Integration of SiPM in a high-pressure noble gas scintillation detector for homeland security Romualdo Santoro Università dell’Insubria M. Caccia, V. Chmill, S. Martemiyanov – Insubria R. Chandra, G. Davatz, U. Gendotti – Arktis MODES_SNM Approved by the European Commission within the Framework Program 7 Modular Detector System for Special Nuclear Material The Main Goal is the development of a system with detection capabilities of “difficult to detect radioactive sources and special nuclear materials” Neutron detection with high γ rejection power γ-rays spectrometry Other requirements Mobile system Scalability and flexibility to match a specific monitoring scenario Remote control, to be used in covert operations R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 2 Baseline technology The Arktis technologies is based on the use of 4He for the neutrons detection The main key features of 4He Reasonably high cross section for n elastic scattering Good scintillating properties Two component decays, with τ at the ns and μs levels Cheaper and easier to be procured wrt 3He 44 cm diameter x 47 cm sensitive length 180 bar 4He sealed system maintaining gas purity Topical Seminar on Innovative Particle and Radiation Detectors R. Santoro et al., 2012 JINST13th R. Chandra 7 C03035 (IPRD13) 7-10 Oct. 2013, Siena, Italy 3 MODES_SNM System overview With γ-ray spectroscopy capability Modular system optimized for: Fast neutron (4He) Thermal neutron (4He with Li converter) Gamma (Xe) All components are being integrated, we are approaching the commissioning and qualification phase R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 4 MODES_SNM R&D The project baseline is based on PMTs coupled with scintillating material R&D activities was planned since the beginning to investigate the possibility of using SiPM as light detector Why SiPM is so appealing? high sensitivity (single photon discrimination) compactness, robustness, low operating voltage and power consumption low cost R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 5 SiPM SiPM is a High density (~103/mm2 ) matrix of diodes with a common output, working in Geiger-Müller regime Common bias is applied to all cells (few % over breakdown voltage) Each cell has its own quenching resistor (from 100kΩto several MΩ) When a cell is fired an avalanche starts with a multiplicative factor of about 105-106 The output is a fast signal (Trise~ ns; Tfall ~ 50 ns) sum of signals produced by individual cells SiPM works as an analog photon detector Signals from SiPM R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 6 SiPM SiPM is a High density (~103/mm2 ) matrix of diodes with a common output, working in Geiger-Müller regime Common bias is applied to all cells (few % over breakdown voltage) Each cell has its own quenching resistor (from 100kΩto several MΩ) When a cell is fired an avalanche starts with a multiplicative factor of about 105-106 The output is a fast signal (Trise~ ns; Tfall ~ 50 ns) sum of signals produced by individual cells SiPM works as an analog photon detector The selected device is a large area (13.6 x 14.3 mm2) monolithic array of SiPM units produced by Hamamatsu: S11829-3344M R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 7 Lab charcterization to fullfill the simulation hints Minimum detectable light Detector sensitivity (i.e. S/N or capability to discriminate an “event” against noise ) Model developed by Arktis We expect 255 photons / matrix for 100 keV deposited energy assuming the 95% of reflectivity R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 8 Lab charcterization to fullfill the simulation hints Firing the matrix with a calibrated photon flux, we measured the minimum detectable light a two different temperatures (different performances due to a combined effect of increased noise and gain drift) ≈ 250 ph @ 25°C ≈ 60 ph @ 21.6°C We expect 255 photons / matrix for 100 keV deposited energy assuming the 95% of reflectivity R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 9 Experimental set-up for proof of principle A short tube (19 cm) used for the proof of principle Filled with 4He at 140 bar, an integrated wavelength shifter and two SiPMs mounted along the wall (by ARKTIS) Two SIPMs read-out through the Hamamatsu electronic board (C11206-0404FB) 2-channels 3-stage amplification with leading edge discrimination (SP5600A – CAEN) Digitizer with a sampling rate of 250 Ms/s 12 bit digitization (V720 – CAEN) R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 10 Counting measurements Test performed measuring: Background, n and γ counting rate using 252Cf and 60Co source in contact Two triggering scheme: Trailing edge discrimination in coincidence Trailing edge and delayed gate of each single SiPM in coincidence 1st Trigger Scheme Few parameters to be optimized: Leading and trailing threshold Delay time (ΔT) Gate aperture typical γ event R. Santoro 2nd Trigger Scheme typical n event 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 11 Counting measurements Result for the different trigger scheme @ 28°C An amazing result, corresponding to a γ rejection power at the 106 level [ 10 counts in 1000s, for a number of γ given by acceptance*activity*time = 1/3 * 3 * 104 * 103 ~ 107 ] R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 12 Off-line data analysis Data recorded with a minimum bias trigger For each triggered events we digitize signal of both SiPMs with sample rate of 125 MS/s and a total duration of 4μs. Three data set: 1. 2. 