Performance of the Liquid Xenon Detector for the

Performance of the Liquid Xenon Detector for the MEG experiment Yasuhiro NISHIMURA Hiroaki NATORI The University of Tokyo MEG collaboration Outline • m → e g and MEG experiment • Design of detector • Calibration • Performance in run 2008 The 12 th Vienna Conference on Instrumentation 19/Feb. /2010

Search for m → e g is a clear 2 body decay Br(m→eg) < 1. 2× 10 -11 (90%C. L. ) given by MEGA experiment Charged lepton-flavor mixing is not observed yet. Many extensions of standard model predict m → e g SU(5) or SO(10) models in SUSY-GUT, etc. � Backgrounds m → enng << Accidental background is dominant Background energy [mm/2] of e+ and g e+ from m → enn g from m → enng Br(accidental) = Rm・fe 0・fg 0・(dweg/4 p)・(2 dteg) � The innovation in MEG experiment for m → e g discovery � Excellent performance of a liquid xenon (LXe) g–ray detector World's most intense ~108/sec DC m beam at Paul Scherrer Institut (PSI) in Switzerland 3 × 107 m+ / sec stop in target Positron spectrometer operational at high rate Aim at 19/Feb. /2010 10 -13 branching ratio in MEG experiment Yasuhiro NISHIMURA Proton ring-cyclotron 590 Me. V, 2. 2 m. A max. The 12 th Vienna Conference on Instrumentation 2

MEG Detector g-ray detector 900 liters LXe scintillator for the g ray detector Positron detector COBRA (COnstant-Bending Radius) magnet ▪ B=1. 27 T, 20% X 0 Drift chamber Timing counter 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 3

Liquid xenon detector Merits of liquid xenon High stopping power 2. 98 g/cm 3 , X 0 = 2. 8 cm No self-absorption of scintillation light High light yield (75% of Na. I(Tl)) Fast scintillation process ▪ 4. 2, 22, 45 ns components HV Feedthru. Developed photomultiplier tube (PMT) Short wave length l~178 nm (VUV) HAMAMATSU Low temperature within 161~165 K Scintillation light absorbed by O 2, H 2 O, etc. LXe is expensive Signal Feedthru. [ns] Difficulties Pulse tube refrigerator Inner cryostat Vacuum vessel Design 900 liters (800 liters active volume) 11% solid angle, 14 X 0 to purifier 846 PMTs of 2'' size on 6 faces Waveform taken for all channel 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 4

Development of liquid xenon detectors � Small prototype � 2. 3 liters active volume 32 PMTs 1998 - LXe detector active 800 liters / Total 900 liters volume (3 t), 846 PMTs Large prototype 68. 6 liters active volume 228 PMTs 2000 - Ready at the end of 2007 Operated in 2008 for MEG run 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 5

Cryogenic System gas pulse-tube refrigerator developed for MEG Storage Gas and liquid Cooling Pulse-tube refrigerator and LN 2 8 × 250 L gas storage tank liquid Cooling pipe with liquid nitrogen LN 2 900 L LXe detector 1000 L liquid storage tank 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 6

Purification system Gaseous purification A getter removes H 2 O, O 2, CO 2, H 2, N 2 and hydro carbon molecules. (10 to 50 l/min) Liquid purification By a molecular sieves and O 2 getter, water and O 2 are removed. (100 l/h) Gas phase Water and oxygen absorb scintillation light Liquid phase 19/Feb. /2010 Yasuhiro NISHIMURA 900 L LXe detector The 12 th Vienna Conference on Instrumentation 7

Calibration and Monitor in 2008 Run 2008 Calibration of PMT Gain Quantum efficiency Monitor Set up Light yield and purification 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 8

purification Run 2008 Engineering run in 2007 p 0 run physics run 2008 - Mar. Jun. Jul. Aug. Sep. Oct. Nov. Dec. Maintenance Installation Purification and Monitor of the light yield p 0 run for the calibration of LXe detector Trigger setup and started physics run Taking physics data, purification again p 0 run again at the end of Dec. LXe detector successfully operated in 3 -month physics run ! 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 9

PMT gain calibration with LEDs at 5 positions on both sides 3 different intensities by attenuation Calculate absolute gain by mean-variance relation 9 steps of different intensity by changing current s 2 = Gain × Mean × e/C(const. ) + const. Slope = gain Gain monitor with stable LED s 2 Absolute gain LED peak 1 month Charge Date 1/2 hour calibration everyday and LED flushing in physics run for the monitor 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 10

