreground Biases on Primordial Non-Gaussianity Measurements from the CMB Temperature Bispectrum: Implications for Planck and Beyond (1807.07324v3)

Abstract: The cosmic microwave background (CMB) temperature bispectrum is currently the most precise tool for constraining primordial non-Gaussianity (NG). The Planck temperature data tightly constrain the amplitude of local-type NG: $f_{\rm NL}^{\rm loc} = 2.5 \pm 5.7$. Here, we compute previously-neglected foreground biases in temperature-based $f_{\rm NL}^{\rm loc}$ measurements, due to the integrated Sachs-Wolfe (ISW) effect, gravitational lensing, the thermal and kinematic Sunyaev-Zel'dovich effects, and the cosmic infrared background. While standard analyses already subtract a significant bias on $f_{\rm NL}^{\rm loc}$ due to the ISW-lensing bispectrum, many other secondary anisotropy terms are present in the temperature bispectrum. We compute the dominant biases on $f_{\rm NL}^{\rm loc}$ arising from these signals. Most of the biases are non-blackbody, and are thus reduced by multifrequency component separation methods; however, recent analyses have found that extragalactic foregrounds are present at non-negligible levels in the Planck component-separated maps. Moreover, the Planck FFP8 simulations do not include the foreground correlations that generate these biases. We compute the biases for individual frequencies; some are comparable to the statistical error bar on $f_{\rm NL}^{\rm loc}$, even for the main CMB channels (100, 143, and 217 GHz). For future experiments, they greatly exceed the statistical error (considering temperature only). Alternatively, the foreground contributions can be marginalized over, but this leads to a non-negligible increase in the error bar on $f_{\rm NL}^{\rm loc}$. A full assessment will require calculations in tandem with component separation, ideally using simulations. We also compute these biases for equilateral and orthogonal NG, finding large effects for the latter. We conclude that the search for primordial NG using Planck data may not yet be over.