Journal of Environmental Radioactivity 102 (2011) 270e282

Contents lists available at ScienceDirect

Journal of Environmental Radioactivity

journal homepage: www.elsevier.com/locate/jenvrad

Monte Carlo based calibration of scintillation detectors for laboratory and in situ
gamma ray measurements
E.R. van der Graaf a, *, J. Limburg b, R.L. Koomans b, M. Tijs b
a
b

Kernfysisch Versneller Instituut, University of Groningen, Zernikelaan 25, 9747 AA Groningen, The Netherlands
MEDUSA Explorations BV, Verlengde Bremenweg 4, 9723 JV, Groningen, The Netherlands

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 1 April 2010
Received in revised form
2 November 2010
Accepted 6 December 2010
Available online 20 January 2011

The calibration of scintillation detectors for gamma radiation in a well characterized setup can be
transferred to other geometries using Monte Carlo simulations to account for the differences between
the calibration and the other geometry. In this study a calibration facility was used that is constructed
from bricks of well-known activity concentrations of 40K and of radionuclides from the 238U- and 232Thseries. Transfer of the calibration was attempted to a Marinelli beaker geometry with the detector inside
a lead shield and to an in situ application with the detector positioned on a sand bed. In general this
resulted in good correspondence (within 5e10%) between the activity concentrations derived using the
transferred calibration and activities that were derived by independent measurements. Some discrepancies were identied that were attributed to coincident summing in the natural decay series and
interference of radon.
2010 Elsevier Ltd. All rights reserved.

Keywords:
Gamma spectrometry
Scintillation detector
In situ measurements
Calibration
Monte Carlo simulations

1. Introduction
Gamma-ray measurements with scintillation detectors are
routinely used in geophysical and environmental applications (Jones,
2001). In most instances, the aim is to obtain an indication of the
concentration of one or more of the natural-occurring radionuclides
(40K and radionuclides from the 232Th and 238U-series) in the
medium (e.g. soil, rock) in the vicinity of the detector. In uranium
prospecting the goal might be to nd elevated uranium concentrations in survey scans or borehole logging (Godin and GalianoRiveros, 2001). In oil and gas exploration, gamma logs are often
found to correlate with petrophysical parameters (Bennett et al.,
2007; van Ditzhuijzen, 1994). In situ gamma surveys that map the
spatial distribution of (natural) radioactivity are also used to monitor
environmental pollution (van der Graaf et al., 2006) and to assess and
chart sediment transport routes (Jones, 2001; Venema and de Meijer,
2001). In all these applications, the same scintillation detector could
be used. However, as the geometry of the detector with respect to the
medium changes, the calibration used in the analysis of the
measured gamma spectra has to be established for the geometry
being used. This is not a straightforward task to perform experimentally. For each situation encountered, a dedicated calibration

* Corresponding author. . Tel.: 31 50 3633562; fax: 31 50 3634003.

E-mail address: vandergraaf@kvi.nl (E.R. van der Graaf).
0265-931X/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvrad.2010.12.001

setup is needed. Traditionally, gamma sensors are often calibrated in

a purely empirical way by placing a gamma spectrometer in a calibration facility (e.g. a borehole, calibration pads) and measuring the
response of the system (see e.g. Stromswold, 1995). This method is
only valid if the activity concentrations of the calibration facility are
precisely known and if the geometry of the calibration facility exactly
resembles the geometry (e.g. detector in borehole, car- or airborne
detector etc.) of the application for which the system is deployed. In
this paper, we describe a procedure in which detectors are calibrated
using a combination of Monte Carlo simulations and measurements
in a dedicated setup having well-known concentrations of radionuclides (Fig. 1). A measurement of the detector in a calibration facility
of well-specied geometry and with well-known activity concentrations is compared with a Monte Carlo simulation to extract
detector-specic parameters. Monte Carlo simulations together with
these detector-specic parameters are than used to obtain concentrations in the geometry of the actual application. In this way, the
calibration in the dedicated setup is extended to other geometries.
2. Monte Carlo simulations
Monte Carlo simulations were used to simulate measurements
of gamma-ray spectra taken with inorganic scintillation detectors.
All Monte Carlo simulations were performed with the code MCNPX
2.5.0 (Pelowitz, 2005). MCNPX is a multipurpose Monte Carlo
radiation transport code that tracks nearly all particles at nearly all

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

271

Fig. 1. Scheme of the strategy to use Monte Carlo simulation to transfer a calibration in a dedicated geometry to an application geometry.