Q. E. calibration with 241 Am a source For the estimation of quantum efficiency (Q. E. ) of PMT 5. 5 Me. V a from 241 Am source (200 Bq, 432 years) 25 sources, 5 on a wire, immersed in liquid xenon Q. E. derived from the comparison between observed charge and expectation in Monte Carlo simulation 100 mm diameter, 2 mm length reconstructed a events � 241 Am Q. E. � 100 mm Separate signal of a from g by the shape of waveform a ~40 mm Shadow of wire makes ring by reconstruction 19/Feb. /2010 Data - MC Yasuhiro NISHIMURA g Charge / height of waveform The 12 th Vienna Conference on Instrumentation 11

Cockcroft-Walton accelerator For the light-yield monitor and to know the uniformity of g energy B Energy of g - ray in LXe detector Accelerator and power module Beam line at opposite of m F B Li LXe detector muon 1 Me. V Cockcroft-Walton accelerator proton Provide proton beam ~1012 / sec Li(p, g)Be 14. 6, 17. 6 Me. V B(p, g)C 4. 4, 11. 7, 16. 1 Me. V Proton target quickly switched from m target ~20 min. 3 times / week beam line Cosmic ray and Am. Be source are also useful for the monitor 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 12

Light yield in 2008 The light yield improved by purification Monitor by Li 17. 6 Me. V, 54. 9 Me. V g from p 0 decay, cosmic ray p- beam Physics data p- Gas phase purification + Liquid phase purification a-g separation This increase of the light yield gives Shape of waveform changing Better a-g separation by waveform Start / End of run 2008 � 19/Feb. /2010 a g Waveform In 2009 the light yield completely recovered and stabilized. Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 13

Performance of g ray measurement Reconstruction Set of g ray up for p 0 run Evaluate the performance around 53 Me. V signal g-ray Energy Timing Position Detection efficiency Measurement 19/Feb. /2010 of m decay Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 14

Reconstruction in 2008 � Position Energy Detected photons = S (weight×charge / Q. E. / Gain)PMT Correct the change of light yield and nonuniformity, eliminate pile-up photons Timing conversion point to avoid effect of shower and pileup 3 d-position fitted with solid angle of each PMT LXe detector Time of each PMT subtracted with ▪ ▪ Light distribution of PMTs Use PMTs around Propagation in LXe Time-walk effect Effect of PMT face by incident angle Offset of channel Average time weighted with the number of photoelectrons of each PMT LXe inner surface Deep event (10 cm) Shallow event (1 cm) LXe detector can reconstruct all property of g ! 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 15

p 0 run for g-ray around signal energy Concept Acquire performance of energy, timing and position around signal energy (53 Me. V) Obtain monochromatic g (55 or 83 Me. V) by tagging another g from p 0 → 2 g at opposite side Determine energy scale near the signal g-ray n p- g Set up ~60% p 0 q (98. 8%) g 54. 9 ~ 83. 0 Me. V ~40% opening angle 8. 9 Me. V + 129. 4 Me. V Na. I energy [Me. V] p LXe energy [Ge. V] Select back-to-back events LXe energy [Me. V] In 2008, p 0 run in Aug. for 1 month and short p 0 run in Dec. 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 16

Detectors to tag g from p 0 Energy : 3× 3 Na. I(Tl) crystals with 9 Avalanche photodiodes (APDs) Control APD temperature by a peltier device and Pt 100 APD enables constant gain wherever in the magnetic field Timing : Pre-shower counter in front of Na. I 2 plastic scintillators with 4 PMTs and lead converter Plastic scintillators and Na. I detector 62. 5 x 305 mm (12 X 0) x 9 bars (18 ℃, DT< 0. 1 ℃) heat Peltier dev. LED APD and amplifier Na. I APDs Mag net 60 x 7 mm Radiator APD, amp. Na. I Thermal control APD 1 cm x 1 cm, HAMAMATSU Light guide How to measure whole acceptance of LXe detector? 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 17

Set up for p 0 run Mover system to scan whole acceptance of LXe Liquid hydrogen target inserted from down stream L = 75 mm f = 50 mm D < 6. 5 o 60 o 120 o p- (1 MHz) m+ target 60 o LXe detector 19/Feb. /2010 Use the same beam line as m+ beam Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 18

