betaFeedback

WorksPhysics and Astronomy

First measurement of high-energy reactor antineutrinos at Daya Bay

Abstract

This Letter reports the first measurement of high-energy reactor antineutrinos at Daya Bay, with nearly 9000 inverse beta decay candidates in the prompt energy region of 8-12~MeV observed over 1958 days of data collection. A multivariate analysis is used to separate 2500 signal events from background statistically. The hypothesis of no reactor antineutrinos with neutrino energy above 10~MeV is rejected with a significance of 6.2 standard deviations. A 29% antineutrino flux deficit in the prompt energy region of 8-11~MeV is observed compared to a recent model prediction. We provide the unfolded antineutrino spectrum above 7 MeV as a data-based reference for other experiments. This result provides the first direct observation of the production of antineutrinos from several high-QβQ_β isotopes in commercial reactors.

Preactor=RRealPowerRFullPower×17.4 GWth,(1)P_{\mathrm{reactor}} = \frac{R_{\mathrm{RealPower}}}{R_{\mathrm{FullPower}}} \times17.4\ \mathrm{GW}_{\mathrm{th}}, \tag*{(1)}

where RRealPowerR_{\mathrm{RealPower}} is the predicted IBD rate using the reactor information provided by the company and RFullPowerR_{\mathrm{FullPower}} is the predicted rate assuming that all reactors were working with full power. It is found that the event rate drops by more than a factor of 20 for the 8-12 MeV energy region compared to that of 6-8 MeV energy region, suggesting a much larger background to signal ratio in the high-EpE_p region. By decomposing RR into two parts, R=RIBDPreactor+RbkgR=R_{\mathrm{IBD}}P_{\mathrm{reactor}}+R_{\mathrm{bkg}}, where RIBDR_{\mathrm{IBD}} represents the IBD reaction rate per unit of reactor power, and RbkgR_{\mathrm{bkg}} represents the background rate which is uncorrelated with PreactorP_{\mathrm{reactor}}, a strong correlation between RIBDR_{\mathrm{IBD}} and PreactorP_{\mathrm{reactor}} is observed for EpE_p in 6 to 8 MeV that is inconsistent with a background-only hypothesis at over 30σ30\sigma. By contrast, the significance of the correlation decreases to <2.5σ<2.5\sigma above 8 MeV, primarily due to the much larger background to signal ratio. This figure illustrates significant challenges in extracting the IBD events with Ep>8E_{\mathrm{p}}>8 MeV.

Figure 1. Relationship between IBD rates without background subtract in four EpE_{\mathrm{p}} bins and the weighted reactor power. The vertical error bars are statistical uncertainties and the horizontal ones are systematic uncertainties. The relationship is fitted with a linear function. Scopes which represent rates of IBD signals are also shown.

The main backgrounds with EpE_{\mathrm{p}} in 8 to 12 MeV are from muon decays, cosmogenic fast neutrons, and cosmogenic isotope decays. Other contributions, such as atmospheric neutrinos, are negligible. A probability distribution function (PDF) is constructed for each event:

F(r;Δt,z,w)=∑prpfp(Δt)hp(z)kp(w),(2)F(\boldsymbol{r};\Delta t,z,w)=\sum_{p}r_{p}f_{p}(\Delta t)h_{p}(z)k_{p}(w), \tag*{(2)}

where pp represents the event types (IBD, cosmogenic isotopes, or fast neutron), rpr_p is the ratio of the number of type-pp events over the total event number in each EpE_{\mathrm{p}} bin, and r\boldsymbol{r} represents the vector of rpr_p. f(Δt)f(\Delta t) is the expected distribution of time difference Δt\Delta t between the IBD candidate and its preceding muon events in the same detector (the vector of Δt\Delta t for each of eight muon categories described below), h(z)h(z) is the distribution of vertical vertex coordinates zz of the prompt signal, and k(w)k(w) is the event distribution of the weighted reactor power ww when the event occurred. r\boldsymbol{r} is determined by minimizing a χ2\chi^2 constructed as

