The Cosmological Lithium Problem: Big Bang’s Missing Lithium

Big Bang nucleosynthesis is one of the great successes of cosmology. Using known nuclear physics, it predicts how much deuterium, helium and lithium the young universe made, and most of those predictions match what astronomers measure. Lithium is the exception. The oldest stars contain roughly three times less lithium-7 than the theory says they should. This is the cosmological lithium problem, and it is still open. This post builds on our explainer How Did the First Elements Form? and is drawn by Primordix, our nucleosynthesis character.

What does Big Bang nucleosynthesis predict?

In the hot, dense early universe, protons and neutrons fused into the lightest nuclei. The Particle Data Group review describes the abundances of deuterium (D), helium-3 (³He), helium-4 (⁴He) and lithium-7 (⁷Li), “synthesized at the end of the first three minutes,” as being “in good overall agreement” with observations, which is “particularly impressive given that these abundances span nine orders of magnitude” (Fields, Molaro & Sarkar, PDG^[1]).

In standard Big Bang nucleosynthesis, with Standard Model physics only, there is just one free number: the baryon-to-photon ratio η, the number of protons and neutrons (baryons) per photon. Cosmologists usually write η₁₀ = η × 10¹⁰. The range of η that fits both deuterium and helium-4 is essentially set by deuterium and gives η₁₀ = 6.040 ± 0.118 (PDG^[1]).

How does the cosmic microwave background check it?

The cosmic microwave background, the leftover glow of the early universe (see What Is the Cosmic Microwave Background?), gives an independent measure of the same quantity. The Planck satellite’s data give Ω_bh² = 0.02230 ± 0.00021, corresponding to η₁₀ = 6.104 ± 0.058 (PDG^[1]; Planck Collaboration 2020^[2]). Here Ω_b is the density of ordinary (baryonic) matter as a fraction of the critical density, and h is the Hubble constant divided by 100 km s⁻¹ Mpc⁻¹.

The two numbers agree, one from nuclear reactions in the first minutes and one from light released about 380,000 years later. The PDG review calls this “a non-trivial success” for deuterium and helium-4. With η fixed by Planck, Big Bang nucleosynthesis predicts every light-element abundance with no free parameters left, and that is where lithium fails.

How much lithium is missing?

Most primordial lithium-7 was actually made as beryllium-7, mainly through the reaction ³He(α,γ)⁷Be, in which a helium-3 nucleus captures an alpha particle (a helium-4 nucleus) and emits a gamma ray. The beryllium-7 later decays to lithium-7. The predicted primordial ratio is Li/H = (4.72 ± 0.7) × 10⁻¹⁰, meaning about 5 lithium atoms per ten billion hydrogen atoms (PDG^[1]).

To measure primordial lithium, astronomers look at old, metal-poor stars in the halo of our Galaxy. In 1982 Monique and François Spite reported that extremely old (Population II) dwarf stars have less lithium than younger Population I stars (Spite & Spite 1982^[3]). The level they found is now called the Spite plateau: ⁷Li “does not vary significantly in halo dwarfs with metallicities ≲ 1/30 of Solar.” From stars in a carefully chosen metallicity range, the PDG review takes Li/H = (1.6 ± 0.3) × 10⁻¹⁰ (PDG^[1]).

Predicted ÷ observed ≈ 4.72 ÷ 1.6 ≈ 3

In the review’s words, the prediction “is a factor 3.1 higher” than the observed value, “which is a 4.4σ discrepancy” (PDG^[1]). The symbol σ is the standard deviation, so 4.4σ means the gap is 4.4 times the combined uncertainty. Brian Fields’s 2011 review also put the plateau a factor of 3 to 4 below the prediction (Fields 2011^[4]). Stars that the Milky Way captured from other galaxies long ago show the same low lithium, so the problem appears to be universal (PDG^[1]).

What could explain the lithium problem?

Fields (2011) lists three classes of possible solution (Fields 2011^[4]): astrophysical systematics in the observations, unknown or poorly measured nuclear reactions that destroy beryllium-7, and physics beyond the Standard Model.

1. Nuclear physics

Perhaps a reaction rate that destroys beryllium-7 has been underestimated. Experiments have chased this. The n_TOF Collaboration at CERN measured the ⁷Be(n,p)⁷Li reaction across a wide energy range and found that its new rate gives “only a relatively minor improvement” to the problem (Damone et al. 2018^[5]). The PDG review concludes that experimental searches have “closed off” the remaining resonance possibilities, “making a nuclear fix increasingly unlikely” (PDG^[1]).

