Unambiguous Oklo natural-reactor constraints on varying fundamental constants: separating $\alpha$ from hadronic parameters and controlling reactor systematics

About two billion years ago, fission chain reactions in uranium ore at Oklo (Gabon) acted as natural nuclear reactors, and the isotopic composition of their rare-earth residues still records the thermal-neutron-capture cross sections of that epoch. Because the huge 97.3-meV capture resonance of $^{149}$Sm lies near the peak of the thermal neutron spectrum, a small past change in the fine-structure constant $\alpha$ would have shifted the resonance and measurably changed the capture rate. Damour and Dyson (1996) turned this into the sharpest terrestrial bound on the variation of $\alpha$, $|\Delta\alpha/\alpha|\lesssim10^{-7}$ over $\sim2$ Gyr, while explicitly flagging what the bound does not establish: it conditions on which constant is assumed to have varied. The open problem is to make the Oklo constraint unambiguous. Concretely: (1) separate a variation of $\alpha$ from variations of the hadronic parameters $X_q=m_q/\Lambda_{\mathrm{QCD}}$ (the ratio of the average light-quark mass to the strong-interaction mass scale) and of $\Lambda_{\mathrm{QCD}}$ (that scale itself) by extracting several independent resonance shifts (samarium together with gadolinium and other rare-earth residues) from high-quality, low-contamination samples, exploiting their different dependence on proton and mass number; (2) remove the reactor systematics — operating temperature and neutron spectrum — that currently limit the analysis; (3) extend the bound from one time-averaged value to oscillatory variation during the $\sim10^{5}$-year reactor episode; and (4) tighten the rough weak-sector bound. Three decades of follow-up work improved the numbers, but every published bound still assumes which parameter varied, and the model-independent disentanglement remains undone.

Background

The Oklo reactor as a detector for varying constants. About 1.8–2 billion years ago, when the natural abundance of the fissile isotope $^{235}$U was still high enough, groundwater-moderated fission chain reactions operated for on the order of $10^{5}$ years inside uranium deposits at Oklo, Gabon. Neutrons from fission were slowed to thermal energies and captured by nuclei in the ore, so the isotopic composition of the rare-earth residues still records the effective neutron-capture cross sections of that epoch. The key nucleus is $^{149}$Sm: it has an exceptionally large neutron-capture resonance — the 97.3-meV excited state of the compound nucleus $^{150}$Sm (the short-lived system formed when $^{149}$Sm absorbs a neutron) — sitting almost on top of the peak of the thermal neutron spectrum, so the observed depletion of $^{149}$Sm in Oklo ore measures an effective cross section $\sigma^{149}$ averaged over the reactor's neutron spectrum and operating history.

If a dimensionless constant such as the fine-structure constant $\alpha$ (the electromagnetic coupling, $\alpha\approx1/137$) had a different value then, every nuclear level, including the 97.3-meV resonance, would be shifted, predominantly through the Coulomb energy of the nuclear protons. Shlyakhter (1983, posthumously eprinted) first pointed out that this makes Oklo an extraordinarily sensitive detector: comparing the inferred ancient cross sections with modern calculations constrains the resonance-energy shift $\Delta E_r$, and hence $\Delta\alpha/\alpha$, far more tightly than any laboratory experiment could over comparable timescales. Damour and Dyson (1996) redid this analysis with a large selection of published rare-earth data, a careful treatment of the poorly known reactor temperature, and isotope-shift systematics — resonance-energy differences across neighboring samarium isotopes, which calibrate the electromagnetic (Coulomb) contribution — for the sensitivity $|\alpha\, dE_r/d\alpha|\approx1$ MeV, obtaining $-0.9\times10^{-7}<\Delta\alpha/\alpha<1.2\times10^{-7}$ (95% confidence level) and a rate bound of order $10^{-17}$ yr$^{-1}$ averaged over $\sim2$ Gyr.

