Design, construction, and initial performance of SciBar detector in K2K experiment Shimpei Yamamoto for the K2K-SciBar group Abstract-- A new near detector in K2K long baseline neutrino experiment, the SciBar, was constructed and started data taking to study neutrino interactions. In K2K, neutrino oscillation is studied by comparing the number of neutrino interactions and energy spectrum between near and far detectors. In order to study neutrino oscillation more precisely, it is necessary to improve the measurement of neutrino spectrum and interaction below 1 GeV, where the latest K2K results suggest maximum oscillation. For that purpose, SciBar is designed to be fully active with fine segmentation. We present the design and basic performance. All detector components have been working as expected. Also presented are the measurements of charged current interactions which are used in the latest K2K oscillation analysis. I. INTRODUCTION A novel neutrino detector, SciBar, was constructed in summer 23 to upgrade the near detectors in KEK-to- Kamioka long baseline neutrino experiment (K2K). In the K2K experiment almost pure muon neutrino beam, with average energy of 1.3 GeV, is produced with KEK 12GeV proton synchrotron toward the direction of Super-Kamiokande (SK) [1] 25km away. By comparing the number of interactions and energy spectrum between near detectors located at KEK and SK, we study neutrino oscillations precisely. A main motivation of the new detector installation is to improve the measurement of neutrino energy spectrum by using Charged Current Quasi-Elastic interaction (CCQE, ν µ +n µ - +p). The latest K2K results indicate existence of neutrino oscillation [2]. The oscillation maximum is expected to be about.6 GeV in K2K. The detector must provide good tracking capability for all charged particles to select CCQE interaction with high efficiency and low background. The SciBar detector consists of plastic scintillator strips to realize fine segmentation. The scintillator itself is a neutrino target and there is no dead region. These features enable the Manuscript received Oct 2, 24. This work was supported by the ministry of Education, Culture, Sports, Science and Technology, Government of Japan, by the Japan Society for Promotion of Science, and by the Korea Research Foundation. S. Yamamoto is with the Department of Physics, Graduate School of Science, Kyoto University, Kitashirakawa Oiwake-cho, Sakyo-ku, Kyoto, 66-852 Japan (e-mail: shimpei@scphys.kyoto-u.ac.jp). detection of a short track down to 1 cm long, especially the proton track from CCQE interaction below 1 GeV region, and allow discrimination between CCQE and inelastic events. The detector also has a capability of particle identification with de/dx information by measuring the energy deposit in each strip. In addition, neutrino interaction can be exclusively reconstructed with SciBar and we expect to provide new and precise measurements of neutrino interaction in the region of 1 GeV, such as CC resonance production cross sections (ν µ +p µ - +p+π + /π ) and neutral current elastic scattering cross section. II. THE SCIBAR DETECTOR The SciBar is a fully active tracker consisting of 14,848 extruded scintillator strips, each of which have a dimension of 1.3 2.5 3 cm 3 with TiO 2 reflecting coating. The scintillator strips are arranged in 64 layers. Each layer consists of two planes, with 116 strips to give horizontal and vertical position. In total the detector weighs 15 tons and the dimensions are 3m in height, 3m in width and 1.7m along the beam direction. Fig. 1 shows a schematic view of SciBar. The scintillator has a 1.8mm diameter hole and each strip is read out by a wave length shifting (WLS) fiber inserted to the each strip. s are attached to a 64-channels multi-anode PMT (MAPMT). Charge and timing information from MAPMT is recorded by custom designed electronics, which is reviewed in the following section and whose schematic view is shown in Fig. 2. Just downstream of the main detector part, an electromagnetic (EM) calorimeter is installed. Electron neutrino contamination in the beam and neutral pions from neutrino interactions, are major backgrounds for ν µ ν e oscillation search [3]. We can improve electron/gamma-ray detection capability in SciBar with EM calorimeter. EM calorimeter consists of two planes, corresponding to 11 radiation lengths. One plane has 32 vertical modules to measure vertical position and the other has 3 for horizontal, covering area of 26 26 cm 2. The modules were originally used for the CHORUS experiment [4], and made of lead sheets and scintillation fibers. Specifications of the SciBar detector are summarized in Table 1. -783-871-5/4/$2. (C) 24 IEEE
