Nuclear Instruments and Methods 172 (1980) 439-446
Q North-Holland Publishing Company
IMPROVEMENT OF THE PERFORMANCE OF A COMPTONSUPPRESSION SPECTROMETER BY
MINIMIZING THE DEAD LAYER OF THE CENTRAL Ge DETECTOR
H.J.M. AARTS, G.A.P. ENGELBERTINK, C.J. VAN DER POEL, D.E.C. SCHERF’ENZEEL and H.F.R.
ARCISZEWSKI
Fysisch Laboratorium,
Rijksuniversiteit
Utrecht, Princetonplein
5, 3508 TA Utrecht, The Netherlands
Received 12 December 1979
The performance of a Compton-suppression
spectrometer is investigated for two different Ge crystals as central detector; a
126 cm3 Ge(Li) with a dead layer of 1.0 mm and a 90 cm3 HPGe with a dead layer of 0.22 mm. The thin dead layer HPGe gives
a 32% improvement in the overall Compton suppression. Predictions of the influence of the dead-layer thickness by means of
Monte Carlo calculations are in good agreement. The spectrometer is further tested with radioactive sources of **Y and t szEu
and the in-beam reaction 24Mg + (45 MeV)160.
1. Introduction
the amount of non-detecting
material between the
central detector and the surrounding shield. As far as
the transmission
is concerned the dead Ge layer
around the central Ge crystal is the main source of
non-detecting
material and in the present paper its
effect
on the background
is investigated
by
comparing
CSS performance
for a conventional
Ge(Li) crystal with a 1 .O mm dead layer and a highpurity germanium (I-PGe) crystal with a 0.2 mm
thick L&diffused outer contact.
To investigate the effect of a continuous variation
of the thickness of the dead layer, Monte Carlo calculations have been performed.
During the last few years there has been an increasing interest
in the use of Compton-suppression
spectrometers
for in-beam -y-ray spectroscopy with
heavy-ion induced reactions, since the y-ray spectra
produced in these reactions are rather complex and
therefore demand spectrometers with high resolution
and low background.
In a Compton-suppression
spectrometer
(CSS)
the background, which normally almost completely
originates from Compton-scattered
Trays, is reduced
by surrounding the primary Ge detector with a scintillator, like NaI(T1) or plastic. Only events in the
central detector which occur without simultaneous
observation of scattered quanta in the surrounding
shield are accepted.
The inherent lower absolute efficiency of such a
spectrometer,
due to the required larger minimum
distance between the target and central detector,
would generally be a serious disadvantage for in-beam
r-ray spectroscopy. However, heavy-ion induced reactions show a reasonably large y-ray yield and a large
solid angle is not always very desirable in view of the
Doppler broadening
of the peaks. In practice the
lower efficiency can be compensated for by the use
of a higher beam intensity.
The actual performance of a CSS is, apart from the
dimensions
of the anticoincidence
shield, mainly
determined by the quality of the central detector and
2. Monte Carlo calculations
A Monte Carlo program has been written to
simulate the histories of photons in a CSS configuration shown in fig. 1. A 115 cm3 cylindrical closedend Ge crystal with a dead core as an inner contact
and a dead layer of variable thickness on the outside
is used as a central detector. The Ge crystal is surrounded by a 1.0 mm thick alumlnium cap. The
influence of the MgO reflector and of the aluminium
cover of the NaI can be neglected in the calculations
since their contribution
is relatively small, Multiple
processes and an appropriate angular distribution for
Compton scattering [l] are taken into account. It is
assumed that in the absorption processes no second439
Nuclear Instruments and Methods 172 (1980) 439-446
© North-Holland Publishing Company
IMPROVEMENT OF THE PERFORMANCE OF A COMPTON-SUPPRESSION SPECTROMETER BY
MINIMIZING THE DEAD LAYER OF THE CENTRAL Ge DETECTOR
H.J.M. AARTS, G.A.P. ENGELBERTINK, C.J. VAN DER POEL, D.E.C. SCHERPENZEEL and H.F.R.
