ARTICLE IN PRESS

Applied Radiation and Isotopes 60 (2004) 167–172

Application of GESPECOR software for the calculation of

Abstract

In this work, coincidence summing correction factors have been measured for 133Ba, 152Eu and 88Y point sources with
a 50% relative efficiency p-type detector and a 25% relative efficiency n-type detector in two close-to-detector
measurement geometries. The experimental data for 133Ba and 152Eu and the results obtained with the GESPECOR
software reveal a complex structure of the conventional dead layer of the p-type detector. The high value of the
coincidence summing correction factor for the 511 keV peak of 88Y, in agreement with the values computed by
GESPECOR, in this case cautions against the application of the semiempirical method for evaluating coincidence
summing effects.
r 2003 Elsevier Ltd. All rights reserved.

Keywords: Gamma-ray spectrometry; Ge detectors; Coincidence summing effects; Detector dead layer

1. Introduction 2. Experimental method

The availability of high-efficiency Ge detectors and Calculations and measurements were performed for a
the tendency to choose measurement geometries appro- p-type coaxial HPGe detector with 50% relative
priate for obtaining very low-detection limits emphasizes efficiency and a n-type coaxial HPGe detector with
coincidence summing effects in present-day measure- 25% relative efficiency. The two detectors have similar
ments. The GESPECOR software (Sima et al., 2001) Al end caps and are shielded with electrolytic copper
was specifically developed for the evaluation of coin- (thickness 7.5 mm) and lead (thickness 20 cm). The
cidence summing effects and matrix effects under typical measured spectra were analyzed with the INTERWIN-
measurement conditions using Ge detectors. In this NER (version 4.1) computer program from Eurisys.
work, experimental values of coincidence summing To emphasize coincidence summing effects, the
effects were measured in specific cases. They included measurements were performed with type PA or PE
coincidences with low-energy photons able to reveal the point sources (PTB 6–11, 2003) placed on a special
structure of the dead layer of the detector, and holder providing a small distance between the active
coincidences with secondary radiation produced by the deposit and the plane of the end cap (2.65 mm for PA
photons emitted by the source in the media surrounding sources and 2.8 mm for PE sources and using GESPE-
the detector. The ability of GESPECOR to deal with COR to calculate the correction for the different
such cases was studied. distances). Both detectors were calibrated in this
geometry (geometry A) using point source activity
standards from PTB which contained single radio-
nuclides showing negligible coincidence summing effects
*Corresponding author. Tel.: +49-531-592-6120; fax: +49- (e.g. 54Mn, 57Co, 65Zn, 85Sr, 109Cd, 137Cs, 241Am). Point
531-592-6109. source activity standards with the radionuclides 88Y,
133
E-mail address: [email protected] (D. Arnold). Ba, 152Eu, respectively, were measured in the same

0969-8043/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apradiso.2003.11.011
 ARTICLE IN PRESS
168 D. Arnold, O. Sima / Applied Radiation and Isotopes 60 (2004) 167–172

geometry. The latter sources were measured also in a where A represents the source activity and pi the
second geometry (B) for which a stainless-steel absorber emission probability. The apparent efficiency values in
(72.5% Fe, 18% Cr and 9.5% Ni, 0.97 mm thick disc) each geometry were divided by the corresponding values
was placed between the sources and detector. The from the efficiency curve to obtain the coincidence
activities of all the sources were in the order of 1 kBq summing correction factors Fc ðEi Þ:
to avoid random coincidences and pile-up; dead time eapp ðEi Þ
corrections were always below 1.5% and for most Fc ðEi Þ ¼ : ð2Þ
eðEi Þ
sources below 1%.
The spectra measured with the two detectors in
geometries A and B display specific features. Selected
3. Experimental results examples are given below.

