SiPM's a very brief review School of Physics & Nepomuk Otte - PowerPoint PPT Presentation

SiPM's a very brief review School of Physics & Nepomuk Otte Center for Relativistic Astrophysics Georgia Institute of Technology

The SiPM MEPhI/Pulsar SiPM 2004 The SiPM concept provides multi-photon resolution: Many passively quenched SPADs are connected in parallel Recover information about number of photons if photons per cell per recovery time <1 Pioneered in the 90's Key persons: Dolgoshein, Golovin, and Sadykov For an extensive review on the history of solid state photon detectors see D. Renker and E. Lorentz (2009) 4 Nepomuk Otte

SiPM with Active Quenching: dSiPM First commercial dSiPM from Philips Individual pixels can be turned on/off Excellent timing → Reduced geometrical efficiency lower PDE (for now...) 5 Nepomuk Otte

Dynamic Range for Light Flashes 20% deviation from linearity if 50% cells fire Andreev et al. (2005) Build-in logarithmic compression Need to pick device with cell density that meets requirements of application Rule of thumb for picking a device: photons per cell <1 Compromise between cell density and geometrical efficiency 6 Nepomuk Otte

SiPMs to detect steady Very-Low-Light Levels Adamo et al. (2013) multiple hits per cell 633nm photons ST Microelectronics 3.5 x 3.5 mm 2 Sensitive to photocurrents of ~10 -15 A Linear regime: Acceptable photon rate for linear response << 1 photon / cell / recharge time 7 Nepomuk Otte

SiPM Advantages and Nuisances Mechanical robust What's being worked on Compact Operating voltages < 100V Radiation hardness Not damaged in bright light Better UV sensitivity No aging Lower optical crosstalk Insensitive to magnetic fields Lower dark rates Excellent SNR Size Excellent single photon timing (<100 ps) Very high photon detection efficiency A near perfect device for many applications 8 Nepomuk Otte

SiPM Applications Cherenkov Telescopes PET calorimeters High Astroparticle Physics Energy RICH physics Fluorescence SiPMs Neutrino telescopes tracker detectors Homeland Security Direct Dark Matter Detection Discussion shifts from device features to how they can be best implemented 10 Nepomuk Otte

You have Choices Number of producers increases from W. Ootani Interactions between producers and users are very productive! 11 Nepomuk Otte

SiPM Parameters User's perspective Nuissance Parameters Photon Detection Overvoltage Gain Efficiency Temperature Afterpulsing Optical Effective Dark Rate Crosstalk 13 Nepomuk Otte

Photon Detection Efficiency = geometrical efficiency * (1-reflection losses) * QE * breakdown probability 14 Nepomuk Otte

Geometrical Efficiency: intra-cell spacing HD 30 µ m 25 µ m 42μm FBK 20 µ m 15 µ m 12 µ m RGB/NUV 2004 MEPhI Hamamatsu 50μm 2014 Hamamatsu 50% to 100% improvements depending on cell size 15 Nepomuk Otte

Geometrical Efficiency: Minimizing Dead Space between SiPMs 3mm 3mm 3mm Hamamatsu SensL Hamamatsu 2008 Elimination of bond wires with through silicon vias thinner guard ring around device Chip packaging with much reduced gaps between chips → >90% efficiency 0.1 to 0.2 mm gap possible between chips The pragmatic and cost-effective approach to arrive at large sensor sizes 16 Nepomuk Otte

Transparent quench resistors Metal film resistors 10μm cells ~30% fill factor Hamamatsu Allows much higher cell densities 17 Nepomuk Otte

Quantum Efficiency Probability that a photon gets absorbed in the device AND that the electron or hole makes it into the avalanche region For UV sensitivity (~100nm absorption length) : Thin entrance window and shallow first implant For Red/IR (>1 μm absorption length): thick depletion region Insensitive surface layer Active volume Non-depleted bulk P abs (x,l abs )=1-e -x/l_abs 21 Nepomuk Otte

Breakdown Probability vs. Bias Pancheri et al (2014) p-on-n structures needed for UV sensitivity → electron initiated breakdown 23 Nepomuk Otte

Parameterization of Breakdown Probability α U Break ] PDE ( U )= PDE max ⋅ [ 1 − e −( U − U Break ) This is a perfect fit of the data!! Saturated regime Breakdown probability > 90% Three free parameters: ● Maximum PDE ● Breakdown voltage ● Constant α Statistical errors on data points are 0.6% All the physics of the breakdown probability is in α 26 Nepomuk Otte

Different Devices and Wavelengths α U Break ] PDE ( U )= PDE max ⋅ [ 1 − e −( U − U Break ) α =0.04-0.06 90% α =0.10 α =0.15 To compare devices Plot breakdown prob. vs. Relative overvoltage x − x Breakdown Probability ( x )= 1 − e α Relative overvoltage = relative electric field above critical field α is the only free parameter Quite different α for the two devices and wavelengths, what is the difference? 27 Nepomuk Otte

