Experimental Verification of a Hand Held Electronically-Collimated Radiation Detector

unkindnesskindΠολεοδομικά Έργα

15 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

84 εμφανίσεις


Experimental Verification of a Hand Held
Electronically
-
Collimated Radiation Detector

Will H. Hill Jr. and Kenneth L. Matthews II,
Member, IEEE

Abstract

A hand
-
held electronically
-
collimated directional
detector is being developed for gamma
-
radiation sour
ce
localization. The system comprises a 6
-
sided box of detectors;
Compton scatter events between pairs of detectors provide the
source localization data. We report on experimental and
simulation results for a partial prototype, with extrapolation to
potent
ial performance of a full detector system.

The partial prototype used CZT detector modules to form two
sides of the box geometry, with external signal processing and
data acquisition components. Coincidence data were collected
with the calibrated detector
from Na
-
22, Cs
-
137 and Co
-
60
sources. The simulation model for charge collection in thick
monolithic CZT accounted for charge trapping and sharing with
estimation of hole and electron mobility, along with a linear
approximation for charge sharing between p
ixels. By using
calibration data from the prototype, the charge trapping model
mimicked the average energy response, including tailing seen in
the CZT energy spectra. Measured Compton
-
scatter data were
compared for the three sources to the corresponding si
mulation
results for a model of the two
-
sided prototype. Simulation results
were also generated for the full 6
-
sided box geometry.

The simulation model and prototype data showed that charge
sharing and trapping contributed significantly to measured
angular

error and resolution for these CZT detector modules. The
angular error typically was less than 5 degrees with angular
resolution of 20
-
40 degrees FWHM, depending strongly on source
location for the two
-
sided prototype. The 6
-
sided box simulation
showed th
at reasonable direction
-
finding capabilities were
possible even with the relatively coarse energy and spatial
resolution of these CZT detector modules. Some discrepancies
between simulation and prototype were due to the simple physics
model used in the sim
ulation; the charge trapping and sharing
model is being incorporated into a more sophisticated GEANT4
simulation of the hand
-
held electronically
-
collimated detector
system. Improvements to energy and spatial resolution of the
detector modules should substa
ntially improve overall system
performance.

I.

I
NTRODUCTION

ITH A

conventional radiation survey meter, one locates
radioactive materials by trial
-
and
-
error sweeps over an area
to find the location that produces the largest signal (Fig. 1a);
these systems a
re only grossly sensitive to the direction to the





Manuscript received November 16, 2007. This work was supported in
part by the Homeland Security Advanced Research Projects Agency
and the
Space and Naval Warfare Systems Center San Diego, under Contract No.
N66001
-
05
-
C
-
6024. Additional support was provided by the Biological
Computation and Visualization Center, HEF (2000
-
2005)
-
01, funded by the
State of Louisiana Board of Regents and

by the Dept. of Physics &
Astronomy at Louisiana State University.


W. Hill (whill3@lsu.edu) and K. L. Matthews II (225
-
578
-
2470,
kipmatth@lsu.edu) are with the Department of Physics and Astronomy at
Louisiana State University, Baton Rouge, LA 70803 USA.

source. To improve the directional ability, one places a
collimator onto the device. The collimator restricts the field of
view of the detector, reducing the volume of space from which
radiation could be d
etected. A collimated survey meter must
initially sweep for a source but it homes in rapidly once the
source emissions are detected (Fig. 1b). The adverse
consequence of a collimator is that it blocks some radiation
from reaching the detector, thus decreas
ing the detection
efficiency of the system. Collimators are more effective with
low energy radiation; high energy radiation penetrates through
the collimator’s walls, thus compromising the directional
ability.

(a)


(b)


(c)


Fig. 1. Illustration of se
arch patterns with (a) a conventional uncollimated
radiation survey meter, (b) a physically collimated detector, and (c) an
electronically
-
collimated detector. The dashed lines indicate the search path
or rate of localization of the radiation source, which

is illustrated as the
arrows emitted from the yellow circle.

Ideally, one would like to have a detector system that
rapidly and directly indicates source position without the need
to sweep the area (Fig. 1c).
We are developing an
electronically
-
collimated

radiation detector (ECRD) to be used
for locating gamma
-
radiation sources in this fashion [1].
Potential applications include radiation safety and nuclear
security, as well as the medical application of intraoperative
localization of sentinel lymph nodes.

