Nuclear Instruments and Methods 187 ( 1981) 451-472 451

North-Holland Publishing Company

CALCULATION OF THE ABSOLUTE PEAK EFFICIENCY OF GAMMA-RAY DETECTORS FOR

DIFFERENT COUNTING GEOMETRIES

L. MOENS 1), j. De DONDER O, LIN Xi-lei .1), F. De CORTE **l), A. De WISPELAERE 1), A. SIMONITS 2)
and J. HOSTE a)
1) Institutefor Nuclear Sciences, Rifksuniversiteit Gent, Proeftuinstraat 86, B-9000 Gent, Belgium
2) Central Research Institute for Physics, H-1525 Budapest 114, P.O.B.49, Hungary

Received 9 January 1981

A new technique is outlined for the calculation of the full-energy peak efficiency of cylindrical gamma-detectors, including
coaxial Ge(Li) detectors. Different source geometries are considered, i.e. point, disk and cylindrically shaped sources. Full
account is taken of gamma attenuation in the source and in any interjacent absorbing layer. No simplifying mathematical model is
adopted arid no Monte Carlo calculations are required. Although this method is to be qualified as semi-empirical, the experimental
work involved in it is simple and of common practice in most nuclear laboratories. For the calculations a FORTRAN IV+ com-
puter program is presented on a VAX 11/780 machine.

1. Introduction i.e. the 7-attenuation and the detector response to

impinging 7-rays. It will be shown in the present
Whenever absolute gamma-counting is performed, paper that these effects in principle cannot be cor-
it is necessary to have knowledge of the absolute peak rected for by separate calculation.
efficiency (ep) of the counting arrangement for the Others [ 3 - 8 ] incorporated this latter factor in
considered gamma energy (E.r). their calculations, which, however, were restricted to
Experimental determination of the peak efficiency the evaluation of the total efficiency (et). This was
as a function of E~/is simple as long as point sources done for simple cylindrical detectors such as NaI(T1)-
and large source-detector separations are concerned. crystals [3,4], planar Ge(Li) or intrinsic Ge-detectors
For extended sources or small separations, experi- [5,6] and for both point and disk shaped sources.
mental determination becomes cumbersome and The total efficiency of large coaxial Ge(Li) detec-
time-consuming. Indeed, it is a hard job to prepare tors, both single and double open ended, was calcu-
accurately calibrated 7-sources of various shape and lated as well, but only with respect to point sources
composition. Sometimes this is even not feasible, for situated on the detector's symmetry axis [7,8]. How-
instance when the efficiency curve valid for a bulky, ever, in none of the above methods attenuation was
metallic 7-emitter has to be determined. included in the calculations.
For this reason it seems preferable to approach the Gunnink and Niday [9] and Cline [10] developed
absolute peak efficiency problem by calculation. semi-empirical methods to find the peak efficiency of
There are many papers dealing with this kind of coaxial Ge(Li) detectors including inlplicit or explicit
work. corrections for gamma attenuation. Points, disks, but
Some authors [1,2] only calculate the geometri- also cylindrical and spherical sources were considered.
cal solid angle presented by the Source to the detec- These techniques, however, are based on approximat-
tor. For 7-counting this is unsatisfactory since it ing mathematical models, representing the detector
leaves undiscussed two major aspects of the problem, [9,10] and the source [9] as physical points.
It is the aim of this work to present a new and
accurate approach to the calculation of the peak effi-
* On leave from: The General Research Institute for Non-
ciency of coaxial Ge(Li) detectors. Evidently the
Ferrous Metals, Beijing, People's Republic of China.
** Research Associate of the "Nationaal Fonds voor Weten- method is a f o r t i o r i applicable to detectors without
schappelijk Onderzoek". p-core [such as NaI(T1), intrinsic Ge, etc.]. Point, disk

0029-554X/81/0000 - 0 0 0 0 / $ 0 2 . 5 0 © North-Holland
452 L. Moens et al. /Absolute peak efficiency of y-ray detectors

and cylinder sources will be considered. No use will where K and u account for the fact that only a limited
be made of any simplifying model, and the geometry fraction of the Compton and pair production effect,
aspect, the count yield of the detector and the gam- respectively, give rise to a full energy count.
ma attenuation in the source and in the interjacent It can be proved that:
absorbers will be treated simultaneously. Although
_ ~f _ ~c _~p
the method aimed for is a semi-empirical one the
experiment involved is short and simple and is com- /2t Pf Pc Pp
mon practice in most nuclear laboratories. where Pt is the total linear absorption coefficient of
the detector material for a gamma of the considered
energy (E7); gf, Pc and pp are the absorption coeffi-
2. Semi-empirical aspects of the method cients for photoelectric, Compton and pair produc-
tion effect respectively.
In order to count an event under the full energy Thus we obtain
peak, a gamma-photon, emitted from the source,
1 (Pf+PcK+PPU)~-~
should satisfy three conditions: ep = ~ ~ Pt
1) It should hit the active zone of the detector
without having undergone any energy degradation in 1P- P
the source itself or in the interjacent materials. - 4n y~2 =~et, (1)
2) It should interact with the detector material in
another way than by coherent scattering. with PIT the "virtual" peak-to-total ratio, referring to
3) The interacting photon should transfer its total the bare detector, without any surrounding material
energy to the detector material thus giving rise to a (e.g. Al-can, Ge dead layer, etc.).
count under the full-energy peak (poor charge collec- When c t is known, the problem is shifted to the
tion is not considered in the present approach). evaluation of the PIT ratio. The values of ~ and u are
Direct calculation of the peak efficiency (ep) leads unknown and can only be approximated by Monte
to very complex computations. The total efficiency Carlo calculations.
(et), on the other hand, is easier to calculate. The pre- Obviously the above defined PIT ratio cannot be
sent method therefore will yield the peak efficiency determined experimentally. It should be realized that
starting from the calculation of the total efficiency. It the usually determined peak-to-total ratio valid for
initially yields ~ , the effective solid angle, which the actual counting set-up-is not applicable here;
accounts for the probability that the above condi- indeed, the total area of the spectrum is contributed
tions (1) and (2) be fulfilled. The total efficiency is to by 7-rays which are scattered incoherently in the
then given by: interjacent and surrounding materials and which
reach the detector with degraded energy. The exper-
et = ~/(4n) .
imental peak-to-total ratio is therefore dependent on
et is related to ep by the following reasoning. The sample and counting geometry [11 ]. The PIT ratio of
peak efficiency can be considered to be composed of eq. (1), on the other hand, is relevant to a bare, active
different contributions: detector body. It is realised that this "virtual" peak-
- epf + e ; + e ~ , to-total ratio, especially in the low energy region, is
ep --
affected by secondary effects, such as poor charge
C
where efp, ep, epp are the fractions of ep corresponding collection, etc. However, it is assumed in this work
to photoelectric effect, multiple Compton interaction that this ratio is independent of counting and sample
and pair production, respectively. geometry and is an intrinsic characteristic of the
Analogous to ~ , we can define the effective solid detector used.
angles ~'2f, ~'-~c and ~2p, describing the probability for The semi-empirical determination of ep for any
the emitted gamma photon to interact by photoelec- sample and counting geometry can then proceed as
tric, Compton or pair production effect, respectively. follows:
Thus we can write: 1)Experimental determination of the ep vs E 7
curve for point sources positioned at a large reference
1 distance (denoted ref) from the detector; this is a
% = ~ ( ~ f + ~ K + fip~'),
standard procedure for which multi-gamma sources
 L. Moens et al. / Absolute peak efficiency oral-ray detectors 453

