Selection of Silicon Photomultipliers for ILC Analogue Hadron Calorimeter Prototype

1. Selection of Silicon Photomultipliers for ILC Analogue Hadron Calorimeter Prototype New Photon Detectors Workshop PD07 University of Kobe, June 27-29, 2007 E.Tarkovsky (ITEP Moscow), in the frame of CALICE collaboration • Lay-out • ILC Hadron Calorimeter prototype with SiPM readout • Selection criteria of photodetectors • Experimental set-up for SiPM test • Result of selection • Other useful applications

2. 900 Results reported in this talk have been obtained in the frame of construction of hadron calorimeter prototype for ILC The physics goals forward requirement of excellent energy resolution for such a calorimeter (~30%/√E) which may be reached using PFLOW. This results in obligatory high granularity in both longitudinal and transversal directions. The cubic meter prototype currently built by CALICE collaboration has 38 planes of 5 mm thick scintillators interleaved with 20 mm thick steel absorbers. Detecting plane with 216 tiles The total number of scintillators in the prototype is 7608. Adding detectors for TCMT and spare ones - overall amount ~ 9000 pcs PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

3. We have chosen for prototype construction novel solid state photo detectors – Silicon PMs. This is a matrix of tiny photo diodes –”pixels” working in Geiger mode and connected to a common bus. Because of big amount of pixels the output signal is equal to the sum of standard signals of individual pixels and is proportional to the number of photons impinging the SiPM area. The recent test of “minical” – prototype with 11 planes (99 tiles read out with silicon photomultipliers) at DESY positron beam in 1-6 GeV/c momentum range showed that SiPM is an adequate photo-detector for the calorimetry. The measured response linearity, energy resolution, longitudinal and transversal shower distributions are in good agreement with Monte Carlo calculations. • Main features of SiPM’s • like ordinary vacuum phototubes they have high gain and PDE • high gain ~106 • PDE ~10-20% in green part of spectrum • they are not sensetive to magnetic field • SiPM operation voltage is low (30-80 V) • match to WLS fiber readout by size and spectral sensitivity • small size ~10 mm3 • Some drawbacks: • high noise – may reach ~MHz at 1/2 pe level • limited dynamic range ~1000 pe • inter pixel cross talk PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

4. We have been using for 1 m3 prototype SiPM’s manufactured by MEPhI-PULSAR collaboration These devices have 34x34 pixel matrix at 1.12 mm2 area Thanks to extremely small size of a photo-detector we have it incorporated in a scintillator without noticeable loss of efficiency. PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

5. The flow chart of photodetector selection • As SiPM is a non linear device it is very convenient to have all photo detectors equalized by response expressed in pixels. We have chosen value for response to MIP to be equal 15 pixels. • This choice is a compromise between requirements to have high (>95%) efficiency to MIP at ~ 1/2 MIP threshold and from other hand requirement to have dynamic range as wide as possible. • Tune Vbias for each SiPM to have response of 15 pixels • At chosen voltage measure main SiPM parameters: • - Gain (G), • - Cross talk (xt) • - Noise at 1/2 pe level (F0) • - Current and its stability (I and RMSI) • - Noise at the ½ MIP level (F1/2MIP) • - Response curve in the ~0.3 - 200 MIP range PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

6. 3. Apply selection criteria - G > 4*105 in 140 ns gate -> 1MIP ~ 1pC - F0 < 3 MHz - I < 2 μA - RMSI < 20 nA - F1/2MIP < 3 kHz - xt < 0.35 - Response(light~200 MIPs) > 900 pixels • - Minimal gain is chosen from requirement to eliminate noise of FE electronics above ½ MIP threshold. • - 3 kHz limit at F1/2MIP – number of noise hits in an event is of order of 1 per 8000 channels • - Limits at xt and F0 correspond to limits at F1/2MIP, requirements of high MIP efficiency and wide dynamic range 4. Keep SiPM’s at elevated (+3.5 V) voltage during at least 40 hours. This allows to detect those devices which due to technological defects have discharge between the common bus and pixel area. 5. Repeat measurement of SiPM parameters to confirm that SiPM parameters have not been changed. PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

7. Test bench for SiPM parameter measurement Set up is realized in CAMAC standard It includes: 16 channel computer driven power supply to feed SiPM’s - 5 mV resolution - 110 V max - 100 μA maximal output current 16 channel computer read-out digital voltmeter to monitor - SiPM bias voltage - SiPM current - temperature during test measurement accuracy voltage – 5 mV current – 5 nA temperature – 0.2O 16 channel 12 bit ADC 0.25pC/count sensitivity PC driven generator to produce LED and random triggers and ignite LED PMT to monitor LED light 15 SiPMs can be tested simultaniously Measurements are done at 2 kHz trigger rate LED driver PC driven generator DATA BASE Digital voltmeter Remote control 16 channel power supply Steering program ……. 16 ch 12 bit ADC gate X~100 PMT 16 ch amp Tested SiPMs 16 ch 12 bit ADC A software package was developed to make easy interface between user and hardware, to perform measurements and to save results in data base PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