The design goals include
compact size, portability and robustness, wide field of view,
and efficiency and accuracy in localization.

W


Intended to be a hand
-
held device, the ECRD front end (Fig.
2) comprises a box with side made of cadmium
-
zinc
-
telluride
(CZ
T) detectors, which provide position
-
sensitive gamma
-
ray
detection with good energy resolution and room
-
temperature
operation. The ECRD design provides high sensitivity to the
presence of radiation because it lacks a physical collimator,
and simultaneously

provides directional information by
processing source photons that undergo Compton scattering in
the detector to calculate direction to the radiation source
(electronic collimation). In practice, a source photon will
Compton scatter in one detector, depos
iting some energy; the
scattered photon then interacts in another detector. From the
measured energy depositions and/or the expected original
photon energy, one calculates the scattering angle. Coupled to
the interaction locations, this forms a cone in spa
ce on which
the source resides [2].


Fig. 2. Illustrations of the ECRD system. The detector (left) is a six
-
sided
box of position
-
sensitive CZT detectors mounted in a handheld unit (right);
Compton scatter events produce source direction information that

is presented
to the user (red star on simulated display).

Thus, the ECRD is essentially a hand
-
held Compton
imaging system. Compton imaging has been utilized in
gamma
-
ray astronomy [3
-
4] and for medical imaging [5
-
7].
Compton imaging systems for nuclear s
ecurity field work [1,8
-
9], have been under development in recent years.

In practice, the ECRD will be used for source localization,
rather than imaging per se. The instrument will provide the
user with an indication of the direction to the source, e.g., b
y
providing azimuth and elevation angles. These direction angles
are obtained by backprojecting the Compton scatter cones and
identifying the brightest intersection or the region of common
overlap of the cones [10], presenting this direction information
to

the user. The backprojected image itself could be presented
if the user desired, perhaps overlaid on a video image of the
area being surveyed for radiation sources.

A partial prototype system and a simple simulation model
have been developed and used to e
valuate the feasibility of the
ECRD concept. In particular, we wished to ascertain the
quality of direction information that could be obtained from
relatively inexpensive, moderate performance, CZT detector
modules in the compact box geometry.

II.

M
ETHODS

This

section describes the partial prototype detector system,
as well as the simulations that were developed to model
detector
-
photon interactions and the gross effects of charge
sharing and charge trapping in the pixelated CZT detector
modules. Simulation dat
a were generated for a detector model
that matched the partial prototype as well as for a model of the
complete ECRD design. The simulation results were compared
to data acquired with the prototype system; comparisons were
made for the Compton scatter data

itself and for
backprojection images of the Compton scatter data.

A.

L
-
shaped prototype detector

The position sensitive CZT detector modules used in the
prototype system were manufactured by Orbotech Medical
Solutions (Rehovat, Israel). Each detector crystal

was
38x38x5
-
mm
3

in size, with a 16x16 array of 2.4
-
mm pitch,
2x2
-
mm
2

contacts on the anode. In this work, the anode was
operated at ground potential while a voltage potential of
-
600

V was applied to the cathode. A printed circuit board
attached to the cr
ystal provided a 104
-
pin header for external
electrical connections and two 128
-
channel ASICs for energy
signal amplification and pixel address determination. The
module was mounted in a socket on a detector carrier board,
which routed power to and signal
lines from the modules. Four
modules were mounted in pairs on carrier boards to form the
“front” and the bottom “side” of the six
-
sided box geometry
(Fig. 3), resulting in an L
-
shaped prototype system.



Fig. 3. Photos from above (left) and from
the radiation source viewpoint
(right) showing the detector modules and carrier boards for the front (green
arrows) and bottom side (purple arrows) of the L
-
shaped prototype. In the
right
-
hand photo, only one of the side modules is visible.