MAPMT with 5 1 5 PMT gain Quantum efficiency 21% ( at 39 nm ) Optical Cross-talk ~3% using 1.5φ Non-linearity 5% at 2-photoelectron equivalent charge injection with 5 1 5 PMT gain EM calorimeter Size 4 8 262 cm 3 Radiation length 11 Energy resolution 14/ E ( GeV )% Table 1. Specification of SciBar detector Fig. 1. Schematic view of SciBar detector. The left figure shows main part of the detector and the right shows EM calorimeters located in the downstream part of the detector. Scintillator LED light injector MAPMT To back-end Front-end electronics (VA/TA) Fig. 2. Schematic view of readout system. Sixty-four s are bundled up by a cookie and mechanical supports, and attached to the surface of MAPMT. The fibers are aligned with the pixel array of the MAPMT within.1 mm accuracy. For gain monitoring, all s are illuminated with blue LED light in the LED light injector. Structure Volume 29 29 17 cm 3 Weight 15 tons Number of strips 14,848 Number of MAPMT 224 ( +8 PMTs for outer strips ) Number of EM modules 32 (vertical) + 3(horizontal) Scintillator Material Polystyrene, PPO(1%), POPOP(.3%) Size of strip 1.3 2.5 3 cm 3 Weight 1 kg / strip Reflective coating TiO 2,.25 mm thickness Emission peak 42 nm Type Y11(2)MS, multi-clad (by Kuraray) Diameter 1.5 mm Absorption peak 43 nm Emission peak 476 nm Refractive index 1.56 (core), 1.49 (inner), 1.42 (outer) Attenuation length 35 cm ( in average ) MAPMT Model Hamamatsu H884 Cathode Bialkali Anode 64(8 8) pixels by 2 2 mm 2 size Number of dynode stage 12 Gain 3 1 5 (at 8V supply) Cathode sensitivity 3-65 nm (max 42 nm) Non-linearity 5% at 2 photoelectron equivalent injection III. READOUT SYSTEM Readout system for MAPMT consists of two components: a front-end electronics board attached to MAPMT and a backend VME module. At the front-end, a circuit board with selftriggering readout using VA/TA ASIC (VA32HDR11 [5] and TA32CG [6] made by IDEAS), was developed. VA has 32-ch pre-amplifier followed by slow-shaper and multiplexer [7], [8]. TA is a fast-shaper and comparator chip, and provides timing information after taking OR of 32 channels. Two packages of VA/TA are mounted on a custom-designed PCB. A serialized analog output from 64 MAPMT anodes and two timing signals are sent from the front-end circuit to back-end modules. A back-end VME module has been also newly developed as a standard VME-9U board, which controls readout of eight front-end boards. On the board eight flash-adcs (FADC) and FIFOs are located, and analog information for each anode is sequentially stored in FIFOs via FADC. Timing information is sent to and recorded by multi-hit TDC with.78 ns resolution and 5 µs range. Programmable integrated circuits are also located and they enable flexible VA/TA control and data taking. For example, we can operate on-board zero suppression data acquisition, or calibrate gain of VA/TA. Specifications are summarized in Table 2, and pictures of these custom-made electronics boards are shown in Fig. 3. Specification of readout electronics Number of front-end boards 224 Number of back-end boards 28 Trigger threshold.4mev Dynamic range 25 pc Dead time 1 µs Trigger rate 1kHz (max) Buffer size 1 events Table 2. Specification of MAPMT readout system with VA/TA front-end board and back-end board. -783-871-5/4/$2. (C) 24 IEEE
reconstruction threshold is 8 cm; it corresponds to a momentum of.4gev/c for a proton and.1 GeV/c for a muon and a charged pion. The track finding efficiency is 99% for an isolated track longer than 1cm. Fig. 3. Pictures of custom-made electronics. The left shows front-end board, with VA/TA packages, the right is back-end module produced by Meisei co.ltd. [9]. Eight front-end bards are connected to one back-end board with 4-pin flat cables with 4m long. IV. BASIC PERFORMANCE We started data taking with SciBar in October 23 and 2 1 18 protons are delivered on target by February 24. The number of neutrino interaction in 9.38-ton fiducial volume is estimated to be 24, for this period. In the operation of SciBar, pedestal, LED, and cosmic-ray data are taken simultaneously with neutrino beam data for calibration and monitoring. PMT gain drift is monitored by a light injection calibration system with.1% accuracy. A pulse of blue LED light is transmitted to the light injector between beam spills. s are illuminated uniformly in the injector. As a reference, LED light yield is monitored relatively by a photo-diode and 2-inch PMT which is calibrated by and Am-NaI source during data taking. Fig. 4 shows the performance of the light injection calibration system for a typical MAPMT channel. The light yield