ARCISZEWSKI
Fysisch Laboratoriurn, Ri/ksuniversiteit Utrecht, Princetonplein 5, 3508 TA Utrecht, The Netherlands
Received 12 December 1979
The performance of a Compton-suppression spectrometer is investigated for two different Ge crystals as central detector; a
126 cms Ge(Li) with a dead layer of 1.0 mm and a 90 cm3 HPGe with a dead layer of 0.22 mm. The thin dead layer HPGe gives
a 32% improvement in the overall Compton suppression. Prgdictions of the influence of the dead-layer thickness by means of
Monte Carlo calculations are in good agreement. The spectrometer is further tested with radioactive sources of 88y and 1S2Eu
and the in-beam reaction ~4Mg+ (45 MeV)I60.
the amount of non-detecting material between the
central detector and the surrounding shield. As far as
the transmission is concerned the dead Ge layer
around the central Ge crystal is the main source of
non-detecting material and in the present paper its
effect on the background is investigated by
comparing CSS performance for a conventional
Ge(Li) crystal with a 1.0 mm dead layer and a highpurity germanium (HPGe) crystal with a 0.2 mm
thick Li-diffused outer contact.
To investigate the effect of a continuous variation
of the thickness of the dead layer, Monte Carlo calculations have been performed.
1. Introduction
During the last few years there has been an increasing interest in the use of Compton-suppression
spectrometers for in-beam T-ray spectroscopy with
heavy-ion induced reactions, since the T-ray spectra
produced in these reactions are rather complex and
therefore demand spectrometers with high resolution
and low background.
In a Compton-suppression spectrometer (CSS)
the background, which normally almost completely
originates from Compton-scattered T-rays, is reduced
by surrounding the primary Ge detector with a scintillator, like NaI(T1) or plastic. Only events in the
central detector which occur without simultaneous
observation of scattered quanta in the surrounding
shield are accepted.
The inherent lower absolute efficiency of such a
spectrometer, due to the required larger minimum
distance between the target and central detector,
would generally be a serious disadvantage for in-beam
T-ray spectroscopy. However, heavy-ion induced reactions show a reasonably large T-ray yield and a large
solid angle is not always very desirable in view of the
Doppler broadening of the peaks. In practice the
lower efficiency can be compensated for by the use
of a higher beam intensity.
The actual performance of a CSS is, apart from the
dimensions of the anticoincidence shield, mainly
determined by the quality of the central detector and
2. Monte Carlo calculations
A Monte Carlo program has been written to
simulate the histories of photons in a CSS configuration shoxvn in fig. 1. A 115 cm 3 cylindrical closedend Ge crystal with a dead core as an inner contact
and a dead layer of variable thickness on the outside
is used as a central detector. The Ge crystal is surrounded by a 1.0 mm thick aluminium cap. The
influence of the MgO reflector and of the aluminium
cover of the NaI can be neglected in the calculations
since their contribution is relatively small. Multiple
processes and an appropriate angular distribution for
Compton scattering [1 ] are taken into account. It is
assumed that in the absorption processes no second439
440
H.J.M. Aarts et al. / A Compton-suppression spectrometer
absorbed in the NaI shield exceeds a lower level of
50 keV.
In fig. 2 we show the calculated Compton suppression averaged over the energy region 1 0 0 - 1 1 0 0
keV. The curve is relative and normalized to unity
for a dead layer of 1.0 mm.
For a reduction of the dead layer from 1 . 0 - 0 . 2 2
mm, the Monte Carlo calculation predicts an improvement of about 35% for the overall Compton suppression.
In the next section the Monte Carlo prediction is
compared with test measurements.
¢48
3. Experimental results
Fig, 1. Schematic drawing of the Compton-suppression
spectrometer. The 3,-rays enter the system through a collimator at the left hand side. At the right hand side two of the
six multipliers are indicated. All dimensions are in mm.
ary electrons or bremsstrahlung escape from the
central crystal, so that the validity of the calculations
is restricted to not too small Ge volumes and not too
high-energy 3,-rays (E~/~< 3 MeV),
In the Monte Carlo calculations spectra of 6°Co
are generated. To obtain a spectrum, l 0 s photons for
each energy (1.17 and 1.33 MeV) are emitted from a
point source in front of the collimator. The signal
from the central detector is vetoed when the energy
The geometry of the NaI(TI) anticoincidence
shield is shown in fig. 1. The shield has been used in
stu.dies of high-spin states in the sd shell as reported
in ref. [2].
Two crystals have been compared as central
detectors; a 126 cm 3 Ge(Li) crystal with a dead
layer of 1.0 mm and a dead core and a 90 cm 3
HPGe crystal with a dead layer of 0.22 m m and a
hollow core. More specific properties of both
detectors are given in table 1.
Table 1
Properties of the two detectors used as central crystal in a
CSS
Z
O_
I
I
I
I
B
I
Manufacturer
1.6
UJ
rt
5)
Ey= 1OO- 11OO keV
1.4
\
~o
n
1.2
O
o
1.O
_..I
Ld
n,'
I
O.0
I
1
O .4
L
L
0.8
I
I
1.2
THICKNESS OF DEAD LAYER (MM)
Fig. 2. Result of the Monte Carlo calculation for the relative
Compton suppression as a function of the thickness of the
dead layer of the central detector. The curve is normalized to
unity for a dead-leayer thickness of 1.0 mm. The statistical
errors are smaller than the points.
