SILICON PHOTOMULTIPLIERS: FROM 0 TO 10000 IN 1 NANOSECOND Giovanni Ludovico Montagnani Giovanniludovico.montagnani@ polimi.it
LESSON OVERVIEW 1. Motivations: why SiPM are useful 2. SiPM applications examples 3. SiPM working principle 4. SiPM parameters 5. SiPM readout strategies
SINGLE PHOTON RESOLUTION: THE HUMAN EYE Human eye can respond to a small number of photons. In very dim light conditions ( for example star gazing ) it can recognize light sources providing to the eye a continuous flux of 8-14 photons summing the signal coming from the rods that are sensible to a single photon. 5000 photons The rods are very low noise but slow detectors compared to the electronics standard (60 Hz bandwith) So how things change if the signal to be detected is a very short pulse? THEY DO NOT CHANGE! 14 photons Human eye and standard imaging detectors can be sensible to very few photons, but they are slow!
SINGLE PHOTON RESOLUTION: THE SPAD The Single Photon Avalanche Diode is an Avalanche Photodiode polarized well above breakdown, once a photons hits the active area, a self substained avalanche is triggered in picosends. The signal provided by the avalanche is well above the noise of the comparator, so the measure is not affected by series noise. Compared to the human eye this device provides a very small dynamic range (1), but a single photon can be easily detected! A SPAD can detect just one photon at the time!
SINGLE PHOTON RESOLUTION: THE SIPM The Silicon Photomultiplier is composed like a matrix of SPAD, where the analog output of each one is summed together. It is like the human retina but composed by SPADs instead of rods. It provides a fast time response, but a wide dynamic range because different photons hit different cells. Photon 3 Photon 2 Photon 1 Photon 4 I TOT =I 1 + I 2 + I 3 =3I I1 I2 I3
SINGLE PHOTON RESOLUTION: THE SIPM These devices are called in different ways: SiPM: Silicon Photo Multipliers MPPC: Multi-Photon Pixel Counter Si-SSPMT: Silicon Solid State PMT Direct challenge with the Photo-Multiplier Tubes (SiPMs are more compact and MR compatible) SiPM: detector for radiation in the light wavelength range. It is composed by a grid of Gaiger Mode (GM) Avalanche PhotoDiodes (APD)
SIPM MATRICES FOR IMAGING Arrays of SiPM for Imaging applications Compact alignment of SiPM Reduced dead area Common bias strategies Tilable on 3 or 4 sides (Possibility to create larger matrices by combining single units) 4 x 4 matrix - AdvanSid 8 x 8 matrix Hamamtsu 12 x 12 matrix SensL (1287 $)
SIPM APPLICATIONS: MEDICAL IMAGING PET/SPECT Tc-99m: bone (140 kev) I-123: thyroid (159 kev) Tl-201: heart (135 and 167 kev) Gamma Medica Tri-Isotope SPECT with CT Energy resolution is needed to acquire different sources
SIPM APPLICATIONS: MEDICAL IMAGING PET/SPECT Fundamental parameters: 1 Spatial resolution 2 Energy resolution (multiple tracers) 3 Others (sensitivity, count rate, etc)
SIPM APPLICATIONS: INTRAOPERATIVE PROBES Fundamental parameters: 1 Spatial resolution 2 Compactness H.Sabat et al., A Hand-Held, Intra-Operative Positron Imaging Probe for Surgical Applications (2015)
SIPM APPLICATIONS: NUCLEAR PHYSICS Fastest and brightest scintillators like Lanthnium Bromide are the most suitable for high resolution energy spectroscopy. Due to their properties SiPM could replace standard PMTs maintaining the readout specifications. Fundamental parameters: 1 Spatial resolution 2 Low DCR and high PDE 3 Magnetic Field Compatibility
GM-APD Signal Shape Avalanche ON/OFF Passive quenching was studied in details in the 60 by McIntyre and Haitz. The GM-APD can be modeled with an equivalent electrical circuit. The model is not entirely valid for SiPM The avalanche charge/discharge can be modelled with an ideal switch GM-APD V BD R S C D R Q V BIAS C D APD capacitance R Q Quenching resistor R S APD series resistance V BD Breakdown voltage V BIAS Applied voltage Some values: C D ~800pF R Q ~300kΩ (poly Si) R S ~1kΩ V BIAS -V BD few V
