Constraining the primordial initial mass function with stellar archaeology

TL;DR

This study uses stellar archaeology and large surveys of the Milky Way to constrain the lower-mass limit of the first stars' initial mass function, providing new insights into early star formation.

Contribution

It introduces a novel method combining semi-analytic modeling and survey data to limit the low-mass cutoff of Population III stars' IMF.

Findings

No Pop III survivors detected in 4 million halo stars excludes IMF below 0.8 M_sun at 68% confidence.

Survey of 20 million halo stars could exclude Pop III stars below 0.8 M_sun at 99% confidence.

Current surveys can tentatively exclude Pop III stars below 0.65 M_sun with 95% confidence.

Abstract

We present a new near-field cosmological probe of the initial mass function (IMF) of the first stars. Specifically, we constrain the lower-mass limit of the Population III (Pop III) IMF with the total number of stars in large, unbiased surveys of the Milky Way. We model the early star formation history in a Milky Way-like halo with a semi-analytic approach, based on Monte-Carlo sampling of dark matter merger trees, combined with a treatment of the most important feedback mechanisms. Assuming a logarithmically flat Pop III IMF and varying its low mass limit, we derive the number of expected survivors of these first stars, using them to estimate the probability to detect any such Pop III fossil in stellar archaeological surveys. Following our analysis, the most promising region to find possible Pop III survivors is the stellar halo of the Milky Way, which is the best target for future surveys. We find that if no genuine Pop III survivor is detected in a sample size of $4 \times 10^6$ ($2 \times 10^7$) halo stars with well-controlled selection effects, then we can exclude the hypothesis that the primordial IMF extended down below $0.8 M_\odot$ at a confidence level of 68% (99%). With the sample size of the Hamburg/ESO survey, we can tentatively exclude Pop III stars with masses below $0.65 M_\odot$ with a confidence level of 95%, although this is subject to significant uncertainties. To fully harness the potential of our approach, future large surveys are needed that employ uniform, unbiased selection strategies for high-resolution spectroscopic follow-up.