development of a counting-type neutron imaging detector
TRANSCRIPT
Development of a counting-type neutron imaging detector for energy-resolved imaging at J-PARC/MLFJoe Parker CROSS-Tokai BL22 Group
1 J-PARC 13 October 2016
RADEN/BL22 and µNID development membersJAEA/J-PARC Center Takenao Shinohara
Tetsuya Kai Kenichi Oikawa (BL10) Masahide Harada (BL10) Takeshi Nakatani Mariko Segawa Kosuke Hiroi Yuhua Su
CROSS-Tokai Hirotoshi Hayashida
Joe Parker (µNID Lead Developer) Yoshihiro Matsumoto Shuoyuan Zhang
Nagoya University Yoshiaki Kiyanagi
RADEN/BL22 – Neutron imaging instrument at the MLF
• World-class, pioneering instrument for pulsed-neutron imaging
• Leading facility for conventional radiography in Japan
• Commissioning from Nov. 2014, user program from April 2015
World’s first instrument dedicated to energy-resolved neutron imaging using pulsed neutrons!
First images from RADEN (7 Nov 2014)
Energy-resolved neutron imaging• Energy-dependence
quantitative information on macroscopic distribution of microscopic quantities
• Pulsed neutrons wide energy range, accurate energy determination by time-of-flight
• Requires detectors with:
• Spatial resolution < 1 mm
• Time resolution < 1 µs
• Count rate > 1 Mcps
• Strong background rejection
Energy-dependent neutron transmission
EnergymeV 1 keV
Wavelength10 10-2
Resonance absorption
Bragg-edge, Magnetic imaging
Moderator
Bulk shield
ShutterOptical Devices
Experimental Space
1st Detector position
2nd Detector position
13m
5m
Large load sample stage
Optical benchMedium sample stage
0m 8m 14m 27m 31m18m 23m
Conventional radiography/pulsed-neutron imaging
Large beam size (up to 30x30 cm2)
High flux (2.6x107 n/s/cm2 @ <0.5eV) Variable L/D (up to 7500)
Wide bandwidth (~9 , / < 0.2%) Large experimental area
Properties of RADEN
RADEN/BL22 – Neutron imaging instrument at the MLF
RADEN computer system
• Computer control of beam line components, sample stages, and detectors using IROHA2 (automated measurements)
• Large data storage capacity (24TB SSD primary, 100TB secondary)
• Fibre channel network (8 Gb/s) for fast data transfer
• GPGPU cluster (12 CPUs, 24 GPGPUs) for data analysis
Detectors available at RADENCamera type Counting type
Andor iKon-L - Cooled CCD - 300µm - No TOF - Automated
system for CT
Neutron Color I.I. - High-resolution
(200 µm) - High-speed (10k,
30k, 100k fps)
µNID - Micro-pattern - 3He (18% eff.) - FOV: 10 × 10 cm2
- x=0.3mm, t=0.6µs, < 1 Mcps
nGEM - Micro-pattern w/ 10B (10% eff.) - FOV: 10 × 10 cm2
- x=1mm, t=15ns, < 1 Mcps
LiTA12
- Li-glass scint. (40% eff.) - FOV: 5 × 5 cm2
- x=3mm, t=40ns,6 Mcps
0.1
1
10
100
0.01 0.1 1 10
Co
unt r
ate
(M
cp
s)
Spatial resolution (mm)
µNID nGEM
LiTA12
Current performance of counting-type detectors at RADEN
µPIC-based Neutron Imaging Detector (µNID)
µPIC-based neutron imaging detector (µNID)
9.0
cm
32.8 cm
400 µm
X (strips)0 10 20 30 40 50 60
Relat
ive cl
ock p
ulse
05
101520253035404550
Entries 28Entries 28Position
X (strips)0 10 20 30 40 50 60Ti
me-
abov
e-th
resh
old
(clo
cks)
0
5
10
15
20
25
30Energy DepositionTOT for proton-triton track
Proton Triton
Neutron
Digital encoder with time-over-threshold (TOT)
ThresholdµPIC
Discriminator
Time-above-threshold (∝ energy dep.)