Type
Active volume
Core
Thickness
"lead
•
layer
Thickness
Al-cap: front
side
Operating
voltage
Efficiency
fwhm
(1.33 MeV,
60C0
Philips
Nederland BV
Ge(Li)
126 cm3
Inactive germanium
1.0 mm
Harshaw Chemic BV
Holland
HPGe
90 cm3
Hollow
0.22 mm
0.5 mm
0.5 mm
4500 V
0.5 mm
0.8 mm
3000 V
27%
2.4 keV
20%
1.9 keV
)
46
Peak-toCompton ratio
50
H.J.M. Aarts et al. / A Compton-suppression spectrometer
441
3.1.
The thickness of the dead layer of the HPGe
detector
The thickness o f the dead layer o f the HPGe
detector was obtained from the observed area ratio o f
the ( 7 9 . 6 + 81.0) keV complex and the 53.1 keV
3'-rays from a collimated ~SaBa source, in the way as
shown in fig. 3.
With this method the thickness o f the dead layer
is given b y
,oo=
As3
/
\ Iss
!
OAt=O.8rnm /el
•
•
~
(80"81~)
A
= 21.5*_o.2
A53
tO
I.--
Z
x104
53,1keY
i
cr
with dGe = the thickness o f the Ge dead layer in cm,
A (8o+81)/-4 s 3 = the observed ratio o f the area o f both
the (80 + 81) keV and 53 keY peaks, I(8o+al)/Is3 =
ratio o f the intensities o f the (80 + 81) keV and 53
keV incoming radiation,
[,"20°/o HPGe
]~ (80,81) =166"-02
I53
2
,4:,
I
Z
J ~ : l
0
81.0 key
keV
I
I
180
200
//
1
i
r
280
CHANNEL
p = #Al(53 keV) - #Al(81 keV) = 0.376 cm - l
3O0
NUMBER
1OO0
(the difference o f the linear attenuation coefficients
for aluminium, see ref. [3]), dAj = the thickness Of
the aluminium can in cm and
r,r"
g
o
w
o
q = #Ge(53 k e V ) - #Ge(81 k e V ) = 10.62 cm -1
(ref. [3]). The term PdAi in eq. (1) is practically
negligible for the present case with dAl = 0.08 cm.
The intensity ratio I(so+al)/Is3 is given b y Gehrke
et al. [4] as 14.8 + 1.1.
Because this value is rather inaccurate, the
intensity ratio has been measured with a 5 cm 3 lowenergy p h o t o n spectrometer (LEPS) with a 2 5 0 / a n
thick Be entrance window, an ion-implanted 0.3 # m
thick front contact and 10 m m depletion depth. The
data are given in fig. 4. They lead to I(8o+81)/Isa =
16.6 ± 0.2. Corrections due to self-absorption in the
source, entrance window, front contact and depletion
depth are taken into account. Their combined
influence is less than 1%. Measurements o f the irradiation at different angles also give consistent results.
With I(ao+al)/Is3 = 16.6 + 0.2 and the present
d/d = 0.08 cm, eq. (1), reduces to
100
20
30
40
50
AREA(80 • 81) / AREA(.53)
Fig. 3. Method to measure the dead-layer thickness of a Ge
detector. The method is based on the difference in transmission of the dead layer for the (80 + 81) keV complex and the
53 keV ?-ray of a laaBa source. For the present HPGe
detector the area ratio amounts to A(80+81)/A53 = 21.5 +
0.2. The lower part gives the thickness of the dead layer as a
function of the measured area ratio.
bO
F-
5 cm 3 LEPS
COLLIMATED 133Ba
z
0
(j
n-
531
79.6
81.0 keY
4000 A
iJ.I
m 2000
E
d o e = LFln(-a('°+s')I-Y
A,3 ] (2.839 + 0.012)]
Z
X 0.0942 c m .
(2)
This relationship is depicted b y the curve in the lower
part o f fig. 3.
The measured area ratioA(ao+s])/Asa = 21.5 + 0.2
O~L
800
J
I
820
840
//
1220 1240 1260
CHANNEL NUMBER
Fig. 4. The 53.1, 79.6 and 81.0 keV ?-rays from laaBa
measured with a 5 cm a LEPS.
as
442
H.J.M. Aarts et al. / A Compton-suppression spectrometer
i.e. the Compton suppression as a function of energy.
