GEM-based Muon Tomography
of Shielded High-Z Materials
10th RD51 Collaboration Meeting, Stony Brook U., Oct 1, 2012
Marcus HohlmannFlorida Institute of Technology
Recent news…Recent news…
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 2
By Douglas Frantz, Updated: Sunday, July 15, 4:05 PM
The Obama administration has failed to meet a legal deadline for scanning all shipping containers for radioactive material before they reach the United States, a requirement aimed at strengthening maritime security and preventing terrorists from smuggling a nuclear device into any of the nation’s 300 sea and river ports.
The Department of Homeland Security was given until this month to ensure that 100 percent of inbound shipping containers are screened at foreign ports.
But the department’s secretary, Janet Napolitano, informed Congress in May that she was extending a two-year blanket exemption to foreign ports because the screening is proving too costly and cumbersome. She said it would cost $16 billion to implement scanning measures at the nearly 700 ports worldwide that ship to the United States.
Instead, the DHS relies on intelligence-gathering and analysis to identify “high-risk” containers, which are checked before being loaded onto ships. Under this system, fewer than half a percent of the roughly 10 million containers arriving at U.S. ports last year were scanned before departure. The DHS says that those checks turned up narcotics and other contraband but that there have been no public reports of smuggled nuclear material.…The DHS says monitors scan 99 percent of the containers for radiation after they arrive at U.S. ports. But experts say the monitors at U.S. ports are not sophisticated enough to detect nuclear devices or highly enriched uranium, which emit low levels of radiation. …
… … on an old problemon an old problem
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 3
Colorado Sen. Eugene Millikin pressed Oppenheimer about how to find a bomb hidden in a city:
Sen. Millikin: “We... have mine-detecting devices, which are rather effective... I was wondering if anything of that kind might be available to use as a defense against that particular type of use of atomic bombs.”
Dr. Oppenheimer: “If you hired me to walk through the cellars of Washington to see whether there were atomic bombs, I think my most important tool would be a screwdriver to open the crates and look. I think that just walking by, swinging a little gadget would not give me the information.”
Transcripts from the National Archives
That candid assessment shocked the senators, who then asked the Atomic Energy Commission to examine the problem. Robert Hofstadter and Wolfgang Panofsky, a veteran of the Manhattan Project team that built the atomic bomb, produced a still-classified assessment, which came to be known as the “Screwdriver Report”.
Panofsky, now the director emeritus of the Stanford Linear Accelerator Center, says the assignment was to detect 1 cubic inch of highly enriched uranium or plutonium hidden inside a crate and smuggled across a land border. "The conclusions of that report are still valid because the laws of physics have not changed one bit," Panofsky tells U.S. News. "You still can't detect a nuclear device unless you are, say, 10 feet away from it - and even then it can be quite easily shielded."
US News & World Report, 2/18/07
μμ
Fe ULarge
ScatteringSmall
Scattering
μIron
Small Scattering
Uraniumμ
LargeScattering
μμ Incoming muons (from natural cosmic rays)
Note: Angles Exaggerated!
)]/ln(038.01[MeV6.130
0
0 XxXx
cp
Tracking detectors
Multiple Coulomb scattering to 1st order produces Gaussian distribution of scattering angles θ with width σ = Θ0 :
𝑋0 = 716.4 g cm-2 ∙AZሺZ + 1ሻln (287 ξ𝑍Τ )
Towards a solution…Towards a solution…
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 4
Muon Tomography Concept:
Growing Interest in MTGrowing Interest in MT
Muon Tomography with•Drift Tubes
– Decision Sciences Int’l Corp., commercial effort, US (now operating full-size MT prototype at Freeport, Bahamas)
•GEMs– Florida Tech, US
•Plastic Scintillators – CRIPT Consortium, Canada– INFN Catania, Italy
•Multi-gap RPCs – Tsinghua U., China
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 5
8 MT presentations scheduled at upcoming IEEE NSS:
1 ft3
active
volume8 30cm 30cm GEMs
Fl. Tech Cubic-Foot MT Prototype
GEM Muon TomographyGEM Muon Tomography
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 6
30cm × 30cmTriple-GEMs
with x-y readout
Event display of 141 reconstructed tracks probing target
Trigger scintillators
X-Z view Y-Z
view
World’s largest RD51 SRS application(12,288 channel APV readout w/ DATE & AMORE)
• Number in Neighboring Pixels (NNP): add number of POCA in 8 voxels surrounding voxel V
• If NNP < some threshold, remove contents of V.