such as e.g. lS2Eu, 226Ra, aa2Ta, 166mHo, etc. can be Zd

1
used, since true coincidence can be neglected.
2) Calculation of ~ for this experimental reference
/
configuration and next for any required s o u r c e -
detector distance and source geometry; so as to
enable the construction of an ep vs. E v curve, this
should be done for a sufficient number of gamma- D×
energies.
3) The peak efficiency for any of the investigated
configurations (x) and for any of the different ener-
gies is then calculated from:

ep, x = 6 p , r e f ( ~ x / ~ " 2 r e f ) , (2) Fig. 2. Calculation of the geometrical solid angle (I2) for off-
axis point sources.
The only assumption made here, is the constancy of
the "virtual" peak-to-total radio. The experipental
results will prove that this assumption is justified with
certainty to within a few percent. The key to the solution for complex source geo-
metries is the expression of the solid angle for a point
source situated off the symmetry axis. For a such
3. Mathematical background of the calculation proce- point (T) the following general expression holds (see
dure fig. 2):

3.1. Geometrical solid angle (~2) =ff TP • n u d o , (4)

s ITp] 3
The calculation of the geometrical solid angle
where
presented by a radiation source of any given shape
to the detector is a very difficult task. For a point TP is the vector from T(XT, YT, ZT) to P(xp,yp, Zp) a
source located on the detector's axis of rotation the variable point in S, the plane surface formed by
problem is highly simplified: thanks to the high the detector top.
degree of symmetry the solution reduces to a single do is an infinitesimal area in S.
integration with respect to one variable i.e. the angle n u is the external unit vector normal to S.
0 (see fig. 1). We thus obtain:
After elaboration, eq. (4) can be written as:
cx
=f 27r sin 0 dO = 27r(1 - cos a ) . (3)
cz =
f ZT dx dy
0 [(Xp -- X T ) 2 + ( y p -- y T ) 2 + Z~]{. (5)

-,-,-,-

"-x

Fig. 1. Calculation of the geometrical solid angle (s2) for Fig. 3. Calculation of the geometrical solid angle (12) for off-
point sources coincident with the detector axis. axis point sources, using polar coordinates.
454 L. Moens et al. / Absolute peak efficiency o f .pray detectors

Introducing polar coordinates in the xy-plane (see Because of tile symmetry, eq. (8) can be simplified to:
fig. 3) eq. (5) is transformed into:
gO Ir

/~(R 2 ZTR dR dO
4 j,d,f
g~ = -~ 2Rr cos q~+ r 2 + z~)-~' (6) 0 0

with ×fo RdR

[R 2 - 2Rr cos¢+ r 2 + z ~ ] $
(9)
0
P: (R,¢, 0)
T: (r, % ZT); due to the symmetry it is allowed to
The relevant expression for a cylindrical radiation
confine T to the xz-plane (~ = 0),
source is given by (see fig. 4):
or
4 L r0 rr
~r no R dR fZ =-;.2o-z f (d + l ) d l / r d r f d(b
~2=2z T f
0
dO f [R=_2Rrcos¢)+?+z~l}'(7)
0
0 0 0

Ro
RdR
where Ro is the detector's radius.
As shown by eqs. (4)--(7) the expression for ~2
×f 0
[R 2 - 2Rr cos ¢ + r 2 + ( d + l) 2 ] ~ ' (10)
requires the solution of an elliptic integral, the value
of which can only be approximated numerically ; this where: d = the distance between the base of the
can nowadays quite easily be done by a computer. source and the top of the detector; L = the height of
The expression for the geometric solid angle pre- the source; l = z T - d.
sented by an extended radiation source to the detec- Analogously ~2 can be calculated for other source
tor can be found by making the point T variable geometries such as spheres and cones but also squares,
within the limits of the source. horizontal cylinders, although in the latter cases the
For a disk source (see fig. 4; D) at a distance z T above simplifications due to symmetry [eq. (9)] can-
from the detector we thus obtain: not be performed.

2Zw / 27r r0
,d, ; 3.2. hyfective solid angle f~)
~2 nr--~- o
- o o
The dependence of the detector efficiency on
o RdR 3'-energy compels the consideration of other factors
X [R ~ - 2 R r c o s d ) + r 2 + z ~ - ] } ' (8) in addition to the geometric solid angle.
0 The first of these factors is the attenuation of the
7-flux by the source and by any material situated
where ro is the radius of the source.
between detector and source, i.e. the source container
and support, the air, the N-can surrounding the
detector, the dead layer on top of the detector, etc.
To correct for this attenuation effect, the inner inte-
grand in eqs. (7), (9) and (10) must be multiplied by:

expt 4 i= I

where: /ai = linear, total narrow beam absorption

×
coefficient, excluding coherent scattering (denoted
"tot, t-coh" by Storm and Israel [12]) of the ith
absorber; 8 i = undisturbed path length of the gamma
through the ith absorber.
Fig. 4. Calculation of the geometrical solid angle (S2) for The total absorption coefficient is to b e used in
coaxial disk and cylinder sources. this context since any interaction occuring in an
 L. Moens et ak / Absolute peak efficiency o f ~,-ray detectors 455

interjacent layer, certainly leads to less of the 7-ray relatively easy. Again the simplicity and the sym-
from the full energy peak. The coherent scattering, metry of the system allow to express ~ii, r7 and A1, A2
however, makes an exception to this statement since as a function of only one independent variable, i.e.
it causes no energy degradation of the quantum, and the plane angle 0 (see fig. 1). The subtended solid
besides the process is strongly peaked forward [ 1 3 - angle is subdivided into different zones [7,8,16], each
15] so that practically the 7 is not significantly of which is characterized by two limitings angles
deflected from its original path. (0rnin and 0max) and by a specific expression for the
The second additional relevant factor is the proba- calculation of A1,2(0) and r/(0); the expression for
bility for a 7-ray impinging on the active zone of the 8i(0) is the same in every zone. Thus the calculation
detector to interact with the detector material before of ~ becomes rather simple and call be performed
leaving it. To account for this factor, the inner inte- even by programmable desk calculators such as HP-97
grand of eqs. (7), (9) and (10) is to be multiplied with or HP-41C [16].
a factor: For an off-axis point (T) it is impossible to follow
the same principles. The zones to be discerned would
Feff =/1 + f 2 f ' (11) be limited by very complicated functions of the vari-
with ables, while the expression of the travelled distances
in terms of geometrical and trigonometrical data
/'1 = 1 - e -udzxl and f2 = 1 - e -vdA2
would become cumbersome.
where:/a d = linear total narrow beam absorption coef- Therefore a decision was made in favour of a
ficient (excluding coherent scattering) of the detector totally different approach. This will be demonstrated
material; A~, A 2 = distance travelled in the detector for a single open-ended Ge(Li) detector counting a
(active zone) by an undisturbed 7-ray along PT cylindrical sample.
(fig. 4), before entering or after leaving the p-c0re Gamma-rays emitted from any extended source
respectively; f ' = e -"(n+al), with ,7 = distance tra- situated within an infinite cylindrical space, which is
velled in the p-core by an undisturbed 7-ray along PT;, coaxial with the detector and which has a radius
for 7s not passing the p-core (and evidently for detec- equal to the detector radius, can theoretically follow
tors without p-core) f2 = 0 and A 1 = distance tra- 8 different undisturbed paths within the detector
velled in the crystal. body. These are pictured in fig. 5 as (a) to (h). In
Note that by introducing the total absorption coeffi- addition the 7-rays can leave the cylindrical source by
cient, eq. (1 I) accounts for the total interaction pro- either the bottom (m) or the side (n), thereby passing
bability.
the sample container's bottom or side wall (see fig. 6)
After introducing the attenuation and efficiency fac-
(minor effects occurring with a 3' passing through the
tors, the expression for the effective solid angle is
edge of the container are not further detailed).
found. E.g. for a cylindrical source we thus obtain:
To distinguish between all of these possibilities,
4 Lj l) : o r : and to choose the correct algorithm for the calcula-
- - - / " (d+ dl dr dc~ tion of the travelled distances in the source (6t), the
- r2°L o o o
container (62), the active detector body (A1, A2) and
the dead p-core (r/), the following procedure was
Ro
FattFeffR dR developed.
×f0 [R 2 - 2 R r c o s , + r~ + (d + 0 = ]~" (12)
When evaluating numerically e.g. the integral

