Case Study 1: The Lithium Problem — The Oldest Unsolved Puzzle in Nuclear Astrophysics

The Discovery: A Plateau That Should Not Be There

In 1982, Monique Spite and Francois Spite, working at the Observatoire de Paris-Meudon, published a paper that would define a problem lasting more than four decades. They measured the lithium abundance in 20 metal-poor halo stars — ancient stars that formed from nearly pristine Big Bang material — and found something remarkable: the lithium abundance was essentially constant, independent of the star's temperature or metallicity.

This Spite plateau was exactly what one would expect if the lithium were primordial. Stars of different masses, ages, and compositions all showed the same lithium abundance, $A(\text{Li}) \equiv \log_{10}(\text{Li/H}) + 12 \approx 2.2$, corresponding to ${}^7\text{Li}/\text{H} \approx 1.6 \times 10^{-10}$. The constancy of the plateau strongly suggested that this was the original Big Bang value, unaltered by stellar processing.

There was just one problem: when improvements in CMB measurements pinned down the baryon-to-photon ratio $\eta$ with high precision, the standard BBN calculation predicted:

$${}^7\text{Li}/\text{H} = (5.24 \pm 0.71) \times 10^{-10}$$

This is a factor of $\sim 3.3$ higher than the observed plateau value. The discrepancy is significant — roughly 4–5 standard deviations when observational and theoretical uncertainties are combined.

The Scope of the Problem

The lithium problem is not a minor bookkeeping discrepancy. It sits at the intersection of three major branches of physics:

1. Nuclear physics: Are the reaction rates correct? The key reactions for ${}^7\text{Li}$ production during BBN are ${}^3\text{He}(\alpha,\gamma){}^7\text{Be}$ (which produces ${}^7\text{Be}$ that later decays to ${}^7\text{Li}$) and ${}^7\text{Be}(n,p){}^7\text{Li}$ (which converts ${}^7\text{Be}$ to ${}^7\text{Li}$ during BBN). The destruction reaction ${}^7\text{Li}(p,\alpha){}^4\text{He}$ depletes lithium. All of these rates have been measured in the laboratory.

2. Stellar astrophysics: Do the old halo stars preserve their primordial lithium, or do internal stellar processes deplete the surface lithium over the $\sim 13\,\text{Gyr}$ lifetime of these stars?

3. Cosmology and particle physics: Is the standard BBN framework correct, or does new physics (late-decaying particles, variation of fundamental constants, exotic energy injection) alter the lithium production during BBN?

Solving the lithium problem requires answering: which of these three pillars is wrong?

Proposed Solution 1: Nuclear Physics

The Smoking Gun That Wasn't

The most conservative explanation would be that one or more of the nuclear reaction rates used in BBN calculations is wrong. If the ${}^7\text{Be}$ production rate were overestimated, or the ${}^7\text{Be}$ destruction rate were underestimated, the predicted lithium could be brought into agreement with observations.

The ${}^3\text{He}(\alpha,\gamma){}^7\text{Be}$ rate. This is the dominant ${}^7\text{Be}$ production channel. The cross section has been measured at multiple laboratories, including LUNA (underground at Gran Sasso, to minimize cosmic-ray backgrounds), ERNA (at the Ruhr-Universitat Bochum), and NABONA. The measurements are consistent with each other and with theory (direct capture into the ground and first excited states of ${}^7\text{Be}$). The experimental uncertainty on this rate is now $\sim 3\%$ — far too small to account for the factor-of-3 discrepancy.

The ${}^7\text{Be}(n,p){}^7\text{Li}$ rate. This reaction converts ${}^7\text{Be}$ to ${}^7\text{Li}$ during BBN. A higher rate would mean more ${}^7\text{Be} \to {}^7\text{Li}$ conversion, but this does not help: the ${}^7\text{Li}$ produced is then available for destruction by ${}^7\text{Li}(p,\alpha){}^4\text{He}$. The net effect of changing this rate is small.

The $d(p,\gamma){}^3\text{He}$ rate. This affects the flow of material from deuterium to helium-3 and hence to ${}^7\text{Be}$. The LUNA measurement by Mossa et al. (2020) at BBN energies is precise to $\sim 3\%$ and has not resolved the problem.

Exotic reactions. Several groups have searched for "missing" reactions that could destroy ${}^7\text{Be}$ during BBN. The most discussed candidates:

• ${}^7\text{Be}(d,p)2\alpha$: This reaction could destroy ${}^7\text{Be}$ if its cross section at BBN energies were large enough. However, measurements by Angulo et al. (2005) and Rijal et al. (2019) show that the cross section is far too small to affect BBN predictions.

• ${}^7\text{Be}(d,\alpha){}^5\text{Li}$: Also measured; insufficient.

• ${}^7\text{Be}({}^3\text{He},p){}^9\text{B}$: Negligible.

Verdict: No plausible change in nuclear reaction rates within experimental uncertainties can resolve the lithium problem. This is perhaps the most important negative result in the entire literature. The nuclear physics is well established.

Proposed Solution 2: Stellar Depletion

Can Stars Destroy Their Surface Lithium?