The loss of pulses in the photopeak due to random
veto's is less than 2%. The average Compton suppression between 100-1100 keV amounts to 11.2. The
Compton edges are reduced to little bumps due to
the small diameter of the entrance hole in the NaI(T1)
shield. Their shape differs appreciably from that of a
normal peak. The single-escape peak of the 1332 keV
?-ray is suppressed with the same factor as the
background while the double-escape peak disappears
due to the absence of collinear holes in the NaI(T1)
shield.
The areas of the 1173 and 1332 keV peaks taken
together amount to 65% of the total number of
registered counts for the suppressed spectrum and
18% for the non-suppressed spectrum.
The Compton suppression for 6°Co with the 1.0
mm dead layer Ge(Li) crystal as the central detector
is obtained similarly as described above. The result is
shown in the middle part of fig. 6. The average
Compton-suppression between 100 and 1100 keV is
now 8.4. The upper part of fig. 6 again shows the
Compton suppression for the 0.2 nun dead layer
HPGe.
implies a thickness of (215 -+ 15) #m for the dead
layer of the HPGe detector.
For the 126 cm 3 Ge(Li), the area ratio is measured as 48.3 + 1.1, which corresponds to a deadlayer thickness of (980 + 25)/am.
From the data of fig. 4 follows as a byproduct
X81.0/I79.6
= 12.5 + 0.2 for the intensity ratio of the
81.0 and 79.6 keV ~,-rays from 133Ba.
3.2. Experimental performance o f the CSS for two
different crystals as the central detector
With the HPGe crystal as the central detector a
6°Co spectrum has been recorded with a PDP 11/40
computer. The HPGe-NaI anti-coincidence pulse is
used to gate two ADCs, such that two spectra, one
coincident and one anti-coincident with the NaI
pulse, are recorded simultaneously. The results are
shown in fig. 5. The upper spectrum is the sum of
the two recorded spectra and represents the result
without Compton suppression (i.e. the normal HPGe
spectrum). The lower spectrum is with Comptonsuppression. The insert gives the ratio between the
number of counts in the upper and lower spectrum,
'
I/}
I-"
z
'
I
20%
'
I
'
I
'
I
1
1173
'
HPGe DETECTOR
'
I
1332
6OAo~(,.. ~ 1.9 keV
FWHM
23 crn x 28 crn NaI (TI)
105
WITHOUT
NaI VETO
O
u
0
I
C OMPTON - SUPPRESSION SPECTROMETER
106
"I"- 3.7 keV
133,?-m
511
'
lO 4
i
FWTM
(
w.T...I
1.1.1
1332'
1
..
_
tr~ 1o 3
I
D
Z
Z
~o
1
h
I
200
,
I
400
,
I
600
i
I
8OO
,
I
1OOO
,
i
1200
,
I
14OO
Ey(keV)
Fig. 5. Spectra of ~°Co for a 90 c m 3 HPGe crystal as central detector. The upper spectrum is without Compton suppression, the
lower spectrum is with Compton suppression. The insert shows the ratio between the number of counts in the upper and lower
spectrum, i.e. the Compton suppression a s a function of the energy.
H.J.M. Aarts et al. / A Compton-suppression spectrometer
25 _
.J
'
I
1
'
I
'
[
'
l
443
'
I
'
/
%.;..;
20
®
20
%
HPGe
02
mm
DEAD
LAYER
15
z
10
0
AVERAGE FOR 1OO-11OO keY is 11.2
w
• -%
o_
8_
O
z
0
15
••
27%
.
I-D_
0
Ge(Li)
1Omm
LAYER
DEAD
10
@
U
AVERAGE
,
18
I
FOR
*
1OO-11OO
J
keY
,
is
I
8.4
,
'%% .%
I
,
I
J
I
*
b
@
AVERAGE FOR 1OO-1100 keY is 1.32
r~
z
<
lO
(~
14
•°
.°
•%'°°°;
°.%.
".-.•
LL
O
• 2.....•_.
12
.
.
.
.
•.
O
°.,., °
=
1
200
l
1
400
t
1
600
,
l
,
800
l
1OOO
t
I
12OO
,
-f
1400
Ev(keV)
Fig. 6. Comparison of the Compton suppression for 6°Co for the 1.0 mm dead layer Ge(Li) and the 0.2 mm dead layer HPGe
crystals as the central detector. The differences for E 7 < 200 keV between the top and the middle curve are probably due to
different timing properties.