χ2(r)=−2∑[log⁡F(r;Δt,z,w)]+g(ϵ),(3)\chi^2(\boldsymbol{r})=-2\sum[\log F(\boldsymbol{r};\Delta t,z,w)]+g(\boldsymbol{\epsilon}), \tag*{(3)}

where Σ\Sigma denotes the sum over all events and g(ϵ)g(\boldsymbol{\epsilon}) constrains the nuisance parameters describing fp(Δt)f_p(\Delta t), hp(z)h_p(z) and kp(w)k_p(w) utilizing information described below.

If a cosmic-ray muon stops in the detector, its energy deposit and the subsequent Michel electron could mimic an IBD event. Muons were tagged with an efficiency larger than 99.7% by the two water Cherenkov detectors surrounding ADs. However, the efficiency decreased after 2017 because of an increase of nonfunctional photomultipliers in the Cherenkov detectors, resulting in the appearance of muon decay background. The background was negligible for previous neutrino oscillation analyses. In this analysis, it was found that a z>1.7z>1.7 m cut applied on the prompt signal can reject more than 99.8% of this background, because stopping muons with less than 12 MeV energy deposits concentrated in the top of AD. This background is not considered in Eq. (2) due to the negligible contribution after the cut.

Neutrons generated by cosmic muon interactions outside ADs could penetrate into ADs and mimic IBD reactions. They are dubbed ”fast neutrons”. Because of the production mechanism, the majority of fast neutrons have downward momentum and deposit energies near the top of ADs. The zz distribution fast neutrons for 8<Ep<128<E_{\mathrm{p}}<12 MeV is inferred from an almost pure fast neutron sample with 12<Ep<2012<E_{\mathrm{p}}<20 MeV. The zz distribution of antineutrinos is uniform and determined using the IBD-enriched sample with 2<Ep<82<E_{\mathrm{p}}<8 MeV that has less than 0.1% fast neutron contamination. Rates of fast neutrons obtained using the vertex information are consistent with an alternative method based on IBD-like events coincident with muons identified using only the outer water Cherenkov detector [39].

The cosmogenic production of 9Li{}^{9}\mathrm{Li} and 8He{}^{8}\mathrm{He} (referred as 9Li{}^{9}\mathrm{Li} in later text) with subsequent β−n\beta-n decay or a coincidence of two 12B{}^{12}\mathrm{B} β\beta decays can generate a correlated pair of signals nearly identical to IBDs. Since cosmogenic isotopes are produced by muons penetrating through the whole detector, their zz vertex distributions are found to be consistent with the uniform distribution of IBDs. The Δt\Delta t distribution is described by f(Δt)=κ⋅e−κΔtf(\Delta t)=\kappa\cdot e^{-\kappa\Delta t}, where κ=Rμ\kappa=R_{\mu} for muon-uncorrelated events such as IBDs and κ=Rμ+nτ\kappa=R_{\mu}+\frac{n}{\tau} for muon-correlated events. Here, RμR_{\mu} is the muon rate, and τ\tau is the life of the isotope. Details are found in Ref [40], Fig. 26 in Ref. [39], and Fig. 2 in Ref. [41]. The factor nn is 1 for 9Li{}^{9}\mathrm{Li} and 2 for the accidental coincidence of two 12B{}^{12}\mathrm{B} β\beta decays. The different Δt\Delta t dependencies facilitate an effective determination of amounts of cosmogenic isotopes. In addition, production of unstable isotopes is usually associated with energetic particle showers attributed to muons [42], which often include spallation neutrons. Thus, muons are divided into eight categories: four EμrecE_{\mu}^{\mathrm{rec}} bins, and each bin with or without accompanying neutrons.

