Introducing the CAEN Silicon Photomultiplier Kit: a flexible, modular system for sensor testing & education

Transcription

1 Massimo Caccia Universita dell Como massimo.caccia@uninsubria.it On behalf of Introducing the CAEN Silicon Photomultiplier Kit: a flexible, modular system for sensor testing & education NDIP July 3, 2014

2 An overview of the kit

3 RAPSODI RAdiation Protection with Silicon Optoelectronic Devices and Instruments!!!! Funded by the EC under the Sixth Framework Program (Co-operative research) Start-time Oct 2006; End-time: Jan 2009 Main objectives: Silicon Photo Multipliers development and optimization for three well defined applications: Dosimetry in Mammography, Radon Monitoring, illicit traffic of radioactive material (homeland security) Consortium composition: 4 Small and Medium Enterprises + 3 R&D performers SensL (IE) PTW (DE) Plch SMM (CZ) ForimTech (CH) UNICO (IT) (Leading organization) AGH (PL) ITEP (RU)

4 A little bit of history: RAPSODI. Prototypes of an easy-to-use, flexible, modular kit for the characterization of SiPM 2009 Licensing of the background knowledge to CAEN. First announcement at the IEEE NSS, Orlando Establishment of a Joint Development Laboratory to finalize the development of the kit and its applications. Kit on the move at IEEE NSS at Knoxville 2011 Early development of the educational project 2013 First Educational Applications completed

5 The building blocks of the kit: 1. The SP5600: a Power Supply and amplification Unit (with a bit of logic on board) " 2 channel mother & daughter architecture " every channel features: - Independent biasing (max 120 V, 100 μa - 2 stage amplification [500 MHz bandwidth, tunable gain up to ~ 50 db] - discriminator ( ±2V) " active feedback control on V bias for Gain stabilization (granularity: 0.1 o C) " coincidence logic A 3 plot qualification:

6 The building blocks of the kit: 1. The SP5600: a Power Supply and amplification Unit (with a bit of logic on board) " 2 channel mother & daughter architecture " every channel features: - Independent biasing (max 120 V, 100 μa - 2 stage amplification [500 MHz bandwidth, tunable gain up to ~ 50 db] - discriminator ( ±2V) " active feedback control on V bias for Gain stabilization (granularity: 0.1 o C) " coincidence logic A 3 plot qualification:

7 The building blocks of the kit: 1. The SP5600: a Power Supply and amplification Unit (with a bit of logic on board) " 2 channel mother & daughter architecture " every channel features: - Independent biasing (max 120 V, 100 μa - 2 stage amplification [500 MHz bandwidth, tunable gain up to ~ 50 db] - discriminator ( ±2V) " active feedback control on V bias for Gain stabilization (granularity: 0.1 o C) " coincidence logic A 3 plot qualification:

8 2. Signal recording: QDC vs Digitization The V792N QDC The 720 desktop Digitizer # 16 channels # VME 6U format # 12 bits # 400 pc range (~2000 Gain 1) # granularity: 100 fc/ count # 2 channels # stand-alone # 250 Ms/s, 12 bits (up to 5 Gs/s) # ±1V input range Featuring the Digital Pulse processor

9 3. The FAST LED, an essential tool for sensor testing Reference LED: " λ peak = 420 nm " peak current 120 ma " luminous intensity = 9500 " 30 o half-view angle τ ~ 5 ns Single Photon Timing spectrum

10 4.The Cosmic (ray) Tile " 150 x 150 x 10 mm 3 plastic scintillator tile " wls fiber => 2 channels in coincidence Single channel Dark Count Rate $ > 0.5 ph Threshold scan 0.5 ph 1.5 ph 2.5 ph > 1.5 ph > 2.5 ph Count rate in coincidence %

11 5.The Gamma Ray Spectrometer Two basic configurations, oriented to EduApplications: " 6 x 6 mm 2 SiPM " 1 CsI crystal, 6 x 6 x 30 mm 3 " 3 x 3 mm 2 SiPM " 3 crystals 3 x 3 x 15 mm 3 [LYSO, BGO, CsI] no source & 30 KeV line from Ba decay 137 Cs [662KeV] 60 Co [1.17MeV] 60 Co [1.33MeV] FWHM [%]

12 5.The Gamma Ray Spectrometer Two basic configurations, oriented to EduApplications: " 6 x 6 mm 2 SiPM " 1 CsI crystal, 6 x 6 x 30 mm 3 " 3 x 3 mm 2 SiPM " 3 crystals 3 x 3 x 15 mm 3 [LYSO, BGO, CsI] no source & 30 KeV line from Ba decay 137 Cs [662KeV] 60 Co [1.17MeV] 60 Co [1.33MeV] FWHM [%]

