aa r X i v : . [ h e p - e x ] O c t DPF2015-149July 10, 2018
Latest nH analysis in the Double Chooz experiment
Guang Yang on behalf of the Double Chooz collaborationArgonne National Laboratory, Lemont, Illinois 60439, USAandIllinois Institute of Technology, Chicago, Illinois 60616, USA Precise measurement of the neutrino mixing angle θ is the primarygoal of the Double Chooz Experiment (DC), which is located in Chooz,France. The inverse beta decay process provides a unique signature ofreactor anti-neutrino interactions, giving prompt signals from positronannihilation and delayed signals from neutron capture by either Gadolin-ium (Gd) or Hydrogen (H). This paper is dedicated to the latest nHanalysis in Double Chooz. Typically, The Gd analysis is primary sincefewer background events are involved. However, with accurate estimatesof backgrounds and a precise reconstruction of energy, the nH analysisgives a powerful independent measurement of θ .PRESENTED AT DPF 2015The Meeting of the American Physical SocietyDivision of Particles and FieldsAnn Arbor, Michigan, August 4–8, 2015 Guang Yang: [email protected] neutrino mixing
Neutrinos propagate in mass states while interact in flavor states in the weakinteraction. This causes the neutrino oscillation phenomenon, which is well describedby a unitary matrix introduced by Z.Maki, M.Nakagawa, S.Sakata and B.Pontecorvo(PMNS). The PMNS matrix is: U P MNS = c c s c s e − iδ CP − s c − c s s e iδ CP c c − s s s e iδ CP c s s c − c s c e iδ CP − c s − s s c e iδ CP c c , (1)where c ij =cos θ ij and s ij =sin θ ij . δ Cp is the CP-violating phase. This matrix canbe broken down into three blocks. Each of them contains one mixing angle thatcan be measured by certain types of experiments. The angle θ has been preciselymeasured by solar neutrino experiments [1, 2]. The angle θ is being measuredby beam and atmospheric neutrino experiments [3, 4] and θ is being measured byreactor antineutrino experiments like Double Chooz [5], Daya Bay [6] and RENO [7].The major remaining puzzles of neutrino oscillation are the sign of ∆ m , the CPviolation phase, the θ octant and the existence of Majorana neutrinos. Double Chooz is located in Chooz, France and it started the data taking in 2012.Two detectors are designed to take data but for the nH analysis presented here,only far detector data are used. The far detector has an average distance 1050 mfrom the reactor cores. The total live time in the data sample is 472.72 days. Thesignal Double Chooz uses is inverse beta decay (IBD). When an electron antineutrinofrom the reactors enters the detector, it interacts with a proton, then a positron and aneutron are created. The positron can be observed as a prompt signal stemming fromits ionization and annihilation with an electron, while the neutron can be observed asa delayed signal if it is captured by hydrogen or gadolinium. The survival probabilityof electron antineutrino travelling around 1 km can be approximated as: P ( ν e → ν e ) = 1 − sin (2 θ ) sin ( ∆ m L E ) , (2)where L is the baseline and E ν is the neutrino energy.The Double Chooz far detector consists of four concentric cylindrical vessels. Theoutermost volume is called the inner veto (IV). It is filled with liquid scintillator andequipped with 78 8-inch photomultiplier tubes (PMTs) so that it operates as a muon1igure 1: Left: The gain calibration. Right: The PE/MeV calibration using thehydrogen capture peak from the radioactive source Cf.veto and a shield. The second outermost volume is called the buffer and it is filledwith mineral oil. The inner detector consists of the neutrino target (NT) and gammacatcher (GC). Both of them are filled with liquid scintillator but NT is also dopedwith gadolinium, which has a large cross section to capture neutrons.In the latest nH analysis, Double Chooz improves several important things overprevious papers to provide more precise measurement, including energy reconstruc-tion, background reduction, detection systematics evaluation and the final fit strategy.These will be briefly introduced one by one in the following sections.
The energy reconstruction for DC nH analysis can be described as: E vis = N pe × f u ( ρ, z ) × f P E/MeV × f datas ( E vis , t ) × f MCnl . (3)The first factor is the gain calibration, which contains the information of transformingthe total charge to the number of photoelectrons. The left panel on fig. 1 shows thegain calibration. The second factor is the spatial uniformity calibration. By usingthe spallation neutron captures in hydrogen, we get uniformity maps for data andMC separately. The third factor is the calibration that transforms the number ofphotoelectrons to visible energy. We use hydrogen captures from radioactive source Cf for this. The right panel on fig. 1 shows this calibration. The fourth factoris the time stability calibration, which is implemented by using natural radioactivesources. The last factor accounts for the light and charge nonlinearity, arising fromthe remaining discrepancy between data and MC. It is only applied to MC, to makeit to be consistent with data. 2igure 2: Left: ANN output values for: Black histogram: on-time data, Red: Off-time data, Black points: on-time minus off-time data, Blue: signal MC. Right: Thepulse shape for a fast neutron event.
Basically, there are three kinds of backgrounds considered in Double Chooz. Theyare accidental background, fast neutron/stopping muon and Li/ He. Artificial neuralnetwork (ANN) is a powerful tool to reduce the accidental background, which is themain contamination in the nH analysis. We use the time difference between theprompt and delayed signals, distance difference and the delayed energy as inputs tothe MLP (Multi-layer Perceptron) network. ANN is trained by using backgroundsamples from data and signal samples from MC. The left panel on fig. 2 shows theANN output values for signal MC, accidental background and data.The fast neutron background can be eliminated by a new technique named MPS(multiple pulse shape). Proton recoils from the fast neutron events can cause asmall shift on the pulse shape. Removing those events will reduce the fast neutronbackground effectively. The right panel on fig. 2 shows the pulse shape of a fastneutron event as an example.The detection systematics is carefully computed by taking three factors into ac-count. First, the hydrogen fraction accounts for the fraction of events captured byhydrogen, rather than captured by gadolinium. This is calculated by using
Cf andIBD events happening at the target center and at positions in the gamma catcherwhich are far away from the target. Those two cases provide two numbers for the tar-get region and gamma catcher region separately. Second, the spill in/out uncertaintyis estimated by calculating the difference between two simulation tools, TRIPOLI-4and GEANT4. Last, the proton number uncertainty is also considered.3igure 3: Left: Observed rates vs. expected rates for the 7 reactor rate modes. Bestfit line is shown as a dashed line. Right: Observation and non-oscillation predictionratio along the visible energy. The red histogram shows the best fit curve.