1. Using Reactor Neutrinos to Study Neutrino Oscillations Jonathan Link Columbia University Heavy Quarks and Leptons 2004 Heavy Quarks and Leptons 2004 June 1, 2004 June 1, 2004

2. Doing Physics With Reactors Neutrinos The original Neutrino discovery experiment, by Reines and Cowan, used reactor neutrinos… Reines and Cowan at the Savannah River Reactor …actually anti-neutrinos. The ν e interacts with a free proton via inverse β-decay: νeνe e+e+ p n W Later the neutron captures giving a coincidence signal. Reines and Cowan used cadmium to capture the neutrons. The first successful neutrino detector

3. Minimum energy for the primary signal is 1.022 MeV from e + e − annihilation at process threshold. Minimum energy for the primary signal is 1.022 MeV from e + e − annihilation at process threshold. Two part coincidence signal is crucial for background reduction. Two part coincidence signal is crucial for background reduction. Nuclear Reactors as a Neutrino Source Arbitrary Flux Cross Section Observable Spectrum The observable e spectrum is the product of the flux and the cross section. The observable e spectrum is the product of the flux and the cross section. The spectrum peaks around ~3.6 MeV. The spectrum peaks around ~3.6 MeV. Visible “positron” energy implies ν energy Visible “positron” energy implies ν energy From Bemporad, Gratta and Vogel Nuclear reactors are a very intense sources of ν e deriving from the  -decay of the neutron-rich fission fragments. Nuclear reactors are a very intense sources of ν e deriving from the  -decay of the neutron-rich fission fragments. A typical commercial reactor, with 3 GW thermal power, produces 6×10 20 e /s A typical commercial reactor, with 3 GW thermal power, produces 6×10 20 e /s E ν = E e + 0.8 MeV ( =m n  m p +m e  1.022)

4. Uses of Reactor Neutrinos Measure cross sections or observe new processes (e.g. neutral current nuclear coherent scattering) Measure cross sections or observe new processes (e.g. neutral current nuclear coherent scattering) Search for anomalous neutrino electric dipole moment Search for anomalous neutrino electric dipole moment Measure the weak mixing angle, sin 2 θ W Measure the weak mixing angle, sin 2 θ W Monitor reactor core (non-proliferation application) Monitor reactor core (non-proliferation application) Measure neutrino oscillation parameters (Δm 2 ’s and mixing angles) Measure neutrino oscillation parameters (Δm 2 ’s and mixing angles)

5. Observations of Neutrino Oscillations Δm 2 ≈3 to 3×10 -2 eV 2 θ ?? Δm 2 ≈2.5×10 -3 eV 2 θ 23 & θ 13 Δm 2 ≈7.5×10 -5 eV 2 θ 12 Short Baseline: 10 to 100 meters Bugey, Gosgen & Krasnoyarsk Medium Baseline: 1 to 2 km Chooz, Palo Verde & future Long Baseline: 100+ km KamLAND Reactor neutrinos can probe oscillations in all three observed Δm 2 regions. Oscillations are observed as a deficit of ν e with respect to expectation.

6. Short Baseline Oscillation Searches Reactor Excluded Experiments like Bugey rule out the low Δm 2, large mixing angle region of the LSND signal. Bugey looked for evidence of oscillations between 15 & 45 meters. Gosgen was closer and Krasnoyarsk farther away. Neutrons capture on Lithium The Bugey Detector

7. Observations of Neutrino Oscillations Δm 2 ≈3 to 3×10 -2 eV 2 θ ?? Δm 2 ≈2.5×10 -3 eV 2 θ 23 & θ 13 Δm 2 ≈7.5×10 -5 eV 2 θ 12 Short Baseline: 10 to 100 meters Bugey, Gosgen & Krasnoyarsk Medium Baseline: 1 to 2 km Chooz, Palo Verde & future Long Baseline: 100+ km KamLAND Reactor neutrinos can probe oscillations in all three observed Δm 2 regions. Oscillations are observed as a deficit of ν e with respect to expectation.