What the bound does not establish. The resonance energy also depends on hadronic physics: on the dimensionless light-quark-mass parameter $X_q=m_q/\Lambda_{\mathrm{QCD}}$ (the ratio of the average $u,d$ current quark mass to the mass scale $\Lambda_{\mathrm{QCD}}$ of quantum chromodynamics, the theory of the strong interaction) and on $\Lambda_{\mathrm{QCD}}$ itself. The Oklo data constrain only some combination of the shifted constants, and each published bound so far assumes which parameter varies. Damour and Dyson closed their paper by acknowledging five unresolved issues that define the follow-up program these items define: (i) the neutron temperature during the reaction episode was known only within a conservative interval ($180^{\circ}\mathrm{C}\leq T\lesssim700^{\circ}\mathrm{C}$), and narrowing it would sharpen the resonance-shift-to-$\alpha$ mapping; (ii) the bound refers to a single time-averaged past value, whereas if $\alpha(t)$ oscillated on a timescale shorter than the reactor episode, translating the isotopic data into a bound on oscillation amplitude or duty cycle was, in their words, more complicated and needed detailed study; (iii) separating an $\alpha$ variation from hadronic-parameter variations was impossible with the existing data set because chiral perturbation theory (the low-energy effective field theory of QCD) could not then reliably estimate the $m_q$-dependence of nuclear resonance energies, and would require extracting several independent resonance shifts, e.g. from both samarium and gadolinium, in low-contamination samples; (iv) their approximate bound on the weak-sector parameter $\beta=G_Fm_p^2c/\hbar^3$, with $G_F$ the Fermi constant of the weak interaction ($|\Delta\beta/\beta|<0.02$, $|\dot{\beta}/\beta|<10^{-11}$ yr$^{-1}$) was model-dependent and could be tightened only with better nuclear-model understanding of weak contributions to the relevant excited-state energies; and (v) the averaged rate depended on the assumed reactor age, taken conservatively as $1.8\times10^{9}$ yr.

Later context. Subsequent work sharpened the numbers and exposed the systematics. Fujii et al. (2000) reanalyzed contamination-controlled samples and found, besides the near-zero solution, a second, nonzero resonance-shift solution that they could only tentatively exclude using gadolinium data. Lamoreaux and Torgerson (2004) triggered a genuine conflict by claiming a $6\sigma$ nonzero $\Delta\alpha/\alpha$ from a realistic neutron spectrum; Petrov et al. (2006) and Gould, Sharapov and Lamoreaux (2006) resolved it with full Monte Carlo particle-transport modeling (the codes MCU and MCNP, validated against Oklo epithermal spectral indices — measured ratios of captures at different neutron energies that discriminate between reactor models), the latter giving $-0.24\times10^{-7}\leq\Delta\alpha/\alpha\leq0.11\times10^{-7}$ but, in their own words, model-dependent in that it assumes only $\alpha$ has varied. On the hadronic side, Dmitriev and Flambaum (2003) and Flambaum and Shuryak (2003) produced the first estimates of the quark-mass dependence, Flambaum and Wiringa (2009) derived $|\delta X_q/X_q|<4\times10^{-9}$ from the samarium resonance assuming the shift was purely hadronic, and Davis (2015) applied in-medium chiral perturbation theory (chiral perturbation theory adapted to nuclear matter at finite density) — precisely the technology Damour and Dyson lacked — finding the quark-mass term enhanced roughly tenfold relative to the $\alpha$ term. The review of Davis, Gould and Sharapov (2014) proposed a combined-sensitivity formula in which resonance shifts depend differently on proton number $Z$ and mass number $A$ through the $\alpha$ and $X_q$ channels, potentially allowing the two to be disentangled if isotopic data on a broader range of nuclei became available; they noted that no Oklo study used more than a couple of nuclei. Davis and Hamdan (2015) reappraised the $\alpha$-sensitivity (increasing the Damour–Dyson estimate by a factor 2.5, with corrections of at most 25%), obtaining $|\alpha_{\mathrm{Oklo}}-\alpha_{\mathrm{now}}|<1.1\times10^{-8}\alpha_{\mathrm{now}}$ (95% CL) — but still subject to a weak and testable restriction on the change of $m_q/\Lambda_{\mathrm{QCD}}$ relative to the change in $\alpha$. Most recently, Flambaum and Mansour (2025) recast Oklo as a bound on $\Lambda_{\mathrm{QCD}}$ variation, $|\delta\Lambda_{\mathrm{QCD}}/\Lambda_{\mathrm{QCD}}|<2\times10^{-9}$ over 1.8 Gyr, again under the minimal assumption that a single gluonic channel is the sole new interaction. The disentanglement itself is what remains open.

Problem Statement

Make the Oklo natural-reactor constraint on the variation of fundamental constants unambiguous, replacing bounds that are conditional on which parameter is assumed to have varied with identified constraints on named dimensionless constants. The scientific objective comprises the following four jointly posed components of the open problem, all originating from Damour and Dyson's own acknowledged limitations:

1. Disentangle the electromagnetic and hadronic channels. Extract several independent resonance shifts from the Oklo data — samarium together with gadolinium ($^{155}$Gd, $^{157}$Gd) and, if data permit, further rare-earth residues — from high-quality, low-contamination samples, and combine them through their different dependences on proton number $Z$ and mass number $A$ so that a variation of $\alpha$ and a variation of $X_q=m_q/\Lambda_{\mathrm{QCD}}$ (the ratio of the average light-quark mass to the strong-interaction mass scale) and of $\Lambda_{\mathrm{QCD}}$ itself are separately constrained, or the residual degeneracy is quantified, without assuming which parameter varied or imposing a unified-theory relation between their variations. This requires sufficiently reliable nuclear-structure calculations of the $m_q$-dependence of the relevant compound-resonance energies (in-medium chiral perturbation theory or equivalent), with a controlled uncertainty budget.
2. Control the reactor systematics to below the physics sensitivity. Resolve the operating-temperature ambiguity — the lutetium thermometer, which reads the $^{176}$Lu/$^{175}$Lu isotope ratio and depends on the branching ratio $B^g$ for $^{175}$Lu neutron capture into the $^{176}$Lu ground state rather than a long-lived excited (isomer) state, currently gives conflicting temperatures at the level of hundreds of degrees, with direct leverage on the samarium effective cross section — and demonstrate that the neutron-spectrum and burn-up (cumulative fissile-fuel consumption) modeling no longer dominates the error budget.
3. Extend beyond a single time-averaged value. Derive what the time-integrated Oklo isotope data actually constrain for oscillatory $\alpha(t)$ or $X_q(t)$ during the $\sim10^{5}$-year reactor episode — a bound on amplitude and duty cycle (the fraction of each period spent in the displaced phase) as a function of oscillation period — rather than only on the difference of epoch-averaged values.
4. Tighten the weak-sector bound. Improve on the model-dependent $|\Delta\beta/\beta|<0.02$ (for $\beta=G_Fm_p^2c/\hbar^3$, with $G_F$ the Fermi constant of the weak interaction) with a better nuclear-model understanding of weak contributions to the specific excited-state energies involved.

The reactor-age uncertainty (the fifth source item) is essentially settled ($\approx1.9$–$2.0\times10^{9}$ yr) and enters the averaged rate bound only linearly; it is recorded here as background rather than as an open target. The target is a constraint on named constants that is not conditional on a single assumed varying parameter, rather than a numerically tighter conditional bound.