The cosmogenic background rates can be effectively determined from the Δt\Delta t distributions except in the near halls (EH1 and EH2) for the lowest muon-energy bin (Eμrec<1E_{\mu}^{\mathrm{rec}}<1 GeV) without accompanying neutrons. The high muon rates in these two cases result in indistinguishable Δt\Delta t distributions for cosmogenic isotope backgrounds and muon-uncorrelated events. Previous analyses in Daya Bay studied the relation between isotope and neutron productions [39, 41]: ϵhi=Yhin/Yhiall\epsilon_{hi}=Y_{hi}^{n}/Y_{hi}^{\mathrm{all}}, where YhiallY_{hi}^{\mathrm{all}} is the total isotope yield from muons in the ithi^{\mathrm{th}} EμrecE_{\mu}^{\mathrm{rec}} bin in experimental hall hh, and YhinY_{hi}^{n} is the yield with accompanying neutrons. Though EH3 is situated deeper underground than EH1 or EH2, ε\varepsilon is similar in all three halls. Thus the fitted YμnY_\mu^n for Eμrec<1E_\mu^{\mathrm{rec}} < 1 GeV can be combined with the determination of ε\varepsilon in EH3 to estimate YallY^{\mathrm{all}} in EH1 and EH2.

An independent method was used to cross-check 9Li{}^{9}\mathrm{Li} yields from low-energy muons in EH1 and EH2. A full GEANT4-based [43] simulation of muons was performed. For each muon track through Gd-doped liquid scintillator, energies from steps with large energy losses are summed to yield a ”shower energy”. The shower energy distribution ShishS_{hi}^{\mathrm{sh}} is obtained for muons in the iith EμrecE_\mu^{\mathrm{rec}} range. The relationship between the known 9Li{}^{9}\mathrm{Li} yields from data (YhiallY_{hi}^{\mathrm{all}}) and ShishS_{hi}^{\mathrm{sh}} is built by: Yhiall=∑jShish×QjY_{hi}^{\mathrm{all}} = \sum_j S_{hi}^{\mathrm{sh}} \times Q_j, where QQ represents the 9Li{}^{9}\mathrm{Li} yield from each muon in the jjth shower energy bin. When extracting the QQ based on YhiallY_{hi}^{\mathrm{all}} and ShishS_{hi}^{\mathrm{sh}}, an unfolding technique is utilized to suppress the amplified fluctuation [10]. Finally, with the determined QQ, the 9Li{}^{9}\mathrm{Li} yield from the lowest muon-energy bin is found to account for 18% (13%) of the total yield in EH1 (EH2), with 50% (55%) relative uncertainty. These predictions are consistent with the results calculated based on ε\varepsilon and are used as constraints in the pull term of Eq. (3).

After the minimization of Eq. (3), using the best-fit values of rr, the probability of being an IBD signal is calculated for each event:

PIBD=rIBDfIBD(Δt)hIBD(z)kIBD(w)F(r;Δt,z,w)(4)P_{\mathrm{IBD}} = \frac{r_{\mathrm{IBD}} f_{\mathrm{IBD}}(\Delta t) h_{\mathrm{IBD}}(z) k_{\mathrm{IBD}}(w)}{F(r;\Delta t,z,w)} \tag*{(4)}

The predictions of PIBDP_{\mathrm{IBD}} for different event types are calculated using toy Monte Carlo datasets generated from the PDFs in Eq. (2). As illustrated in Figure 2, the reactor antineutrinos can be statistically separated and the differentiating capability decreases as the prompt energy increases. The PIBDP_{\mathrm{IBD}} distributions from data and predictions are consistent within statistical uncertainty, which demonstrates good agreement between data and model.

Figure 2. (a-c) Distributions of probabilities of being an IBD signal based on the best-fit results for different event types in different EpE_{\mathrm{p}} bins ((a): 8-9, (b): 9-10, and (c): 10-11 MeV). The filled histograms show the expected distributions which are calculated using toy Monte Carlo datasets generated from the PDFs of the fitting model. (d) Ratios of the distributions between data and predictions in (a) and (b). The error bars are statistical only. The data points for EpE_{\mathrm{p}} in the 9-10 MeV bin are shifted slightly left for visual clarity.