The efficiency curve eA ðEÞ for geometry A was 3.1. p-type detector
established directly using the measured spectra for
nuclides with negligible coincidence summing effects. The 133Ba spectrum measured with the p-type detector
The efficiency curve eB ðEÞ for geometry B was evaluated in geometry A reveals the presence of a tail on the high-
using the efficiency transfer factor from geometry A to B energy side of the peaks (Fig. 1). This tail is absent for
computed by the GESPECOR software. The reliability measurement in geometry B. In fact, besides the decrease
of this method has been demonstrated elsewhere (Sima in the count rate in the peaks resulting from photon
and Arnold, 2002). From the count rate Ni in the peak attenuation in the stainless-steel absorber in geometry B,
with energy Ei from the measured spectra the apparent the prominent difference between spectra A and B is the
efficiency eapp ðEi Þ at energy Ei was computed for each behavior on the high-energy side of the peaks. The ratio
peak of 88Y, 133Ba and 152Eu by the formula of the count rates (spectrum A divided by spectrum B)
shows more conveniently the typical shape of this tail for
Ni various 133Ba peaks. Practically, the tail is associated
eapp ðEi Þ ¼ ; ð1Þ
pi A with each peak.

100 1000
geometry A
geometry B
ratio A/B
10
100

Ratio of countrates
Countrate in 1/s

10

0.1

1
0.01

0.001 0.1
0 50 100 150 200 250 300 350 400 450 500
Energy in keV
133
Fig. 1. The spectra of the Ba source measured with the p-type detector in geometry A and B (left scale); the ratio of the count rate in
spectrum A to the count rate in spectrum B (right scale). The tail on the high-energy side of the peaks from the spectrum measured in
geometry A is clearly seen both in the spectrum itself and in the ratio.
 ARTICLE IN PRESS
D. Arnold, O. Sima / Applied Radiation and Isotopes 60 (2004) 167–172 169

1.0E+06

778 keV
964 keV
Counts (scaled)

1.0E+05

1.0E+04
0 10 20 30 40 50 60
Channels (translated numbers)
Fig. 2. The peak shape for the 778 and 964 keV peaks of the 152Eu source measured with the p-type detector in geometry A. The data
have been vertically scaled to obtain similar peak heights. The tail on the high-energy side of the 964 keV peak is obvious.

The behaviour of the 152Eu spectrum measured in of K X-rays are clearly seen (Fig. 3). There is no tail on
geometry A is more complex. The tail is present on the the high-energy side of the peaks. This fact is more
high-energy side of the peaks produced by photons from evident from the ratio of the count rates (spectrum A
EC decay of 152Eu, but it is absent in the case of peaks divided by spectrum B) measured in geometries A and B
corresponding to b decay. This is evident in Fig. 2 (compare the ratio plots in Figs. 3 and 1). The ratio
showing the shapes of the peaks for the gamma rays reflects peaks at the energies of the X-ray peaks and at
with energy 778 keV (b decay branch) and 964 keV (EC the energies of the sum peaks of gamma rays and
decay branch). The two peaks have been chosen because X-rays.
they are close in energy, so the energy dependence of the As in the case of 133Ba, the 152Eu spectrum measured
peak shape should be relatively unimportant, and in geometry A reveals the presence of sum peaks
because they have comparable count rates. In order to associated with X-rays; there is no tail on the high-
emphasize the differences between the peak shapes, the energy side of any peak. In the 152Eu spectrum measured
peaks were scaled in Fig. 2 to obtain similar peak in geometry B, the sum peaks associated with X-rays are
heights. (the difference in the background in the low- absent and there is no tail on the high-energy side of any
energy part is due to the background around 964 keV peak.
which is lower than that around 778 keV; the difference
in the width of the peaks is the result of the increase in
the peak width with energy). 4. Discussion
In the case of 152Eu measured in geometry B, there is
no tail on the high-energy side of any peak. 4.1. Effect of the dead layer on coincidence summing
effects with low-energy photons
3.2. n-type detector
Theoretical values of the coincidence summing
In the 133Ba spectrum measured with the n-type correction factors Fc have been computed using the
detector in geometry A, the Ka and Kb peaks and also GESPECOR software (Sima et al., 2001) for all peaks of
133
the sum peaks of gamma rays with various combinations Ba and 152Eu appropriate to measurements in
 ARTICLE IN PRESS
170 D. Arnold, O. Sima / Applied Radiation and Isotopes 60 (2004) 167–172

