How precisely do we know the antineutrino source spectrum from a nuclear reactor? Klaus Schreckenbach (TU München) Klaus Schreckenbach Gran Sasso 3.5.2011

1) Origin of reactor antineutrinos 2) Experimental determination of the source spectrum Measurements of Nß in 1981-1989 at ILL with ’BILL’ Conversion into correlated Nν 5) Discussion of the revised Nν obtained by a new conversion procedure (TH. Mueller et al, PRC in press) 6) Conclusion/outlook

Origin of the reactor antineutrinos -> antineutrinos from the beta decaying fission products 0…10 MeV 2 1017 antineutrinos/s for 1MW reactor power Research reactors with uranium highly enriched in 235U: pure spectrum from 235U fission Power reactor: composition of several fissile material Typical composition of a PWR fuel example Bugey reactor (average over measurement periode „Bugey 4“ (PL B338(1994)383): fissile isotope fraction of fission fk σ (v p -> e+ v) per fission in 10-43 cm2 fk ∙ σ 235U 54 % 6.6 58.4 % 239Pu 33 % 4.3 23.2 % 238U 7.8 % 10.1 12.9 % 241Pu 5.6 % 6.0 5.5 %

2) Experimental determination of the Nv source spectrum measurement of the integral beta spectrum in units of betas per fission and MeV for individual fissile isotopes (online measurement fissile sample in neutron flux) deduction of the correlated Nv spectrum from the measured beta spectrum Emitted antineutrinos from the reactor core according to the known composition and power of the reactor core Challenge of the procedure: filter the betas from the gamma and fission neutron background absolute calibration of the spectrum per fission in the range 2- 9 MeV intensity response function of the detection device conversion Nß -> Nv ; depends on nuclear charge Z of the fission products Details of the reactor core Early experiment on Nß +conversion: R.E. Carter et al, PR 113(1959)280: σ (v p -> e+ n) = (6.1 ±1.0) 10-43 cm2 per fission of 235U Precise experiments 1981-1989 with the BILL magnetic spectrometer at ILL and conversion into Nv

3) Measurements on beta spectra with BILL at the ILL and conversion to antineutrino spectra Exposure time to constant neutron flux more than 12 …36 h 235U at full reactor power of 57 MW PL 99B(1981)251 (235U at 8 MW) test 235U at 4 MW PL 160B(1985)325 (final on 235U) 239Pu full power PL118B(1982)162, final Nv PL 218B(1989)365 241Pu full power PL 218B(1989)365 (final) In progress: 238U with fission neutrons (fission neutron beam at FRM2, Garching, using a beta telescope) Absolute calibration Nß in betas per fission and MeV via internal conversion electron lines of known partial cross section per neutron capture:

BILL spectrometer, ILL

BILL spectrometer ILL, 1974-1990

Signal to background conditions for the two different 235U experiments at 57 MW and 4 MW reactor power at ILL

Uncertainty for the ratio from absolute calibrations: 4 % at 68% CL Statistical uncertainty <1% below 7 MeV

(50 keV bins, calibration by internal conversion electrons) 207Pb(n,γ)208Pb σ absolut 115In(n,γ)116mIn (ß decay) σ absolut 113Cd(n,γ)114Cd relativ (50 keV bins, calibration by internal conversion electrons)

4) Conversion into the correlated antineutrino spectrum ; method used 1982 -1989 with BILL data Beta spectrum in 100 keV bins (average of 50 keV points) Description of the measured beta spectra by about 30 virtual branches (allowed beta decay shape) of different endpoint energy and intensity, endpoint energy E0 dependent mean Z values and radiative correction antineutrino spectrum of the individual virtual branches by mirror in E0 = Eß + Ev Sum-up of these antineutrino branches Repetition of the procedure with somewhat different set for virtual branches Comparison of these different conversion results Smoothing over 250 keV Global correction due to coulomb and weak magnetism term (P. Vogel PRD 29(1984)1918)

Illustration of one step in the deconvolution of the experimental Nß into virtual branches (as Kurie plot) - Pi(Eo(i)) :

235U data conversion 1985 (total error in Nv 90% C.L.)

