Daisuke Kaneko, ICEPP, Univ. of Tokyo on behalf of MEG collaboration Performance of UV-sensitive MPPC for liquid xenon detector in MEG experiment Daisuke Kaneko, ICEPP, Univ. of Tokyo on behalf of MEG collaboration
μ+→e++γ MEG experiment MEG is searching for charged lepton flavor violating decay Highly suppressed in the standard model. (ℬ∼ 10 −50 ) But, sizable decay rate (ℬ∼ 10 −12 ⋯ 10 −14 ) is predicted in theories beyond the standard model. We published the latest result with 2009-2011 data. 𝓑<𝟓.𝟕× 𝟏𝟎 −𝟏𝟑 (90% CL), the most stringent experimental upper limit. Phys. Rev. Lett. 110, 201801 (2013) Data taking finished in 2013 August. μ+→e++γ 2013 Nuclear Science Symposium and Medical Imaging Conference
MEG upgrade MEG is planning a major upgrade. Our proposal was approved by Paul Scherrer Institute (Switzerland) in 2013 January. (arXiv: 1301.7225) Upgrade items Increase beam intensity Active and thinner target Unified volume, stereo angle wire drift chamber Pixelated, SiPM read out timing counter Radiative μ decay counter to remove accidental background SiPM readout liquid xenon γ detector Sensitivity goal is 𝟓× 𝟏𝟎 −𝟏𝟒 , 10 times higher than current MEG. 2013 Nuclear Science Symposium and Medical Imaging Conference
Upgrade of liquid xenon detector Present Improved granularity by smaller photo-sensor. MPPC PMT γ-ray Upgraded CG image 12mm PMT 46mm γ-ray Resolution was limited by sensor size, especially for shallow events. 2013 Nuclear Science Symposium and Medical Imaging Conference
Performance : Energy resolution depth < 2cm 40 % of events depth ≧ 2cm 60 % of events Upgraded Upgraded σup 2.4% ↓ 1.1% σup 1.7% ↓ 1.0% Present Present Response to 52.8MeV γ (MEG signal) 2013 Nuclear Science Symposium and Medical Imaging Conference
Performance : Position resolution Red : Present Blue : Upgraded Depth from inner face [cm] Position resolution in σ [mm] 2013 Nuclear Science Symposium and Medical Imaging Conference
Cross-sectional image of MPPC UV sensitive MPPC Development is performed in collaboration with Hamamatsu Photonics. Requirements Sensitivity to liquid xenon scintillation (λ = 175 nm) Large active area (12×12 mm2) Single photon counting capability Moderate trailing time constant (τ < 50 ns) In order to improve sensitivity Remove protection layer Matched refractive index to liquid xenon Anti reflection coating Cross-sectional image of MPPC 2013 Nuclear Science Symposium and Medical Imaging Conference
Achieved performance ↑Charge spectrum with 12×12㎟ MPPC 12×12㎟ MPPC ↑17% of PDE (crosstalk & after pulse removed) is already achieved. Even higher PDE is expected after new technologies of Hamamatsu will be included. 3×3㎟ MPPC 2013 Nuclear Science Symposium and Medical Imaging Conference
Series connection of MPPC Remaining issue was long tail in waveform (τ ~ 200ns) due to large capacitance of large-area sensor. In order to shorten long decay time, reduce sensor capacitance by subdividing active sensor region and connect them in series. 6mm 12mm Signal is expected to be sharpened due to smaller capacitance. However there is anxiety that S/N becomes worse. 2013 Nuclear Science Symposium and Medical Imaging Conference
How many segments to divide We simulated the series connection using four independent small MPPCs (6×6㎟ each). C is capacitance when all segments are connected in parallel. 4 segments (C/16) 4 × 6mm × 6mm 2 segments (C/4) 2 × 12mm × 6mm 2013 Nuclear Science Symposium and Medical Imaging Conference