8. Medium Baseline Oscillation Searches Chooz Nuclear Reactors, France Experiments like Chooz looked for oscillations in the atmospheric Δm 2. At the time they ran the atmospheric parameters were determined by Kamiokande, not Super-K (larger Δm 2 ). Gadolinium loaded liquid scintillating target. 1050 m baseline

9. sin 2 2  13 < 0.18 at 90% CL (at  m 2 =2.0×10 -3 ) sin 2 2  13 < 0.18 at 90% CL (at  m 2 =2.0×10 -3 ) Future experiments should try to improve on these limits by at least an order of magnitude. Future experiments should try to improve on these limits by at least an order of magnitude. Down to sin 2 2  13 0.01 In other words, a measurement to better than 1% is needed! < ~ No evidence found for e oscillation. No evidence found for e oscillation. This null result eliminated  → e as the primary mechanism for the Super-K atmospheric deficit. This null result eliminated  → e as the primary mechanism for the Super-K atmospheric deficit. Medium Baseline Oscillation Searches

10. Observations of Neutrino Oscillations Δm 2 ≈3 to 3×10 -2 eV 2 θ ?? Δm 2 ≈2.5×10 -3 eV 2 θ 23 & θ 13 Δm 2 ≈7.5×10 -5 eV 2 θ 12 Short Baseline: 10 to 100 meters Bugey, Gosgen & Krasnoyarsk Medium Baseline: 1 to 2 km Chooz, Palo Verde & future Long Baseline: 100+ km KamLAND Reactor neutrinos can probe oscillations in all three observed Δm 2 regions. Oscillations are observed as a deficit of ν e with respect to expectation.

11. Long Baseline Oscillations The KamLAND experiment uses neutrinos from 69 reactors to measure the solar mixing angle (θ 12 ) at an average baseline of 180 km. Neutron Capture on Hydrogen results in a 2.2 MeV gamma Scatter plot of energies for the prompt and delayed signals In 145 days of running they saw 54 events where 86.8±5.6 events where expected. The fit energy confirms the oscillation hypothesis.

12. Long Baseline Oscillations Eliminates all but the large mixing angle (LMA) solution. The best fit sin 2 2θ 12 = 0.91 The best fit Δm 2 = 6.9×10 -5 eV 2 KamLAND Results The Δm 2 sensitivity comes primarily from the solar measurements

13. Future Experiments to Search for a Non-zero Value of sin 2 2θ 13 Subject of a lot of interest because of it relevance to lepton CP violation and neutrino mass hierarchy. See Whitepaper: hep-ex/0402041

14. νeνe νeνe νeνe νeνe νeνe νeνe Distance Probability ν e 1.0 E ν ≤ 8 MeV 1200 to 1800 meters Sin 2 2θ 13 Reactor Experiment Basics Unoscillated flux observed here Well understood, isotropic source of electron anti-neutrinos Oscillations observed as a deficit of ν e sin 2 2θ 13 Survival Probability

15. Proposed Sites Around the World SitePower (GW thermal ) Baseline Near/Far (m) Shielding Near/Far (mwe) Sensitivity 90% CL Krasnoyarsk, Russia 1.6115/1000600/6000.03 Kashiwazaki, Japan 24300/1300150/2500.02 Double Chooz, France 8.9150/105030/3000.03 Diablo Canyon, CA 6.7400/170050/7000.01 Angra, Brazil 5.9500/135050/5000.02 Braidwood, IL 7.2200/1700450/4500.01 Daya Bay, China 11.5250/2100250/11000.01 Krasnoyarsk, Russia 1.6115/1000600/6000.03 Kashiwazaki, Japan 24300/1300150/2500.02 Double Chooz, France 8.9150/105030/3000.03 Diablo Canyon, CA 6.7400/170050/7000.01 Angra, Brazil 5.9500/135050/5000.02 Braidwood, IL 7.2200/1700450/4500.01 Daya Bay, China 11.5250/2100250/11000.01Status

16. The combination of these two plus a complex analysis gives you the anti-neutrino flux What is the Right Way to Design the Experiment? Start with the dominate systematic errors from previous experiments and work backwards… CHOOZ Systematic Errors, Normalization Near Detector CHOOZ Background Error BG rate 0.9% Statistics may also be a limiting factor in the sensitivity. Identical Near and Far Detectors Movable Detectors, Source Calibrations, etc. Muon Veto and Neutron Shield (MVNS)

17. Backgrounds There are two types of background… 1. Uncorrelated − Two random events that occur close together in space and time and mimic the parts of the coincidence. This BG rate can be estimated by measuring the singles rates, or by switching the order of the coincidence events. This BG rate can be estimated by measuring the singles rates, or by switching the order of the coincidence events. 2. Correlated − One event that mimics both parts of the coincidence signal. These may be caused fast neutrons (from cosmic  ’s) that strike a proton in the scintillator. The recoiling proton mimics the e + and the neutron captures. These may be caused fast neutrons (from cosmic  ’s) that strike a proton in the scintillator. The recoiling proton mimics the e + and the neutron captures. Or they may be cause by muon produced isotopes like 9 Li and 8 He which sometimes decay to β+n. Estimating the correlated rate is much more difficult!