Interpretation of alpha α=0.045 α=0.027 α=0.19 α=0.19 Pancheri et al (2014) α ~ 0.03-0.05 pure electron injected α ~ 0.2 pure hole injected Looks like α does not strongly depend on technology → α can be used to reverse engineer avalanche structure :) 28 Nepomuk Otte

α the Otte number ;) Interpretation of alpha α=0.045 α=0.027 α=0.19 α=0.19 Pancheri et al (2014) α ~ 0.03-0.05 pure electron injected α ~ 0.2 pure hole injected Looks like α does not strongly depend on technology → α can be used to reverse engineer avalanche structure :) 29 Nepomuk Otte

PDE: Spectral Response Breakdown probability and QE Are both functions of wavelength → Both determine the spectral response FWHM SensL Hole dominated breakdown FWHM Electron dominated breakdown probability Bonnano et al. (2015) submitted 31 Nepomuk Otte

Gain Dependence on Temperature Breakdown voltage changes typically between 20-40 mV/°C Early devices For 1 Volt overvoltage → 2% - 4% gain change per °C ( less change for PDE ) Present generation can operate at much higher voltages For 5 Volt overvoltage → 0.4% - 0.8% gain change per °C ( less change for PDE ) For 10 Volt overvoltage → 0.2% - 0.4% gain change per °C ( less change for PDE ) Compare to 0.1 % to 0.2 % change in QE per °C for PMTs (Burle/Hamamatsu photomultiplier handbook) 35 Nepomuk Otte

Gain and Temperature SiPMs are considered low power devices But operation in high background environments can dramatically increase temperature → Temperature management can become a problem and needs dedicated application specific solutions Where are devices with small Adamo et al. 2013 effective cell capacitances? 38 Nepomuk Otte

Optical Crosstalk Photons are emitted during breakdown Photon emission mechanism not well understood Photons with λ = 900nm – 1100nm have the right absorption length to produce optical crosstalk ~3·10 -5 photons per charge carrier in the breakdown C. Merck Direct optical crosstalk Instantaneous <<1ns → pile up of signals Indirect optical crosstalk Delayed 10 - 100 ns → contribution to afterpulsing and effective dark rate 40 Nepomuk Otte

Direct Optical Crosstalk Most vendors do produce SiPMs with trenches or implement structures to reduce optical crosstalk Hamamatsu 9V FBK Direct cross-talk Hamamatsu Optical crosstalk of a few 4V percent now achievable Delayed even for high overvoltages correlated 2V noise 42 Nepomuk Otte

SensL ES 30035 TSV Array A model to fit optical crosstalk vs. bias voltage 1-exp[-(U-U break )/(U break *α)] * * ΔG/ΔU*(U-U break )*ε OC transmission Breakdown probability Photons produced Optical crosstalk during breakdown transmission factor α=0.31±0.07 ε=3*10 -5 photons/charge carrier OC transmission =0.48 Pure hole injected 43 Nepomuk Otte

Afterpulsing Two contributions Delayed release of trapped charge carrier → breakdown of the same cell proportional to gain (ΔU) (filling traps) and breakdown probability (1-exp(-ΔU(t)/A)) Cova 2003 (detecting released trapped carriers) Solution: improvements in technology Delayed optical crosstalk photons → breakdown of a neighboring cell Solution: potential barrier between epitaxial layer and bulk 45 Nepomuk Otte

Slide from Hamamatsu 46 Nepomuk Otte

Effective Dark Rates Contributions 1. thermal generated 2. tunneling 3. afterpulsing 48 Nepomuk Otte

Dark Rate FBK Measurements NUV-HD 30 μm cells at Room Temp. Sub 100 kHz/mm 2 is the new standard Sub 50 kHz/mm 2 standard in reach Achieved already by SensL, Hamamatsu, ... Cattaneo et al. (2014) 49 Nepomuk Otte

Summary of Key SiPM Parameters Parameter 2005 Now Wish List Spectral Response Green Blue and Tailored to application Sensitive Green n-on-p p-on-n structure structure Photon Detection Efficiency ~10% ~45% >70% Dark Noise 1MHz/mm 2 <100kHz/mm 2 As low as possible Optical Crosstalk >20% <10% As low as possible Afterpulsing >20% <1% As low as possible Sensor Size 1mm 2 1mm 2 -36mm 2 SiPMs are ready for prime time due to rapid improvements in the past 10 years 50 Nepomuk Otte

What else is new out there? 51 Nepomuk Otte

Fast SiPM Signals SensL development Tapping the signal between the quench resistor and diode 52 Nepomuk Otte

FBK: Linearly-graded SiPM (LG-SiPM) Flood map T = 25 o C SiPM with integrated charge division readout → X-Y resolution