Search for hidden high-z materials inside containers with the Muon Portal Project Paola La Rocca for the Muon Portal Collaboration Department of Physics, University of Catania
Presentation overview Cosmic-ray muons and muon tomography Ongoing projects The Muon Portal Project Overall Design Strip and module Photosensors Readout and DAQ Simulation studies Image reconstruction Present status of the Project
The problem Containers: employed since 50 years as a standard way to transport goods by ships or trucks Estimated yearly traffic: 200 M containers (in Catania ~ 15000 container /year) Safety regulations are more and more demanding a better and fast way to inspect such containers, not possible by traditional techniques
The problem Presently only about 1% inspected X-ray tomography Based on X-ray absorption (requires high fluxes) Problems to traverse large thickness (Mean free path ~ 25 g/cm 2 ~ 2 cm Pb) Mono or two-dimensional, not 3D (require multiple projections) Other possibilities Directional imaging with gamma rays Neutron radiography
Why Muon Tomography Secondary cosmic muons: highly penetrating radiations (even more than X-rays) Natural radiation, hence no additional dose delivered to users and goods Muon flux at sea level relatively large: 1 cm -2 min -1 Muonic interactions well understood Muon scattering strongly dependent on the Z of the material Each muon may contribute to determine the overall imaging result (contrary to muon absorption techniques)
Muon Tomography l L. Schultz IEEE NSS Conference Record (2006)
Muon Tomography Muon tomograph: basically made by several detection planes, above and below the volume to be inspected Reconstruction of muon tracks allows, by different methods, to produce a 3D tomographic image Performance of the system are given in terms of Sensitive area / volume to be inspected Spatial and angular resolution Time to scan a volume Sensitivity & Efficiency to high-z objects Sensitivity to false positive Discrimination between high-z vs low & medium-z
On-going Projects Several Projects (INFN Padova, Los Alamos, Florida State Univ., ) worldwide interested to Muon Tomography A few detector prototypes already built and tested Various detection techniques employed Drift chambers Drift tubes GEM Use of the same technique for similar applications (nuclear waste, orphan sources, )
The Muon Portal Project in Catania Basic architecture based on 8 physical detection planes (4 XY logical planes) segmented into 48 detection modules (1 m x 3m) Modules segmented into 100 strips of extruded scintillator with double WLS fibre readout High PDE, high fill-factor Silicon photomultipliers as readout sensors 9600 channels with a readout compression technique Distance between top and bottom planes 5-7 m Angular resolution around 3 mrad Plane X 6 m Plane Y 3 m
Mechanical structure Mechanical structure under control with PLC Monitoring and storage of various parameters Sensors for alignment and alarms Assembly tools
Test measurements with strips and WLS fibres To maximize the light yield and reduce the dark current for SiPM, extensive tests carried out under different conditions with Various designs of scintillator strips (Fermilab, Amcrys, Uniplast) Different WLS fibres (Kuraray, St.Gobain) Silicon photomultiplier prototypes from STMicroelectronics SiPM Lab. measurements in a dark box with an external scintillator trigger Dark current vs threshold Light yield at various distances (0-3 m) Charge spectra
Strip design Various designs of extruded strips tested and simulated by GEANT4 Threshold 2 p.e. Threshold 1 p.e.