100 1000
geometry A
geometry B
ratio A/B
10
100

Ratio of countrates
Countrate in 1/s

10

0.1

1
0.01

0.001 0.1
0 50 100 150 200 250 300 350 400 450 500
Energy in keV
133
Fig. 3. The spectra of the Ba source measured with the n-type detector in geometry A and B (left scale); the ratio of the count rate in
spectrum A to the count rate in spectrum B (right scale). The peaks do not show tails on the high-energy side. Sum peaks of gamma-ray
photons with various combinations of X-rays are evident in spectrum A and in the ratio.

geometry A and B with the p- and n-type detectors. For the above procedure in the case of p-type detector
a more accurate description of coincidence summing yielded a value d=0.46 mm for the thickness of the
with X-rays, the Monte Carlo simulation part of conventional dead layer. With this value, important
GESPECOR was updated, including now the possibility differences (second and last column in Table 1) between
of independently dealing with each individual X-ray the computed and the measured values of Fc for 133Ba
group (Ka1 ; Ka2 ; Kb1 ; Kb2 and L X-rays) instead of a and 152Eu in geometry A have been observed for p-type
single X-ray group. The database was also updated, but detector. Less important differences, generally below
for the moment it allows computation only for two 4% (the only exception being the 160 keV peak with
groups (Ka and Kb ). Generally the values of Fc for 6.5%) were noted for geometry B. For the n-type
coincidence losses due to X-rays computed using a single detector no systematic differences were observed using
X-ray group are very close to the values computed with the appropriate thickness of the conventional dead
separate Ka and Kb groups; the main advantage of the layer.
updated version lies in the possibility of computing Fc We believe that the differences observed in measured
for sum peaks with Ka or Kb X-rays. and calculated coincidence summing correction factors
The computed values of Fc are sensitive to the values with the p-type detector in geometry A are due to the
of several detector parameters. In particular, the higher complexity of the conventional dead layer
contribution of the coincidences with low-energy structure, with a gradual change from the true dead
photons to Fc depends on the characteristics of the layer to the completely sensitive volume of the detector.
dead layer of the detector. Because reliable data about In the transition layer the charge collection is incom-
the dead layer are hardly available either from the plete. As a result of this, the photon interactions in this
detector manufacturer or from direct measurement, the region do not contribute to the peak, i.e. this layer
thickness of the dead layer (hereafter called conventional behaves like a true dead layer for the values of FEPE.
dead layer) is usually inferred by comparison of the On the other hand, the interactions taking place in the
experimental values of the full-energy peak efficiency transition region contribute with some charge to the
(FEPE) at low energy with the corresponding values detector signal, so they contribute to the total efficiency
computed for various input dead layer data. Applying and, accordingly, to the coincidence summing effects.
 ARTICLE IN PRESS
D. Arnold, O. Sima / Applied Radiation and Isotopes 60 (2004) 167–172 171

Table 1
Experimental values Fcexp of the coincidence summing correction for the peaks at energy E of the 133Ba source measured with the p-type
detector in geometry A, compared with the values FcMC ðkÞ computed by GESPECOR for various values of k (the ratio of the true to
the conventional dead layer)

E (keV) Fcexp FcMC ðkÞ

k=0 k=0.05 k=0.10 k=0.13 k=0.15 k=0.2 k=0.4 k=0.6 k=0.8 k=1.0

53 0.618(22) 0.565 0.590 0.611 0.621 0.626 0.640 0.673 0.691 0.701 0.708
81 0.666(15) 0.631 0.653 0.671 0.679 0.685 0.695 0.722 0.736 0.742 0.747
160 0.879(17) 0.849 0.875 0.895 0.904 0.911 0.923 0.955 0.968 0.975 0.979
223 0.606(11) 0.490 0.536 0.575 0.594 0.606 0.633 0.706 0.745 0.770 0.787
276 0.634(7) 0.549 0.589 0.623 0.639 0.650 0.672 0.733 0.766 0.786 0.800
302 0.719(7) 0.614 0.660 0.696 0.715 0.725 0.751 0.814 0.847 0.865 0.877
356 0.792(8) 0.695 0.732 0.761 0.776 0.784 0.803 0.852 0.878 0.893 0.902
383 1.069(11) 0.941 0.984 1.02 1.04 1.05 1.07 1.12 1.15 1.16 1.17
437 0.057(1) 0.055 0.563 0.058 0.058 0.058 0.059 0.060 0.061 0.061 0.062
S104 1186 396 84 33 42 146 1034 1853 2448 2885

S represents the sum of the squares of the relative deviations of the computed values of Fc from the experimental values.