References

1. Thibault Damour and Freeman Dyson (1996). The Oklo bound on the time variation of the fine-structure constant revisited. Nuclear Physics B 480, 37-54. DOI: 10.1016/S0550-3213(96)00467-1. https://doi.org/10.1016/S0550-3213(96)00467-1 (preprint arXiv:hep-ph/9606486, https://arxiv.org/abs/hep-ph/9606486)
2. Alexander I. Shlyakhter (1983). Direct test of the time-independence of fundamental nuclear constants using the Oklo natural reactor. ATOMKI Report A/1 (1983), posthumously eprinted as arXiv:physics/0307023. https://arxiv.org/abs/physics/0307023
3. Yasunori Fujii, Akira Iwamoto, Tokio Fukahori, Toshihiko Ohnuki, Masayuki Nakagawa, Hiroshi Hidaka, Yasuji Oura, and Peter Möller (2000). The nuclear interaction at Oklo 2 billion years ago. Nuclear Physics B 573, 377-401. DOI: 10.1016/S0550-3213(00)00038-9. https://doi.org/10.1016/S0550-3213(00)00038-9
4. Yasunori Fujii, Akira Iwamoto, Tokio Fukahori, Toshihiko Ohnuki, Masayuki Nakagawa, Hiroshi Hidaka, Yasuji Oura, and Peter Möller (2002). Nuclear data in Oklo and time-variability of fundamental coupling constants. arXiv:hep-ph/0205206. https://arxiv.org/abs/hep-ph/0205206
5. Keith A. Olive, Maxim Pospelov, Yong-Zhong Qian, Alain Coc, Michel Cassé, and Elisabeth Vangioni-Flam (2002). Constraints on the variations of the fundamental couplings. Physical Review D 66, 045022. DOI: 10.1103/PhysRevD.66.045022. https://doi.org/10.1103/PhysRevD.66.045022
6. V. F. Dmitriev and V. V. Flambaum (2003). Limits on cosmological variation of quark masses and strong interaction. Physical Review D 67, 063513. DOI: 10.1103/PhysRevD.67.063513. https://doi.org/10.1103/PhysRevD.67.063513 (preprint arXiv:astro-ph/0209409, https://arxiv.org/abs/astro-ph/0209409)
7. V. V. Flambaum and E. V. Shuryak (2003). Dependence of hadronic properties on quark masses and constraints on their cosmological variation. Physical Review D 67, 083507. DOI: 10.1103/PhysRevD.67.083507. https://doi.org/10.1103/PhysRevD.67.083507
8. Yasunori Fujii (2003). Time-variability of the fine-structure constant expected from the Oklo constraint and the QSO absorption lines. Physics Letters B 573, 39-45. DOI: 10.1016/j.physletb.2003.08.036. https://doi.org/10.1016/j.physletb.2003.08.036
9. S. K. Lamoreaux and J. R. Torgerson (2004). Neutron moderation in the Oklo natural reactor and the time variation of α. Physical Review D 69, 121701. DOI: 10.1103/PhysRevD.69.121701. https://doi.org/10.1103/PhysRevD.69.121701 (solo-authored preprint arXiv:nucl-th/0309048, https://arxiv.org/abs/nucl-th/0309048)
10. Charles L. Steinhardt (2005). Constraints on field theoretical models for variation of the fine structure constant. Physical Review D 71, 043509. DOI: 10.1103/PhysRevD.71.043509. https://doi.org/10.1103/PhysRevD.71.043509 (preprint arXiv:hep-ph/0308253, https://arxiv.org/abs/hep-ph/0308253)
11. Yu. V. Petrov, A. I. Nazarov, M. S. Onegin, V. Yu. Petrov, and E. G. Sakhnovsky (2006). Natural nuclear reactor at Oklo and variation of fundamental constants: Computation of neutronics of a fresh core. Physical Review C 74, 064610. DOI: 10.1103/PhysRevC.74.064610. https://doi.org/10.1103/PhysRevC.74.064610 (preprint arXiv:hep-ph/0506186, https://arxiv.org/abs/hep-ph/0506186)
12. C. R. Gould, E. I. Sharapov, and S. K. Lamoreaux (2006). Time variability of α from realistic models of Oklo reactors. Physical Review C 74, 024607. DOI: 10.1103/PhysRevC.74.024607. https://doi.org/10.1103/PhysRevC.74.024607 (preprint arXiv:nucl-ex/0701019, https://arxiv.org/abs/nucl-ex/0701019)
13. V. V. Flambaum and R. B. Wiringa (2009). Enhanced effect of quark mass variation in 229Th and limits from Oklo data. Physical Review C 79, 034302. DOI: 10.1103/PhysRevC.79.034302. https://doi.org/10.1103/PhysRevC.79.034302 (preprint arXiv:0807.4943, https://arxiv.org/abs/0807.4943)
14. M. S. Onegin, M. S. Yudkevich, and E. A. Gomin (2012). Investigation of the fundamental constants stability based on the reactor Oklo burn-up analysis. Modern Physics Letters A 27, 1250232. DOI: 10.1142/S021773231250232X. https://doi.org/10.1142/S021773231250232X (solo-authored preprint arXiv:1010.6299, https://arxiv.org/abs/1010.6299)
15. C. R. Gould and E. I. Sharapov (2012). 176Lu/175Lu thermometry for the Oklo natural reactors: A new examination of old data. Physical Review C 85, 024610. DOI: 10.1103/PhysRevC.85.024610. https://doi.org/10.1103/PhysRevC.85.024610 (preprint arXiv:1202.3682, https://arxiv.org/abs/1202.3682)
16. Edward D. Davis (2015). Implications of the Oklo phenomenon in a chiral approach to nuclear matter. Few-Body Systems 56, 431-437. DOI: 10.1007/s00601-014-0909-0. https://doi.org/10.1007/s00601-014-0909-0 (published online September 2014; preprint arXiv:1406.7394, https://arxiv.org/abs/1406.7394)
17. E. D. Davis, C. R. Gould, and E. I. Sharapov (2014). Oklo reactors and implications for nuclear science. International Journal of Modern Physics E 23, 1430007. DOI: 10.1142/S0218301314300070. https://doi.org/10.1142/S0218301314300070 (preprint arXiv:1404.4948, https://arxiv.org/abs/1404.4948)
18. Edward D. Davis and Leila Hamdan (2015). Reappraisal of the limit on the variation in $\alpha$ implied by the Oklo natural fission reactors. Physical Review C 92, 014319. DOI: 10.1103/PhysRevC.92.014319. https://doi.org/10.1103/PhysRevC.92.014319 (preprint arXiv:1503.06011, https://arxiv.org/abs/1503.06011)
19. Leila Hamdan and Edward D. Davis (2015). Bound on the variation in the fine structure constant implied by Oklo data. arXiv:1510.00856. https://arxiv.org/abs/1510.00856
20. V. V. Flambaum and P. Munro-Laylim (2023). Spacetime variation of the s and c quark masses. Physical Review D 107, 015004. DOI: 10.1103/PhysRevD.107.015004. https://doi.org/10.1103/PhysRevD.107.015004 (preprint arXiv:2211.03325, https://arxiv.org/abs/2211.03325)
21. E. D. Davis (2023). Oklo natural fission reactors and dynamical models of dark energy. Radiation Protection Dosimetry 199, 2288-2292. DOI: 10.1093/rpd/ncad020. https://doi.org/10.1093/rpd/ncad020
22. V. V. Flambaum and A. J. Mansour (2025). Constraints on the variation of the QCD interaction scale $\Lambda_{\text{QCD}}$. Journal of High Energy Physics 2025, 86. DOI: 10.1007/JHEP11(2025)086. https://doi.org/10.1007/JHEP11(2025)086 (preprint arXiv:2508.07266, https://arxiv.org/abs/2508.07266)