Whether any long-lived, low-mass, metal-free Population III stars survive today in the Milky Way and its satellite galaxies - and what their presence or absence implies for the low-mass end of the first-star initial mass function - is an open question posed by the ELT-ANDES science case (Roederer et al. 2023). No star without metals has ever been found: the most iron-poor stars known are second-generation objects, and statistical non-detections in the Galactic halo and ultra-faint dwarf galaxies already favor a top-heavy first-star IMF (minimum mass above 0.8 M⊙0.8\,M_\odot or characteristic mass above 1 M⊙1\,M_\odot at 99% confidence in one model), but these limits are model-dependent, and theory permits survival of low-mass primordial fragments that were ejected early from their birth clumps. The problem is to determine whether a metal-free survivor can be confirmed directly (hydrogen Balmer absorption with no metal or molecular lines), or whether a well-characterized non-detection can be converted into a high-confidence exclusion and a quantitative constraint on the low-mass cutoff of the first-star IMF.

Background

The first, metal-free generation of stars (Population III, or Pop III) formed from primordial hydrogen and helium gas at redshifts z≳15z\gtrsim 15-30, in small dark-matter halos of about a million solar masses (minihalos) where gas cooling relied on molecular hydrogen. Because no metals (elements heavier than helium) were available to aid cooling, theoretical work has long predicted a top-heavy initial mass function (IMF - the distribution of stellar masses at birth, 'top-heavy' meaning weighted toward high masses), with characteristic masses of order 10-100 solar masses. Such massive stars live only a few million years, so Pop III stars are usually sought either directly at high redshift or indirectly, through the abundance patterns they imprinted on later stellar generations - an approach called stellar archaeology. Stellar metallicity is commonly quoted as [Fe/H], the logarithmic iron abundance relative to the Sun; 'metal-poor' survey samples reach [Fe/H] < -3, and the most iron-poor stars known have measured [Fe/H] near -5. The record object, SMSS J031300.36-670839.3, shows no detectable iron lines at all but does show calcium and magnesium, so it is a second-generation star enriched by a single Pop III explosion (Clarkson, Herwig and Pignatari 2017); every star examined to date contains at least some heavy elements. A crucial exception motivates the local search: if the Pop III IMF extended below roughly 0.8 M⊙0.8\,M_\odot, those low-mass stars would still be shining today, since their hydrogen-burning main-sequence lifetimes exceed the age of the Universe, and they would be observable as unevolved or barely evolved stars in the Milky Way halo and nearby dwarf galaxies, including the ultra-faint dwarfs (the smallest, most metal-poor Milky Way satellites, thought to preserve the earliest enrichment episodes). Finding one would deliver an intact specimen of the first stellar generation; a well-characterized failure to find any constrains the low-mass cutoff of the first-star IMF, a core input to models of cosmic reionization, the earliest chemical enrichment, and seed black-hole formation. The observational signature of a true survivor, as specified in the ELT-ANDES science case, is absorption in the hydrogen Balmer series (the visible-wavelength spectral lines of atomic hydrogen) together with the complete absence of metal and molecular absorption - the CH G band (a molecular carbon feature) near 430 nm, the Na I D doublet near 589 nm, the Mg I b triplet near 518 nm, the Ca II H and K doublet at 393.3 and 396.7 nm, and Fe I lines at 358.1 and 382.0 nm - which requires high-resolution, high-S/N spectroscopy on the [Fe/H] < -4 candidate population. Existing statistical non-detections already limit the possibilities: models producing more than ten low-mass Pop III stars per minihalo are excluded (Ishiyama et al. 2016), the Hamburg/ESO survey tentatively excludes an IMF extending below 0.65 M⊙0.65\,M_\odot at 95% confidence (Hartwig et al. 2015), and ultra-faint-dwarf data imply that either the characteristic mass exceeds 1 M⊙1\,M_\odot or the minimum mass exceeds 0.8 M⊙0.8\,M_\odot at 99% confidence, strongly favoring a top-heavy IMF (Rossi et al. 2021). Three complications keep the question open. First, surface pollution: accretion of interstellar material, including kilometer-scale interstellar objects, can raise the surface iron abundance of a 0.8 M⊙0.8\,M_\odot survivor to [Fe/H] of order -6, so a survivor may masquerade as a mildly metal-poor star and evade the no-metal-lines signature (Tanikawa et al. 2018). Second, discrimination: genuinely second-generation, extremely metal-poor stars are common in the same surveys, and only the most carbon-enhanced, iron-poor ones carry a pure Pop III imprint (Vanni et al. 2023). Third, theory is split: fragmentation of primordial accretion disks produces a logarithmically flat mass function with low-mass fragments (Glover and Klessen 2025), but most fragments accrete up to high masses unless ejected from their birth clump early, which semianalytic models show is possible (Dutta et al. 2017).