neutron proton
triton
3He
• 3-dimensional tracking of decay pattern
• Energy via time-over-threshold (TOT)
• Compact ASIC+FPGA data encoder
→ Good spatial resolution, strong background rejection, high data rates possible
Neutron detection via 3He
Track length ~8 mm in gas
E
µPIC
µPIC-based neutron imaging detector (µNID)
• FPGA-based data encoders • FPGA-based DAQ controller • Data transfer via Ethernet
EncodersDAQ PC
SiTCP
EncodersµPIC
Control box DAQ controller system monitor DC power
External timing signals
Vessel pressure
Ethernet
GbE × 4
±2.5V, +3.3V
Ambient temperature
Network
DAQ control
Monitoring
Power
Add 10GbE hub to reduce cables
Sensor power
µNID performance
Image data taken at NOBORU in Feb. 2011
Distance from interaction point (mm)-2 0 4 8Ti
me-
over
-thre
shol
d (n
s)
0
50
100
150
200
250
62
Template for fit
Measured TOT distribution
Proton Triton
• Strong gamma rejection using TOT information
• Template fit for position analysis
Avg. time (clocks)0 100 200 300 400 500
Coun
ts/h
r/3.75
cloc
ks0
1
2
3
4
5
6
7
Track-length cut
Avg. time (clocks)0 100 200 300 400 500
Coun
ts/h
r/3.75
cloc
ks
0
0.5
1
1.5
2
2.5
3
3.5
Track length and PID > 2
137Cs No source
‘Energy’ cut
Neutrons
γ’s
Total TOT (clocks)
µNID performance characteristics
Area 10 x 10 cm2
Spatial res. 0.3 mm
Time res. 0.6 µs
TOF/TOF < 0.07% @18m
-sensitivity < 10-12
Efficiency Up to 26%
Count rate 0.6 Mcps
Development of µNID• Objectives:
• Improve count rate and spatial resolution
• Improve data analysis; reduce processing time
• Integration into RADEN control system
• Count rate • Throughput of data encoder modules
• Drift velocity, stopping power of filling gas
• Readout geometry
• Spatial resolution • Electron diffusion, stopping power of filling gas
• Readout strip pitch
• Improvement of count rate and data analysis is most pressing
Data encoder• FPGA-based encoder modules
• CMOS ASICS
• Spartan6 FPGA
• Ethernet transfer (SiTCP)
• 128 ch/encoder (4 encoders total)
• Original encoder throughput limited by 100BASE-T Ethernet transfer
• Upgrades
• Gigabit Ethernet PHY (1st revision)
• On-board DDR3 memory (2nd revision, not yet tested)
22 cm FPGA
CMOS ASICS
GbE
DD
R3 m
em
ory
Originally developed by Kyoto U. and KEK (Open-it)
Gas optimization• Change to CF4-based mixture
• Increased drift velocity (count rate) • Decreased electron diffusion (spatial resolution) • Increased stopping power (both)
Previous gas New gas
Mixture Ar-C2H6-3He (67:7:30 @ 2atm)
CF4-iC4H10-3He (45:5:50 @ 2atm)
Drift velocity 23 µm/ns 58 µm/ns
Diffusion 275 µm/cm1/2 80 µm/cm1/2
Efficiency @25.3meV 18% 26%
Proton-triton track length 8 mm 5 mm
Gas characteristics simulated with MAGBOLTZ, GEANT4
Rate testing at RADEN
• Control incident intensity using B4C slits
• Testing of
• rate capacity of hardware
• rate linearity of detector
Detector
B4C slits
Neutron beam
Rate testing at RADEN• Test of revised encoders
with GbE
• Ar-Ethane gas mixture
• Compared with original encoder
• Rate capacity increased by more than factor of 6
• Mostly linear up to more than 3 Mcps
0
1
2
3
4
5
0 500 1000 1500 2000 2500
Co
unt r
ate
(M
cp
s)
Slit area (mm2)
