LAUSR

Standard analyses of the reionization history of the universe from Planck cosmic microwave background (CMB) polarization measurements consider only the overall optical depth to electron scattering ($\tau$), and further assume a step-like reionization history. However, the polarization data contain information beyond the overall optical depth, and the assumption of a step-like function may miss high redshift contributions to the optical depth and lead to biased $\tau$ constraints. Accounting for its full reionization information content, we reconsider the interpretation of Planck 2015 Low Frequency Instrument (LFI) polarization data using simple, yet physically-motivated reionization models. We show that these measurements still, in fact, allow a non-negligible contribution from metal-free (Pop-III) stars forming in mini-halos of mass $M \sim 10^5-10^6 M_\odot$ at $z \gtrsim 15$, provided this mode of star formation is fairly inefficient. Our best fit model includes an early, self-regulated phase of Pop-III star formation in which the reionization history has a gradual, plateau feature. In this model, $\sim$20\% of the volume of the universe is ionized by $z \sim 20$, yet it nevertheless provides a good match to the Planck LFI measurements. Although preferred when the full information content of the data is incorporated, this model would spuriously be disfavored in the standard analysis. This preference is driven mostly by excess power from E-mode polarization at multipoles of $10 \lesssim \ell \lesssim 20$, which may reflect remaining systematic errors in the data, a statistical fluctuation, or signatures of the first stars. Measurements from the Planck High Frequency Instrument (HFI) should be able to confirm or refute this hint and future cosmic-variance limited E-mode polarization surveys can provide substantially more information on these signatures