KamLAND, a culmination of half century of reactor neutrino studies. Petr Vogel, Caltech

Selected references to this lecture: Kamland papers: K.Eguchi et al., Phys. Rev. Lett.90, (2003), K.Eguchi et al., Phys. Rev. Lett.92, (2004), T. Araki et al., hep-ex/ General review: C. Bemporad, G. Gratta, and P. Vogel., Rev. Mod. Phys. 74, 297 (2002).

Pontecorvo already in 1946 suggested to use nuclear reactors in order to perform neutrino experiments. Indeed, in Reines and Cowan showed that neutrinos are real particles using nuclear reactors as a source. Since then, reactors, powerful sources with ~6x10 20 /s electron antineutrinos emitted by a modern ~3.8 GW thermal reactor, have been used often in neutrino studies. The spectrum is well understood….

Electron antineutrinos are produced by the  decay of fission fragments

Reactor spectrum : 1) Fission yields Y(Z,A,t), essentially all known 2)  decay branching ratios b n,i (E 0 i ) for decay branch i, with endpoint E 0 i, some known but some (particularly for the very short-lived and hence high Q-value) unknown. 3)  decay shape, assumed allowed shape, known P(E,E 0 i,Z) or for electrons E e = E 0 – E  dN/dE =  n Y n (Z,A,t)  i b n,i (E 0 i ) P(E,E 0 i,Z) and a similar formula for electrons. If the electron spectrum is known, it can be `converted’ into the antineutrino spectrum.

Spectrum Uncertainties Theory only Klapdor and Metzinger, 1982 Beta calibrated Schreckenbach, 1985 Hahn, 1989 Bemporad, Gratta, and Vogel, RMP 74, 297 (2002) Results of Bugey experiment (1996)

Reactor spectra

Detecting reactor antineutrinos; low detection threshold required

Detector reaction e + p -> e + + n, positron spectrum measured Cross section known to ~0.2%, see Vogel & Beacom, Phys. Rev. D60,053003

The survival probability of electron antineutrinos of energy E produced at the distance L from the detector is P ee (E,L) = 1 – sin 2 (2  )sin 2 (  m 2 L/4E  The experiment become sensitive to oscillations if  m 2 L/E ~ 1, proof of oscillations is P ee (E,L) < 1. To study oscillations, use the disappearance test:

Probability of oscillations is proportional to sin 2 (  m 2 L/4E). Since for the reactors E~4MeV, the sensitivity to  m 2 is inversely proportional to the distance L. History of reactor neutrino oscillation search:

Discovery of oscillations of atmospheric neutrinos implies  m 2 ~ (2-3)x10 -3 eV 2, thus indicating that reactor experiments with L ~ (1-3) km should be performed (Chooz and Palo Verde). Also, the preferred `solution’ to the solar neutrino deficit implies  m 2 ~ (5-10)x10 -5 eV 2, thus indicating that reactor experiments with L ~ 100 km should be performed (KamLAND)

~180 km

~80 GW : 6% of world nuclear power ~25GW : most powerful station in the world

Ｃｏｌｌａ ｂｏｒａｔ ｏｒ KamLAND Collaboration 13 institutions & 93 members

KamLAND Experiment 180 km 300 antineutrinos from the Sun...

A brief history of KamLAND Dates Live time (days) Start data taking Jan Run A (data-set of 1 st paper) Mar 9 – Oct * Electronics upgrade & 20” PMT commissioning Jan/Feb Run B Oct - Jan Data-set presented here † Mar 9, 2002 – Jan 11, * Was with old analysis † T.Araki et al, arXiv:hep-ex/ Jun 13, 04 submitted to Phys Rev Lett submitted to Phys Rev Lett

A limited range of baselines contribute to the flux of reactor antineutrinos at Kamioka Over the data period Reported here Korean reactors 3.4±0.3% Rest of the world +JP research reactors 1.1±0.5% Japanese spent fuel 0.04±0.02%

 - Induced Neutrons & Spallation- 12 B/ 12 N 12 B 12 N  T (T-T  ),  L  L < 3m

Tagged cosmogenics can be used for calibration 12 B 12 N Fit to data shows that 12 B: 12 N ~ 100:1 τ=29.1ms Q=13.4MeV τ=15.9ms Q=17.3MeV μ

Radioactivity in liquid scintillator 238 U: 214 Bi → 214 Po → 210 Pb β+γ α τ=28.7 m τ=237 μs E=3.27 MeV E=7.69 MeV 232 Th: 212 Bi → 212 Po → 208 Pb β+γ α τ=87.4 m τ=440 ns E=2.25 MeV E=8.79 MeV

238 U: (3.5±0.5)∙ g/g needed g/g 232 Th: (5.2±0.8)∙ g/g τ=(219±29) μs Expected: 237 μs

Note: The best background in 76 Ge  decay detectors is at present ~0.2 counts/(keV kg y). Expressing the background in the liquid scintillator in KamLAND in the same units, and for energies 2-3 MeV, one finds value ~10 times smaller going out to 5.5 m radius and ~20 times smaller for 5 m radius