18. Veto Detectors p n   n 1.Go as deep at you can (300 mwe → 0.2 BG/ton/day at CHOOZ) Reducing Background 2.Veto  ’s and shield neutrons (Big effective depth) 3.Measure the recoil proton energy and extrapolate into the signal region. (Understand the BG that gets through and subtract it) 6 meters Shielding

19. Isotope Production by Muons Source 300 mwe (/ton/day) 450 mwe (/ton/day) Comments: 9 Li+ 8 He 0.17 ± 0.03 0.075 ± 0.014 E ≤ 13.6 MeV, τ ½ = 0.12 to 0.18 s 16% to 50% correlated β+n 8 Li 0.28 ± 0.11 0.13 ±0.05 E ≤ 16 MeV, τ ½ = 0.84 s 6 He 1.1 ± 0.2 0.50 ± 0.07 E ≤ 3.5 MeV, τ ½ = 0.81 s 11 C 63 ± 9 28 ± 4 E ≤ 0.96 MeV, τ ½ = 20 m 10 C 8.0 ± 1.5 3.6 ± 0.6 E ≤ 1.98 MeV, τ ½ = 19 s 9C9C9C9C 0.34 ± 0.11 0.15 ± 0.05 E ≤ 16 MeV, τ ½ = 0.13 s 8B8B8B8B 0.50 ± 0.12 0.22 ± 0.05 E ≤ 13.7 MeV, τ ½ = 0.77 s 7 Be 16.0 ± 2.6 7.2 ± 1.1 E ≤ 478 keV, τ ½ = 0.53 d 12 B 113 E ≤ 13.4 MeV, τ ½ = 0.02 s A ½ second veto after every muon that deposits more that 2 GeV in the detector should eliminate 70 to 80% of all correlated decays. The vetoed sample can be used to make a background subtraction of in a fit to the energy spectrum.

20. Movable Detector Scenario The far detector spends about 10% of the run at the near site where the relative normalization of the two detectors is measured head-to-head. Build in all the calibration tools needed for a fixed detector system and verify them against the head-to-head calibration. 1500 to 1800 meters

21. Reactor Sensitivity Sensitivity to sin 2 2θ 13 ≤ 0.01 at 90% CL is achievable. Sensitivity to sin 2 2θ 13 ≤ 0.01 at 90% CL is achievable. Combining with off-axis some of the CP phase, δ, range can be ruled out. Combining with off-axis some of the CP phase, δ, range can be ruled out. Unexpected results are possible & might break the standard model. Unexpected results are possible & might break the standard model.

22. Is the mixing angle θ 23 is not exactly 45º then sin 2 θ 23 has a two-fold degeneracy. Combining reactor results with off-axis breaks this degeneracy. With the 0.03 precision of the Double Chooz experiment the degeneracy is not broken. Reactor Sensitivity

23. 2003 2004200520062007200820092010 Run 2011 Aggressive Experiment Timeline Construction Years 1 year 2 years 2 years 3 years (initially) Site Selection: Currently underway. The early work on a proposal is currently underway. With movable detectors, the detectors are constructed in parallel with the civil construction Run Phase: Initially planned as a three year run. Results or events may motivate a longer run. Site SelectionProposal