It is found that cosmogenic isotopes are the dominant background for Ep<9.5E_{\mathrm{p}}<9.5 MeV and fast neutrons for Ep>9.5E_{\mathrm{p}}>9.5 MeV. As a sanity check, removing reactor power information from the fitter leads to a similar result because the reactors are almost always operating at full power. Removing the Δt\Delta t information or reconstructed vertex information yields consistent results with significantly larger uncertainties. Analysis with the data of only one EH also shows consistent results within 1σ1\sigma. The uncertainty of the IBD yields for Ep>8E_{\mathrm{p}}>8 MeV is dominated by systematic uncertainties from the background estimation. Uncertainty from cosmogenic isotopes takes up to ∼60%\sim60\% of the total uncertainty for EpE_{\mathrm{p}} in 8 to 9.5 MeV, while the uncertainty from fast neutrons accounts for more than 70% for Ep>10E_{\mathrm{p}}>10 MeV.

The above analysis has extracted the antineutrino yield for Ep>8E_{\mathrm{p}}>8 MeV. An unfolding technique is needed to remove the detector response and convert the EpE_{\mathrm{p}} spectrum to the EνE_\nu spectrum. This requires the EpE_{\mathrm{p}} spectrum in lower-energy regions. The prompt energy spectrum below 8 MeV has

been updated with the new 9Li{}^{9}\mathrm{Li} background estimation in this analysis. The result is consistent with those released in previous publications [44, 45] within uncertainties.

The prompt energy spectrum per fission above 6 MeV and the comparison with theoretical predictions are shown in Figure 3. An extended neutrino spectrum for Eν>10E_\nu>10 MeV was kindly provided by the authors of a recent summation model (SM2018) [20] for the comparison. The measured IBD yield is 3% larger than the SM2018 prediction for EpE_{\mathrm{p}} in 6-8 MeV (similar to the conclusion from Ref. [45]) but is 29% smaller than the prediction for EpE_{\mathrm{p}} in 8-11 MeV. Interestingly, for Ep>10E_{\mathrm{p}}>10 MeV, while a nonzero signal rate of greater than 5σ5\sigma statistical significance is predicted by the SM2018 model, the significance is only 1σ1\sigma for the data. The large deficit in the observed high-energy IBD yields is consistent with the presence of pandemonium-affected decay data for high-QβQ_\beta isotopes such as 94,96,98Rb{}^{94,96,98}\mathrm{Rb} [19, 31] in the SM2018 summation prediction, which would result in overstated feeding to low-lying states and an overprediction of high-energy antineutrinos. In this case, overprediction of high-energy antineutrinos would be accompanied by underprediction of beta-delayed neutron release by these isotopes [46, 47, 48].

Figure 3. (a) Measured prompt energy spectrum compared with the prediction from the SM2018 model. The components from different isotopes (235U{}^{235}\mathrm{U}, 238U{}^{238}\mathrm{U}, 239Pu{}^{239}\mathrm{Pu}, and 241Pu{}^{241}\mathrm{Pu}) are shown for fission fractions (0.564, 0.076, 0.304, and 0.056 respectively). (b) A enlarged plot of (a) above 8 MeV with logarithmic vertical scales. (c) Ratio of the measurement over the prediction from the SM2018 and the extrapolated Huber-Mueller (HM) model. The HM model is not valid above 8 MeV as described in the text. The error bars in the data points are the square root of the diagonal elements of the covariance matrix, including both the statistical uncertainty and systematic uncertainty. The result above 11 MeV is not shown due to the larger than 100% uncertainty.

The measured high-energy IBD yields were also compared to predictions derived from the Huber-Mueller beta conversion model [15, 16]. A polynomial extrapolation was used to obtain the predictions for Eν>8E_\nu> 8 MeV [16]. The extrapolated result is larger than the measurement by 30% or more for Ep>7.5E_{\mathrm{p}} > 7.5 MeV, and gives worse agreement with data than the SM2018 model. For this reason, we recommend not using this extrapolation in future high-energy reactor antineutrino studies.