Lithium is fragile. The ${}^7\text{Li}(p,\alpha){}^4\text{He}$ reaction destroys lithium at temperatures above $T \sim 2.5 \times 10^6\,\text{K}$ — well below the central temperature of any star, but above the temperature at the base of the convective envelope for many stars. If material from the stellar surface is mixed down to hot layers, lithium can be destroyed.

The convective depletion problem. In standard stellar models of warm ($T_{\text{eff}} \sim 6000\,\text{K}$) metal-poor halo stars, the base of the convective zone is at $T \sim 2 \times 10^6\,\text{K}$ — just below the lithium-burning temperature. These models predict negligible lithium depletion. This is consistent with the existence of the Spite plateau (lithium is preserved), but not with its value (it should be at the BBN level, which is $3\times$ higher).

Non-standard mixing mechanisms. Several mechanisms have been proposed to produce "just right" lithium depletion — enough to reduce the surface lithium by a factor of 3, but not enough to destroy the plateau:

1. Atomic diffusion. Gravitational settling and radiative levitation cause helium and heavy elements (including lithium) to sink below the convection zone over gigayear timescales. Richard et al. (2005) showed that models including atomic diffusion (but no turbulent mixing) predict too much depletion and destroy the plateau.

2. Atomic diffusion + turbulent mixing. Adding a mild turbulent mixing below the convection zone (parameterized by a single free parameter) can balance the gravitational settling and produce a plateau with a depleted value. Korn et al. (2006) found evidence for this in the globular cluster NGC 6397, where iron (also affected by diffusion) shows a trend with evolutionary state consistent with diffusion models. The required depletion factor is $\sim 0.2$–$0.3\,\text{dex}$, consistent with reducing the primordial lithium by a factor of 2–3.

3. Rotationally-induced mixing. Pinsonneault et al. (2002) showed that rotational mixing can deplete lithium, but the amount of depletion depends on the (unknown) initial rotational velocity distribution. Getting a uniform factor-of-3 depletion across all stars requires fine-tuning.

4. Internal gravity waves. Mixing driven by internal gravity waves excited at the base of the convective zone could transport lithium to hot regions. The efficiency is uncertain.

The ${}^6\text{Li}$ test. If stellar depletion is responsible, it should affect both ${}^7\text{Li}$ and ${}^6\text{Li}$ (with ${}^6\text{Li}$ depleted more efficiently because it burns at lower temperatures). Some observations have claimed detections of ${}^6\text{Li}$ in metal-poor halo stars, which would be very difficult to explain if significant depletion has occurred. However, subsequent analyses (Lind et al. 2013) showed that the ${}^6\text{Li}$ "detections" were likely artifacts of inadequate treatment of stellar atmospheres (NLTE, 3D convection effects). The ${}^6\text{Li}$ test remains inconclusive.

Verdict: Stellar depletion is currently the leading "conventional" explanation, but it requires non-standard physics (turbulent mixing or diffusion at just the right level) and is difficult to test unambiguously.

Proposed Solution 3: New Physics

Beyond the Standard Model

If neither nuclear physics nor stellar astrophysics can explain the lithium problem, perhaps the standard BBN framework itself needs modification.

Late-decaying particles. A hypothetical massive particle ($m \sim 100\,\text{GeV}$, lifetime $\tau \sim 10^3$–$10^4\,\text{s}$) that decays during or just after BBN could inject energetic photons or hadrons into the plasma. These could photodisintegrate ${}^7\text{Be}$ (via $\gamma + {}^7\text{Be} \to {}^3\text{He} + {}^4\text{He}$) or trigger neutron-producing reactions that enhance ${}^7\text{Be}(n,p){}^7\text{Li}$ followed by ${}^7\text{Li}(p,\alpha){}^4\text{He}$. Jedamzik (2004, 2006) showed that such scenarios can reduce lithium while leaving deuterium and helium unaffected. The required particle properties are not wildly implausible in supersymmetric extensions of the Standard Model (e.g., a gravitino or a long-lived NLSP).

However, these models are tightly constrained by the requirement of not spoiling the D and ${}^4\text{He}$ predictions. The allowed parameter space is narrow.

Variation of fundamental constants. If the nuclear reaction rates were different in the early universe (due to a time-varying fine-structure constant or a different QCD scale), the lithium production could be altered. Coc et al. (2012) explored this and found that changes in the binding energies at the $\sim 1\%$ level could potentially resolve the problem, but such variations are tightly constrained by other observations.

Dark matter-baryon interactions. Non-standard interactions between dark matter and baryons during BBN could modify the expansion rate or the nuclear reaction rates. Constraints are model-dependent.

Verdict: New physics solutions are possible but require specific (and often fine-tuned) modifications to the standard framework. No compelling candidate has emerged.

The Current State of the Art (2025)

After more than 40 years, the lithium problem remains unsolved. The key developments in the past decade include:

1. LUNA measurements have eliminated remaining nuclear physics uncertainties for the key reactions. The problem is not in the cross sections.