2 types of series connection There are two options for series connection of segments; "simple" and "hybrid“. Serial – signal Parallel - bias Simple series connection Hybrid connection amplifier ○ Less parts ○ Automatic over voltage adjust × Higher voltage × Different potential between each sector × Extra parts × Gain uniformity is required ○ Same voltage as single MPPC ○ All sectors have common potential bias Both connection types work in LXe, but hybrid is more advantageous. 2013 Nuclear Science Symposium and Medical Imaging Conference
Set up for liquid xenon test 6mm MPPC LED 10 kΩ 10 nF 2013 Nuclear Science Symposium and Medical Imaging Conference
Waveform of 1 photo electron Signal tail is successfully reduced down to 30-50ns ! The tail for 4 segmentation is as short as that for 3×3㎟ single sensor as expected. 1 p.e. signal can be resolved 4segments V → Vov = 3.0V 2segments Vov = 3.0V 0 p.e. 1 p.e. 2 p.e. charge → Parallel Vov =1.5V ×10 amp t → Baseline RMS have no significant difference, ~2mV 2013 Nuclear Science Symposium and Medical Imaging Conference
Decay time and connection type Calculated decay constant by fitting waveform by event. 4 segments 25ns 2 segments 46ns ↕ Parallel 200ns 50ns of target time constant is achieved in both cases. 2 segments 4 segments 2013 Nuclear Science Symposium and Medical Imaging Conference
Gain (multiplication factor) Gain is 4~10×105 Gain is lower when capacitance is smaller, while pulse amplitude is almost the same. Lower than that of parallel connection, but 1 p.e. peak is resolved with proper voltage. 2 segments 4 segments 2013 Nuclear Science Symposium and Medical Imaging Conference
Effect of long cable In MEG LXe detector, long coaxial cable of about 10m exists between sensor and readout electronics. 2 segments 4 segments Cable Length [m] Gain does not change, while decay time slightly increases (1ns/m). Constant is 56ns at 11m of 2 segments, but still acceptable. 2013 Nuclear Science Symposium and Medical Imaging Conference
Summary of serial connection Gain (@ 3V) Peak separation Noise rms [mV] Decay const [ns] Rise 4 segments 4×105 ○ 2.0 – 2.5 25 ~ 33 2 – 3 2 segments 8×105 ◎ 46 ~ 56 4 - 5 * 1 segment 1.2×106 3~5 ~200 ~10 * estimation from previous measurement Gain is larger and peak is clearer when number of segment is few. This makes charge calibration easier and relative gain error smaller. On the other hand, signal is sharper with finer segmentation. Timing resolution is expected to be better, and pile-up will be reduced. 4 segments is favorable, because gain error is small enough, and timing resolution should be as good as possible. In addition, if single p.e. can not be resolved, it is easier to switch 2 segments. 2013 Nuclear Science Symposium and Medical Imaging Conference
Prototype detector γ We are planning to make prototype of MPPC read-out detector 576 MPPCs on incident face. Beam test will be performed in 2014 CG image of prototype detector 2013 Nuclear Science Symposium and Medical Imaging Conference
Summary We are developing UV-sensitive MPPC for upgrade of liquid xenon detector in collaboration with Hamamatsu Photonics. >17% PDE was already obtained by 12×12㎟ and 1p.e. countable MPPC prototype. In order to solve problem of long waveform due to the carge capacitance of large-area MPPC, we proposed series connection. Serial connection was tested in LXe environment. 30 – 50 ns of decay constant was achieved Single p.e. can be resolved Prospects Prototype detector with ~600MPPCs is being prepared. Detector construction will start in 2014 for next MEG experiment from 2016. 2013 Nuclear Science Symposium and Medical Imaging Conference
2013 Nuclear Science Symposium and Medical Imaging Conference Fin. 2013 Nuclear Science Symposium and Medical Imaging Conference
