1. Study of the MPPC for the GLD Calorimeter Readout Satoru Uozumi (Shinshu University) for the GLD Calorimeter Group (KNU, Kobe, Niigata, Shinshu, ICEPP Tokyo, Tsukuba) Introduction Performance of 1600-pixel MPPC GLD calorimeter prototype Summary & Plans

2. International Linear Collider GLD LDC SiD 4th Center-of-mass energy at 500 ~ 1000 GeV. Four detector concepts are currently proposed. Various precision measurements with clean e + e - collision. –e + e - -> Higgs, SUSY, W, Z, tt … -> Multi-jet final states Precise jet energy measurement is crucial !

3. Particle Flow Algorithm (PFA) - A powerful method for precise jet energy measurement - Charged particles (~60% in a jet) are measured by central tracker. Photons (mainly from  0 decays, ~30%) are measured by electromagnetic calorimeter. Neutral hadrons (mainly K L 0, ~10%) are measured by hadron calorimeter. E TOT = p e + p  + p charged hadron + E  + E neutral hadron [ tracks only] [calorimeter only] For the PFA, separation of jet particles in the calorimeter is indispensable. Key : Momentum resolution > Energy resolution (central tracker) (calorimeter)

4. Sampling calorimeter with W/Pb - scintillator sandwich structure with WLSF readout Particle Flow Algorithm (PFA) requires fine granularity for the calorimeter Employ well-established plastic scintillator. Scintillator stirp structure to achieve fine granularity. Huge number of readout channels. –~10M (ECAL) + 4M (HCAL) Used inside 3 Tesla solenoid,. The MPPC is feasible for the GLD calorimeter readout ! The GLD Calorimeter

5. What are required for the photon sensor? Gain, Photon Detection Efficiency (P.D.E.) comparable to PMTs. –Gain at least 10 5. –P.D.E. ~ 20% (capable to distinguish MIP signal). Dynamic range enough to measure ~100 GeV photons. –corresponds to ~5000 photoelectrons. –MPPC is non-linear device. -> necessary to know right response curve for correction. Stability & Robustness. Tolerance to temperature variation, long-term use, magnetic field, neutron radiation. Uniform performance among many pieces. Acceptable dark noise rate, inter-pixel cross-talk probability. –Noise rate < 1 MHz, Cross-talk probability ~ a few per cent. Low cost, compactness. –Price order of $1~5, package fitting to 2 mm thick scintillator strip.

6. 4 mm 1 x 1 mm 1.3 mm Side 3 mm Small package 1600 pixel MPPC 1600 pixel MPPC with compact plastic package (S10362-11-025MK) Sensor size = 1 x 1 mm 2. suitable for attaching to scintillator strips. Front MPPC Scintillator strip (1 x 4 x 0.3 cm) WLS fiber Side view

7. 30 o C 25 o C 20 o C 15 o C 10 o C 0 o C -20 o C –C … Pixel capacity –V 0 … Breakdown voltage 30 o C 25 o C 20 o C 15 o C 10 o C 0 o C -20 o C Gain, Dark Noise Rate, Cross-talk 30 o C 25 o C 20 o C 15 o C 10 o C 0 o C -20 o C These basic properties are almost OK. Still need to improve temperature dependence of breakdown voltage  V (over-voltage)  V 0 /  T = (56.0±0.1) mV/ o C Noise > 0.5 p.e.

8. Variation of the Gain over 700 MPPCs Device-by-device variation is less than a few %.  No need for further selection or categorization on massive use ! Just need small tuning of operation voltages. Variation = 0.45 V Variation < 3.3 %

9. Variation of the Dark Noise (400 pieces) Noise rate is far less than 1 MHz with all the samples. Device-by-device variation is order of ~10 %.  V = V bias – V 0 (V) Noise Rate (kHz) Noise > 0.5 p.e.

10. Photon Detection Efficiency (P.D.E.) MPPC 0.5 mm  holes PMT LED WLSF ~ 16 % P.D.E. = Q.E. x  geometry x  Geiger Measured by injecting same light pulse into both MPPC and PMT, and comparing light yield. MPPC PMT The 1600-pixel MPPC has comparable P.D.E. with conventional photomultipliers (15~20%).