The detector mo
dules provided energy information from
only one pixel per gamma
-
ray interaction; the energy signal
was a pair of analog differential current signals. Only 2D pixel
addresses were provided as position information (no depth of
interaction). The pixel number
(0
-
255) within an ASIC, the
ASIC identifier (0
-
1) and the user
-
assigned bit pattern for each
detector module were encoded as a 16
-
bit pixel address; this
address was decoded later by a software look
-
up table to
determine the x
-
y coordinate of the hit pixel
. The pixel and
energy signals were processed further by comparators and
shaping amplifiers, respectively, on a detector access board
(DAB) obtained from the detector manufacturer. The DAB
provided control communications to the detector modules; it
also pr
ovided digitization of the energy signals and a digital
I/O data interface, although these were not used for this
project. Instead, the differential analog energy signals and 16
-
bit digital pixel address lines were tapped from the DAB. The
energy signals r
outed to a NI
-
6143 analog I/O board while the
pixel address went to a NI
-
6533 digital I/O board (National
Instruments, Austin, Texas). Trigger signals, also tapped from
the DAB, were routed first to a NIM coincidence module
(Model 418A, Ortec, Oak Ridge, T
ennessee) and then used to

trigger simultaneously the I/O boards. Acquisition and control
software for the detector system was written in LabVIEW 7
running on a Windows 2000 computer. The Compton scatter
coincidence data were saved in list
-
mode for later p
rocessing
to determine cone axis, vertex and opening angle; this
information was used for assessing angular error and angular
resolution, as well for creating backprojection images.

B.

Simulation model

To simulate the ECRD, we developed a model for charge
col
lection in thick CZT. Thick detector CZT simulations must
take into account charge trapping and sharing, which reduce
signal amplitude and broaden recorded energy distributions.
Charge trapping in the material reduces the charge that will be
induced on col
lecting electrodes. Charge trapping, like charge
induction, is a function of depth of interaction. Charges
produced between pixels can induce signals on more than one
electrode; charge sharing is a function of both depth and planar

location. Statistical an
d electronics noise contribute further to
variations in the measured signals.

An infinite plane Hecht's relation was used to model charge
trapping; Hecht's relation describes charge collection as a
function of depth in a planar detector [11]. The charges
i
nduced in the detector's electrode are a function of the
distance traveled, described by the Shockley
-
Ramo theorem.
Electrons have greater mobility than holes in CZT; typical
values are 1350 cm
2
/V
-
s for electrons and 120 cm
2
/V
-
s for
holes [12]. Values of a
verage electron lifetime have been
reported between 100 ns and several microseconds. Hole
lifetimes vary between 50 ns and 300 ns. Simulation output
was compared to detector calibration data to select simulation
parameters that resulted in average energy r
esponse
comparable to the calibration data.

A linear approximation was used for charge sharing between
pixels. The full magnitude of induced charge was assigned to a
single pixel when the simulated photon interacted within the
2x2
-
mm
2

electrode area. For a
n interaction between two
adjacent pixels, the amount of charge assigned to each pixel
was linearly interpolated based on the distance to the edges of
the pixel electrodes, with a fraction of 0.5 assigned to each
pixel for an interaction midway between the

pixels. For an
interaction in the region between four adjacent pixels, bilinear
interpolation was used. The fraction of charge assigned to each
pixel was 0.25 for an interaction midway between the four
pixels.

The charge sharing and trapping model was int
egrated into
FreeMonte, a Monte Carlo radiation transport simulator
developed in house [13]. In practice, the charge sharing and
trapping model could be utilized with any common Monte
Carlo package such as GEANT4 or MCNP. The FreeMonte
simulator modeled p
hotoelectric and Compton scatter
interactions in CZT; interaction cross
-
sections were obtained
from the NIST XCOM database [14]. In FreeMonte, the
interaction type and the location in the first detector were
selected randomly. For Compton scatter interacti
ons, the
Klein
-
Nishina distribution provided weighted selection of the
scattering angle; energy deposition was then calculated. The
interaction location of the scattered photon in the second
detector was determined from the propagation direction of the
sca
ttered photon.

Two versions of the ECRD system were simulated: an L
-
shaped version with only front and bottom side detectors,
matching the partial prototype system, and a full version with
detectors on all six sides of the box geometry. In both models,
two

CZT detector modules comprised each side.

C.

Data acquisition and analysis

The prototype system provided energy measurements in the
range of 50 keV to 200 keV. The prototype was calibrated in
this energy range with Ba
-
133 (81 keV), Cd
-
109 (88 keV), Co
-
57 (12
2 keV), Tc
-
99m (140 keV), and I
-
123 (159 keV). For
Compton scatter measurements, the point sources were Na
-
22
(511 keV and 1274 keV), Cs
-
137 (662 keV) and Co
-
60 (1173
keV and 1333 keV). Measurements were made at eight
locations in one quadrant of the forwa
rd hemisphere: local
source azimuth


was 0° or 45°, while local source elevation
z

was 0°, 15°, 30°, or 45°; Fig. 4 illustrates the definition of
local azimuth and elevation for source location.