PCB based feedthrough Slope (Charge/Voltage) = 100.8 ± 1.2 [fF] LEMO = 103.0 ± 2.0 [fF] MMCX Charges of 1 p.e. event are compared between usual LEMO-connector and PCB feedthroughs. About for charge, there is no difference. Another properties are being analyzed. 2013 Nuclear Science Symposium and Medical Imaging Conference
Serial connection test at room temp. Set up for room temp. test Observed Waveform LED light, averaged LED MPPC mask 4 serial 30ns 2serial,2parallel 50 ns 4parallel 200ns 2013 Nuclear Science Symposium and Medical Imaging Conference
Reminder 1 etc. How to reduce long tail ? Application to LXe detector Sensitivity to VUV (λ=175nm) 1 p.e. resolve with 12x12㎟ 0 p.e. 1 p.e. 2 p.e. Long tail (τ~200ns) Worse S/N Increase pileup How to reduce long tail ? Target is about 50ns (~Scintillation decay) etc. 2013 Nuclear Science Symposium and Medical Imaging Conference
Reminder 2 How to reduce long tail of 12×12㎟ MPPC ? Reduce quench resistance as low as possible. Reduce input resistance of amplifier Divide and connect in series active sensor area Tail is not be reduced than expected with lower quench Waveform is distorted with long coaxial cable Rs = 33Ω, 10m cable 200ns Most realistic plan The effect was confirmed at room temperature test. 2013 Nuclear Science Symposium and Medical Imaging Conference
Raw-Waveform 25μ Low Rq 50μ Low Rq 25μ Mid Rq 50μ Mid Rq 25μ High Rq 2013 Nuclear Science Symposium and Medical Imaging Conference
pre amplifier 2013 Nuclear Science Symposium and Medical Imaging Conference
Result : Waveform and Quench-R Fitting waveforms with a double-exponential function works well. ・The tail time constant do not depend on Rq so much ・25µm pitch MPPC is not so different from 50µm. 1 Rising & 1 Trailing component, τr ,τt MPPC Type Quench R [kΩ] Trail time constant τt [ns] Rise time constant τr [ns] 50 Over voltage [V] 1.0 1.2 1.5 R1 Low 349 246 255 276 19.6 21 24.3 R2 Mid 606 277 288 314 19 19.4 20.7 R3 High 8867 - 783 16.5 25 Over voltage [V] 2.0 2.5 3.0 719 214 1170 218 20.1 21433 538 23.5 2013 Nuclear Science Symposium and Medical Imaging Conference
… Vb Rp Rs Rq Cs Cd Cause of long waveform AMP side Vb Waveform can not be shortened only by reducing the quench resistance and cell capacitance. bias voltage Rp protect resistance MPPC side amplifier quench resistance Rs Rq … shunt resistance Cs Cd stray capacitance Rs,Rp×Cs term is dominant against Rq×Cd term under small Rq condition? diode (sensor) 2013 Nuclear Science Symposium and Medical Imaging Conference
How to shorten waveform Smaller Rs → Effective, but only in limited situation. ↓Data taken with large-area MPPC at room temperature↓ Rs = 50Ω short cable Rs = 33Ω short cable Rs = 33Ω long cable τt= 192 ns τt= 138 ns Tail is reduced with small Rs, but the waveform is distorted with a long read-out cable because of the impedance mismatch. Smaller Rp → Not effective 2013 Nuclear Science Symposium and Medical Imaging Conference
2013 Nuclear Science Symposium and Medical Imaging Conference
Width of 1 p.e. peak 2013 Nuclear Science Symposium and Medical Imaging Conference
Leading edge and connection type 2 segments 4 segments Leading time constant is also shorter in smaller capacitance. Leading time constant seems to depend on o.v. , fitting for small waveform was difficult. ↓ To use pico sec pulse laser, Hamamatsu PLP-10 400nm. Available soon. 2013 Nuclear Science Symposium and Medical Imaging Conference
Detailed PDE calculation of G-type MPPC Voltage 1.4 1.7 Errors Gain 3.1% 1.6% Alpha peak 0.26% 0.21% Correction 3.2% 1.7% Model syst. 14% 22% total 15.3% 22.2% In 2013Feb-1 test 1301-50UM-R2 voltage 1.4 1.7 PDE +/- 15.5 2.5 16.9 3.7 2013 Nuclear Science Symposium and Medical Imaging Conference