A data-based reactor antineutrino energy spectrum is determined with an unfolding technique, similar to the earlier analysis in the low-energy region [45]. We constructed a fitter by removing the detector response based on

χ2=(P−M)Cov⁡−1(P−M)T.(5)\chi^2 = (P-M)\operatorname{Cov}^{-1}(P-M)^{T}. \tag*{(5)}

Here the MM and Cov⁡\operatorname{Cov} are the measured prompt energy spectrum per fission above 6 MeV and its covariance matrix, respectively. The prediction of the prompt energy spectrum per fission (PP) is calculated by: P=RSfitP = RS^{\mathrm{fit}}, where RR is the detector response matrix at Daya Bay, which maps the predicted antineutrino energy spectrum (SfitS^{\mathrm{fit}}) to prompt energy. Here Sifit=Siinit×ηiS_i^{\mathrm{fit}} = S_i^{\mathrm{init}} \times\eta_i, with free parameter ηi\eta_i on the initial values (SiinitS_i^{\mathrm{init}}) in iith energy bin. The starting point of the antineutrino energy spectrum in the unfolding is set at 6 MeV to ensure that resolution-induced feedup of 6–7 MeV antineutrinos into the high-energy region is properly accounted for.

The unfolded antineutrino energy spectrum is shown in Figure 4. In the unfolding process, the postfit prediction (PP) is the same as the measurement (MM), with the best-fit χ2\chi^2 value equal to 0. Mathematically, this method is equivalent to the matrix inversion method, but this procedure has the advantage of enabling a variety of statistical tests. While this method could suffer from amplified statistical fluctuations and big bin-to-bin anticorrelation in the unfolding process [45], these problems are mitigated by the large bin widths used in this analysis. Assuming no reactor antineutrinos above specific energies (10, 10.5, or 11 MeV) in SfitS^{\mathrm{fit}}, the best-fit χ2\chi^2 values (38.3, 1.6, and 0.03, respectively) are obtained. Therefore, the significance of our result in rejecting the hypothesis of no reactor antineutrinos above 10 MeV is 6.2σ6.2\sigma. Above 10.5 MeV, the ability to reject the background-only hypothesis is marginal. The prompt energy spectrum, unfolded antineutrino energy spectrum and their covariance matrices are included in Supplemental Material.

Figure 4. (a) Reactor antineutrino energy spectrum above 7 MeV and its comparison with the prediction from SM2018. The IBD cross section [27] has been removed. The fission fractions of 235U^{235}\mathrm{U}, 238U^{238}\mathrm{U}, 239Pu^{239}\mathrm{Pu}, and 241Pu^{241}\mathrm{Pu} are 0.564, 0.076, 0.304, and 0.056 respectively. The central values have larger deviation from SM2018 compared with Figure 3 due to the amplified statistical fluctuations with bin-to-bin anticorrelation in the unfolding process [45]. (b) A enlarged plot of (a) above 8.5 MeV with a logarithmic vertical scale. (c) Ratio of the unfolded antineutrino energy spectrum over the prediction from SM2018.

In summary, the Daya Bay experiment has determined the prompt IBD energy spectrum up to 11 MeV. A 29% difference in IBD rate in the prompt energy region of 8–11 MeV is found compared with a recent summation model. An antineutrino energy spectrum is then obtained from these data using an unfolding procedure that removes the detector response effects. The hypothesis of no reactor antineutrinos with energy above 10 MeV is rejected with a significance of 6.2σ6.2\sigma. For the first time, this work extends the energy region of reactor antineutrinos above 10 MeV by direct measurement. The combination of very high statistics and low cosmogenic backgrounds of Daya Bay suggests that the precision of this measurement is unlikely to be surpassed in the foreseeable future.