Fig. 4. Definition of local azimuth and elevation for so
urce location; axis
x
3

is normal to and directed out from the front detector face.

For the high
-
energy photons evaluated here, many of the
scattered photons are likely to undergo a second scattering
interaction, rather than photoelectric absorption, so the

source
energy was assumed known and this value was used along with
the energy deposition in the first interaction in the Compton
scatter angle calculation. For Na
-
22, each data set was
analyzed by assuming both source energies in turn; for Co
-
60,
the assu
med energy was taken as 1253 keV, the average of the
two emission energies, because the actual photon energy
cannot be inferred from the scattering interactions. Because of
the partial range of scattering angles subtended by the side
detector and the 50
-
20
0 keV measurement range of the front
detector, not all source locations provided geometric scattering
angles that matched the measurable Compton scatter angles.
The ranges of geometric and Compton scatter angles that were
expected to match satisfactorily a
re summarized in Table 1.

Global angular error and angular resolution for the
prototype system and the simulations were assessed from
histograms of the difference between the Compton scatter
angles calculated from energy deposition in the front detector
an
d the geometric angles obtained from the pixel coordinates
and the known source location. Peak location in the histogram
determined angular error while peak full
-
width
-
at
-
half
-
maximum (FWHM) provided angular error. Two
-
dimensional
histograms of Compton ang
les vs. geometric angles were used
to assess error and resolution as a function of angle.

Comparison of mean and mode measured Compton scattering
angles vs. geometric angle was made between the L
-
shaped
prototype and the simulations of the L
-
shaped geometr
y.

To judge the quality of source localization in space, the
Compton scatter data from both the prototype system and the
simulations were backprojected as the intersection of cones
with a plane located at the known source distance. The peak
intersection po
int relative to the known source location was
used to determine angular error. The half
-
widths
-
at
-
half
-
maximum in the horizontal and vertical directions were
assessed for angular resolution.

III.

R
ESULTS

A.

Gross detector performance

For sources located 50
-
cm from

the L
-
shaped prototype
system, the measured coincidence count rates for valid
-
angle
producing events were 0.051 s
-
1

Ci
-
1

per front module for the
511
-
keV photon of Na
-
22, 0.018 s
-
1

Ci
-
1

per front module for
Co
-
60, and 0.018 s
-
1

Ci
-
1

per front module for C
s
-
137. The
count rates should increase several fold for a full 6
-
sided box
geometry compared to the L
-
shaped prototype.

B.

Angular error histograms

Fig. 5 shows a typical histogram for Cs
-
137 of the
differences between Compton scatter angles and geometric
an
gles; the source was located at (0°,0°). The angular error for
the sources and locations given in Table 1 were less than 5° in
all cases; angular resolution for this data ranged from 9°
-
32°
FWHM, with resolution being better at higher energies for the
larg
e azimuths and elevations because the forward
-
peaked
scattering probability resulted in more scatter photons incident
on the bottom side detector. For sources located at positions
that did not provide a good match between measurable
Compton scatter and geo
metric angles (e.g., combinations not
listed in Table 1), the corresponding histograms did not show
clearly defined peak locations or widths. In all cases, the
simulation of the L
-
shaped geometry produced histograms
similar to those of the prototype data.

Fig. 6 shows the corresponding 2D histogram (right) for Cs
-
137 at (0°,0°) of Compton scatter angles vs. geometric angles,
as well as two profiles showing the range of Compton scatter
angles that could corresponded to the same geometric angle
(top left) and

the range of geometric angles that produced the
same calculated Compton scatter angle. The profiles show how
the distributions are peaked around the correct true scattering
angle (corresponding to a small angular error), with the width
of the peak indicat
ing the angular resolution. In the 2D
histogram, a perfect detector should produce a 45° diagonal
line.

C.

Backprojections

Backprojection images for sources at various locations are
compared for the prototype, the L
-
shaped simulation, and the
6
-
sided box simu
lation in Fig. 7
-
12. Backprojections are
shown for the 0°, 15° and 30° source elevations at 0° azimuth
(Fig. 7, 9, and 11) and for all elevations at 45° azimuth (Fig. 8,
10, and 12). The sources were Cs
-
137 (Fig. 7
-
8), Co
-
60 (Fig.
9
-
10) and Na
-
22 (Fig. 11
-
12). The assumed energy was 1253
keV for the Co
-
60 data and 511 keV for the Na
-
22 data.