The distances 8i, ml, A2 and 77 (if relevant) are

depending on the position of both the points T and P
and thus on the integration variables R, ¢, r and l.
side
Their calculation will be the subject of the next sub- •

• bose
section.

3.3. Calculation o f 6i, A1, A2 and ~7

\
For a point source coincident with the detector's Fig. 5. a-h show eight possible paths to be followed by a
axis of rotation the calculation of 8i, A1, A2 and r/is ")'-rayimpinging on the detector top.
456 L. Moens et al. /Absolute peak efficiency of ,/-ray detectors

// - // - ts 2) calculation of the intercepts of TP with the

/
/
/
/
planes, denoted as 1,2 and 3 (see fig. 7);
/
/
/
/
3) calculation of the intercepts of TP with the
/
infinite cylindrical surfaces, denoted in fig. 7 as 1,2,
/// // 3 and 4;
4) the intercepts are next critically tested to find
out which type of path is followed by TP, with
I,, \ respect to the source, its container and the detector
body;
Fig. 6. Two possible paths for a ,/-ray, impinging on the 5) following the above conclusions, the relevant
detector, to leave the source and pass its container walls. distances (6~ = path length in the source, 82 = path
length in the container wall, A1, A 2 and r~) are calcu-
lated most of them analytically;
represented by eq. (12) (i.e. for cylinders) the point 6) the distances travelled in the other interjacent
P is varied within the limits of the detector top area; materials are next calculated trigonometrically;
T is varied within the area of intersection of the 7) the factors Feff and Fatt are calculated so that
xz-plane and the source body. Therefore the values of the actual value of the inner integrand of eq. (12) can
R , 4), r and l are changed discontinuously (see fur- be found.
ther). By a set of these four variables, the straight The procedure for point and disk sources is analo-
line TP is fixed in space. The problem to be solved gous.
can next be treated completely analytically as follows
(see fig. 7): 3.4. Numerical in tegration procedure
1) calculation of the relevant coefficients of the
two-point representation of the straight line TP; The most complicated multiple integration proce-
dure to be considered in this paper is the one that cal-
culates the efficiency for radiation originating from a
Iz cylindrical source [eq. (12)]. This case will be treated
here; the procedure for disk and point sources can be
=r, 2 (cyt 11 found by simplification.
The actual program makes use of the Gauss-Le-
=(r,+t,) 2 ( e y l /~)
gendre quadrature. In this method integration accu-
racy is increased over other methods by evaluating
the function at n specially picked sample points. Thus
z=d (plane 1) the evaluation of a single integral can be performed
by:
b

ff(x) dx =½(b- a) ~
i=1
w ~ {~ [zi(b - a) + b +a])
a

?1

i
= l ( b - a) ~ wif(xi) ,
z=-D (plane 2) i=1

where: f(x) = any function (continuous in the finite

interval [a, b]) of the independent variable x; a, b =
the lower and upper integration limit respectively;
z:-H (plane 3)
[zi]n=l are n constant quantities, the values of which
are listed [17] for n = 2 , 3 , 4 , 5,6, 7, 8,9, 10, 12, 16,
20, 24, 32, 40, 48, 64, 80 and 96; [wi]i=l n are n cor-
'l I [Link]=R,2 (cyt 3)
responding weighing factors [17].
',[Link]=R; z (cyt 2)
This method can also be applied to the numerical
Fig. 7. Analytical solution to the evaluation of Fat t and Fef f. evaluation of multiple integrals. This can be demon-
 L. Moens et al. / Absolute peak efficiency of q/-ray detectors 457

strated by the computer routine, written in 4. Program description

FORTRAN IV, which is actually included in the pro-
gram SOLANG (see the appendix). This routine 4.1. Input/output features
allows the number of base-points to be used for each
variable to be chosen separately, by assigning the For the calculation of ~ a program, named
required number to IDPI, IDPJ, IDPK and IDPL. In SOLANG, was written in FORTRAN IV+. The pro-
this work however, for each variable the same number gram reads values for the source and detector dimen-
of points was systematically taken, except for point sions, the source-detector distance, the thickness of
sources where IDPI= IDPJ = 1 and disk sources the interjacent layers and the container walls and
where IDPI = 1 ; this assignment indicates the com- finally for the gamma energies and the corresponding
puter to skip integration with respect to I and/or r. linear total attenuation coefficients in any relevant
material. Options can be taken with respect to the
wanted number of base points for the integration.
The output consists in a complete labelled record of
the input data and of a list of the calculated ~-values.

DETECTOR EFFECTIVE ~OLID ANGLE

DETECTOR HEIGHT e.5~4~E÷~t RADIUS ~,2615eE÷et

P-CO~E RADIUS ~,65~E+~ TIGAP ~,194~eE*~!

SOURCE RADIUS P.I~7~E÷~$ HEIGHT '~,134~E+P%

CONTAINER S I D E ~.162~E+~ BOTTCM ~,~VQBmE÷PF

EXTERNAL ABSOmBER-THICWNESS
~.6~BE-BI

SDURCE-DETECTO~ DISTANCE ~.13943E÷~! AIR LAYER ~.3P~3~E+~¢

GAUSS COEFF, O~OER ~ ~ 2~ ~

ENERGY MU-DETECTeR H L i - $ O U R C E ~U-CONTAINE~ EFFErTIVE SOLID aNGLE

EXTERNAL ABS~RPTIO~ CDEFFIqIENTS

~.1~3~E+~
~°16~E+~ ~o1173gE÷~1

mol~aaaE+~l
~.14~E+~

¢.t~qa3E+~1
P.85~R~E-~1

Fig. 8. Typical output of the program SOLANG.

