Neutron backgrounds in KamLAND Tadao Mitsui Research Center for Neutrino Science, Tohoku University (For the KamLAND collaboration) 12-14 December, 2004 Low Radioactivity Techniques 2004, Sudbury, Canada Neutron backgrounds in KamLAND e.g. 13C(a, n)16O (a from 210Po) Effects on Dm2 measurement

Neutron: serious BG for inverse b decay The oldest and the strongest technique for ne detection ne p n p d e+ g Dt ~ 200 ms DR < 1.5 m ~ Ed = 2.2 MeV (in the KamLAND scintillator) Prompt Delayed Delayed coincidence ~102~103 BG suppression Three tags: Dt, DR, and Ed seems independent, but all are neutron feature

Neutron: serious BG for inverse b decay The oldest and the strongest technique for ne detection n p d ? g Dt ~ 200 ms DR < 1.5 m ~ Ed = 2.2 MeV (in the KamLAND scintillator) Prompt Delayed If only prompt is faked  perfect delayed coincidence event e.g. fast neutron: p from np elastic scattering fakes prompt

Possible neutron sources Cosmic-ray m Fast neutrons Long-lived spallation products emitting neutrons Radioactivity Spontaneous fission (g, n) (a, n) Atmospheric n Solar n

Fast neutrons: m v.s. n CHOOZ Simple n/m flux ratio: > KamLAND > Pala Verde Very thick shield of KamLAND (see Inoue’s talk) ~50-cm water (active (Che)) 2.5-m mineral oil 1.0-m scintillatior (active to recoil proton) Kamioka CHOOZ full paper (arXiv:hep-ex/0301017) Sudbury

Fast neutrons are determined from data Fast neutron sample < 5 fast n’s in the 5.5-m fiducial (for data set of 2nd reactor result) OD 92% efficient: < 0.4 for OD muon For rock muon < 0.5 from MC (MC only for relative contribution) Total < 0.89 fast n (258 events in n sample) Scintillator balloon Fiducial volume Selection: same delayed coincidence criteria as neutrino events, but with Outer Detector hit

(a, n)

a sources: 238U series 2.5  106 decay/livetime (234Pa) KamLAND single spectrum 1.2  104 decay/livetime (214Bi214Po) 1.3  109 decay/livetime (210Bi, 210Po)

a sources: 232Th series 3.2  105 decay/livetime (212Bi212Po) KamLAND single spectrum

5.3 MeV a from 210Po ( 210Pb, T1/2=22y) S. Enomoto, in the KamLAND collab. meeting

Target: 13C is dominant (a, n) cross section  abundance in KamLAND Abundances in KL scintillator 13C & total nuclei Abundance in number 13C 0.37 % 14N 0.012 % 15N 4.6105 % 17O 2.1106 % 18O 1.1105 % Cross section from JENDL

13C(a, n)16O events · · · prompt, delayed fake “genuine” n capture (2.2-MeV g) What fakes prompt signal: 16O ground state fast n  proton recoil fast n  12C excitation 16O excited (e+e-) 16O excited (g) 16O 13C d 206Pb 210Po ~200ms g Prompt Delayed a p e+ n e-

13C(a, n)16O events · · · prompt, delayed fake “genuine” n capture (2.2-MeV g) What fakes prompt signal: 16O ground state fast n  proton recoil fast n  12C excitation 16O excited (e+e-) 16O excited (g) 16O 13C d 206Pb 210Po ~200ms 12C g Delayed a p e+ Prompt n e-

13C(a, n)16O events · · · prompt, delayed fake “genuine” n capture (2.2-MeV g) What fakes prompt signal: e+e- g 16O ground state fast n  proton recoil fast n  12C excitation 16O excited (e+e-) 16O excited (g) 13C 16O d 206Pb 210Po ~200ms g Delayed a Prompt e+ e- p e+ n e-

13C(a, n)16O events · · · prompt, delayed fake “genuine” n capture (2.2-MeV g) What fakes prompt signal: 16O ground state fast n  proton recoil fast n  12C excitation 16O excited (e+e-) 16O excited (g) 16O 13C d 206Pb 210Po ~200ms g Delayed a Prompt p e+ n e-

Estimate the number of (a, n) events in the final data set measure 210Po and 210Bi rates Number 210Po decay a propagation and (a, n) rate: dE/dx and range of a, and 13C(a, n)16O (or 16O*) cross section Neutron energy spectrum obtained n propagation (np, n12C scattering, diffusion of thermal n) Scintillation quenching for low energy p Detector resolution and off-line selection (vertex, energy) Delayed coincidence rate and prompt energy spectrum obtained numerical integral Geant4 based MC from a and g quench data measured efficiency

210Po decay rate Number 210Po decay measure 210Po and 210Bi rates Number 210Po decay a propagation and (a, n) rate: dE/dx and range of a, and 13C(a, n)16O (or 16O*) cross section Neutron energy spectrum obtained n propagation (np, n12C scattering, diffusion of thermal n) Scintillation quenching for low energy p Detector resolution and off-line selection (vertex, energy) Delayed coincidence rate and prompt energy spectrum obtained numerical integral Geant4 based MC from a and g quench data measured efficiency