Neutron rates vs slit area
GbE
100Mbps
Rate testing at RADEN• Test of CF4-based gas
mixture
• Encoders with GbE
• Rate capacity over 8 Mcps
• Nearly factor of 2 increase over Ar-based gas mixture
0
2
4
6
8
10
0 1000 2000 3000 4000
Co
unt r
ate
(M
cp
s)
Slit area (mm2)
Neutron rates vs slit area
Ar-C2H6
CF4-iC4H10
20 30 40 50 60 70
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Sample
Spatial resolution with CF4
• Image of Gd test pattern • L/D: 5000 • Exposure time: 1.5 hours • 16% contrast at 2.5 lp/mm
(200µm line width) • Improvement over Ar-
Ethane mixture
8 cm
0.2
mm
0.3
mm
0.4
mm
0.5
mm
0.6
mm
Bin size: 40 x 40 µm2
20 30 40 50 60 70
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Sample
Tra
nsm
issio
n
Distance from top (mm)
Remaining issues• Data analysis: performance
of event clustering
• Event ‘pile-up’ at rates above 100~300 kcps
• Developing new algorithm
• Working with software company to improve speed and ease-of-use of analysis software
TOF (ms)0 5 10 15 20 25 30 35 40
Effic
ienc
y
0
0.2
0.4
0.6
0.8
1
Neutron reconstruction efficiency
5.6 Mcps
2.5 Mcps
<0.4 Mcps
Efficiency of analysis determined by comparing numbers of raw hits and reconstructed neutron events
Other ongoing development• New µPIC readout geometry
• Additional strip plane at 45° to x,y strips
• Aid in reconstruction of simultaneous events
• Now testing at Kyoto U.
• Reduced strip pitch
• Manufactured using MEMS (structures down to 10µm)
• µPIC with 280, 215µm pitches
• Performed preliminary testing at RADEN
280 µm215 µm
y1
y2
x2 x1
u2
u1
First on-beam test of MEMS µPIC at RADEN
280µm pitch (192×192 strips)
215µm pitch (64×64 strips)
Time (ms)0 5 10 15 20 25 30 35 40
sµ
Coun
ts/p
ulse
/25
0
0.02
0.04
0.06
0.08
0.1
0.12
Neutron TOF (MEMS uPIC test)
MEMS µPIC test board
Neutron TOF spectrum measured on 215µm section
• No signal measured on 280µm section (gain too low)
• Signal confirmed on 215µm section
• Further testing to study gain stability, imaging capability
µNID with Boron converter
µNID with Boron converter
• 10B-coated drift cathode (t=1µm)
• 3-axis µPIC
• Encoder with on-board memory
• CF4-based gas at 2 atm
• On-beam test at RADEN early next year
~5 mm
Al drift cathode (t=1mm)
3-axis µPIC readout
Expected performance
[email protected] 3~5%
Time resolution 10 ns
Spatial resolution 200~400 µm
Peak count rate 20~30 Mcps
10B coating (t=1µm)
0.1
1
10
100
0.01 0.1 1 10
Co
unt r
ate
(M
cp
s)
Spatial resolution (mm)
µNID GEM
LiTA12
Current and projected performance
µNID (GbE and
optimized gas)
Boron µNID
µNID (GbE/memory
and reduced pitch)
Summary• Development of µNID detector is proceeding
• Increased rate capacity to 8 Mcps through hardware upgrades, optimization of gas mixture
• Need to adapt analysis algorithms to higher rate, new gas characteristics
• Testing of ew µPIC readout boards for increased rate, higher spatial resolution has begun
• Starting development of faster off-line data processing software
• µNID with Boron converter • Expect greatly improved rate (20~30 Mcps) and similar spatial
resolution thanks to smaller event size
• Will perform on-beam test at RADEN in 2016B