of the scintillator is monitored with cosmicray and is found to be stable within 1% after the correction of gain fluctuation. The average light yield is measured to be 18.7 photoelectrons/cm for a minimum ionizing particle at the edge of the detector near to the MAPMT. On the performance of readout system, pedestal width is equivalent to.3 MeV (.3 photoelectron) and trigger threshold is.7mev/strip. The timing resolution is 1.3 ns. Fig. 5 shows an event display of CCQE interaction candidate in real data. There are a long track penetrating SciBar and a short track with large energy deposit. We can clearly find tracks from the interaction vertex, and identify a proton track and a muon in CCQE events. For a neutral pion event shown in Fig. 6, we can reconstruct pion mass by summing up all energy deposits from electro-magnetic showers. For charged track reconstruction, tracks which penetrate at least three layers are reconstructed. The SciBar provide us with projections of tracks in two views. We apply Cellular Automaton algorithm for 2-dimensional (2-D) track finding. The 3-dimensional track is reconstructed by matching 2-D tracks with track end point and timing information. The Fig. 4. Cosmic-ray light yield for a typical channel. Upper figure shows a variation of ADC value for cosmic rays during data taking from Oct 24 to Dec 24. There is a deviation in the latter half of the period. The lower shows ADC value for cosmic rays after the correction by the light injection calibration. The PMT gain follows on within 1% deviation. Run 5189 Spill 27384 TRGID 1 SBEv 3721 23 11 12 19 5 29 Nvtx 1 cm p K2K Fine-Grained Detector (Top View) µ TOP VIEW Run 5189 Spill 27384 TRGID 1 SBEv 3721 23 11 12 19 5 29 Nvtx 1 cm K2K Fine-Grained Detector (Side View) SIDE VIEW Fig. 5. Event display of ν µ interaction. The figure shows a typical CCQE event in each view. The longer track is muon and the shorter is proton. The area of red circle is proportional to energy deposit in a strip. π γ e γ Fig. 6. π production event candidate. Gamma-rays from π decay converted in SciBar, and activity around vertex can be observed. The detector has a capability of particle identification, especially for protons and pions or muons, with de/dx e p µ -783-871-5/4/$2. (C) 24 IEEE
information. Fig. 7 shows measured de/dx for muon and proton samples. Almost pure muon sample is selected by requiring tracks to extend into Muon Range Detector (MRD) [1] located downstream of SciBar. Proton sample is selected from second (short) tracks in 2-track events with the kinematics of the events. The purity of proton in this sample is more than 9%. Fig. 8 shows the muon likelihood distribution which is constructed using energy deposit distribution of cosmic-rays as probability density function. Using this likelihood, misidentification probability for protons below 1GeV/c is found to be less than 1.7% with 9% efficiency. 18 16 14 12 1 8 6 4 2 2 4 6 8 1 photoelectron Fig. 7. de/dx distribution of muon and proton enriched sample. The histogram with lower peak is muon, the higher is proton. 2 mn Eµ mµ / 2 Eν = mn Eµ pµ cosθ where m N, E µ, m µ, p µ and θ µ is the nucleon mass, the muon energy, the muon mass, the muon momentum, and the muon angle, respectively. Here we present the results of charged current event analysis, which are used in the latest K2K oscillation analysis [11]. We select charged current (CC) events by requiring at least one of the tracks extend to MRD. We use events in which one or two tracks are reconstructed. For 2-track events, we categorize them into QE enhanced and non-qe enhanced samples using kinematic information of the second track. Reconstructed muon momentum and angular distributions with respect to the beam direction for each sample are shown in Fig. 9. The energy of muon is reconstructed with its range through SciBar, EM calorimeter, and MRD. Muon energy and angular resolutions are 8MeV and 1 degree, respectively (Fig. 1, 11). Fig. 12 shows the difference of reconstructed and true neutrino energy in CCQE interaction by MC simulation. The neutrino energy resolution in CCQE events is.15gev. With current criteria, the muon momentum threshold is 45 MeV/c. To lower muon threshold in future analysis, we are investigating muons contained in SciBar, where particle identification would be crucial. 