2023MNRAS.522.3809V

Despite extensive search efforts, direct observations of the first [Population III (Pop III)] stars have not yet succeeded. Theoretical studies have suggested that late Pop III star formation is still possible in pristine clouds of high-mass galaxies, co-existing with Pop II stars, down to the epoch of reionization. Here, we re-assess this finding by exploring Pop III star formation in six 50 h^–1 cMpc simulations performed with the hydrodynamical code dustyGadget. We find that Pop III star formation (∼10^–3.4-10^–3.2 M_⊙ yr^–1 cMpc^–3) is still occurring down to z ∼ 6-8, i.e. well within the reach of deep JWST surveys. At these epochs, $\gtrsim\!{10}~{{\ \rm per\ cent}}$ of the rare massive galaxies with M_⋆ ≳ 3 × 10⁹ M_⊙ are found to host Pop III stars, although with a Pop III/Pop II mass fraction $\lesssim\!0.1~{{\ \rm per\ cent}}$ . Regardless of their mass, Pop III-hosting galaxies are mainly found on the main sequence, at high star-formation rates, probably induced by accretion of pristine gas. This scenario is also supported by their increasing star-formation histories and their preferential location in high-density regions of the cosmic web. Pop III stars are found both in the outskirts of metal-enriched regions and in isolated, pristine clouds. In the latter case, their signal may be less contaminated by Pop IIs, although its detectability will strongly depend on the specific line of sight to the source, due to the complex morphology of the host galaxy and its highly inhomogeneous dust distribution.