Energy resolution Map of s (upper tail) on LXe inner surface 55 Me. V energy peaks at each position Lower tail of energy spectrum from : Escape of shower in a shallow event Interaction with material before reaching LXe Fit with (Exponential + Gaussian + Difference of pedestal between p- and m+) Deeper events than 2 cm, in acceptance 2. 0± 0. 15% s of upper tail, important to identify the signal 5. 8± 0. 35% FWHM 55 Me. V g-ray 0 1 2 3 4 s [%] of upper tail 19/Feb. /2010 Yasuhiro NISHIMURA 5 The 12 th Vienna Conference on Instrumentation 19

Linearity and energy scale Good linearity p 0→ 2 g Within 0. 5% B (4. 4 Me. V, 12. 0 Me. V), Li (17. 6 Me. V), Li p 0 decay (54. 9 Me. V, 83. 0 Me. V) Include the uncertainty of the light-yield correction Energy B scale Determined at only 55 Me. V Uncertainty < 0. 4% Due to the correction of light yield, gain, etc. 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 20

Timing LXe detector Pre-shower counter (2 plastic scintillator ) g g Use pre-shower counter as a reference Difference of LXe – pre-shower counter 135 ps s at 55 Me. V 127 ps s at 83 Me. V However… Time difference of 2 detectors contains Spread of p 0 decay point in target : 58 ps s Resolution of reference counter : 93 ps s Time difference of 2 PMTs attached with the same scintillator Timing resolution 78 ps s = 135 Ө 58 Ө 93 ps at 55 Me. V Better resolution (68 ps s) at the end of run 2008 due to the light-yield recovery 80± 6 ps s at signal energy (53 Me. V) after correction of energy dependence 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 21

Position Use Monte Carlo simulation for whole acceptance Partially confirmed by p 0 run with lead collimators in front of LXe 18 mm thickness with 10 mm slits 55~83 Me. V in p 0 decay Projection on surface lower – upper [cm] lead slit along beam axis [cm] 5 mm s along the LXe surface Fit with 2 error-functions + 3 gaussians + floor 6. 8 mm s average Beam spread of ~8 mm s in target → ~2 mm s on detector surface Projection of slit ~ 14 mm 6 mm s along radial direction from Monte Carlo simulation data simulation [cm] 19/Feb. /2010 Yasuhiro NISHIMURA lower – upper [cm] The 12 th Vienna Conference on Instrumentation 22

Detection efficiency 3 methods to estimate the detection efficiency Monte Carlo simulation of signal g ray Count g from radiative m decay p 0 → 2 g tagged with Na. I opposite to liquid xenon detector Count g around 55 Me. V tagged with around higher 83 Me. V at tagging Na. I detector Subtract neutron background which comes from tail of 129 Me. V(p - → n g) in Na. I neutron ? Monte Carlo simulation of p 0 83 Me. V g 55 Me. V g All consistent within a few percent Efficiency estimated by Monte Carlo simulation Including analysis efficiency (pile up, cosmic ray, etc. ) Eg > 46 Me. V Detection efficiency : 63± 4% 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 23

g measurement in physics run 2008 g ray from m+ decay pile up g background in m decay Successfully 19/Feb. /2010 Time difference between e+ and g (m → enng) operated during physics run ! Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 24

Conclusions Successful construction of the 900 L liquid xenon detector First evaluation of the performance around signal g Energy 2. 0% s, 5. 8% FWHM Studying discrimination of pile-up and reconstruction, etc. Timing 78 ps s Almost enough for our requirement Position 5 mm s on surface, 6 mm s along depth Better estimation of Q. E. and gain is the candidate for the improvement Took first physics data for 3 months in 2008 with stable operation and proper calibration ! MEG experiment will run for next few years to reach 10 -13 branching ratio sensitivity. 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 25

Our MEG collaborators 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 26

End of slides 19/Feb. /2010 Yasuhiro NISHIMURA 2008 data in analysis box The 12 th Vienna Conference on Instrumentation 27

Back up slides 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 28

Light yield in 2009 Preliminary look at 2009 data a - g separation Light yield • Cosmic ray, Li 18 Me. V ▪ Enabled identification in trigger level 2009 stable 2008 start / end a g 2009 2008 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 29

Correction for non-uniformity Detected scintillation photons in current reconstruction is not independent on the position Estimated by Li 18 Me. V peak Prepared 2 set of correction in 2008 separated by the light yield Peak dependence along radial direction Peak map on LXe inner surface Lower light yield Higher light yield Better uniformity after the correction 19/Feb. /2010 Yasuhiro NISHIMURA The 12 th Vienna Conference on Instrumentation 30