Output signal R Q I I L V BD C D V BIAS R S
GM-APD Signal Shape τ dd = RR SS CC DD = 1kkΩ 10ffff = 100pppp τ QQ = RR QQ CC DD = 300kkΩ 10ffff = 3nnnn II LLLLLLLLLL = VV BBBBBBBB VV BBBB = 3 VV RR QQ + RR SS 300kkΩ = 10µAA VV OOOO = VV BBBBBBBB VV BBBB is usually called overvoltage
GM-APD Gain R Q compromise between Recovery time and P 10 MM = QQ qq ~ II LLLLLLLLLL τ QQ qq Charge collected per event is the area of the exponential decay which is determined by circuital elements and bias. = VV OOOO CC DD qq = 10µAA 3nnnn 1.6 10 19 CC = 1.875 105 VV OOOO After the recovery time, the GM-APD can detect another photon.
GM-APD Gain MM~ VV OOOO CC DD qq The multiplication factor M (or gain G) is proportional to the applied voltage. The gain increases reducing temperature because electron/hole mobility is higher at lower temperatures. Impact ionization is more effective.
SiPM Signal Shape The real output current is more complex, because many GM-APD are connected in parallel. The model should include: the equivalent circuit of the other microcells the metal grid capacitance related to the parallel short-circuit Connections of the device to the front-end Front-end electronics input impedence In any case the concept is still valid: a fast rise time and a slow (with one or two components) slow decay.
Photodetection efficiency The photodetection efficiency (PDE) of a SiPM is the product of 3 factors: 1. the geometrical efficiency (Fill Factor) 2. the quantum efficiency (QE) 3. the turn-on probability (P 01 ) 1. Geometrical efficiency (FF) The ratio between the active area and the total device area is a critical issue in SiPMs. Each GM-APD cell is surrounded by a dead region determined by the guard ring and the structure preventing optical cross-talk. (DOMINANT CONTRIBUTION!) PPPPPP = FFFF QQQQ PP 0000 FF: depends on the size of the cell. Value: 50 µm cells, 45% fill-factor
Photodetection efficiency 2. Quantum efficiency (QE) wavelength dependent The quantum efficiency (QE) represents the probability for a photon to generate an e/h pair in the active thickness of the device. It is given by the product of 2 factors: the transmittance of the dielectric layer on top of the silicon surface the internal QE Both are wavelength dependent Wavelength of interest? Visible Near-UV PPPPPP = FFFF QQQQ PP 0000 QE limitation Shallow junction form short λ Thick epi-layer for long λ
Photodetection efficiency Turn-on probability (P 01 ) wavelength dependent Is the finite probability for a carrier to initiate an avalanche when passing through a high-field region. In case of a photogeneration event, 2 carriers are created travelling in opposite directions. Both contribute to the triggering probability that can be evaluated from the following: Pt =Pe +Ph PePh P(A B)=P(A)+P(B)-P(A B) where Pe and Ph are the electron and hole breakdown initiation probabilities Example with constant high-field: (a) only holes may trigger the avalanche (b) Both electrons and holes may trigger (c) only electrons may trigger PPPPPP = FFFF QQQQ PP 0000
Photodetection efficiency Fill factor
NOISE IN SIPMS Noise in SiPMs is given by: Primary source: dark count pulses triggered by non-photo-generated carriers (thermal / tunneling generation in the bulk or in the surface depleted region around the junction) Secondary sources: After pulse carriers can be trapped during an avalanche and then released triggering another avalanche Cross talk Photo-generation during the avalanche discharge. Some of the photons can be absorbed in the adjacent cell possibly triggering new discharges
