Reappraisal of the limit on the variation in $\alpha$ implied by Oklo

Abstract

We reconsider the analysis of the sensitivity of neutron resonance energies $E_i$ to changes in $\alpha$ with a view to resolving uncertainties that plague earlier treatments. We point out that, with more appropriate choices of nuclear parameters, the standard estimate (due to Damour and Dyson) of the sensitivity for resonances in ${}^{150}$Sm is increased by a factor of 2.5. We go on to identify and compute excitation, Coulomb and deformation corrections. To this end, we use deformed Fermi density distributions fitted to the output of Hartree-Fock (HF) + BCS calculations (with both the SLy4 and SkM$^*$ Skyrme functionals), the energetics of the surface diffuseness of nuclei, and thermal properties of their deformation. We also invoke the eigenstate thermalization hypothesis, performing the requisite microcanonical averages with two phenomenological level densities which, via the leptodermous expansion of the level density parameter, include the effect of increased surface diffuseness. Theoretical uncertainties are assessed with the \emph{inter-model} prescription of Dobaczewski et al. [J. Phys. G: Nucl. Part. Phys. {\bf 41}, 074001 (2014)]. The corrections diminish the revised ${}^{150}$Sm sensitivity but not by more than 25\%. Subject to a weak and testable restriction on the change in $m_q/\Lambda$ (relative to the change in $\alpha$) since the time when the Oklo reactors were active ($m_q$ is the average of the $\text{u}$ and $\text{d}$ current quark masses, and $\Lambda$ is the mass scale of quantum chromodynamics), we deduce that $|\alpha_{\text{Oklo}}-\alpha_{\text{now}}|<1.1\times 10^{-8}\alpha_{\text{now}}$ (95\% confidence level). The corresponding bound on the present-day time variation of $\alpha$ is tighter than the best limit to date from atomic clock experiments.