% Total LS mass2.1 Fiducial mass ratio4.1 Energy threshold2.1 Tagging efficiency2.1 Live time 0.07 Reactor power2.1 Fuel composition1.0 Time lag0.28 e spectra2.5 Cross section0.2 # of target protons <0.1 Total Error6.4 % Systematic Uncertainties E > 2.6 MeV 5 % : goal 4. 6

Very clean measurement Accidentalbackground Expect 1.5 n- 12 C captures Second KamLAND paper

2003 saw a substantial dip in reactor antineutrino flux

90% CL Good correlation with reactor flux Expected for no oscillations (But a horizontal line still gives a decent fit with χ 2 =5.4/4 ) Fit constrained through known background χ 2 =2.1/4 ~0.03 for 3TWhypothetical Earth core reactor

In CHOOZ it was possible to determine background by this effect.

Inconsistent with simple 1/R 2 propagation at % CL BackgroundEvents Accidentals2.69± He/ 9 Li 4.8±0.9 μ-induced n <0.89 Total7.5±1.3 Observed events 258 No osc. expected 365±24(syst) Background 7.5±1.3 Results (Observed-Background)/Expected = 0.686±0.044(stat)±0.045(syst) Caveat: this specific number does not have an absolute meaning in KamLAND, since, with oscillations, it depends on which reactors are on/off (766.3 ton·yr, ~4.7  the statistics of the first paper) ~4.7  the statistics of the first paper) Second paper

Decay chain leading to 210 Po: 222 Rn  (3.8d) 218 Po  (3.1m) 214 Pb  (27m) 214 Bi, 214 Bi  (20m) 214 Po  (164  s) 210 Pb  (22.3y) 210 Bi, 210 Bi  (5d) 210 Po  (138d) 206 Pb(stable) The long lifetime of 210 Pb causes its accumulation. The  from 210 Po decay then interact with 13 C in the scintillator by 13 C( ,n) 16 O making unwanted background. There is only ~ g of 210 Pb in fhe fiducial volume, enough however to cause 1.7x10 9  decays in 514 days.

Ratio of Measured to Expected e Flux from Reactor Neutrino Experiments LMA:  m 2 = 5.5x10 -5 eV 2 sin 2 2  = G.Fogli et al., PR D66, , (2002)

Energy spectrum now adds substantial information Best fit to oscillations: Δm 2 =8.3·10 -5 eV 2 sin 2 2θ=0.83 Straightforward χ 2 on the histo is 19.6/11 Using equal probability bins χ 2 /dof=18.3/18 (goodness of fit is 42%) A fit to a simple rescaled reactor spectrum is excluded at 99.89% CL ( χ 2 =43.4/19) Second paper

 m 2 = 6.9 x eV 2 sin 2 2  = 1.0 This result Δm 2 =8.3·10 -5 eV 2 sin 2 2θ=0.83 First KamLAND result

Combined solar ν – KamLAND 2-flavor analysis Includes (small) matter effects

KamLAND uses a range of L and it cannot assign a specific L to each event Nevertheless the ratio of detected/expected for L 0 /E (or 1/E) is an interesting quantity, as it decouples the oscillation pattern from the reactor energy spectrum no oscillation expectation Hypothetical single 180km baselineexperiment  

Conclusions Data consistent with large flux swings in 2003 Welcome to precision neutrino physics ! Together with solar ν Spectrum distortions now quite significant, shape-only very powerful Best KamLAND fit to oscillations Δm=8.3·10 -5 eV 2, sin 2 2θ=0.83 LMA2 is now excluded KamLAND reactor exposure: ton·yr (470% increase)

What’s next? Purification of liquid scintillator; remove 85 Kr and 210 Pb (low energy background) and attempt to observe solar 7 Be e (feasibility studies under way). Determine or constrain the flux of solar antineutrinos and of the geoneutrinos. Study the neutron production by muons. In a different reactor experiment (two detectors, one close and another at ~1-2 km) determine or constrain the unknown mixing angle  13. That is a whole different story. What next in reactor neutrino studies?

Future Reactor Measurements Apollonio et al., EPJ C27, 331 (2003) Chooz reached at ~1 km 2.8% statistical error 2.7% systematic error Next generation search for Theta-13 needs to achieve ~1% errors

A high sensitivity search for e from the Sun and other sources at KamLAND hep-ex/ , Phys. Rev. Lett. 92, (2004) No events found between MeV for 0.28kt-y exposure. Assuming the 8 B spectrum shape, this limits the antineutrino flux to 2.8x10 -4 of the SSM 8 B flux. This represents a factor of 30 improvement over the best previous limit.

Thanks to Atsuto Suzuki, Patrick Decowski, Gianni Fiorentini, Andreas Piepke and Giorgio Gratta who made some of the figures used in this talk.