8. Response to MIP, pixel Bias voltage, V Extraction of SiPM parameters from LED and random spectra F0 and F1/2MIP are extracted from random trigger spectrum F0 = F1/2MIP = -ln(N0/Ntot) N>1/2MIP Tgate Ntot*T gate Fit of LED spectrum with Poisson distribution distorted by cross talk gives values of G, xt, Npe Choice of operation Vbias For each value of bias voltage the number of pixels per MIP is calculated and operating bias voltage is found from the fit of response points by a power function pix/MIP = A * ( V - Vbd )B and requirement Npix/MIP=15 At chosen bias voltage the saturation curve of a SiPM is measured in light range ~0.3-200 MIPs as well as SiPM current and its stability PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

9. Result of selection Rejected : Gain – 2.8% Noise at ½ pixel – 5.5% Noise at ½ MIP – 22.6% Cross talk – 3.5% Current – 0.5% Current RMS – 1.4% Yield of good SiPM’s > 70% SiPM parameter distributions PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

10. Accuracy of parameter measurement Multiple measurement of the same SiPMs in the same test bench channels allows to estimate statistical errors in determination of parameters which are subject to selection cuts: Operational bias voltage ~0.1% Number of pixels per MIP ~1.5% Noise ~2% Gain ~ 2% SiPM current ~2.5% Cross talk ~3% Noise at ½ MIP level ~20% for limit F0 the accuracy is higher – 1./√(3*103*2*10-7*3.5*105)=7% R M S / m e a n , % C h a n n e l n u m b e r PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

11. From measurements of parameters at various bias voltages it is possible to derive value of of parameter variation at 100 mV Vbias variation One has to know what stability of bias voltage and temperature has to be kept in order to have SiPM parameters stable at necessary level, for example in order to have gain stable at 1% level one needs to keep bias voltage with accuracy 30 mV. PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

12. Deviation of operational bias V from fit RMS= 0.052 V Temperature factor of Vb mean=0.057 V/oC RMS=0.008 V/oC Long term stability of measurements Measurements of SiPM parameters and determination of operational bias voltage have been repeated many times (~100) during almost 2 years with set of 15 SiPM’s. Analysis of these data shows the stability of the set up and gives opportunity to study the temperature dependence of main SiPM parameters as temperature was not stable during measurements. Operational bias V vs temperature PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

13. mean=12.9oC RMS=0.9oC F0 temperature factor, oC Temperature dependence of SiPM noise (F0) F0 = A * exp( T / T0 ) T0 = (12.9 ± 0.2)oC PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

14. Temperature dependence of SiPM noise at ½ MIP (F1/2MIP) mean=6.1oC RMS=0.6oC F1/2MIP temperature factor, oC F1/2MIP = B * exp( T/ T1/2MIP ) T1/2MIP = (6.1 ± 0.2)oC PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

15. Comparison of parameters of multi pixel Geiger mode photo-detectors New photodetectors manufactured by Hamamatsu Corp. recently became available. We have done the comparative test of such MPPCs with Russian MRS APD from CPTA enterprise with ~550 pixels/mm2 and SiPMs fromMEPhI-PULSAR collaboration. Designed to work in blue region of spectrum MPPC have 2-3 times more PDE in blue light. MRS APD with 2x2mm2 areahas PDE close to MPPC. At green light PDE of MPPC and MRS APD are close to each other (15-25%). Gain of 1600 pixel MPPC is more than twice less than gain of SiPM and MRS APD. MPPC1600 and MRS APD have rather low cross talk – less than 0.2 MPPC400 has larger efficiency and gain but larger cross talk PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

16. Comparison of noise at level ½ pixel shows that noise frequency of single samples of 1600 pixel and 400 pixel MPPCs is smaller by at least one order of magnitude. PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

17. Test of radiation hardness of Geiger mode multi pixel photon detectors with Co60. • Dose rate ~330 kRad/hour • energy – continuous spectrum up to 1.33 MeV All devices except one are operational after 800krad but with increased dark current and deteriorated single photoelectron resolution For most of tested devices the increase of current is seen to be proportional to accumulated dose: ΔI ~(1-3)μA / 500 kRad But 1600 pixel MPPC shows several times worse proportionality: current increased up to 14 μA after the first 200 kRad and up to 60 μA after the second 200 kRad PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

18. Current of 1600 pixel MPPC (after 200 kRad+200 kRad) vs annealing time 200 kRad 200 kRad 1600 pixel MPPC shows big annealing after doses of 200 kRad, the current is not constant even after several hours after bias voltage is on PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

19. After the first irradiation pedestal is very broad and individual pixels are not seen Situation improves with time and is better for smaller Vbias Only one sample was tested and it is too early to make conclusions More studies are necessary with reasonable dose rate PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

20. Conclusions • We have tested and made selection of more than 9000 SiPMs for prototypes of • hadron calorimeter and tail catcher for future linear collider experiments • The computerized LED test bench for SiPM test and selection was designed, • constructed and used for measurements at ITEP • Parameters of selected SiPMs meet requirements from the physical performance • Variation of SiPM parameters at variable bias voltage and temperature have been • measured • Comparison of various multi pixel Geiger mode photo detectors has been done • This work was supported by ISTC grant # 3090 PD07 workshop University of Kobe June 27-29, 2007 E.Tarkovsky, ITEP, Moscow

21. Backup slides