2. The stars destroyed some lithium

Lithium is fragile and burns at temperatures found inside stars. If material from a star’s surface slowly sinks or mixes down to hotter layers over billions of years, the surface would show less lithium than the star was born with. In 2006 Andreas Korn and colleagues found that surface abundances change with evolutionary stage among stars in the old globular cluster NGC 6397. They concluded that diffusion, which transports lithium deep into the star, is predominantly responsible for the low lithium seen at the surface of old stars (Korn et al. 2006^[6]). The PDG review agrees that diffusion, rotationally induced mixing and pre-main-sequence depletion “certainly occur,” but notes that reducing lithium to the required level “generally requires some ad hoc mechanism and fine tuning of the initial stellar parameters” (PDG^[1]).

Recent data add twists. The GALAH survey has lithium abundances for more than 100,000 stars, and warmer stars of moderate metal content form an elevated plateau consistent with the Big Bang prediction; it has been suggested that the oldest stars may once have started there too. Meanwhile, the most metal-poor stars, with [Fe/H] below −3, show “a puzzling drop” below the Spite plateau (PDG^[1]). [Fe/H] is the logarithm of a star’s iron-to-hydrogen ratio relative to the Sun, so −3 means a thousandth of the Sun’s iron.

3. New physics

The third option is that physics beyond the Standard Model changed the lithium yield. Fields discusses scenarios such as decaying supersymmetric particles and time-varying fundamental constants, with a firm condition: deuterium and helium-4 “must remain unperturbed” (Fields 2011^[4]). Any such idea has to cut lithium without spoiling the match that already works.

So is the lithium problem solved?

No. This is an open question. The PDG review states plainly: “The cosmological lithium problem remains an unresolved issue in BBN” (PDG^[1]). Stellar explanations need fine tuning, nuclear fixes are running out, and new-physics scenarios remain unconfirmed. Primordix also appears on our On Cosmic Time sticker sheet, and more of the early universe lives in our Relativity & Cosmos collection.

FAQ

What is the cosmological lithium problem?

Big Bang nucleosynthesis, with the baryon density measured by the cosmic microwave background, predicts about three times more lithium-7 than is found in the oldest stars.

Why only lithium and not helium or deuterium?

Deuterium and helium-4 predictions agree well with observations. Lithium-7 alone is off, which is one reason stellar effects are taken seriously as an explanation.

What is the Spite plateau?

The nearly constant lithium abundance seen in old, metal-poor halo dwarf stars, first reported by Monique and François Spite in 1982.

Could new physics explain it?

Possibly, and it remains one of three proposed classes of solution, but no proposal has been confirmed.

References

1. B. D. Fields, P. Molaro & S. Sarkar, “Big-Bang Nucleosynthesis,” in Particle Data Group (2024), Review of Particle Physics (revised August 2023). https://pdg.lbl.gov/2024/reviews/rpp2024-rev-bbang-nucleosynthesis.pdf
2. Planck Collaboration, N. Aghanim et al. (2020), “Planck 2018 results. VI. Cosmological parameters,” Astron. Astrophys. 641, A6 (arXiv:1807.06209). https://arxiv.org/abs/1807.06209
3. M. Spite & F. Spite (1982), “Lithium abundance at the formation of the Galaxy,” Nature 297, 483–485 (DOI 10.1038/297483a0). https://www.nature.com/articles/297483a0
4. B. D. Fields (2011), “The Primordial Lithium Problem,” Annu. Rev. Nucl. Part. Sci. 61, 47 (arXiv:1203.3551). https://arxiv.org/abs/1203.3551
5. L. Damone et al. (n_TOF Collaboration) (2018), “⁷Be(n,p)⁷Li Reaction and the Cosmological Lithium Problem,” Phys. Rev. Lett. 121, 042701 (arXiv:1803.05701). https://arxiv.org/abs/1803.05701
6. A. J. Korn et al. (2006), “A probable stellar solution to the cosmological lithium discrepancy,” Nature 442, 657 (arXiv:astro-ph/0608201). https://arxiv.org/abs/astro-ph/0608201

Written by Pixelated Physics. Every factual claim is linked to the source we checked; points that are still debated are labelled open.