1-1 1-2 1-3 1.1.2.3.4.5.6.7.8.9 1 muon likelihood 1-1 1-2 1-3.1.2.3.4.5.6.7.8.9 1 muon likelihood Fig. 8. Muon likelihood distributions for a muon enriched sample (upper) and a proton enhanced sample (lower). The likelihood is close to 1 for minimum ionizing particles, and for other particles. Momentum (GeV/c) Angle (degree) Fig. 9. Reconstructed muon momentum and angle distribution with respect to the beam direction for each event categories. Open circles with error bars show data, a solid line shows the MC, and hatched region shows CCQE contributions. V. ANALYSIS OF CHARGED CURRENT EVENTS In the K2K experiment, signature of the neutrino oscillation is energy dependent muon neutrino disappearance. To observe it, we need to measure the neutrino energy spectrum at the near detector using CCQE events and check the Monte Carlo simulation (MC) of neutrino interactions. In CCQE interaction, neutrino energy is reconstructed from muon momentum and muon angle using two-body kinematics: 4 3 2 1 Constant 391. Mean -.551E-2 Sigma.8272E-1 -.4 -.2.2.4 Muon energy resolution (GeV) Fig. 1. Muon energy resolution. The difference between reconstructed energy and true energy in MC sample are shown. -783-871-5/4/$2. (C) 24 IEEE
Mean 1.484 Constant 6515. 5 (a) 3-D 8 (b) X Mean.8299E-2 Sigma.9598 4 3 6 2 4 1 2 5 1-5 5 3-D angle resolution (degree) 2-D angle resolution (degree) 8 6 4 2 (c) Y Conantt 6557. Mean -.377E-2 Sigma.9624-5 5 2-D angle resolution (degree) Fig. 11. Muon angle resolution. (a) shows the angle between MC true direction and reconstructed direction of a muon. The angular resolution is defined as the value at which 68% of events are covered. In this case, the 3- dimentional angle resolution is 1.6 degrees. (b) and (c ) show angular resolutions in 2-D views, and it is one degree for each projection. Constant 628. 39.86.14 Mean -.132E-1.7E-3 Sigma.1498.6639E-3.12.1.8.6.4.2-2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5 E ν resolution (GeV) Fig. 12. Neutrino energy resolution. The difference of reconstructed and true neutrino energy is shown in CCQE interaction by MC simulation. Neutrino energy is reconstructed with.15 GeV resolution. VIII. REFERENCES [1] S. Fukuda, Y. Fukuda, T. Hayakawa, E. Ichihara, M. Ishitsuka, Y. Itow, et al., The Super-Kamiokande detector, Nucl. Instrum. Meth. A 51, pp. 418-462, Apr. 22. [2] M. H. Ahn, S. Aoki, H. Bhang, S. Boyd, D. Casper, J. H. Choi, et al., Indications of Neutrino Oscillation in a 25 km long baseline experiment, Phys. Rev. Lett., vol. 9, no. 4181, 23. [3] M. H. Ahn, S. Aoki, Y. Ashie, H. Bhang, S. Boyd, D. Casper, et al., Search for electron neutrino appearance in a 25 km long-baseline experiment, Phys. Rev. Lett., vol. 93, no. 5181, 24. [4] S. Buontempo, A. Capone, A. G. Cocco, D. De Pedis, E. Di Capua, U. Dore, et al., Construction and test of calorimeter modules for the CHORUS experiment, Nucl. Instrum. Meth. A 349, pp. 7-8, Sep. 1994. [5] Ideas ASA, (2, Oct.), VA32_HDR11 specifications. [Online]. Available: http://www.ideas.no/products/asics/pdf/va32_hdr11.pdf [6] Ideas ASA, (21, Feb.), TA32CG specifications. [Online]. Available: http://www.ideas.no/products/asics/pdf/ta32cg.pdf [7] E. Nygård, P. Aspell, P. Jarron, P. Weilhammer and K. Yoshioka, CMOS low noise amplifier for microstrip readout Design and results, Nucl. Instrum. Meth. A 31, pp. 56-516, Mar. 1991. [8] O. TokerS. Masciocchi, E. Nygård, A. Rudge and P. Weilhammer, VIKING, a CMOS low noise monolithic 128 channel frontend for Sistrip detector readout, Nucl. Instrum. Meth. A 34, pp. 572-579, Mar. 1994. [9] Meisei electric co.,ltd. 2-5-7, Koishikawa, Bunkyo-ku, Tokyo, 112-8511 Japan (http://www.meisei.co.jp). [1] T. Ishii, T. Inagaki, J. Breault, T. Chikamatsu, J. H. Choi, T. Hasegawa et al., Near Muon Range Detector for the K2K Experiment - Construction and Performance-, Nucl. Instrum. Meth. A 482, pp. 244-253, 22. [11] T. Nakaya, New K2K Results presented at the 21 th Int. Conf. NEUTRINO, Paris, France, 24. VI. CONCLUSIONS A novel neutrino detector, SciBar, was designed and constructed to improve the measurement of neutrino energy spectrum and neutrino interaction in K2K long baseline neutrino oscillation experiment. The SciBar is a fully active, fine-gained detector made of plastic scintillator strips. Data taking was started in October 23 and the performance of the detector was found to be extremely good. The data from SciBar were already used in the latest neutrino oscillation analysis at K2K. More results on neutrino oscillation and interaction are expected in near future. VII. ACKNOWLEDGMENT This work has been funded by the ministry of Education, Culture, Sports, Science and Technology, Government of Japan, by the Japan Society for Promotion of Science, and by the Korea Research Foundation. The authors gratefully acknowledge KEK technical staffs assistance in fabricating the detector and developing electronics. We are also grateful volunteers for their hard work on the construction. -783-871-5/4/$2. (C) 24 IEEE