In general, the source location was the brightest spot in the
backprojected image, especially for the expected good source
locations listed in Table 1. The backprojecti
ons also tended to
produce brighter smaller intersections for sources located at 0°
azimuth, while the sources at 45° azimuth produced streaks or
arcs in the backprojections. The arcs were more pronounced
for the L
-
shaped prototype and simulation, and less

pronounced for the 6
-
sided box geometry; the box geometry
provided a more extensive collection of scatter cones which
resulted in a more coherent spot at the source location.


Fig. 5. Histogram for Cs
-
137 showing the measured Compton scatter
angles, the

differences between geometric angles and Compton scatter angles
calculated from the first interaction energy and the assumed source energy,
and (blue) the differences calculated using both measured interaction energies
instead of the assumed source energy
.


Fig. 6. For Cs
-
137 at (0°,0°), 2D histogram (right) of Compton scatter
angle (vertical axis) vs. geometric angle (horizontal axis). (Top left) Profile
showing reported Compton scatter angles for a given geometric angle of 35°;
(bottom left) profile sh
owing the range of geometric angles that resulted in the
same calculated Compton scatter angle of 35°.


Fig. 13 illustrates the effect of making a gross error in the
assumption of the source photon energy, for Na
-
22 when the
source emission energy was assum
ed to be 1274 keV. Most of
the detected photons were 511 keV; the large error in assumed
energy resulted in a systematic bias of the backprojected
cones, clearly visible in the images. By comparison, the small
error that occurs for every event for Co
-
60 (a
ssuming 1253
keV for the actual 1173 keV and 1333 keV photons) still
resulted in clearly defined source locations. Thus, a field
ECRD instrument should be able to indicate the occurrence of
a gross error but be relatively immune to small errors in
assumed
energy.

IV.

C
ONCLUSIONS

The 6
-
sided box simulation showed that reasonable
direction
-
finding capabilities were possible even with the
relatively coarse energy and spatial resolution of these CZT
detector modules. Some discrepancies between simulation and
protot
ype were due to the simple physics model used in the
simulation; the charge trapping and sharing model is being
incorporated into a more sophisticated GEANT4 simulation of
the hand
-
held electronically
-
collimated detector system.
Improvements to energy and
spatial resolution of the detector
modules should substantially improve overall system
performance.



Fig. 7. Backprojections for Cs
-
137 located at 0° azimuth.

Fig. 8. Backprojections for Cs
-
137 located at 45° azimuth.

A
CKNOWLEDGMENT

We thank the staff

of the Electronics Development Group in
the LSU Physics department for their assistance with
electronics development for the prototype system.

R
EFERENCES

[1]

Matthews II K.L., Smith B.M., Lackie A.W., Hill W.H., Wang W.
-
H.,
and Cherry M.L. An electronically
-
c
ollimated portable gamma
-
ray
detector for locating environmental radiation sources.
Proc. SPIE
Vol.
6319, X
-
Ray and Gamma
-
Ray Detector Physics and Penetrating
Radiation Systems VIII; article 63190H; doi: 10.1117/12.683568, 2006.

[2]

Rogers W.L., Clinthorne N.H
., and Bolozdynya A. Compton Cameras
for Nuclear Medical Imaging. in
Emission Tomography, The
Fundamentals of PET and SPECT

(Wernick M.N. and Aarsvold J.N.,

eds). San Diego: Elsevier Academic Press, pp. 383
-
419, 2004.

[3]

Kurfess J.D. Compton scatter imaging
in astrophysics.
IEEE Trans.
Nucl. Sci.
, 45(3):936

942, 1998.

[4]

Ryan J., Andritschke R., Bloser P.F., et al. MEGA: the next generation
medium energy gamma
-
ray telescope.
Proc. SPIE

Vol. 5488, UV and
Gamma
-
Ray Space Telescope Systems, pp. 977

987, 2004.

[5]

Boloz
dynya A., Ordonez C., and Chang W. A concept of cylindrical
Compton camera for SPECT.
Conf. Rec. of 1997 IEEE Nucl. Sci.
Symp. and Med. Imag. Conf.
, 2:1047

1051, 1997.