458 L. Moens et al. / Absolute peak efficiency of y-ray detectors

Table 1
calculated with an increasing number (n) of Gauss-Legendre base points (Ey = 1000 keV, d : 14.143 cm).

n Point Disk Cylinder

percentile ~ percentile percentile

error error error

2 0.067379 + 13 0.064714 + 8.6 0.056401 + 9.0

4 0.060152 + 0.46 0.060062 + 0.78 0.052176 +0.81
8 0.060577 + 1.2 0.059522 -0.12 0.051698 -0.12
16 0.060412 + 0.90 0.059605 + 0.015 0.051771 + 0.025
20 0.059737 0.23 0.059591 0.0084 0.051753 -0.0097
24 0.059882 + 0.013 0.059605 + 0.015 0.051763 + 0.0097
32 0.059844 - 0.050 0.059597 + 0.0017 0.051758 0
40 0.059890 + 0.027 0.059596 0 0.051758 ~0
48 0.059861 - 0.022 0.059596 0
64 0.059870 - 0.0067 0.059596 0
80 0.059871 - 0.0050 0.059596 0
96 0.059874 ~0 0.059596 ~0

Fig. 8 shows a typical output for ten gamma energies source-detector distance and "r-energy the results are
representating the information needed for the con- shown in table 1.
struction of an ep vs. Ey curve for the given source It can be seen from table 1 that, when assuming
and detector geometry. the result for n = 96 (point and disk) or n = 40 (cylin-
der) to be correct, the percent error made by using
4.2. Accuracy o f the integration procedure fewer base points is below 0.1% for n as low as 24
(point source) or even 16 (disk and cylinder).
Special attention was paid to the investigation of In addition the same test was done for a disk
the accuracy of the integration. Therefore ~ was cal- source with a diameter of 2" (=5.08 cm) considering a
culated with an increasing number of base points. hypothetical bare 3" X 3 " NaI(T1)-detector and a
This was done for different counting geometries and source-detector distance of 0.2 cm. Between many
different energies; a single open ended coaxial Ge(Li) others, the effective solid angle for this arrangement
detector (fig. 10) was considered. For a representative can be found from the work of Grosjean [ 18], so that

. . . . . . . . . . . . . . . # . . . . . . . .
6.00 t.~- Isr)

5.00F •
[ 1 +10./. " • •
t..O0~- 1-10% • • • • • 0 0 0 0 0 0 0 0 3.9081 --

3.00 ~- • • •

2"00F • G r o s j e a n iii : 3.9131

,.o r . . . . . . . . . . . . . .

20 24 32 40 /dl 64 80 96 n

Fig. 9. ~ , calculated with an increasing number of Gauss-Legendre base points (n). Source: 2" diam. disk; detector: hypothetical
bare 3" X 3" NaI(T1); source-detector distance: 0.2 cm; E, r = 2070 keV.
 L. Moens et al. / Absolute peak efficiency of'r-ray detectors 459

results can be compared. For a representative Ptexigtass hotder Point Source

7-energy of 2070 keV the results are pictured in
126
fig. 9. It turns out that for this rather extreme case
the values are converging to a result which is con-
firmed by the value of Grosjean. Besides it is shown
that a 40 points Gauss-legendre quadrature for the I
variables ( R , ¢ and r) produces a result which is i 0.20
within 0.1% of the correct value (for n = 96).
i
i
4.3. Computation time
i
i
The program SOLANG was executed on a VAX I
11/780 machine with virtual memory operating sys- i
tem (VMS); it has a 258 kbyte memory and is i
i
equiped with a RM03 disk drive heaving 67 Mbyte
capacity. SOLANG occupies 60 blocks (of 512 bytes
i
i
each) on the disk. i
Typical calculation times are as follows: to obtain i
a set of 10 values of ~ i.e. for 10 different energies
(appropriate for a 10 points efficiency curve in the
7 0 - 3 5 0 0 keV range) the required cpu-time is not At : 0.051 J Air _ _ g a 0.099
p:
longer than 32 s, for point sources and n = 24. For a
disk source with n = 16, the elapsed cpu-time is 3 min
/ i/
,.-- a J iJ ..,.t 0.2g
41 s, while for a cylindrical source with n = 16, 30 i G =0,5
rain 28 s is needed. Of course the calculation time
g ! R, = 2.615
required by smaller computers would be significantly
ii
higher, but even an increase with a factor of 10 would o i
n- la rer : 0.06
keep the calculations still feasible. g i
II J
.j R,'=0.65
5. E x p e r i m e n t a l control

To check the accuracy of the present method, the i

ratio of the peak efficiency was experimentally deter-
i
i
mined:
1) for point sources measured at four different dis-
tances to the detector, one of these distances being Fig. 10. Characteristic dimensions (in cm) of detector 6;
chosen as reference (experiment 1); active volume = 104 cm3.
2) for disk and cylindrical sources with respect to
a point source measured at the same distance; this
was done also for four different distances (experi-
ment 2); two source diameters were considered and 5.1. Sources
for each diameter cylinders with two different heights
were measured. For the experiments we have in view, the sources
The experimental ratio of the peak efficiencies was to b e made must also be measurable close to the
compared to the corresponding X-ratio, which was detector. Therefore only isotopes emitting practically
calculated by the present method, in order to check non,coincident gamma-lines were considered. No cor-
the validity of eq. (2). Two different single open- rection was made for X - 7 coincidences, since in most
ended coaxial Ge(Li) detectors were involved, the cases the accompanying X-rays were soft enough to
parameters of which are pictured in figs. 10 and 11, be absorbed completely before entering the detector.
together with the dimensions of their respective plexi- The isotopes used are listed in table 2.
glass sample supports. These isotopes were either bought as aqueous solu-
460 L. Moens et al. / Absolute peak efficiency of y-ray detectors

Ptexic Lass holder Table 2

Practically coincidence-free gamma lines.

Isotope T1/2 E,y (keV)

Point Source l°9cd 453 d 88.04
57Co 271 d 122.06
o.1o/, 136.47
141Ce 32.50 d 145.44
117ms n 14.0 d 158.56
139Ce 137.2 d 165.85
iI 0.20
2°3Hg 46.60 d 279.19
51Cr 27.71 d 320.08
F 113Sn/113mi n 115.0 d 391.69
85Sr 65.19 d 514.00
I 137Cs 30.17 y 661.64
9SZr 64.03 d 724.18
1 756.72
i 95Nb 35.15 d 765.79
I 54Mn 312 d 834.83
I S6Rb 18.60 d 1076.77
i 65Zn 243.8 d 1115.52
142pr 19.13 h 1575.80
I
28A1 2.2408 min 1779.00
I 37S 5.06 min 3103.84
I
I
At : 0.0508 I Air 9o,e : 0.0,&5
/ , /
, ,.~. / I 0.307 Scotch tape. For the sources to be used in experi-
j G=0.5 ment 2, equal volumes of active solution were spotted
i~.. on Scotch tape, or equally distributed on a circular
R0=2.39 -,J
filter paper or put in a polyethylene vial and diluted
"
O i with water to obtain the wanted cylindrical shape.
i
,~
i*
] n - la' er:O.06 5.2. Measurements
_1 Ro'=0.60
-I J In experiment 1, point sources were measured at 4
different distances to the active zone of the Ge(Li)
I
detector (excluding the top n-layer). The activity of
i
the sources was kept low in order to avoid too high
count rates when measuring at low distance, which
implicates however long counting time at high dis-
tances.
Fig. 11. Characteristic dimensions (in cm) of detector 3; In experiment 2, the sources were measured at 4
active volume = 82.6 cm 3. distances as well. The cylinder sources were posi-
tioned such that the base of the liquid column cor-
responded to the position of the point and disk
tions from the National Physical Laboratories (NPL) sources.
at Teddington, Middlesex, England, or produced by
irradiation of suitable target materials. In the latter 5.3. Experimental inaccuracy
case radiochemical purifications were carried out, if
necessary. To the authors' experience, accurate experimental
For experiment 1 point sources were prepared by determination of efficiency ratios is not a simple task.
spotting some active solution or by deposition a small Rather large inaccuracies can be caused by source
amount of solid, activated compound or metal on preparation and by slightly deviating source position-
 .... I ....... I 461