210Po decay rate 210Pb  210Bi  210Po  206Pb a b 13C (a, n) 16O T1/2 = 22.3y 5.013d 138.4d stable 210Pb  210Bi  210Po  206Pb a b 5.3 MeV Kinetic energy = 1.2 MeV 13C (a, n) 16O BG in KamLAND-II (solar) see Kishimoto’s talk

210Po, 210Bi decay rate KamLAND single spectrum Ph.D thesis by I. Shimizu, RCNS Tohoku (being written)

210Po, 210Bi decay rate 210Po a 210Bi b run by run Run 3607 (2-hr low-th run) R < 550 cm R < 550 cm Evis~260 keV gaussian+ax+b NsumMax For fiducail cut: low-th (th=35) run For all volume: history run Theoretical

Results Bi, and Po agree within error Bi, R < 550 cm Bi, and Po agree within error Stable, and almost in equilibrium ~ 33 Hz 2004/happy new yr y/m/d 2002/Jul./2 Po, R < 550 cm 2004/May/2

210Po non-equilibrium Po all volume Master thesis by K. Ichimura, RCNS Tohoku (being written in Japanese)

210Po non-equilibrium Fit with 210Po life time KamLAND filling (May-Sep, 2001) Master thesis by K. Ichimura, RCNS Tohoku (being written in Japanese)

210Po non-equilibrium Fit with free life time T1/2 = 129 day (fit) (210Po = 138 day) KamLAND filling (May-Sep, 2001) Master thesis by K. Ichimura, RCNS Tohoku (being written in Japanese)

a propagation and n yield measure 210Po and 210Bi rates Number 210Po decay a propagation and (a, n) rate: dE/dx and range of a, and 13C(a, n)16O (or 16O*) cross section Neutron energy spectrum obtained n propagation (np, n12C scattering, diffusion of thermal n) Scintillation quenching for low energy p Detector resolution and off-line selection (vertex, energy) Delayed coincidence rate and prompt energy spectrum obtained numerical integral Geant4 based MC from a and g quench data measured efficiency

a propagation and n yield All sources and targets are included in actual calculation s is actually differential cross section to obtain neutron energy spectrum (see next) dE/dx table from GEANT3 range ~ 0.04 mm 5.3 MeV  s (dE/dx)1dE

S. Enomoto & K. Inoue

16O excited state JENDL gives only theoretical cross sections The absolute number of events from 16O excited state is treated as a free parameter in final oscillation analysis.

Neutron yield and energy spectra 3 to 7 MeV neutrons from ground state events For excited-state events, neutron energy is negligible (prompt energy is from g or e+e)

n propagation, detector effects measure 210Po and 210Bi rates Number 210Po decay a propagation and (a, n) rate: dE/dx and range of a, and 13C(a, n)16O (or 16O*) cross section Neutron energy spectrum obtained n propagation (np, n12C scattering, diffusion of thermal n) Scintillation quenching for low energy p Detector resolution and off-line selection (vertex, energy) Delayed coincidence rate and prompt energy spectrum obtained numerical integral Geant4 based MC from a and g quench data measured efficiency

n propagation, detector effects Genat4 based MC, cross-check by GENAT3 Birk’s quenching is included (see next) Low-energy (< 2.6 MeV) results are very preliminary (more study is needed for quenching) 4.4-MeV g from 12C excitation is clearly seen

Birks constant: quenching effect Determined from 10 data points Real Energy [MeV] a quench neutrons g, e- quench

Prompt energy spectrum (w/o resolution) with quenching (“visible energy”)

Prompt energy spectrum (with resolution) expected number of events in the data sample low-energy part is preliminary ~10 events above the analysis thr. of 2.6 MeV

With a-n

Without a-n

With a-n

Summary 13C(a, n)16O : main neutron source in KamLAND Estimation of rate and energy spectra has been done ~10 BG events from 13C(a, n)16O (total n candidates: 258 events) Effects on oscillation analysis (Dm2 measurement) is very small More study needed for low energy region below 2.6 MeV

Discussion

Birks constant: quenching effect Determined from 10 data points Real Energy [MeV] a quench neutrons g, e- quench

Monte Carlo for GoF

6-MeV b.g. (free): best-fit v.s. input Scaled no oscillation Oscillation 6-MeV b.g. (free): best-fit v.s. input Good correlation between best-fit and input 6-MeV b.g. can essentially be extracted (excluded) from the reactor spectra Neutrino decay Neutrino decoherence

6-MeV b.g. vs Reactor component Scaled no oscillation Oscillation 6-MeV b.g. vs Reactor component Horizontal axes: 6-MeV b.g. (best-fit) - (input of MC) Vertical axes: Dm2, neutrino life time etc Shows how “misfit” of 6-MeV b.g. affects analysis of reactor component Neutrino decay Neutrino decoherence

6-MeV b.g. vs Reactor component Oscillation

6-MeV b.g. vs Reactor component Oscillation -1: our previous preprint (“truth” is 7, we “fitted” it as 0, then (fit-input)/7=-1 In this case, LMA-II: disfavored, LMA-I: higher Dm2, LMA-0 favored Just as we experienced.