Photosensor design Silicon PhotoMultipliers by STMicroelectronics as photosensors Compactness Cost-effective Low voltages required High Photon Detection Efficiency to light from WLS fibres High Fill Factor Several prototypes built, customized for this application MUON70N Prototype Number of cells: 548 Cell fill factor: 73.8 % PDE ~ 40 % (λ 500 550 nm @ OV = 5 V)
Basic SiPM characterization
The SiPM test station @ INAF in Catania fotodio do
Front-end Readout Combining the information of two WLS fibres Reduction factor for module: 2 N (N number of channels) (8 planes) x (6 modules) x (100 x 2 WLS) = 9600 channels After channel reduction (8 planes) x (6 modules) x (2 100 ) = 960 channels
Front-end Readout Strips WLS SiPM MAROC MAROC (Multi Anode Read Out Chip) MAROC No. of channels: 64 For each channel: 8 bit variable gain, preamplifier, shaper
Data Acquisition Boards based on FPGA Flex RIO National Instrument programmable by LabVIEW FPGA module Acquisition board GPS tagging of the events for possible correlation to other detectors
Simulation procedures Full GEANT4 replica of the detector Cosmic muons modeled with realistic energy and angular distribution by CORSIKA air shower simulations Transport of optical photons fully simulated for a subset of events and then parameterized to save CPU time Reconstruction of hits and cluster in each plane Single muon tracks and e.m. showers taken into account After track reconstruction, tomographic images built by several methods Various scenarios considered
Simulated scenarios A MUON shape with letters of different materials (U, Fe, Pb, Al) As before, with a heavy scenario with washing machines elements (iron, concrete,..) surrounding the MUON shape About 1 M events simulated for each scenario, with realistic energy and angular distributions of muons
Event reconstruction and selection Event reconstruction and selection Hit selection (threshold on deposited energy) Cluster finding from hit strips Track finding by Kalman Filter algorithm, valid also for high-multiplicity events Selection Single cluster events for tomographic imaging Multicluster events recorded for physics analysis
The POCA algorithm Simplest approach, fast and easy to implement Geometrical point of closest approach between incoming and outcoming tracks P poca = ½ (P in + P out ) Spatial distribution of the scattering centres, weights given by some power of the scattering angle However: Neglects multiple scattering within the material Poor resolution images Critical behaviour for material located close to volume borders Several improvements may be implemented: - Density based clustering algorithms - Two points (2P) correlation analysis S.Riggi et al., NIM A 728 (2013) 59-68
The EM-ML method Better statistical treatment of scattering processes by a log-likelihood method Volume to be inspected divided into voxels Scattering density defined for each voxel Iterative estimation with some stopping criterion
Density based Clustering Algorithm The FOF is a percolation algorithm normally used to identify dark matter halos from N- body simulations It defines uniquely groups that contain all the particles separated by a distance smaller than a given linking length. Once the lenght is defined, the algorithm identifies all pairs of particles which have a mutual distance smaller than the linking one and defines these as friends. Clusters are defined as sets of particles that are connected by one or more of the friendly relations, so that they are friends of friends. Another parameter in FOF algorithm is the minimum number of particles Nmin, in a cluster. The aim is to reject spurious clusters, that is groups of friends who do not form persistent objects in the simulation. Choosing Nmin sufficiently large allows to eliminate spurious clusters. In fact it is much more likely that a spurious cluster (noise) involves a small number of points and not viceversa.
Results from the 3 algorithms Poca Volume rendering EM-ML Volume rendering Clustering Volume rendering
Tools for image processing Intercommunication between computer resources for monitoring, data acquisition and image processing User-friendly tools to handle image reconstruction, under development
Demo
On-going activity and outlook A large area muon tomograph currently under construction after extensive R&D phase Large number (10 4 ) of channels involved, with corresponding number of SiPM photosensors New solutions exploited for channel reduction, electronic readout and data acquisition Monitoring and image reconstruction with different algorithms and tools implemented Construction of the first 2 (of 48) modules in progress End of construction expected by end of 2014
Outreach and dissemination activities Due to the nature of the Project, also outreach activities are believed to be important: Public meetings and colloquia targeted to students, high-school and citizen people Posters, articles on local newspapers Stands during public events Construction of an exhibition box located in a public area for outreach activities
The partners Dept. Of Physics & Astronomy, University of Catania INAF, Astrophysical Observatory, Catania STMicroelectronics S.r.l. Catania Insirio SPA Meridionale Impianti Welding Technology
Thank you for your attention and BUON APPETITO!
Back-up slides
Fake probability and efficiency EM-ML method
Momentum filter Additional prototype detector to reject low momentum muons being tested After cutting low momenta (< 1 GeV/c) muons
The Muon Portal for cosmic ray physics It is also planned to employ such detector prototype for cosmic ray studies, due to its large area (18 m 2 ) and tracking capabilities. GPS tagging used on both detectors to correlate events Trigger on air showers Muon bundles in the Portal by 3-fold coincidences
Muon tomography project* INFN and University of Padua Drift chambers to track muons, similar to CMS Momentum filter Spatial resolution 200 μm * S.Pesente et al, NIMA 604(2009)738
LOS ALAMOS* 12 detection planes with drift tubes Spatial resolution 400 μm Angular resolution 2 mrad * K.N. Borozdin et al, Nature 422(2003)277
Muon Tomography Station Using GEM Detectors* Florida Institute of Technology, Melbourne 30x30 cm triple GEM 50 μm resolution for perpendicular tracks XY readout with 400 μm pitch * K.Gnanvo et al, NIMA 652(2011)16