(Ep)d o
'(~p)d
(Ep)ref ({plref . . . . I . . . . . . . I
34.0(
40.00

32.00

36.0C

30.00

~ . 2 /°
32.00

' ' ' ' i i

(tog)
i

I00 I000 E~ keY

. . . . , . . . . . . . i;2\
tepid I00 1000 E~. key (tog)
(Ep)ref . . . . I . . . . . . . . I
6.50 .... I ....... I ' " "
tepid
d = 5.726 cm (Ep)ref
5.5G

-~ 5.663 cm -t
6.00

6.00

5.50
oo
5.50
. . . . I . . . . . . . . I J , i=2,
I00 I000 E .~ keV (tog)
I00 I000 E T keV(tog)

.... I . . . . . . . . I
tepid "1 . . . . . . . . I " '
(Ep)ref (£p)d
I%--~f
2.10
d = 10.753 cm 2.10 d =10.608 cm --

o
2.00 .2~" 2.00

1.90 -- ~ 2 " , ~ .... I . . . . . . . . I

1.90
1oo 1ooo k~ kgv (t;g)
. . . . I . . . . . . I t ; J
I00 1000 E I keV (logl

Fig. 12. Comparison of calculated and experimentally deter- Fig. 13. Comparison of calculated and experimentally deter-
mined values of the efficiency ratio for point sources at dif- mined values of the efficineey ratio for point sources at dif-
ferent distances (d) to detector 6; reference distance (ref) = ferent distances (d) to detector 3; reference distance (ref) =
15.742 cm; experimental points: circles. 15.653 cm; experimental points: circles.
462 L. Moens et al. / Absolute peak efficiency oral-ray detectors

ing. The points shown in figs. 1 2 - 1 7 could only be ables:

obtained as an average of repetitive experimental
point sources n = 96 (cpu-time for a 20-points
determinations. Although no error bars are shown,
~-curve = 7 min 22 s),
the inaccuracy on the resulting points is estimated to
disk sources n = 40 (cpu-time for a 20-points
be about 2% for experiment 1 and 3--4% for experi-
~-curve = 54 rain 21 s).
ment 2.
cylinder sources n = 20 (cpu-time for a 20-
points ~-curve =
5.4. Results and discussion
2 h 23 rain 30 s).

The experimental results were next compared to Figs. 1 2 - 1 7 show this comparison. It turns out
the corresponding ratios of the calculated ~2-values. clearly that, as a general rule, the experimental
For the numerical integration the following number check points are within 2% or less from the curves
of base points was used for each of the relevant vari- obtained using the present calculation method. The

Cp, geo I .... I . . . . . . . . i .... I . . . . . . . I

Ep, point Ep, point

n n d=l.394crn _ co Q a
1.0 -- 1.0 - 0 0
G O

d=1.429 c m
0.8 - D [] D -- 0.8

d =1./~29 c m

d=4.116 cm
0.6 -- 0.6

.... I . . . . . . . . I • • .... I . . . . . . . . I i i i
100 I000 EE keY (log) I00 1000 E-II keY (Log)

¢ p, geo .... I . . . . . . . . I .... I . . . . . . . . I

£p, point

1.0 ~.j v w v 1.0 - °,o~ o o o -

[] rl O --

0.8 0.8 _ -

d=9.153 c m d=14.143 c m
0.6 0.6

.... I . . . . . . . . I .... I . . . . . . . . I
100 1000 Elf key Itog) I00 I000 EI~ keV flog)

Fig. 14. Comparison of calculated and experimentally determined values of the efficiency ratio for disk and cylinder sources (ro =
0.738 cm) vs. point sources measured at the same distance (d) to detector 6; experimental points: disk o; cylinder with L = 0.59
cm D;withL = 1.18 cm z~.
 L. Moens et al. /Absolute peak efficiency of ,,/-ray detectors 463

.... i ' ' ''"1 .... I ...... I

£:p,geo
~:p,point Ep,point
1.0 1.0
d=1.216 cm __n ~ o o o
--o o o o

0.8 _ d =1.394 c m _ 0.8

d =1.39/, cm
_ , ~ ~ ~ ~ ~ _ d=4.116 cm
0.6 0.6

.... I . . . . . . . . I . , , .... I . . . . . . . . I
100 1000 E ~.keV (tog) 100 1000 E~.keY (log)

.... I ...... I .... I . . . . . . I

£p,geo
(:[Link]
1.0 O O ,.0 - 0:' o ~ o
0
...~.....~..m------B-~

0.8 0.8 - - ~

d=9.153 cm
0.6 0.6lI d=14,143 cm_

.... I . . . . . . . . I I , ,,I . . . . . . . . I
100 lOgO Ell keY (tog) I00 I000 F'8 key (log)
Fig. 15. Comparison o f caluclated and experimentally determined values of the efficiency ratio for disk and cylinder sources (ro =
1.375 cm) vs. point sources at the same distance (d) to detector 6; experimental points: disk o; cylinder, with L = 0.67 cm t~; with
L = 1.34 cm a.

results of experiment 2 (figs. 14-17) show no rele- 6. Systematic errors in the calctdation o f ep
vant discrepancy between theory and experiment,
even when large cylindrical sources are measured The use of data such as the dimensions of the
close to the detector. When considering the results of detector and especially of its p-core, might call forth
experiment 1, a 2% bias can be observed in fig. 12 serious criticism of this method. It is indeed generally
(for d = 1.326 cm) and in fig. 13 (for d = 5.663 cm). assumed that such parameters are badly specified.
This can be due to either a positioning error, or an The experiments described in the former paragraph
inaccurate specification of the detector parameters however prove that for the considered Ge(Li) detec-
(see further). However, in view of the time-consuming tors the uncertainty on these parameters is either
and tedious work involved in the alternative experi- small, or has no effect on the final ratio of ~-values
mental determination and the errors inherent to it, (and thus of ep-values). Nevertheless it seems neces-
such bias is definitely acceptable. sary to study whether any important uncertainty of
It can therefore be concluded that the present the above mentioned parameters can make the appli-
method, applied to single open-ended Ge(Li) detec- cation of this method questionable.
tors can yield the absolute peak efficiency for any Therefore ~-values were calculated for different
counting geometry to within 3% (including 1% uncer- energies, using parameters which were deviating from
tainty on the experimental reference curve). the originally specified ones, as follows:
464 L. Moens et al. /Absolute peak efficiency of "r-ray detectors

.... I . . . . . . . I
.... i ' ' ' ' .... I
F-p, geo
C-p,point
1.0 ~ 0 0 d=1./,56 cm 1.0