• Repeat for all voxels
Track & Image ReconstructionTrack & Image Reconstruction• Detector hits formed from readout strip
clusters• All hit combinations within each of the
four tracking stations (t, b, l, r) → track segment candidates
• Track with smallest distance of closest approach (DOCA) of incoming and exiting segments in 3d is selected as best track
• Detector alignment using tracks crossing an empty MT station
• Remove low-angle scattering (< 2o)• Scattering point reconstruction using
Point Of Closest Approach of incoming and exiting segment in 3d
• Find <scatter> in voxels in volume• Remove isolated scattering pixels; keep
clustered scattering pixels
7
DOCA
Object
DOCA
unbiasedresiduals
Vmeasuredscattering
angle
Ref.: Michael Staib, M.S. thesis
Number of POCA points Mean Scattering Angle Removing isolated pixels (NNP)
Reco & Image Processing StepsReco & Image Processing Steps
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 8
Not aligned
Aligned
M. Staib, M.S. thesis
Current R&D FocusCurrent R&D FocusSystematic Performance Studies of GEM-based MT:• Imaging Studies
– How well can we resolve material shapes?– What is the (ultimate) imaging resolution?– What is the image quality with (substantial) shielding?– How much information are we gaining by having added the side detectors?– How do results depend on the target location within the MT volume?
•Z-discrimination– Can we tell U from medium-Z material?– Can we distinguish high-Z materials from each other? (U from W, Pb)?
•Less emphasis on “How fast can you detect”? – Typical MT images shown here take 24-48 hrs. of running– Other MT efforts have shown that detection presence “of some high-Z
material” can be done in a few minutes; will address later
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 9
Unshielded MaterialsUnshielded Materials
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 10
U
WPb
FeSn
Top View
Side views in 3 vertical planes:
→→→
MeanScatteringAngle (2mm×2mm×40mm voxel)
6mm Al cladding for U
Pb W U Sn Fe
• 164,323 total reconstructed tracks
• Number Neighboring Pixel Cut > 5
Horizontal Imaging ResolutionHorizontal Imaging Resolution
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 11
0 mm 2 mm
115,834 Tracks
94,719 Tracks
111,036 Tracks
Gap starts to become visible with ≥ 6 mm spacing
4 mm
Lead Lead
Tungsten Tungsten
0 mm 8 mm
2 mm increments
8 mm6 mm
107,506 Tracks 121,634 Tracks
Gap begins to become visible
with ≥ 6 mm spacing in y
115,834 Tracks
M. Staib, M.S. thesis
Horizontal Imaging ResolutionHorizontal Imaging Resolution
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 12
No significant signal with 0 mm, 2 mm, or 4 mm gap spacing
Significant signal for a gap begins to develop with 6 mm spacing:
We conclude that the lower limit on the spatial imaging resolution in the XY plane with ~100k total reconstructed MTS tracks is currently 6 mm.
M. Staib, M.S. thesis
Vertical Imaging ResolutionVertical Imaging Resolution
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 13
Analogous test with vertical gaps (Pb cube stacked on top of W cube):
15 mm vertical gap 30 mm vertical gap 45 mm vertical gap
Gap begins to become visiblewith ≥ 45 mm spacing in z
160,096 tracks 235,836 tracks 201,836 tracks
We conclude that the lower limit on the spatial imaging resolution in the ZX plane with ~200k total reconstructed MTS tracks is currently 45 mm.