11. The MPPC is a non-linear device, because one pixel can detect one photon at once. For a short light pulse input, response to input light can be theoretically calculated as However for the 1600-pixel MPPC, it is not the case ! Since recovery time is order of a few ns, one pixel can detect a photon several times. Dynamic range Recovery time >> light input width ? Recovery time << light input width

12. Recovery Time Measurement MPPC delay LED Inject blight laser pulse (width=52 ps) into the MPPC After delay of  t, inject blight LED light pulse, and measure MPPC outpu for the LED pulse. Compare the MPPC output for the LED pulse with and without the first laser pulse. tt Black … MPPC output for Laser pulse Green … MPPC output for LED pulse Red … Laser + LED Blue … (Laser+LED) - Laser Ratio of Blue / Green shows recovery fraction. t (  sec) Oscilloscope view

13. Recovery Time Result The curve is fitted by a function t D : dead time  : recovery time Delay Recovery fraction (%) V bias (V)  (nsec) t D (nsec) 71.04.1 + 0.11.9 + 0.1 71.54.0 + 0.11.7 + 0.1 72.04.2 + 0.11.3 + 0.1 Recovery time of the 1600-pixel MPPC ~ 4 ns. The shape does not depend on bias voltage. V bias = 71.0 V 71.5 V 72.0 V at 25 o C

14. Response Curve Response curves taken with various width of LED light pulses. (gate width = 100 ns) Dynamic range is enhanced with longer light pulse, Time structure of the light pulse gives large effects in non-linear region. No significant influence with changing bias voltage. Knowing time structure of scintillator light signal is crucial -> study is ongoing. 8 ns 16 ns 24 ns w = 50 ns 24 ns 8 ns 16 ns 1600 PMT LED w

15. MPPCs (1600 pixels) The GLD ECAL prototype Scintillator strip (1 x 4.5 x 0.3 cm) Frame WLS fibre Tungsten (3.5 mm thick) Scintillator layer (3 mm thick) e+e+ Beam test has been performed at DESY using 1-6 GeV e+ beams.

16. Beam position (mm) Signal (ADC counts) MPPC MIP events (data without absorber) WLS Fibre readout strip Scintillator strip Direct readout strip MPPC WLS fibre

17. 6 GeV e + center injection 1 2 3 4 5 6 GeV Results with tungsten absorber Deviation from Linearity < 4 % The calorimeter with the MPPC is working! Dynamic range seems to be enhanced (expect more study with higher beam energy in 2008). No MPPC saturation correction

18. Summary We are developing the 1600-pixel MPPC for the GLD calorimeter collaborating with Hamamatsu Photonics. Basic properties are almost sufficient for the requirement. –Comparable gain / P.D.E. with photomultipliers. –Acceptable noise rate (~100kHz) and cross-talk probability (~10 %) Variation of the gain among 800 pieces is small enough (<4%). Dynamic range is enhanced thanks to the short recovery time! However need more study for time structure of light signal from scintillator. Result of the beam test demonstrates that the MPPC is feasible for the GLD calorimeter readout. Further improvements of the MPPC performance is still underway with Hamamatsu (package, temp. dependence, dynamic range, etc…).

19. Backups

20. Jet Reconstruction @ ILC Particle Flow Algorithm (PFA) Z → qq @ 91.2GeV The tracks are used instead of making a calorimetric measurement of the energies of the charged hadrons (with large errors).

21. Excellent photon counting ability 0,1,2,3,4,5,6,7,... Photoelectrons ! 1 photoelectron 2 photoelectrons 4 mm 1.3 mm Side 3 mm Front

22. PhotomultiplierMPPC Gain~10 6 10 5 ~10 6 Photon Detection Eff.0.1 ~ 0.2~0.2 for 1600 pix. MPPC Responsefast Photon countingYesGreat Bias voltage~ 1000 V~ 70 V SizeSmallCompact B fieldSensitiveInsensitive CostVery expensive !Not very expensive Dynamic rangeGoodDetermined by # of pixels Long-term StabilityGoodUnknown Robustnessdecent Unknown, presumably good Noise (fake signal by thermions) QuietNoisy (order of 100 kHz) The MPPC has lots of advantages The MPPC is a promising photon sensor, and feasible for the GLD Calorimeter readout !