13 The Educational project The SiPM kit offers the possibility to perform a series of experiments well suited to undergraduates (& possibly beyond): 1. Hands-on photon counting statistics (it can be introduced and performed at different levels, from an introduction to stochastic processes for [bright] high school students to advanced data analysis aimed for doctoral students) [completed][arxiv: [physics.ins-det]] 2. γspectrometry (with a series of classical small experiments) [completed] 3. A simple method for measuring after-pulsing [completed][arxiv: [physics.ins-det]] 4. Measuring the maximum counting frequency in a SiPM based system for Poissonian distributed events [completed; note in preparation] 5. Hands-on Poissonian processes (counting statistics & time domain) [next in line]

14 More experiments in the pipeline: 6. Cosmic ray experiments [proof of concept] 7. Introduction to PET and TOF-PET [tbc] 8. Beta gauging of thin layers 9. Exemplary illustration of NonDestructiveTesting [requires an X-ray tube] and the best is possibly yet to come... N.B.: once the exercise is completed, the analysis software [MATLAB] is also available!

15 1. Hands-on photon counting statistics Can we tell by an analysis of the multiphoton spectrum what is the underlying statistics of the emitted light? (M. Ramilli, M. Caccia et al., J. Opt. Soc. Am. B/Vol. 27, No. 5, )

16 Step 1: estimate the area under every peak By a point&click procedure (P&P for Pick&Play) or By a MultiGaussian Fit (MGF)

17 Step 2: turn the spectrum in the probability density function for the number of observed photoelectrons estimate the mean number of Ph.e. (model independent): compare it to what you get by the 0 th photon peak fit the distribution to a simple Poissonian

18 A few more details: model independent estimate of the mean number (nothing but the mean): estimate by the 0 th Ph.e. peak (presuming a Possonian Distribution): A comparison of the results: As the errors get smaller, a small discrepancy pops us

19 Step 3: introduce the effect of the X-talk % Probability to observe m Ph.e. %' Probability for the incoming photons to fire (m-k) cells Binomial Probability for the (m-k) primary cells to trigger k cells by optical cross-talk The P B distribution is characterized by the first & second order momenta given by:

20 Step 4: fit the distribution with the PxB model [2014 update: Xtalk accounted for at all orders, following Vinogradov et al., DOI: /NSSMIC ] Simple Poissonian Poisson & Xtalk Definitely showing a better agreement and confirming the validity of the model (and the relevance of detector effects)

21 Gamma spectrometry: introducing the SNIP, an algorithm for background subtraction under the photopeak " The SNIP (Statistics sensitive Non-linear Iterative Peak clipping) is not new and it is actually a quite popular [and implemented in Root (Tspectrum) & R] algorithm for automated (or semi-automated) background subtraction " originally introduced for the treatment of PIXE (Proton Induced X-ray Emission) (ref.1), has been adapted for bckg elimination in coincidence γray spectra (ref.2) " developed to account for spectra with poor & large statistics (extended dynamic range), searching for a solution with the minimal number of parameters, aiming for a full automation 1. C.G. Ryan et al., NIM B34 (1988) M Morac et al., NIM A401 (1997)

22 SNIP fundamentals 1. Start by a spectrum y(i), where i is the ADC/Energy bin identifier and y is the corresponding number of events 2. transform the original spectrum in v(i) = log[log( y(i) +1 +1) +1] where the log(s) compress the dynamic range and the sqrt enhances the small peaks 3. Replace v(i) with v p (i) = min(v p 1 (i), 1 2 (v (i p) + v p 1 p 1 (i + p))) 4. Iterate on p till when it is convenient [this is the hard part] 5. anti-transform & subtract from the original spectrum

23 SNIP in action On a peak, p = 3 On a valley, p = 3

24 The SNIP has been implemented in a MATLAB routine for the subtraction of the smeared Compton shoulder under the photopeak, to improve the estimate of the Peak Position, Peak Width and Peak area Followed by a measurement of the linearity & resolution:

25 What else? So far, quite a number of kits were sold to Universities, Research Centres & sensor producers and the community starts growing: where No. of kits Europe 18 Asia (but China) 11 China 10 US 8 South America 2 + a non negligible number of single components.

26 What else? The system is also growing up: a UV fast LED is also available (248 nm) an add-on to read out an array (4x4) is on the way

27 To conclude: I hope you found all this intriguing and you would like to Join the club! Thank you!