24. Conclusions and Prospects Reactor neutrinos are relevant to oscillations in all observed Δm 2 regions. Reactor neutrinos are relevant to oscillations in all observed Δm 2 regions. The KamLAND experiment has been crucial to resolving the oscillation parameters in the solar Δm 2 region. The KamLAND experiment has been crucial to resolving the oscillation parameters in the solar Δm 2 region. There are many ideas for reactor θ 13 experiments around the world and it is likely that more than one will go forward. There are many ideas for reactor θ 13 experiments around the world and it is likely that more than one will go forward. Controlling the systematic errors is the key to making this measurement. Controlling the systematic errors is the key to making this measurement. With a 3+ year run, the sensitivity in sin 2 2  13 should reach 0.01 (90% CL) at  m 2 = 2.0×10 -3. With a 3+ year run, the sensitivity in sin 2 2  13 should reach 0.01 (90% CL) at  m 2 = 2.0×10 -3. Reactor sensitivities are similar off-axis and the two methods are complementary. Reactor sensitivities are similar off-axis and the two methods are complementary. The physics of reactor neutrinos is interesting and important. The physics of reactor neutrinos is interesting and important.

25. Question Slides

26. With Gd Without Gd With Gd Without Gd Why Use Gadolinium? Gd has a huge neutron capture cross section. So you get faster capture times and smaller spatial separation. (Helps to reduce random coincidence backgrounds) Also the 8 MeV capture energy (compared to 2.2 MeV on H) is distinct from primary interaction energy. ~30 μs ~200 μs

27. From CHOOZ  interactions ? Characterizing BG with Vetoed Events Matching distributions from vetoed events outside the signal region to the non-veto events will provide an estimate of correlated backgrounds that evade the veto. Proton recoils Other Useful Distributions: Spatial separation prompt and delayed events Faster neutrons go farther Radial distribution of events BGs accumulate on the outside of the detector.

28. Medium Baseline Oscillation Searches Homogeneous detector Homogeneous detector 5 ton, Gd loaded, scintillating target 5 ton, Gd loaded, scintillating target 300 meters water equiv. shielding 300 meters water equiv. shielding 2 reactors: 8.9 GW thermal 2 reactors: 8.9 GW thermal Baselines 1115 m and 998 m Baselines 1115 m and 998 m Used new reactors → reactor off data for background measurement Used new reactors → reactor off data for background measurement Chooz Nuclear Reactors, France

29. Palo Verde 32 mwe shielding (Shallow!) Segmented detector: Better at handling the cosmic rate of a shallow site 12 ton, Gd loaded, scintillating target 3 reactors: 11.6 GW thermal Baselines 890 m and 750 m No full reactor off running Palo Verde Generating Station, AZ

30. Exelon has agreed to work with us to determine the feasibility of using their reactors to perform the experiment. “We are excited about the possibility of participating in a scientific endeavor of this nature” “At this time we see no insurmountable problems that would preclude going forward with this project.” They have given us reams of geological data which we are currently digesting.

31. Quantitative Analysis of Movable vs. Fixed Detectors Both the Kashiwazaki and Krasnoyarsk proposals assume that they can get the relative normalization systematic down to 0.8% with fixed detectors. Even if you halve the relative normalization, fixed detector are not as sensitive for a two year (or longer) run. All fixed detector scenario quickly become systematics limited. Double Chooz believes that 0.6% is achievable.

32. At the preferred  m 2 the optimal region is quite wide. In a configuration with a tunnel connecting the two detector sites, one should choose a far baseline that gives the shortest tunnel (1200 to 1400 meters). One must consider both the location of the oscillation maximum (~2200 m at Δm 2 =2×10 -3 ) and statistics loss due to 1/r 2 flux. Optimal Far Baseline Kinimatic Phase ≡ 1.27Δm 2 L/E for E=3.6 MeV

33. Sensitivity Scaling with Systematic Error For a rate only analysis The optimal baseline is very sensitive to the level of systematic error. The standard assumption of 0.8% relative efficiency error for fixed detectors is ~250% of the statistical error after 3 years at Braidwood.

34. Comparison of Shape & Rate Systematics Limited Systematic Error = 0%Systematic Error = 200% Systematic Error = 600% The optimal Baseline for the systematics limited shape analysis is ~40º. The optimal baseline for the systematics limited counting experiment is at the least optimal spot for a shape analysis! You better know what regime your working in. Statistics Limited

35. Sensitivity wrt Near Baseline Ultimately the location of the near detector will be determined by the reactor owners. The main question here is what can we live with? There is a 1/r 2 dependence in statistics (a small effect) and increasing oscillation probability with distance. Sensitivity degrades with increasing near baseline. When L near =L far the sensitivity is about the same as CHOOZ.