2. 3D NLTE stellar atmosphere models (Lind et al. 2013; Wang et al. 2021) have shown that the observed Spite plateau may be slightly modified when non-LTE effects and 3D convection are properly accounted for, but the factor-of-3 discrepancy persists.

3. New observational campaigns (Sbordone et al. 2010; Bonifacio et al. 2015; Aguado et al. 2019) have confirmed the Spite plateau with larger samples and have extended it to lower metallicities ($[\text{Fe/H}] < -4$). Some extremely metal-poor stars show lithium below the plateau ("meltdown" below $[\text{Fe/H}] \sim -3$), suggesting that depletion does occur in at least some stars.

4. Globular cluster studies (Korn et al. 2006; Nordlander et al. 2012; Gruyters et al. 2016) provide evidence for atomic diffusion at the $\sim 0.1\,\text{dex}$ level, but whether this can fully account for the factor-of-3 discrepancy remains debated.

5. Beryllium observations (Boesgaard et al. 2011) show that ${}^9\text{Be}$ (produced by cosmic-ray spallation, not BBN) is not depleted in Spite-plateau stars, suggesting that any mixing mechanism cannot extend to the ${}^9\text{Be}$-burning temperature ($\sim 3.5 \times 10^6\,\text{K}$).

A Hierarchy of Likelihoods

Most experts currently assign the following rough probabilities:

These numbers are subjective and reflect the author's reading of the literature. The important point is that no single explanation is universally accepted.

The Observational Challenge

Measuring lithium in old, metal-poor stars is far from straightforward. The Li I 6707.8 A resonance line is weak in these stars, with equivalent widths of only $\sim 20$–$40\,\text{mA}$. Extracting an accurate abundance requires:

1. High-resolution spectroscopy. Resolving powers $R = \lambda/\Delta\lambda > 40{,}000$ are needed to measure the line profile accurately. Modern echelle spectrographs on 8-10m telescopes (Keck/HIRES, VLT/UVES, Subaru/HDS) routinely achieve this.

2. Accurate effective temperatures. The lithium abundance derived from the line strength depends sensitively on the adopted stellar effective temperature ($T_{\text{eff}}$). A systematic error of $+100\,\text{K}$ in $T_{\text{eff}}$ changes the derived $A(\text{Li})$ by $\sim +0.07\,\text{dex}$. Different temperature scales (photometric vs. spectroscopic vs. infrared flux method) sometimes disagree at this level.

3. 3D and NLTE effects. The standard assumption of one-dimensional, local thermodynamic equilibrium (1D LTE) model atmospheres introduces systematic errors. Three-dimensional hydrodynamic atmosphere models with non-LTE line formation (3D NLTE) typically give higher lithium abundances by $\sim 0.05$–$0.1\,\text{dex}$ compared to 1D LTE analyses. This goes in the right direction but is far too small to resolve the factor-of-3 discrepancy.

4. Sample selection. The Spite plateau is defined by warm ($T_{\text{eff}} > 5900\,\text{K}$), metal-poor ($[\text{Fe/H}] < -1.5$), unevolved (subgiant or turn-off) stars. Cooler stars show lithium depletion from deep convective envelopes. Giant stars have destroyed their surface lithium. The plateau sample must be carefully selected to avoid these effects.

Despite these challenges, multiple independent groups using different instruments, temperature scales, and analysis methods consistently find the same plateau value: $A(\text{Li}) \approx 2.2 \pm 0.1$. The robustness of this result across independent analyses strengthens the case that the discrepancy is real.

Why It Matters

The lithium problem is not just a numbers game. It sits at a critical juncture:

• If stellar depletion is the answer, we learn something important about mixing in old, metal-poor stellar interiors — something that has implications for the age dating of the oldest stars, the chemical evolution of the Galaxy, and the interpretation of stellar abundances in general.

• If nuclear physics is the answer (a new reaction, a previously unrecognized resonance), it would be a discovery in fundamental nuclear physics — a process operating at BBN temperatures that we have somehow missed despite decades of measurements.

• If new physics is the answer, it would be a window into beyond-Standard-Model particles or interactions that operate on timescales of $\sim 1000\,\text{s}$ after the Big Bang — a regime that is almost impossible to probe by any other means.

The lithium problem is thus one of those rare scientific puzzles where every possible resolution teaches us something profound. It is the oldest unsolved problem in nuclear astrophysics, and it awaits the next generation of observations, experiments, and theoretical insights.

Discussion Questions

1. The BBN predictions for D/H and $Y_p$ agree beautifully with observations, but the ${}^7\text{Li}$ prediction does not. Does the success of D and ${}^4\text{He}$ make the lithium discrepancy more or less puzzling? Explain your reasoning.

2. If you had unlimited resources, what experiment or observation would you propose to resolve the lithium problem?

3. The factor-of-3 discrepancy corresponds to $\sim 0.5\,\text{dex}$ (a half-order of magnitude). In astrophysics, discrepancies of this size are sometimes dismissed as "within the uncertainties." Why is the lithium problem taken more seriously?

4. The Spite plateau's remarkable uniformity across diverse stellar types is both the strongest evidence for its primordial origin and the biggest challenge for stellar depletion models. Explain this tension.