The experimental data presented in Figs. 1 and 2 layer of thickness d=0.46 mm, the majority of the Cs
support this interpretation. The fact that the high- X-rays interact in this layer, only about 5% ðKa Þ and
energy tail (in p-type detector peaks), observed for each 12% ðKb Þ penetrating through it. The shape of the
of the 133Ba peaks, but only for the 152Eu peaks measured spectra indicates that although practically
following EC decay paths (compare the 964 and each X-ray deposits its total energy in a region with
778 keV peak shapes) indicates that the tail is due to spatial extension much smaller than d, only a fraction of
X-rays. More exactly, the tail is produced by coincidence this energy and, accordingly, of the charge locally
summing between the events which otherwise would produced contributes to the detector signal. This means
have been registered in the peaks and the X-rays that the charge locally produced in the layer of thickness
accompanying the corresponding decay paths. Indeed, d is incompletely collected.
for each of the peaks with tail there is a great chance for In GESPECOR the above model has been implemen-
the photons corresponding to the peak to be emitted ted by introducing a new parameter k as the ratio of
together with X-rays. On the contrary, the photons from the thickness of the true dead layer to the thickness d of
the peaks of 152Eu following b decay paths have a the conventional dead layer. In this model, the conven-
much smaller probability of being emitted together with tional dead layer is composed of a true dead layer of
X-rays. Furthermore, in the high-energy tail of the 133Ba thickness kd and a partially active dead layer of
sum peak at the energy 437 keV (with dominant thickness ð1  kÞd: While the interactions taking place
combination 81+356 keV) the only possible contribu- in the first layer do not contribute to the detector signal
tion is from the X-rays. The similar shape of the at all, the interactions in the second layer contribute to
spectrum extending from 437 keV to 437 þ Ka;b with the the total efficiency but not to the full-energy peak
shape above each 133Ba peak is an indication of their efficiency.
common nature. The absence of the tails in measure- In Table 1 the computed values FcMC ðkÞ of the
ments in geometry B, in which the X-rays are highly coincidence summing correction for the main peaks of
133
attenuated in the absorber, is a further proof of the Ba measured in geometry A with the p-type detector
above interpretation. are listed for several values of k: The relative uncertainty
The X-ray photons have a much higher probability to of the computed values FcMC ðkÞ is in the range 2–4%.
produce photoeffect than Compton scattering in Ge The complete uncertainty budget can be found in Sima
(Berger and Hubbell, 1987); this is directly proven by the and Arnold, (2000). The experimental values are also
spectra measured with the n-type detector. Conse- included in Table 1. For each value of k; the sum S of
quently, most of the X-rays deposit their entire energy the squares of the relative deviations of the computed Fc
in Ge. As the range of photoelectrons produced by values from the experimental values has been added. As
the Cs X-rays (about 3–4 mm) or the mean-free path of can be seen from Table 1, good agreement between the
the Ge K X-rays (about 0.05 mm) is much shorter than computed and the experimental values is obtained for
the Cs X-ray mean-free path (0.15 mm (Ka ) and 0.22 mm k ¼ 0:13; accordingly, the true dead layer is about
(Kb )) in Ge, the cases in which part of the energy is 0.06 mm thick. When the differentiation between the
transported off the detector occur seldomly. For a Ge true dead layer and the conventional dead layer is
 ARTICLE IN PRESS
172 D. Arnold, O. Sima / Applied Radiation and Isotopes 60 (2004) 167–172