DARK COUNT RATE This noise is measured in khz/mm 2 (e.g. 30-500) It increases with OV(higher probability to start an avalanche) and decreases with T (less thermal energy to break the bond). Counteract: Quality of epi-silicon
Dynamic Range Ph 2 Ph 3 Ph 4 Ph 1 The output signal is proportional to the number of of fired cells as long as the number of photons in a pulse (Nph) times the PDE is significant smaller than the number of cells Ncell. Hp: photons are emitted all at the same time, so each single cell has no time to detect another photon. Scintillation decay time << recovery time Equation is not exact but describes the data very well. NN ffffffffff = NN tttttttttt 1 ee NN ppppppppppppp PPPPPP NN tttttttttt
CROSSTALK Light is produced during cell discharge. Effect is known as a hot-carrier luminescence: 10 5 carriers produce ~3 photons with a wavelength less than 1 μm. Photons can induce avalanches in neighboring cells. Counteract: Optical isolation between cells by trenches filled with an opaque material Low voltage operation
PARAMETERS SUMMARY SiPM parameters: Breakdown voltage [V] - VV BBBB (TT) affects Gain, GG~ (VV bbbbbbbb VV BBBB ) CC DD qq Micro-cells (APDs) dimension [μm] affects CC DD (Gain and timing) and dynamic range, but also FF (then PDE) Photodetection efficiency [%] - PPPPPP = FFFF QQQQ(λ) PP 0000 (λ, T, VV OOOO ) affects the SNR Dark Count Rate [Hz/mm2] - DDDDDD(TT, VV OOOO ) affects the SNR (energy resolution and spatial resolution) System parameters: Temperature (T): needs to be stable because of the high GG/GG in SiPMs Wavelength (λλ): depends on application. Scintillation crystals emissions spectrum has to match well with the SiPM absorption Overvoltage (VV OOOO ): optimal SNR is generally reached by setting the VV OOOO as suggested by the vendor
IMPACT OF PARAMETERS ON APPLICATIONS Imaging and spectroscopy optimization (i.e. SPECT and PET): PDE(λλ, TT, VV OOOO ) and DCR(TT, VV OOOO ) must be optimized to have the best SNR which directly impacts on the spatial resolution and energy resolution performance Depending on the energy range and scintillation time constant, the dynamic range has to be optimized by choosing an appropriate dimension of the micro-cell Very bright scintillator are implemented for these applications timing optimization (i.e TOF PET and ranging for automotive): SiPMs with low Cd are preferred (less interest in high FF) Fast readout electronics is mandatory to read and sample the SiPM signal PDE(λλ, TT, VV OOOO ) and DCR(TT, VV OOOO ) are still really important to maximize the SNR, reduce time jitter and optimize timimg resolution Very fast scintillator are implemented for these applications
SIPM READOUT STRATEGIES: VOLTAGE INPUT I-V conversion is realized by mean of Rsense R sense affects the gain and the signal waveform The charge could be integrated directly on C detector http://www.weeroc.com/en/products/petiroc-2
SIPM READOUT STRATEGIES: CURRENT INPUT I-V conversion by means of the filtering stage. R input is small, the circuit is inherently fast. The current can be replicated through the use of current mirrors. Less problems of dynamic range. http://ideas.no/products/ide3380/
SIPM READOUT STRATEGIES: DIGITAL INPUT Cost effective and low power front end. I V conversion by means of Sigma Delta modulation. Clock period must be a lot faster than the pulse period. Successful results in terms of charge readout but with Dynamic range limitations http://ieeexplore.ieee.org/document/7807257/
CONCLUSIONS SiPM are a cost effective solution if fast and dim light signals are to be read. Dynamic range is larger than SPADS and they provide faster response than APS or CCD. They can be tiled in large area arrays with high fill factor. Different techniques are used for the readout, being a novel technology there is still space for smart ideas!