SiPM for CBM Michael Danilov ITEP(Moscow) Muon Detector and/or Preshower CBM Meeting ITEP SiPM main characteristics R 50 h pixel U bias Al Depletion Region 2 m Substrate Resistor R n =400 k 20 m 42 m Pixel size ~20-30 m Electrical inter-pixel cross-talk minimized by: - decoupling quenching resistor for each pixel - boundaries between pixels to decouple them reduction of sensitive area and geometrical efficiency Optical inter-pixel cross -talk: -due to photons from Geiger discharge initiated by one electron and collected on adjacent pixel Working point: V Bias = V breakdown + V ~ V V ~ 3V above breakdown voltage Each pixel behaves as a Geiger counter with Q pixel = V C pixel with C pixel ~50fmF Q pixel ~150fmC= 10 6 e Dynamic range ~ number of pixels (1024) saturation SiPM Spectral Efficiency Depletion region is very small ~ 2 m strong electric field (2-3) 10 5 V/cm carrier drift velocity ~ 10 7 cm/s very short Geiger discharge development < 500 ps pixel recovery time = (C pixel R pixel ) ~ 20 ns Photon detection efficiency (PDE): - for SiPM the QE (~90%) is multiplied by Geiger efficiency (~60%) and by geometrical efficiency (sensitive/total area ~30%) - highest efficiency for green light important when using with WLS fibers Temperature and voltage dependence: -1 o C +4.2% in Gain * PDE*Xtalk +0.1 V +4.5% in Gain * PDE*Xtalk WLS fiber emission Photon detection efficiency = QE geom U breakdown =48V operating voltage one pixel gain (exp. data) one pixel gain (linear fit) One pixel gain M, 10 5 Overvoltage U=U-U breakdown, V detection efficiency ( =565nm) Efficiencyof light registration , % ,pixels : SiPM signal saturation due to finite number of SiPM pixels Very fast pixel recovery time ~ 20ns For large signals each pixel fires about 2 times during pulse from tile Average number of photoelectrons for central tiles for 6 GeV SiPM Noise random trigger 1p.e. 2p.e. 3p.e. Ped. noise rate vs. threshold Optimization of operating voltage is subject of R&D at the moment. 1p.e. noise rate ~2MHz. threshold 3.5p.e. ~10kHz threshold 6p.e. ~1kHz Experience with a small (108ch) prototype (MINICAL) Tile with SiPM cassette 3x3 tiles SiPM Moscow Hamburg Light Yield from Minical tiles (5x5x0.5cm 3) Using triggered Sr source and LED at ITEP Using electron beam at DESY (SiPM signals without amplification) Good reproducibility after transportation from Moscow to Hamburg N pixel LED LED and 90 Sr signals Individual photoelectrons are clearly seen Light Collection Uniformity, Y11 MC 1mm fiber, Vladimir Scintillator, mated sides, 3M foil on top and bottom Reduction in light yield near tile edges is due to finite size of a source Sufficient uniformity for a hadron calorimeter even for large tiles Acceptable cross-talk between tiles of ~2% per side Sufficient light yield of 17, 28, 21 pixels/mip for 12x12, 6x6, and 3x3 cm 2 tiles (quarter of a circle fiber in case of 3x3 cm 2 tile) 5x5x0,5cm 3 Light yield drop between tiles acceptable (Calorimeter geometry is not projective) Cross-talk between tiles ~2% - acceptable Light collection efficiency mm Shower Shape After single tile calibration and smearing MC describes well shower shape 3 GeV e + Black data Yellow -MC NON-LINEARITY and ENERGY RESOLUTION for DIFFERENT BEAM POSITIONS Beam position SiPM For all cases the same fit function is used ! Energy resolution With universal SiPM calibration curve there is no differences in response for different beam positions (different SiPM saturation) Cassette, Absorber and Support Structure Cassette contains 220 tiles with SiPM and electronics There are 40 cassettes between 2cm iron absorber Altogether about 8000 tiles of 3x3, 6x6 and 12x12 cm 2 with SiPM and individual readout Scan of Strip Using Cosmics Setup LEDcosmics Poiss XtalkG(x 0 +ix, 0 + 1 i) Center of strip, N pixels (peak) = 9.7 groove depth 2.5mm Scan of Strip Using Cosmics Setup (2) Poisson mean for MIP at normal incidence Strip 200x2.5x1cm 3 MC: light yield for 4x1cm 2 strip will be ~25% smaller. In the middle of 280x4x1cm 3 strip we expect ~ 9p.e. With threshold 4.5p.e. noise rate 99% Similar detector - MRS APD - with smaller x-talk and similar efficiency is under investigation now. Expect efficiency >99.5% with noise