0 0

0.8 0.8

A d=1.498 cm

0.6 0.6
.... I . . . . . . . . I
t00 I000 E~ keV (tog) d=1./,56 cm

.... I . . . . . . . . I i =
.... I . . . . . . . I 100 I000 F~ll keV (Iogl
(p,geo
rp, point
d = 5.8/.,6 cm .... I . . . . . . . . I
1.0 - o O O O O rp,geo
Ep,point
1.0 R

d =5.888 cm 0 CO 0 u 0

0.8

0.8 ~....~._~--.------- -

0.6
.... 1 . . . . . . . . [ i l l

)00 1000 E~ keY (tog)

0.6 _ m

.... I . . . . . . . . I d = 5.8t,6 cm
Ep, geo
.... I . . . . . . . . 1
d=15.836 cm ,oo ,ooo ~'~,,v' .o'~1
I.(3 - 0 .c~ 0 0
Fig. 17. Comparison of calculated and experimentally deter-
m i n e d values of the efficiency ratio for disk and cylinder
m sources (r o = 1.375 cm) vs. point sources measured at the
same distance to detector 3; experimental points: disk o;
0.~ cylinder (L = 1.34 cm) zx.

0~-, , , . I . . . . . . . . I i i i This was performed for one small (1.394 cm) and one
I00 I000 E 11"keV (tog) large 14.143 cm) source-detector distance. Taking the
Fig. 16. Comparison of calculated and experimentally deter- experimental (ep)poin t VS. E I, curve at 14.143 cm as a
mined values of the efficiency ratio for disk and cylinder reference, ep vs. E v curves were calculated [eq. (2)]
sources (r 0 = 0.738 cm) vs. point sources measured at the for:
same distance to detector 3; experimental points: disk o;
cylinder (L = 1.18 cm) A.
1) a point source at 1.3943 cm,
2) disk and cylinder sources at both distances
(1.3943 cm and 14.143 cm).
The results can next be compared to the correspond-
+10% for the detector dimensions, for the distance ing curves obtained with the originally specified (cor-
between the detector top and the M-cap and for the rect) parameter data. The results for detector 6 are
literature #-values, presented in figs. 1 8 - 2 0 and table 3. It is reassuring
+20% for the parameters referring to the p-core or that in the considered arrangement the data referring
the n-layer. to the p-core are not critical at all (figs. 18b, 19b,
 L. Moens et al. /Absolute peak efficiency of'r-ray detectors 465

.... I . . . . . . . . I .... I . . . . . . . . I
~+15%
8 .,s.,.I
,., + IOOh .10"/.~
-If) %
I::
•, +5o/, + 5%1
m

(u ,,lO%
.~ o'/, 0%1 -

~ _SOb - 5 "/,~
÷ I0 %
- I0"I, -10%!
"~Ge Ro
-15°Io -15°/.;
.... I . . . . . . . . I . . . . .... I. . . . . . . . . I
100 I000 EllkeV (tog) IOO IOOO E~.keV (tog)

~÷15°I, .... I . . . . . . . . I .... I ....... I

u + I0"1,

E
® +S'/,
-I00 "/.
-10%
--- Oe/~ O*/o1--
*10 "/. .10%
0- . 5"/,
-5% I
.10o/, -10%
G H
-15°/, -is,/,I- I
.... I . . . . . . . . I t i i .... t . . . . . . . . I
I00 I000 Ell keY (log] I00 I000 Eli keV (tog)

Fig. 18a. Percentile systematic error introduced on the calculated ep curve for point sources (d = 1.3943 cm to detector 6) by
inaccurate values; (a) linear attenuation coefficient of Ge(/~Ge), detector radius (Ro), detector-A1 window distance (G), detector
height (H); (b) p-core radius (R~), p-core top/detector top distance (D), thickness of top n-layer (see fig. 10).

Table 3
20b, table 3). Further it turns out that for disk and Error in ep by inaccurate detector parameters; results for disk
cylinder sources measured at the large reference dis- and cylindrical sources at d = 14.143 cm -= reference dis-
tance ep is quite insensitive to inaccurate parameter tance.
knowledge (table 3). The estimated uncertainty on
Source of error Error in ep
the data, concerning the top n-layer, the height of the (70 keV < E'r < 3500 keV)
detector body (H), the gGe-Values and the gap (%) (absolute value)
between the crystal and its can (G), yields only 1 - 3 %
disk cylinder
uncertainty on ep. The only really sensitive parameter
is the radius of the detector body (Ro); therefore /4Ge (+10%) <0.01 <0.04
manufacturers must pay the utmost attention to the top n-layer (+10%) .<0.01 <0.03
measurement of the external crystal diameter and for R0 (+10%) <0.02 <0.05
Ge(Li) detectors to the evaluation of the thickness of H (+10%) <0.01 <0.06
the n-layer surrounding the sensitive zone. Of course R~) (+20%) <0.01 <0.03
D (+20%) <0.01 <0.01
it is to be recommended also that all other parameters G (+20%) <0.02 <0.01
be specified correctly and the experimentor should
466 L. Moens et al. / Absolute peak efficiency o f z,-ray detectors

7. Inseparability of geometry, efficiency and attenua-

''"I ' ' ' ' ' '"l
+3% tion calculations

,2% It is tempting to suppose that the peak efficiency

.-
o
E of the counting arrangement can be separated into
*1*/, three individually evaluable factors.
+20%
1) A geometry factor, which takes full account of
-~ 0%
the geometrical aspects of the source-detector system,
-20%
and is invariant with gamma-energy.
.ioA
2) A 7-attenuation factor representing the count
losses due to incoherent interaction in the source and
-2%
Ro in any interjacent absorber; this factor depends on
-3%
"),-energy and counting geometry.
, , i ii , i i i i i [ i i i 3) An efficiency-factor, a mere function of energy,
lO0 tO00 E~ keY (tog}
which is characteristic for the detector but is indepen-
i i , i ' ~Ji i i
dent of counting geometry.
' ' "l
+3% In this reasoning several principal falsities can be dis-
covered, which lead to smaller or larger errors.
*2% First it is obvious that the geometrical solid angle
alone cannot fully account for the geometrical
E
+I % aspects of the peak efficiency: the curves shown in
- 20% figs. 12-17 leave no doubt in this respect. Besides the
: 0%
*20 % attenuation factor is inevitably dependent on the
geometry.
a. -I*/, Secondly, it turns out that there exists not such
a quantity as a geometry independent efficiency fac-
-2*/,
D tor. This can be demonstrated here: it is indeed pos-
sible to evaluate this "intrinsic efficiency factor",
-3%
i ~ ,11 i i , i illl i i i using the actual calculation technique, by calculating
I00 I000 Eli keY (log) for a bare detector and an attenuation-free source
i ' ''l
(Pi = 0) - thus excluding 7-attenuation (fat t = 1 ) -
' ' ' ' ' ' " l ' ' '
*3%
and by next dividing it by ;2 [calculated from eq. (3),
(9) or (10)], thus normalising ~ for pure geometrical
*2% effects.
This was done for point, disk and cylinder sources
E
÷1% at two different source-detector distances. The result
m
- 20%
is most striking when comparing the found efficiency