Bugey result 1995: Perfect agreement in shape and rate with the BILL derived data for no oscillation

5) Discussion of the revised Nv (as published recently: Th.Mueller, D. Lhuillier, et al, PRC in press; G. Mention et al., PRC in press) Description of the measured BILL integral beta spectrum to about 90% with real branches (tremendous work!) and 10% virtual branches to fill up the spectrum to equal the BILL result. -> better knowledge of the Z values of the involved beta decays Nuclear charge Z, radiative correction and ACW term included at the level of each branch individual branches mirrored by E0 = Eß + Ev into antineutrinos Sum up of the anti neutrino branches yields cumulated Nv Observed differences, although using the same integral beta spectra: Few percent higher Nv intensities increasing difference with energy Mean cross section per fission for v p -> e+ n higher by 2.5 % for 235 U, 3.2% for 239Pu, 3.9% for 241Pu In addition: prediction for the yet unmeasured 238U case: Almost 10% higher than former predictions! Origin of these differences?

X A_CW term A_C term is Z dependent (~Z) Corrections in % on integral spectrum ( smaller than on shape of individual branches!) 2011 1985 X Integral cross section per fission for v p -> e+ n relative to no A_CW term (multiplication with detected spectrum’): 1985 2011 range 2-4 MeV -0.5% +0,9% range 2-8 MeV +0.1% +1.3% -> only about 1.2% correction, but uncertain theory behind!

Investigation for the bugey 4 result (integral cross section per fission for v p -> e+ n): σmeasured. = 5.752 ± 1.4% 10-43 cm2 per fission at 15 m distance from reactor core Prediction for no-oscillation for bugey4 reactor core σold = 5.82 ± 2.7 σnew =6 .102 ± 2.7% difference: 4.7 % (with new n lifetime: 5.1%, 6.3%-> 2.3σ to measured) As I understand: Estimation for the claimed changes on σold for the bugey4 result: A_CW +1.2 % uncertain theory detailed Z 1.3 …1.5 % save n lifetime +0,2% (to 2010 value); 0,64% to 2011 value save 238U +1.25% from nuclear libreries off equilibrium +0.7 …1.0% save, although from libreries _________________________________________________________________________ total +4.7% +5.1%

neutron lifetime values used by PDG 2011: neutron lifetime values, without the two most recent values (on the left) 860 870 880 890 900 910 1 2 3 4 5 6 7 8 9 10 lifetime (s) ) PDG 2010 value 885.7 (0.8) s including Serebrov et al 2005 and MAMBO II 2010: PDG 2011: 881.8(1.4) s (0.44% lower) [further correction 4th value?-> 880.3(1.1) s (0.61% lower)]

New predicted source spectra Nv are reasonable 6) Conclusion Possible uncertainties in beta spectra: calibration σf,, σcalpartiell: no significant changes since the 80ties (confusion 207Pb(n,γ)?) 238U missing data for experimental integral beta spectrum Uncertainties in conversion: A_CW term, check theory! Z dependence well treated by new approach! Neutron life time in cross section integral: new 2011 value additional +0.44% (+0.61%) New predicted source spectra Nv are reasonable and the best you can do at the moment But: Anomaly is only 2.1 sigma! A_CW and 238U spectrum are not savely determined the predicted Nv is based on only one set of Nß measurements ->Spectral shape change at a close neutrino detector may be decisive Some further comments: Fission of 235U, 239Pu by faster neutrons important? ( slightly different fission product distribution) Radiative correction on antineutrinos small enough? 235U fuel, research reactors (ILL, FRM II): reactor power known well enough? (Power reactor: power better known, precise composition and geometry of fuel!?)