3. Low threshold on the pulse height discrimination No coincidence between the two SiPMs 400 events without radioactive sources 6000 events with 60Co source in contact 10000 events with 252Cf source in contact Analysis strategies: Identify an observable allowing to measure the ratio between noise & particle induced triggers in samples 2 & 3 Filter noise from particle induced events through a multivariate analysis Identify the ratio between γ and n events in sample 3 Filter γ from n through a multivariate analysis Measure the rejection power of interacting γ and the selection efficiency of interacting neutrons R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 13 Sample composition: (% of Background and signal) Each signal is baseline subtracted in the integrated time windows FAST and SLOW component is calculated as the integral of the signal to the left / right side of the peak Definition of Fast and Slow Component The areas underneath the fits are used to measure sample composition These numbers are used to estimate the selection efficiency and bkg rejection power R. Santoro BKG Co60 Cf252 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 14 A multivariate Bayesian analysis The strategy: Select 4 no-correlated variables where bkg, γ and n appear to be “reasonably” different TOT_Diff, Charge Diff, Charge Skewness, Full_charge Bkg data-set is used to build the experimental probability density functions (p.d.f.) The corresponding cumulative distributions function (c.d.f.) Ii is then constructed: Ii (~ xi ) ~ xi h i ( x i ) dx i h i ( x i ) : signal over noise distributions The four Ii’s are combined to get the final distribution: 3 P i 0 ( log ) i! i , I1 I 2 I 3 I 4 Two step procedure: 1st step: the selection criteria based on P is used to remove the background from n and γ induced events 2nd step: the procedure is reiterated to define the n/γ selection criteria (60Co sample used to build the p.d.f.) R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 15 1st Step c.d.f. obtained for one of the selected variables (total charge), based on the bkg data-set bkg data-set As expected we have a random quantity with a flat distribution R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 16 1st Step c.d.f. obtained for one of the selected variables (total charge), based on the bkg data-set Moving towards a different p.d.f (60Co or 252Cf), we have an accumulation of events on the right part of the histogram which allows as to separate the signal form bkg R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 17 1st step: bkg rejection Repeat the exercise for all the quantities and combine the c.d.f. as follow 3 P ( log ) i 0 i! i , I1 I 2 I 3 I 4 These two plots show that almost all the signal is on the right part of the histograms (peak) while the bkg is a flat component on the left The bin at P>0.995 contains ~78% of the γ events The bin at P>0.995 contains ~99% of the γ +n events for Cf Selection for the 2nd step R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy Selection for the 2nd step 18 2nd step after the Bkg has been filtered out, the ratio between γ and n events in 252Cf can be measured: γ- events in the 60Co data-set n and γ composition in the 252Cf data-set R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 19 2nd step: γ rejection The procedure is reiterated using the 60Co data-set to build the cumulative distributions to identify γ over n in the 252Cf data-set Results for a P cut of 0.995 Results are well beyond the expectation which are pushing us to continue with Further test to measure the neutron detection efficiency and to qualify the γ/n separation using TOF technique Tube layout optimization Improved electronics (see next slide) R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 20 Customized electronics Adjustable bias voltage for each of 16 channels SiPM temperature readout and gain compensation The thermo-chip placed onto the SiPM generates a digital pickup noise which cannot be removed The frequency of the temperature readout is settable The board include a lemo connector that can be used for veto trigger Improved minimum detectable light and dynamic range R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 21 Customized electronics Adjustable bias voltage for each of 16 channels SiPM temperature readout and gain compensation The thermo-chip placed onto the SiPM generates a digital pickup noise which cannot be removed The frequency of the temperature readout is settable The board include a lemo connector that can be used for veto trigger Improved minimum detectable light and dynamic range ≈ 30 ph @ 19.6°C Pedestal R. Santoro ≈ 60 ph @ 19.6°C 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 22 Conclusion The Proof of Concept of SiPM arrays in the tube has been successfully completed The tube design optimization is certainly required A new front-end electronics (designed at Uni. Insubria) has been characterized in the lab and fulfils the requirements New test campaign is on the way to optimize the on-line and off-line analysis R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 23 Spares R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 24 Typical events with PMTs Neutron/γ discrimination is based on the difference between the fast/slow component of the scintillation light Typical neutron and gamma events Typical plot from the 4He detectors showing the discrimination between neutrons (Am-Be) and gamma (60Co) R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 25 Bkg_rejection Bkg flat distribution (1st selection) Co60 flat distribution (2nd selection) R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 26 2nd step Distributions of the four discriminant variables used in the second step of the procedure after having filtered the noise induced events R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 27 2nd step: γ rejection n _ select N cf _ sig _ sel bkg _ rej * N cf N cf _ total _ bkg bkg _ rej _ total _ signal N co 60 _ sel N co 60 _ total _ signal measured with a data-set with pure bkg rej _ power 1 bkg _ rej P_value=0.995 Neutron eff = 94% γ rejection power= 92% R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 28 Few spectra used in the linearity plot R. Santoro 13th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD13) 7-10 Oct. 2013, Siena, Italy 29