/
o~
+20%
factor for point sources at different distances. This is
shown in fig. 21 at distances of 1.3943 and 14.143
o_ -I% cm respectively from detector 6. It is clearly demon-
strated that this efficiency factor is not at all constant
-2% with counting geometry. The edge-effect comes out
top n- layer
quite clearly as a geometry dependent factor which is
-3%
i i ii1 i , i i t ti i i i
not included in neither the geometrical solid angle
I00 1000 Eli keV (log) nor the attenuation factor. This edge effect is owing
Fig. 18b. Percentile systematic error as in fig. 18a: p-core to 3'-rays escaping the crystal without interaction
radius (R~), p-core t o p / d e t e c t o r top distance (D), thickness through its edges, it is increasingly prominent with
of t o p n-layer (see fig. 10). increasing gamma-energy.
A third remark concerns the separate calculation
make sure that the source-detector cap distance is of the attenuation factor. Generally, the attenuation
measured with an accuracy of at least 0.1 mm. in the source, its container, the support, the window
 L. Moens et al. / Absolute peak efficiency o f v-ray detectors 467

..... I . . . . . . . . I ' ' .... I . . . . . . . . I

.15% +15%[
o

.o *10% ,10%
cI
E -10%
*5'/, +5%1

w +10%
0"h 0%
-,o'%
=- -5"h -5%
+ 10%

- 10% -10%
.~lc.-e Ro

- 15% -15'/,I
.... I . . . . . . . . I , , , ,I ...... I
100 I000 Eli keV (tog) I00 I000 .g keY (log)

.... I . . . . . . . . I ' '1 ' ' ' '"1

"•*lS% +15°/,
O
la

._o *10% +10%

0
I::
.5"/, +5°I,
¢ -10%
m -10%
ip
•-- 0"/, 0%

,.10% +I0%
~ -5% -5°/,

-I0% -10%
G H

-15% -15°/,
.... i , , , ,,,,,I , , , , ,I i 0 i i ill i i

I00 I000 E ] keV (tog) I00 I000 Ell keV (log)

Fig. 19a. Percentile systematic error introduced on the calculated ep curve for disk sources (ro = 1.375 cm;d = 1.3943 cm to detec-
tor 6) by inaccurate values; (a) linear attenuation coefficient for Ge(PGe), detector radius (R0), detector-A1 window distance
(G), detector height (H).

of the detector, etc. are calculated approximately, it is hit by radiation and thus attaches equal weight
assuming e.g. that the gammas pass the source and to the attenuation factor for any ?-direction within
enter the absorbers in a direction normal to the detec- the solid angle sustained by the detector surface.
tor face; sometimes an average 7-direction is intro- In reality the efficiency factor ( F e f f ) for e.g. a
duced, occasionally related to an effective sample Ge(Li) detector such as detector 6 (see fig. 10), with
radius and penetration depth [9]. Whatever simplifi- respect to an on-axis point source is a function of the
cation is used, it turns out to fail to more or less distance from the point of impact to the center of the
extent especially when close-in geometries are con- detector and thus is non-uniform over the detector
cerned. The most serious separate approach consists top, as shown in fig. 22. An analogous remark holds
in evaluating the integral, the integrand of which is when calculating the detector response by integrating
the attanuation factor as a function of the coordi- the efficiency factor over the detector surface; the
nates of the points were the radiation originates and attenuation should be considered here, since the
were it hits the detector surface. Even this approach gamma-flux, hitting the detector is no longer iso-
is principally incorrect, since it considers the detector tropic due to the unequally distributed attenuation in
as uniformly efficient with respect to the point where the source and the interjacent absorbers.
468 L. Moens et al. /Absolute peak efficiency o f ~-ray detectors

Neglect of these effects induces no errors larger

. 3 % I .... I . . . . . . . . I
than a few percent on the calculated attenuation or
efficiency factors but this does not alter the fact that,
._u +2%
to be principally correct, the efficiency, attenuation
E and inevitably also the geometry factor must be
+20% treated simultaneously by including them in the same
0%
integrand. Besides one should realize that even when
-20%
separate calculations are considered these cannot
~ _1%I evade numerical integration for neither of the three
factors under consideration [excluding the pure
-2% geometry factor for point sources: eq. (3)], unless
Ro simplifying models are judged satisfactory. By conse-
-3% quence it is obviously preferable to combine these
100 1000 E~- keV (log)
integrations into one, thus saving time and meeting
the requisites of theoretical precision.
.... I . . . . . . . . I
-- *3%

8. Conclusion
,~ +2%

The method described in this work has proved to

*1%
yield the absolute full-energy peak efficiency for the
-20% different counting geometries to within 3%. Geom-
O%
*20 % etry, 7-attenuation and detector response are treated
o- - I %
as inseparable elements of the peak efficiency, which
is proved to be a fundamentally correct approach.
-2%
Although the technique is essentially semi-empirical it
D needs only a little experimental work and calculations
- 3*/.
can be done within a feasible space of time even by
.... ~ . . . . . . . . J , L , small computers. The only supposition is the invaria-
I00 I000 EI~ key (log)
bility of the "virtual" peak-to-total ratio with count-
ing geometry; from the obtained results it can be con-
.... I . . . . . . . . I ' ' '
÷3./. cluded that this prerequisite is practically met.
The computer program developed applies to any
u +2"/ type of cylindrical 7-detector and to point, disk and
cylinder-shaped sources. It can easily be modified to
+1°/, evaluate the peak efficiency for other source geome-
= -20% tries; this will be shown in a forthcoming paper
0% R
+ 20 % together with additional experimental evidence,
including different detector types [19].
-I%

Thanks are due to Dr. R. Van Keer (Seminarie

-2./
top n-layer voor Wiskundige Analyse, Rijksuniversiteit Gent) for
his many enlightening remarks which smoothed the
-3*/
. . . . 1 , , , ,,,,,I , , , way of the development of this technique. The
100 I000 tr~ keV (log)
authors are also indebted to the "Nationaal Fonds
voor Wetenschappelijk Onderzoek" for financial sup-
Fig. 19b. Percentile systematic error as in fig. 19a: p-core
port (F.D.C.) and to the "Interuniversitair Instituut
radius (R~), p-core top/detector top distance (D), thickness
of top n-layer (see fig. 10). voor Kernwetenschappen".
 L. Moens et al. /Absolute peak efficiency of ?-ray detectors 469

I I . I
i i i l t 1, I
' '"1
*15% ,15"/,

+10% ~.10%

E -10%
o +5%
+5*/,
*10%
0"/,
-10%

o_ _5°/, -5"/, *10%

-IO*A -I0"/,
R o
JLIGe
-15'/, -15°/
i llll I I I I I I ,I I I I ,I I I I I I I II I I I
IOO I000 E T keY (log) IOO I000 E~ keY (log)

' '"1 '1 '1 . . . . . . . . I ' '

~,,15%
+15°/°I
,., +10°/o

+5%
-I0%
ZI - 10%

:,7. 0% o O/oh--
+10 % *10%
~ -5O/o

-10%
G H

-15"/, - 15%~- i
i,, f I I I I I I ,I I I I
IOO El keY(log} I00 I000 Elf keV (log)