[6]

Du Y.F., He Z., Knoll G.F., Wehe D.K., and Li W. Evaluation of a
Compton scattering came
ra using 3
-
D position sensitive CdZnTe
detectors.
Nucl. Instrum. Meth. Phys. Res.
, A457:203

211, 2001.

[7]

Chelikani S., Gore J., and Zubal G.. Optimizing Compton camera
geometries.
Phys. Med. Biol.
, 49:1387

1408, 2004.

[8]

Seifert C.E., Jordan D.V., and Myjak M.J
. Simulated performance of the
GammaTracker CdZnTe handheld radioisotope identifier.
Conf. Rec. of
2007 IEEE Nucl. Sci. Symp. and Med. Imag. Conf.
, Paper N17
-
2,
2007.

[9]

Vetter K., Burks M., Cork C., et al. High
-
sensitivity Compton imaging
with position
-
sensi
tive Si and Ge detectors.
Nucl. Instrum. Meth.
Phys. Res.
, A579:363

366, 2007.

[10]

Lackie A.W., Matthews II K.L., Smith B.M., Hill W.H., Wang W.
-
H.,
and Cherry M.L. A directional algorithm for an electronically
-
collimated gamma
-
ray detector.
Conf. Rec. of 2006

IEEE Nucl. Sci.
Symp. and Med. Imag. Conf
., paper N14
-
19, 264
-
269, 2006.

[11]

Knoll G.F.
Radiation Detection and Measurement
, 3
rd

ed. John Wiley
& Sons; New York, 2000.

[12]

Schlesinger T. and James R.
Semiconductors for Room Temperature
Nuclear Detector Applicatio
ns
, volume 43 of
Semiconductors and
Semimetals
. Academic Press, San Diego, 1995.

[13]

http://www.phys.lsu.edu/~willhill/free_monte.html

[14]

http://physics.nist.gov/PhysRefData/Xcom/Text/XCOM.html.


Prototype
L
-
shaped box
simulation
6
-
sided box
simulation
Elevation:
0
°
15
°
30
°
azimuthal
direction
elevational
direction

Fig. 9. Backprojections for Co
-
60 located at 0° azimuth, with an
assumed
energy of 1253 keV.

Prototype
L
-
shaped box
simulation
6
-
sided box
simulation
Elevation: 0
°
15
°
30
°
45
°
azimuthal
direction
elevational
direction

Fig. 10. Backprojections for Co
-
60 located at 45° azimuth, with an
assumed energy of 1253 keV.

Prototype
L
-
shaped box
simulation
6
-
sided box
simulation
Elevation:
0
°
15
°
30
°
azimuthal
direction
elevational
direction

Fig. 11. Backprojections for Na
-
22 located at 0° azimuth for 511 keV.

Prototype
L
-
shaped box
simulation
6
-
sided box
simulation
Elevation: 0
°
15
°
30
°
45
°
azimuthal
direction
elevational
direction

Fig. 12. Backprojections for Na
-
22 located at 45° azim
uth for 511 keV.

Prototype
L
-
shaped box
simulation
6
-
sided box
simulation
Elevation:
0
°
15
°
30
°
azimuthal
direction
elevational
direction

Fig. 13. Backprojections for Na
-
22 located at 45° azimuth for 1274 keV.


Table 1. Source locations expected to match L
-
shaped prototype’s measurable Compton scatter range, by isotope and assumed emission energy.

Isotope

Assumed energy
(keV)

Compton scatter range
b

Source location (

,z)

Geo浥tric 獣sttering angles
c

Co
-
60

a
1253

15°
-

25°

0°,30°


-

21°
-

53°

Cs
-
137

662

30°
-

50°

0°,0°


-

44°
-

80°




0°,15°


-

30°
-

66°




45°,30°


-

41°
-

86°




45°,45°


-

35°
-

80°

Na
-
22

5
11

40°
-

70°

0°,0°


-

44°
-

80°




0°,15°


-

30°
-

66°




45°,0°

15°
-

60°
-

98°





45°,15°


-

49°
-

92°




45°,30°


-

41°
-

86°

Na
-
22

1274

15°
-

25°

0°,30°


-

21°
-

53°

a

average of 1173 keV and 1333 keV source emissions

b

for 100
-
200 keV

energy deposition in front detector

c

range of angles subtended by side detector, listed as minimum


average
-

maximum