Possible origins and implications of the shoulder in reactor neutrino spectra

We analyze within a nuclear database framework the shoulder observed in the antineutrino spectra in current reactor experiments. We find that the ENDF/B-VII.1 database predicts that the antineutrino shoulder arises from an analogous shoulder in the aggregate fission beta spectra. In contrast, the JEFF-3.1.1 database does not predict a shoulder for two out of three of the modern reactor neutrino experiments, and the shoulder that is predicted by JEFF-3.1.1 arises from ${}^{238}\mathrm{U}$. We consider several possible origins of the shoulder, and find possible explanations. For example, there could be a problem with the measured aggregate beta spectra, or the harder neutron spectrum at a light-water power reactor could affect the distribution of beta-decaying isotopes. In addition to the fissile actinides, we find that ${}^{238}\mathrm{U}$ could also play a significant role in distorting the total antineutrino spectrum. Distinguishing these and quantifying whether there is an anomaly associated with measured reactor neutrino signals will require new short-baseline experiments, both at thermal reactors and at reactors with a sizable epithermal neutron component.

Figure 1

Approximately 5% of the fission fragments have unknown beta-decay spectra that can only be modeled. The figure shows the measured versus modeled contributions to the total aggregate antineutrino spectrum for Daya Bay, as predicted using the JEFF-3.1.1 fission yields (left panel). The same spectra folded over the neutrino detection cross section are shown in the right panel.

Figure 2

The ENDF/B-VII.1 and JEFF-3.1.1 predictions, including the beta-decay database update [19], for the ratio of the Daya Bay [31], RENO [32] and Double Chooz [33] antineutrino spectra to the Huber-Mueller or Huber-Haag models [4, 5, 22, 34], as labeled on the y-axis. In all cases, the spectra are normalized to the same number of detectable antineutrinos in the energy window $E_{\nu }=2–8\mathrm{MeV}$ ($E_{\nu }\approx E_{\text{prompt}}+0.782\mathrm{MeV}$) as the Huber-Mueller (or Huber-Haag) spectra when folded over the antineutrino detection cross section [35]. The database uncertainties shown are only for the beta-decay branches. The uncertainties arising from the fission-fragment yields are large, as is evident from the difference between the ENDF/B-VII.1 and JEFF-3.1.1 predictions. The large difference between the two database predictions for the shoulder, particularly for Daya Bay and RENO, arises entirely from a difference in the evaluated fission-fragment yields. The predicted shoulder for JEFF 3.1.1 relative to the Huber-Haag prediction for Double Chooz arises because the Haag prediction for ${}^{238}\mathrm{U}$ is appreciably smaller in the shoulder region than JEFF 3.1.1.

Figure 3

The absolute ratio of the ENDF/B-VII.1 aggregate beta spectrum for ${}^{235}\mathrm{U}$ to that of Schreckenbach [20]. The shoulder in the energy window $E_{\beta }\approx 4–6\mathrm{MeV}$ corresponds to the same shoulder in the ENDF/B-VII.1 antineutrino spectrum shown in Fig. 2.