neglected (i.e. k ¼ 1), important discrepancies between p-type detector and with a 25% n-type detector in two
the computed and the experimental values are observed. close-to-detector measurement geometries. The results
The sensitivity of the computed values of Fc to the value obtained in geometry A with the p-type detector
of k for the geometry B measurement with the p-type revealed a complex structure of the conventional dead
detector is reduced; for k ¼ 0:13 good agreement between layer of the detector, including a partially active layer.
the computed and the experimental values is obtained. The model of this structure was included in GESPE-
In the case of measurement with the n-type detector, COR; also, an improved procedure allowing to deal
the thickness of the conventional dead layer is very small separately with the Ka and Kb X-ray groups was
and the measured data do not evidence a distinction implemented. The coincidence summing correction
between the true and the conventional dead layer. factors due to the contributions of low-energy photons
are sensitive to the structure of the dead layer. The
4.2. Coincidence summing effects for the 511 keV comparison of the experimental values of the efficiency
peak of 88Y and of the coincidence summing correction factors for
133
Ba with the computed values suggests for the p-type
In the case of point sources, the computation of the detector a thickness of the conventional dead layer of
coincidence summing correction factors is usually 0.46 mm, of which 0.40 mm is a partially active dead
possible with reasonable accuracy using the experimen- layer. The results obtained using GESPECOR are in
tally measured peak and total efficiency curves. On the close agreement with the experimental values for all
contrary, for volume sources the distinction between the peaks of 133Ba and 152Eu measured with the two
total efficiency and the effective total efficiency is detectors in geometries A and B.
important for the computation of coincidence summing The high experimental values of Fc for the 511 keV
corrections (Arnold and Sima, 2001). The Fc value for peak of 88Y (1.41 in geometry A and 2.93 in geometry
the 511 keV peak of 88Y was chosen to illustrate that the B with the p-type detector) caution against the applica-
method based on the measured peak and total efficien- tion of the simplified procedure for the evaluation of
cies, lead to wrong values even for the measurement of Fc based on the experimental peak and total effi-
point sources. The experimental value of Fc for this peak ciency curves. Good agreement with the values
is 1.41 for geometry A and 2.93 for geometry B for computed with GESPECOR was obtained, including
measurements with the p-type detector. The coincidence the production of 511 keV photons by positron decay
summing effects for the 511 keV peak are due to due to pair production by the 1836 keV photon in the
1836 keV photons. For the method based on the peak detector surroundings.
and total efficiency, the value of Fc should be lower than
1 because of the coincidence losses due to simultaneous
detection of the 1836 and the 511 keV photons. The References
experimental values are higher than 1 because additional
signals in the peak are produced by positron annihila- Arnold, D., Sima, O., 2000. Coincidence-summing in gamma-
tion following pair production by the 1836 keV photon ray spectrometry by excitation of matrix X-rays. Appl.
interactions in the source and the surrounding materials. Radiat. Isot. 52 (3), 725–732.
Arnold, D., Sima, O., 2001. Total versus effective total
The higher value for geometry B is easily understood
efficiency in the computation of coincidence summing
from the additional interactions of the 1836 keV photons corrections in gamma-ray spectrometry of volume sources.
in the absorber, followed by positron annihilation and J. Radioanal. Nucl. Chem. 248 (2), 365–370.
detection of the 511 keV photons. The production and Berger, M.J., Hubbell, J.H., 1987. XCOM: photon cross
transport of 511 keV photons were included in GESPE- sections on a personal computer. NBSIR 87-3597, 1987.
COR as previously for the study of coincidence PTB 6-11, 2003. Activity Standards; Internet information:
summing effects with matrix X-rays (Arnold and Sima, www.ptb.de/en/org/6/61/611/katalog/allgemeines en.htm.
2000). The values computed by GESPECOR (1.33 for Sima, O., Arnold, D., 2000. Accurate computation of
geometry A and 2.80 for geometry B) are in reasonable coincidence summing corrections in low level gamma-ray
agreement with the experimental values. spectrometry. Appl. Radiat. Isot. 53 (1–2), 51–56.
Sima, O., Arnold, D., 2002. Transfer of the efficiency
calibration of germanium gamma-ray detectors using
the GESPECOR software. Appl. Radiat. Isot. 56 (1–2),
5. Summary and conclusions 71–75.
Sima, O., Arnold, D., Dovlete, C., 2001. GESPECOR: a
In this work, coincidence summing correction factors versatile tool in gamma-ray spectrometry. J. Radioanal.
have been measured with a 50% relative efficiency Nucl. Chem. 248 (2), 359–364.