The lower parts o f fig. 6 gives the ratio of the two
Compton suppressions and thus shows the improvement in performance of the spectrometer as a
function of energy. The improvement is especially
notable for the (high-background) region around the
Compton edges ( 8 0 0 - 1 1 5 0 keV), i.e. the energy
/A
H.J.M. Aarts et al.
444
'
I
Compton.suppression spectrometer
'
I
8ay
'
'
I
898
I
1836
1o~
.,,- 2.2 keY
FW 0,.SM
I-
z
lO4
0
(J
511
1~'
--',,-4.1 keV
FW0.1M
1~36'
103
•.=. ~ 6AkeV
FWO~IM:
0
,=,,
10 2
,a
~11~
.~.5 - "
.
.
[
.
.
.
.
.
,4eI
.
l.. A._._
.4J
Z
10
I
I
I
I
5OO
I
1000
I
I
1500
2OOO
E¥(keV)
Fig. 7. Compton-suppressed spectrum o f a 8 S y source.
3.3. Performance of the CSS with the thin-deadlayer HPGe as central detector for radioactive sources
Say and lS2Eu and the in-beam reaction 24Mg+
(45 MeVf160
region corresponding to the escape of Comptonscattered photons of low energy from the central
detector. Reduction of the dead layer surrounding
the central detector makes it more transparent for
these low-energy ~(-rays, which can now reach the
surrounding shield more easily to produce a veto
signal.
The average improvement for 100-1100 keV is
32% in good agreement with the Monte Carlo prediction of 35%.
'
,
In fig. 7 we give the spectrum from a SSy source.
The peak-to-Compton ratio for the 1836 keV peak is
650, i.e. the residual Compton background is lower
by almost three orders of magnitude. The sum of the
I
I
I
I
122
245
Ul
I--
344
779
411 444
867
10E
lO4
0
Is_
0
w
IE
1112
2j
z
0
964
1408
1213
1299
1458 1528
152
Eu
z
I
1
I
I
I
I
I
200
400
600
800
1000
1200
14OO
E¥ (keV)
Fig. 8. Compton-suppressed spectrum of a I s 2Eu source.
445
H.J.M. Aarts et al. / A Compton.suppression spectrometer
z
5Olr 7
~
~
COMPTON
~"~,
20~-
t
°"~. . . . . .
o
0
SUPPRESSION SI=ECTROMET
W,TH A gO CM 3 HPGE DETECTOR
152Eu
182Ta
05
10
20
15
25
30
35
Ey(MeV)
Fig. 9. The relative full-energy peak efficiency for the CSS
with the 90 cma HPGe crystal as the central detecter.
areas of the 898 and 1836 keV peaks amount to 62%
of the total number of registered counts.
In fig. 8 we show the more complicated 7-ray
spectrum of a lS2Eu source. The relative full-energy
peak efficiency for the CSS has been measured with
S6Co (ref. [5]), 133Ba (ref. [4]), 152Eu (ref, [5]) and
182Ta (ref. [4]). The relative intensities are taken
from the references indicated. The result is shown in
fig. 9. The slope of the efficiency curve for higher
energies is less steep, when compared with the HPGe
detector outside the NaI shield, This reflects the
thickness (48 mm) of the selectively irradiated part of
the central detector (see fig. 1).
In fig. 10 we show the low-energy part (E7 < 2
MeV) of a singles ?-ray spectrum of the heavy-ion
reaction 24Mg + 160 at E(160) = 45 MeV and 0,t =
90 °. The measuring time was 5 h with a 1606÷ beam
current of 100 nA (electrical) on a 320/zg. cm -2,
99,94% enriched, ~4Mg target on thick Au. The
upper spectrum is without the NaI veto. The lower
one is the Compton-suppressed spectrum. The
suppression varies from about 7 to 12.
This spectrum clearly shows the advantage of a
CSS for in-beam 7-ray spectroscopy.
Weak and hardly visible peaks in the upper
spectrum stand out clearly in the lower spectrum, see
e.g. the 350 and 1400 keV regions.
24Mg ", (45 MeV) 160
670
i
8.y =
WITHOUT
90 °
"1
123
ffl
f-
! I
Z
o
A,#v
(.)
1
o
200
400
600
n~
800
I
1000
1642
w
m
1822
z
I .
1200
. .
L
1400
I
'~00
14800
2000
E~ (keY)
Fig. 10. Suppressed and non-suppressed v-ray spectrum observed with the CSS at 0~ = 9 0 ° in the bombardment of a 320 pg
cm -2, 99.94% enriched, ~4Mg target on thick Au with a 45 MeV, 100 nA, t 6 0 e~ beam after 5 h measuring time.