Problem Statement

Determine whether any long-lived, low-mass, metal-free Population III stars exist today in the Milky Way halo or its satellite dwarf galaxies, and thereby constrain the low-mass end of the Population III initial mass function - a single coupled objective in which the observational outcome and the IMF constraint are two faces of one result. A confirmed survivor is a star showing absorption in the hydrogen Balmer lines (the visible-wavelength spectral series of atomic hydrogen) with no absorption from any metal or molecule, with quantitative upper limits on the CH G band (a molecular carbon feature) near 430 nm, the Na I D doublet near 589 nm, the Mg I b triplet near 518 nm, the Ca II H and K doublet at 393.3 and 396.7 nm, and the Fe I lines at 358.1 and 382.0 nm, together with evidence ruling out surface pollution, internal mixing, and second-generation or non-stellar contaminants. A high-confidence exclusion of the existence of such stars is one that rests on the survey's own selection function, the spatial coverage of the predicted survivor habitats (the Galactic halo and its dwarf satellites), and a statistical model converting the non-detection into a stated-confidence lower bound on the Pop III IMF minimum mass (or an upper bound on the present-day number of survivors), propagating the degeneracy between the minimum mass and the IMF shape and accounting for the possibility that survivors are spectroscopically hidden by surface pollution or otherwise fail the no-metal-lines signature.

References

  1. Roederer, Ian U.; Alvarado-Gómez, Julián D.; Allende Prieto, Carlos; et al. (ANDES Science Team) (2023). 'The discovery space of ELT-ANDES. Stars and stellar populations.' arXiv:2311.16320, doi:10.48550/arXiv.2311.16320. https://arxiv.org/abs/2311.16320
  2. Hartwig, Tilman; Bromm, Volker; Klessen, Ralf S.; Glover, Simon C. O. (2015). 'Constraining the primordial initial mass function with stellar archaeology.' arXiv:1411.1238, doi:10.48550/arXiv.1411.1238. https://arxiv.org/abs/1411.1238
  3. Ishiyama, Tomoaki; Sudo, Kae; Yokoi, Shingo; Hasegawa, Kenji; Tominaga, Nozomu; Susa, Hajime (2016). 'Where are the Low-Mass Population III Stars?' arXiv:1602.00465, doi:10.48550/arXiv.1602.00465. https://arxiv.org/abs/1602.00465
  4. Rossi, Martina; Salvadori, Stefania; Skúladóttir, Ása (2021). 'Ultra-faint dwarf galaxies: unveiling the minimum mass of the first stars.' arXiv:2103.09834, doi:10.48550/arXiv.2103.09834. https://arxiv.org/abs/2103.09834
  5. Dutta, Jayanta; Sur, Sharanya; Stacy, Athena; Bagla, Jasjeet Singh (2017). 'Modeling the survival of Population III stars till present day.' arXiv:1712.06912, doi:10.48550/arXiv.1712.06912. https://arxiv.org/abs/1712.06912
  6. Tanikawa, Ataru; Suzuki, Takeru K.; Doi, Yasuo (2018). 'Metal Pollution of Low-Mass Population III Stars through Accretion of Interstellar Objects like `Oumuamua.' arXiv:1804.08200, doi:10.48550/arXiv.1804.08200. https://arxiv.org/abs/1804.08200
  7. Vanni, Irene; Salvadori, Stefania; Skúladóttir, Ása (2023). 'Are all metal-poor stars of second-generation?' arXiv:2305.02358, doi:10.48550/arXiv.2305.02358. https://arxiv.org/abs/2305.02358
  8. Glover, Simon C. O.; Klessen, Ralf S. (2025). 'The first stars.' arXiv:2509.14799, doi:10.48550/arXiv.2509.14799. https://arxiv.org/abs/2509.14799
  9. Ferreira, Thiago; Bellinger, Earl P.; Farag, Ebraheem; Lindsay, Christopher J. (2026). 'Red-Giant Asteroseismology of Low-Mass Population III Stars.' arXiv:2603.08804, doi:10.48550/arXiv.2603.08804. https://arxiv.org/abs/2603.08804
  10. Clarkson, O.; Herwig, F.; Pignatari, M. (2017). 'Pop III i-process Nucleosynthesis and the Elemental Abundances of SMSS J0313-6708 and the Most Iron-Poor Stars.' arXiv:1710.01763, doi:10.48550/arXiv.1710.01763. https://arxiv.org/abs/1710.01763
  11. Visbal, Eli; Hazlett, Ryan; Bryan, Greg L. (2025). 'LAP1-B is the First Observed System Consistent with Theoretical Predictions for Population III Stars.' arXiv:2508.03842, doi:10.48550/arXiv.2508.03842. https://arxiv.org/abs/2508.03842
  12. Fujimoto, Seiji; Naidu, Rohan P.; Chisholm, John; Atek, Hakim; Endsley, Ryan; Kokorev, Vasily; et al. (2025). 'GLIMPSE: An ultra-faint ~10^5 M_sun Pop III Galaxy Candidate and First Constraints on the Pop III UV Luminosity Function at z ~ 6-7.' arXiv:2501.11678, doi:10.48550/arXiv.2501.11678. https://arxiv.org/abs/2501.11678
  13. Cai, Sijia; Li, Mingyu; Cai, Zheng; Wu, Yunjing; Yu, Fujiang; Dickinson, Mark; et al. (2025). 'A Metal-Free Galaxy at z = 3.19? Evidence of Late Population III Star Formation at Cosmic Noon.' arXiv:2507.17820, doi:10.48550/arXiv.2507.17820. https://arxiv.org/abs/2507.17820