We thank Magali Estienne and Muriel Fallot for their generosity to extend their summation model in Ref. [20] above 10 MeV and their helpful discussions.

The Daya Bay experiment is supported in part by the Ministry of Science and Technology of China, the U.S. Department of Energy, the Chinese Academy of Sciences, the CAS Center for Excellence in Particle Physics, the National Natural Science Foundation of China, the Guangdong provincial government, the Shenzhen municipal government, the China General Nuclear Power Group, the Research Grants Council of the Hong Kong Special Administrative Region of China, the Ministry of Education in Taiwan, the U.S. National Science Foundation, the Ministry of Education, Youth, and Sports of the Czech Republic, the Charles University Research Centre UNCE, the Joint Institute of

Nuclear Research in Dubna, Russia, the National Commission of Scientific and Technological Research of Chile, We acknowledge Yellow River Engineering Consulting Co., Ltd., and China Railway 15th Bureau Group Co., Ltd., for building the underground laboratory. We are grateful for the ongoing cooperation from the China Guangdong Nuclear Power Group and China Light & Power Company.

References

References

  1. [10]H. Almazán et al. (STEREO), J. Phys. G 48, 075107 (2021), arXiv:2010.01876 [hep-ex].DOI
  2. [15]T. Mueller et al., Phys. Rev. C 83, 054615 (2011), arXiv:1101.2663 [hep-ex].DOI
  3. [16]P. Huber, Phys. Rev. C 84, 024617 (2011).
  4. [19]D. A. Dwyer and T. J. Langford, Phys. Rev. Lett. 114, 012502 (2015), arXiv:1407.1281 [nucl-ex].DOI
  5. [20]M. Estienne et al., Phys. Rev. Lett. 123, 022502 (2019), arXiv:1904.09358 [nucl-ex].DOI
  6. [27]P. Vogel and J. F. Beacom, Phys. Rev. D 60, 053003 (1999), arXiv:hep-ph/9903554.DOI
  7. [31]E. Valencia et al., Phys. Rev. C 95, 024320 (2017), arXiv:1609.06128 [nucl-ex].DOI
  8. [39]F. P. An et al. (Daya Bay), Phys. Rev. D 95, 072006 (2017), arXiv:1610.04802 [hep-ex].DOI
  9. [40]L.-j. Wen, J. Cao, K.-B. Luk, Y.-q. Ma, Y.-f. Wang, and C.-g. Yang, Nucl. Instrum. Meth. A 564, 471 (2006), arXiv:hep-ex/0604034.DOI
  10. [41]D. Adey et al. (Daya Bay), Phys. Rev. Lett. 121, 241805 (2018), arXiv:1809.02261 [hep-ex].DOI
  11. [42]S. Abe et al. (KamLAND), Phys. Rev. C 81, 025807 (2010), arXiv:0907.0066 [hep-ex].
  12. [43]S. Agostinelli et al. (GEANT4), Nucl. Instrum. Meth. A 506, 250 (2003).
  13. [44]D. Adey et al. (Daya Bay), Phys. Rev. Lett. 123, 111801 (2019), arXiv:1904.07812 [hep-ex].DOI
  14. [45]F. P. An et al. (Daya Bay), Chin. Phys. C 45, 073001 (2021), arXiv:2102.04614 [hep-ex].DOI
  15. [46]R. B. Roberts, L. R. Hafstadt, R. C. Meyer, and P. Wang, Phys. Rev. 55, 664 (1939).
  16. [47]D. Abriola, B. Singh, and I. Dillmann, “Beta-delayed neutron emission evaluation,” IAEA Report INDC(NDS)-0599 (2011), https://www-nds.iaea.org/publications/indc/indc-nds-0599.pdf.
  17. [48]M. Birch, B. Singh, D. Abriola, I. Dillmann, T. D. Johnson, E. A. McCutchan, and A. A. Sonzogni, Nucl. Data Sheets 120, 66 (2014).

Paper details

vlnc:W_mdddysyk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents