AI-generated illustration for Galileo Whispers of a generic stellar photosphere. Surface texture and color are illustrative; the image does not measure lithium abundance or show primordial nucleosynthesis.
Old stars contain a clue to the early universe, but their visible surfaces are not untouched samples of it. Warm, metal-poor stars generally show less lithium than standard Big Bang nucleosynthesis predicts when the cosmic baryon density is supplied independently. The mismatch is the cosmological lithium problem.
The comparison involves two inferences: a nuclear-reaction calculation for the early universe and a stellar-atmosphere calculation translating a spectral feature into an abundance. A discrepancy between them deserves attention, without automatically identifying either new physics or one settled stellar explanation.
What the primordial prediction means
Big Bang nucleosynthesis describes the formation of light nuclei in the early hot universe. Given the baryon density and the standard physical ingredients, it predicts primordial element ratios within nuclear and cosmological uncertainties. Much of the eventual primordial lithium-7 is initially produced as beryllium-7, which later becomes lithium-7.
The 2025 Particle Data Group review, especially its lithium discussion, compares that prediction with stellar evidence. Deuterium and helium provide crucial cross-checks: changing early-universe physics to reduce lithium must also preserve their successful constraints. One cannot repair a single abundance while ignoring the others.
A stellar spectrum measures an atmosphere
Observers often use the neutral-lithium feature near 670.8 nanometers in warm halo dwarfs. They measure the absorption and compare it with calculated spectra. Effective temperature, surface gravity, composition and the treatment of line formation affect the inferred lithium abundance.
Low metallicity makes these stars valuable because they formed from material with relatively little heavy-element enrichment. It does not guarantee that the present surface lithium is exactly the natal value. Diffusion can redistribute elements below the observable layers, and mixing can carry lithium to hotter regions where it is destroyed.
In Sbordone and colleagues’ 2010 study, the analysis used several temperature scales and three-dimensional, non-local-thermodynamic-equilibrium lithium calculations. Below very low iron abundances, their sample did not maintain a simple flat lithium plateau. The observed behavior therefore cannot be reduced to one universally identical surface abundance in all ancient stars.
Keep three abundances separate
A useful diagnostic chain is: primordial gas abundance, abundance in the gas from which a particular star formed, and abundance now inferred in its photosphere. Those quantities can differ. Galactic chemical evolution acts before star formation; transport and destruction operate during stellar life; atmospheric modeling governs the final measurement.
For a deliberately assumed example, suppose the primordial calculation gives three lithium nuclei per arbitrary reference parcel, while a present stellar surface yields one. The ratio is three, but that arithmetic alone cannot tell whether two nuclei were destroyed, moved out of the observed layer, or whether one of the inferred starting values is biased. The example illustrates the inference problem, not an observed stellar parcel or an exact universal discrepancy.
What would make a resolution convincing?
A stellar explanation should reproduce patterns with temperature, mass and metallicity, including small scatter where a plateau exists and greater scatter where it breaks down. An early-universe explanation must fit other light elements and cosmological evidence as well. Merely raising all inferred stellar abundances by an arbitrary offset does not accomplish either test.
The PDG review discusses depletion as an important possible resolution while retaining the problem as unresolved. Old stars can constrain primordial lithium only through a physical account of what happened between their birth and the light reaching a telescope. Their usefulness comes from testing that account across many stars, rather than assuming their surfaces never changed.