Z Z Z
Uranium Shielded w/ BronzeUranium Shielded w/ Bronze
• 187,731 reconstructed tracks• Number Neighboring Pixel Cut > 10• 2 mm x 2 mm x 40 mm voxels
40 mm XY slices descending in Z by 5 mm per frame
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 14
Shielding made of tin-bronze(83% Cu, 7% Sn, 7% Pb, 3% Zn) with X0 = 1.29 cm & 1.7 cm walls
1.7cm
DU
M. Staib, M.S. thesis
With Lead ShieldingWith Lead Shielding
LeadTantalum
Tungsten
UraniumTin Iron
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 15
Lead box with 3.4mm thick walls
placed inside
Muon TomogramMuon Tomogram
The shielded targets are clearly visible in the reconstruction
LeadTantalum
Tungsten
UraniumTin Iron
• 292,555 reconstructed tracks• NNP cut = 5• 2 mm x 2 mm x 40 mm voxels
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 16
40 mm XY slices descending in Z by 5 mm per frame
M. Staib, M.S. thesis
More Pb ShieldingMore Pb Shielding
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 17
Add two more Pb plates to the top of the box, for a total of 10mm of Pb top shielding:
Well imaged in x-y plane• 397,362 reconstructed tracks• NNP cut = 5
3mm vs. 10mm Pb top shielding3mm vs. 10mm Pb top shielding
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 18
Side view:
LeadTantalum
Tungsten
UraniumTin Iron
Vertical slice in this plane
3.4 mm top shielding 10 mm top shielding
Pb WTa
outline of Pb shielding box
Z-discrimination for shielded cubesZ-discrimination for shielded cubes
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 19
Define a “simple scattering density” to discriminate shielded target materials:
Sum of all scattering angles measured within a target / volume of target cube (normalized to # of rec. tracks)
Fit to 1/X0
X0 ( )
)]/ln(038.01[MeV6.130
0
0 XxXx
cp
normalized to1000 rec. tracks
Where is the limit? Where is the limit?
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 20
Place U cube inside a shielding structure with ~ 2cm Pb plates at top, bottom, and 2 sides:
• 335,410 rec. tracks• NNP cut = 5
U cube still discernible
Spatial sensitivity biasSpatial sensitivity bias
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 21
Top View
MeanScatteringAngle
U
WPb
FeSn
• 164,323 total reconstructed tracks
Now flip position of U and Pb:
• 183,051 total reconstructed tracks
Pb
PbU
U
right tomogramleft tomogram
Towards an unbiased sensitivityTowards an unbiased sensitivity
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 22
Mainly a geometric acceptance effect
Beginnings of mapping out sensitivity within the volume of the MTS to correct for bias towards center…
Three identical lead-acid UPS batteries in MTS:
• 102,679 reconstructed tracks• NNP cut = 1
Benefit from side detectorsBenefit from side detectors
10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 23
Three identical lead-acid UPS batteries in MTS
Tracks through top & bottom GEMs onlyTracks through top & bottom GEMs
plus top & side GEMs plus bottom & side GEMs
• Side detectors help extend acceptance towards edges of MTS (as expected)• GEM-MTS is only MTS operating with side detectors
Summary & ConclusionsSummary & Conclusions• GEM-based MTS prototype taking lots of data!• Without shielding, U can clearly be discriminated
from Pb, quite possibly from W• High-Z materials clearly discriminated from medium-Z
materials even when (moderately) shielded• Imaging resolutions measured to be ~6 mm in the
horizontal and ~45 mm in the vertical (w/o shielding)• First measurements of expected spatial biases on
sensitivity• Side detectors help mainly near the edge of the MTS
as expected10/1/2012 10th RD51 Coll. Meeting, Stony Brook U. - Marcus Hohlmann 24
Thank you for your time!Thank you for your time!
Acknowledgements:
This work is currently being pursued mainly by our students.
Thanks to grad students Mike Staib, Vallary Bhopatkar, and Lenny Grasso, and undergraduates Mike Phipps, Jessie Twigger, and Christian Zelenka!