Fig. 20a. P e r c e n t i l e s y s t e m a t i c error i n t r o d u c e d o n the c a l c u l a t e d ep_ curve for c y l i n d e r sources (r 0 = 1.375 cm, L = 1.34 c m ; d =
1 . 3 9 4 3 c m t o d e t e c t o r 6) b y i n a c c u r a t e values; (a) linear a t t e n u a t i o n c o e f f i c i e n t s for G e ~ G e ) , d e t e c t o r r a d i u s ( R 0 ) , d e t e c t o r - A 1
w i n d o w d i s t a n c e (G), d e t e c t o r h e i g h t (H).
470 L. Moens et al. / Absolute peak efficiency of'r-ray detectors

. . . . . I I
.... I ....... I
+3"/,
1.0

,~ +2°/ 0.9
"6 t,
u
-~ +1"/. 0.8
+ 20 % =
,~ (17
~: 0% ._u

-20 % (t6
a_ -I'/, -~
c m
"~ os --
-2%
R0 0.4

-3%
.... I . . . . . . . I l J I
0.3
..... I . . . . . . . . I . . . . .
lO0 I000 E~ keY (tog)
I00 I000 Elf keY (log)

"• Fig. 21. Geometrical dependence of the "intrinsic detector

.... I . . . . . . I
+3'/, efficiency".
a
u .2'/ I
d : 14.143 c m
E T : 7 0 keY

=
C
0%
-20%
+20%
_/ E30okv
E ~ = 3500 keY

-2%

- 3%
i

i
iiii
I00

i I Lj
i

,
I

,
i

i
i

, 11
ill
I000

I
D

I I
E~ keY (log)
I J 2
+3%
o.o . . . . I . . . . . . . . . I . . . . . . . .
1 2 Ro
RIcm)
•~ +2"I

'=
+I~

0%
L_ -20%
+20%
Feff l

1"0I
/
I
E 1-=70 keY
I d :1.3943 cm

a. -1%

- 2%

top n - layer 0.~ -

-3%
i i tlJ I I I I I I] I I I
)00 1000 E'~ keV {tog)

Fig. 20b. Percentile systematic error as in fig. 20a: p-core E~:3500 keY

radius (R~), p-core top/detector top distance (D), thickness

of top n-layer (see fig. 10). 0.0' I l . ,
0 I 2 Ro
Rlcm)

Fig. 22. F e f f as a function of R for different distances (d)

and g a m m a energies (E3,).
 L. Moens et aL / Absolute peak efficiency of ~-ray detectors 471

Appendix

Routine from the program SOLANG

LOOP OVER LT
SEFLT = 0
DO 8000 I = 1, IDPI
BLT(I) = (VBLT * SOUHE + SOUHE)/2
C
C LOOP OVER RT calculates the
SEFRT = 0 values o f the
DO 6 0 0 0 I = 1, IDPJ base points with respect to l, r,
RT(J) = (VRT * S O U R A + S O U R A ) / 2 q~and R, respectively
C
C LOOP OVER FI
SEFFI = 0
DO 4 0 0 0 K = 1, IDPK
FI(K) = (VFI * PI + PI)/2
C
C LOOP OVER RP
SEFRP = 0
DO 2000 L = 1, IDPL
RP(L) = (VRP * D E T R A + D E T R A ) / 2
C
C calculates Feff, Fatt,
C FUNCTION COMPUTATION A l 2 f o r a given set o f the variables

EFRP(L) = OMEGA * E A T F A C * C A T F A C * ATFACS * E F F A C

SEFRP = S E F R P + E F R P ( L ) * WRP evaluation of the integral in R, for
2OO0 CONTINUE given set/, r,
~C
EFFI(K) = SEFRP * D E T R A / 2
SEFFI = SEFFI + EFFI(K) * WFI evaluation of the integral in ¢, for
4000 CONTINUE given set I, r
C
E F R T ( J ) = SEFFI * PI/2 * RT(J)
S E F R T = S E F R T + E F R T ( J ) * WRT evaluation of the integral in r, for
6000 CONTINUE given l
C
EFLT(I) = S E F R T • S O U R A / 2 * (HAD + BLT(I))
S E F L T = SEFLT + EFLT(I) * WBLT evaluation of the integral in I
8000 CONTINUE
C
4
ESAPTE = S E F L T * SOUHE/2
ESAPTE = ESAPTE * 4 / S O U H E / S O U R A / S O U R A X Lr--~o

I n t h e a b o v e : B L T ( I ) = l~, E F L T ( I ) = f(l~) a n d S E F L T = y n= 1 wi f(l'i) a n a l o g o u s l y f o r F I ( K ) , R T ( J ) , R P ( L ) e t c .

SOUHE=L; SOURA=ro; DETRA=Ro; EFFAC=fl; ATFACS*EATFAC*CATFAC=f2; OMEGA=J2
E S A P T E = f2
VBLT, VRT, VFI, VRP are zrvalues, while WBLT, WRT, etc. are the corresponding wi-factors.
472 L. Moens et al. / Absolute peak efficiency of 7-ray detectors

References [11] L. Moens, X. Lin, F. De Corte, A. De Wispelaere,

J. Hoste and A. Simonits, to be published.
[12] E. Storm and H.I. Israel, Rept. LA-3753 of the Los Ala-
[1] I.R. Williams, Nucl. Instr. and Meth. 44 (1966) 160. mos Scientific Laboratory of the University of Califor-
[2] U. Pr6sch, M. Luthardt and L. Kolditz, Isotopenpraxis nia (Nov. 1967).
15 no. 10 (1979) 297. [13] B.T. Brice, C.C. Horton and K.T. Spinney, Radiation
[3] C.C. Grosjean, Nucl. Instr. and Meth. 17 (1962) 289. shielding (Pergamon Press, London, 1957) pp. 30, 31
[4] R.L. Heath, Scintillation spectrometry, gamma-ray spec- and 35.
trum catalogue, 2nd ed., USAEC Report IDO-16880 [14] J.W.T. Spinks and R.J. Woods, An introduction to
(1964). radiation chemistry (J. Wiley, New-York, 1964) p. 55.
[5] [Link]. Graudynya, Yu.R.H. Kalnyn and L.L. Pelekis, J. [ 15] W.E. Burcham, F.R.S., Nuclear physics-an introduction
Radioanal. Chem. 9 (1971) 341. (Longman, London, 1970).
[6] S.N. Kaplanis, Int. J. Appl. Rad. Isotopes 29 (1978) [ 16] L. Moens, unpublished work.
543. [17] M. Abramowitz and I.A. Stegun, Handbook of mathe-
[7] D.C. Camp and A.L. Van Lehn, Nucl. Instr. and Meth. matical functions (NBS, Applied Mathematics Series
76 (1969) 192. no. 55; 9th printing 1970).
[8] R. Griffiths, Nucl. Instr. and Meth. 91 (1971) 377. [18] C.C. Grosjean and W. Bossaert, Table of aboslute detec-
[9] R. Gunnink and J.B. Niday, USAEC Rept. UCRL- tion efficiencies of cylindrical scintillation gamma-ray
51061 (yol. 1) (Univ. of California, NTIS, 1972). detectors, A publication of the Computing Laboratory
[10] J.E. Cline, Computers in activation analysis and gamma- of the university of Ghent, Belgium (1965).
ray spectroscopy, Proc. ANS Topical Conf. at Maya- [19] L. Moens, X. Lin, F. De Corte, A. De Wispelaere,
guez, Puerto Rico (1978). J. Hoste and A. Simonits, to be published in Nucl. Instr.
and Meth.