energies. It is the next in the series of Monte Carlo transport codes

the development of which started at Los Alamos National Laboratory nearly sixty years ago.
2.1. Detector models
Two detectors were used in this study, a 3  3 inch NaI and
a 4  4  16 inch CsI. Both detectors were purchased from Scionix,
The Netherlands and both consisted of a crystal coupled to a photomultipliers that are integrated in an aluminium housing. These
two detectors were modelled based on the exact dimensions of the
crystal and the surrounding aluminium protective casing (Fig. 2).
Other features like the photomultiplier and thin linings with
reecting material or m-metal were not modelled. For the 3  3 inch
and 4  4  16 inch detectors, aluminium casings with thicknesses
of 0.5 mm and 1 mm, respectively, were included in the model
geometry. The crystals of the detectors were modelled as pure
crystals with densities of 3.67 and 4.51 kg dm3, respectively
(Melcher, 1989).
2.2. Spectrum simulation
2.2.1. Gamma radiation
Gamma radiation considered in the simulations in this article
either originates from 40K, from the natural decay series of 238U and
232
Th or from the anthropogenic radionuclide, 137Cs.
The gamma-ray energy for 40K is taken as 1461 keV with an
emission probability of PK 0.1067 gammas per decay. It is
assumed that radionuclides from the 238U- and 232Th-series are in
secular equilibrium. All 458 (238U) and 346 (232Th) gamma-ray
emissions are implemented, resulting in a total emission probabilities of PU 2.197 and PTh 2.628 gammas per decay. The
gamma-ray energy of 137Cs is 661.66 keV with an emission probability of PCs 0.8521 gammas per decay. The data for gamma-ray
energies and emissions probabilities are taken from Firestone
(1996). For the implementation in the MCNPX input le, the
module prepared by Hendriks (2003) was used.

2.2.2. Geometry truncation

In this article, the response of a scintillator detector to gamma
radiation originating from a specic volume source is simulated.
Geometries of the volume source that are studied are either nite
or semi-innite. For a semi-innite source to be accessible for
simulations, the geometry is truncated (restricted). Truncations
are based on simulations with the highest gamma energy
encountered in practice (2.6 MeV of 208Tl from the 232Th-series)
and the lowest medium (formation) density expected, which is
taken as 1.2 kg dm3 (very watery mud). In these simulations, the
size of the source volume is systematically enlarged and the
increase of count rate in the detector is calculated as a function of
this volume. As gamma radiation is attenuated by the material
between its point of origin and the detector, the count rate in the
detector will saturate as a function of the size of the source
volume. The geometries are truncated such that the volume
considered provides 99.9% of the signal for a homogeneous,
distributed source of 2.6 MeV gamma radiation in a formation
with density 1.2 kg dm3. In reality, this truncated geometry
provides even more than 99.9% of the signal, as densities are
usually larger and most of the relevant gamma-ray energies are
lower than the value used for the estimate of the truncated
geometry size.
2.2.3. Detector response
Detector response was simulated using the MCNPX pulse height
tally (F8-tally). This tally estimates the distribution of the energy
deposition (energy spectrum) in a model of a physical detector. In
all simulations, an unbroadened spectrum was calculated. These
unbroadened spectra (histograms) were adapted to t the energy
resolution of the detector by performing a Gaussian energy
broadening procedure on these histograms. In the simulations, the
energy deposition in the detector was recorded in three hundred
10 keV bins to span an energy range from 0 to 3 MeV.
2.2.4. Construction of standard spectra
Spectra directly resulting from MCNPX are expressed as count
rates in the energy bins in units of counts per Starting Particle.

272

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

Fig. 2. Schemes of 3  3 inch ( 7.6  7.6 cm) (left) and 4  4  16 inch (10.6  10.6  40.6 cm3) (right) detectors.

These spectra are converted to so-called standard spectra, which

give the detector response in the energy bins in cps per Bq kg1
material of the source. In an earlier study it was found that the
inuence of porosity and water content of the material of the
source have negligible inuence on the shape of the gamma-ray
spectra (De Groot et al., 2009). Spectral intensity, however, is
inuenced by water content and, unless specied otherwise, all
density indications will refer to the bulk density of the material as it
is found in situ (sometimes called fresh or wet mass density). These
standard spectra are used to decompose measured spectra in the
weighted sum of the standard spectra (one standard spectrum for
each radionuclide to be measured), where the weights are the
activity concentrations in the source material (in Bq kg1). This
decomposition procedure is referred to as Full Spectrum Analysis
(FSA) and has been described by Hendriks (2001).
To elucidate the construction and use of these standard spectra,
we will describe the construction of standard spectra for the situation of a detector and a source (either nite or semi-innite). The
source has a bulk density of rb (kg m3).
The relation between a standard spectrum used in FSA (SSFSA )
and the simulated spectrum directly from the Monte Carlo calculation (SSsim ) is given by:

SSFSA
X rb

rb V sim PX SSsim
X rb

(1)

where:
1
SSFSA
X rb : standard spectrum (cps per Bq kg ) for radionuclide,
X, to be used in FSA for a certain detector e source situation with
bulk density, rb, of the source.
SSsim
X rb : simulated standard spectrum (counts per Starting
Particle) for radionuclide, X, directly from Monte Carlo calculation
for a certain detector e source situation with bulk density, rb, of the
source.
Vsim: volume (m3) of the source used in the simulation.
Px: emission probability in number of gammas per decay for
radionuclide, X.
In Eq. (1) the factor, rbVsim, represents the total mass of the
source (in kg). Assuming an activity concentration of 1 Bq kg1 of
radionuclide, X, the number of gammas produced per second in,
Vsim, is given by, rbVsimPx. This can be considered in terms of the
Monte Carlo simulations as the number of Starting Particles per
second per Bq kg1.
For a semi-innite geometry Eq. (1) is only valid if it is independent of the size of Vsim. This volume has to be large enough to
account for all parts of the source from which gamma radiation

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

273

emerges that deposits energy in the detector. This implies that the
product, V sim SSsim
X rb , should not change when the source size is
increased. As mentioned earlier, this size is estimated by performing calculations for a source with low density, in combination
with a mono-energetic source of gamma radiation with the highest
energy encountered in the decay of the radionuclides considered
(2.6 MeV of 208Tl from the 232Th-series).
3. Spectrum analysis
The unbroadened simulated standard spectra, SSFSA
X , are the
starting components to decompose measured spectra using FSA.
However, we have to take into account two features that are related
to the characteristics of the detector that was used for the
measurement. Firstly, the standard spectra have to be broadened to
match the resolution of the detector. Secondly, the energy represented by the channels of the measured spectrum will not exactly
match the energy represented by the bins of the simulated spectra.
Moreover, the number of energy channels traditionally used in the
spectrum acquisition (516, 1024, etc.) usually differs from the
number of energy bins used in the simulation (300). To account for
the latter mismatch, the measured spectrum is rst redistributed
over 300 channels (i 1.300). Thereafter, the spectrum is
decomposed in contributions of the various standard spectra by
minimizing:

c2

in
m
 XX


1 X
S a0 a1 i a2 i2 
CX SSX i  Bgi
n  m i1
X 1

!2
(2)

where n is the number of channels used in the analysis, m is the

number of adjustable concentration parameters, i is the channel
number, Cx is the activity concentration (Bq kg1) of radionuclide X,
S is the measured spectrum, a0, a1 and a2 are adjustable parameters
to compensate the spectrum for possible gain drift (this shift is
mostly linear and consequently, a2, can often be set to zero), Bg(i) is
a background spectrum and SSX are the Gaussian broadened
standard spectra:

SSX i

ji2

jX
n  2sj

j1

SSFSA j
p X
2psj

(3)

Fig. 3. Photograph of the Stonehedge calibration facility.

Inside the square gap a 20 cm diameter and 80 cm long PVC tube

(wall thickness 3.9 mm) is inserted. The bricks are normal
10  20  6.5 cm3-sized Dutch paving bricks that consist of baked
clay and were purchased in one batch to minimize inter-brick
variations.
In all simulations, the bricks were assumed to be pure SiO2 with
a density of 2.32 kg dm3 (from mass determinations of 10 bricks;

where SSFSA
X is the unbroadened standard spectra (from Eq. (1)) and
s(j) is the standard deviation of the Gaussian broadening function
at channel j. The standard spectra are broadened to the energy
resolution of the detector by using a parameterization (Debertin
and Helmer, 1988) of the full width at half maximum (FWHM):

FWHMj 2

q
p
2ln2sj b0 j

(4)

where b0 is an adjustable parameter. The value for FWHM resulting

from Eq. (4) will have the dimensions of channels and is based on
a 300 channel spectrum.
The minimisation of c2 is done using CX (X K, Th, U, Cs) and a0,
a1, a2 and b0 as free parameters. Usually the energy range for the
spectrum analysis is set between 300 keV and 2750 keV; this
implies that 245 channels are used in the spectrum decomposition.
4. The Stonehedge calibration facility
4.1. Geometry and composition
Stonehedge is a brick castle (height and depth: 100 cm; width
120 cm) with a square (20  20 cm2) and 80 cm deep opening at the
front in which a detector can be placed for testing (Fig. 3 and Fig. 4).

Fig. 4. Schematic view of Stonehedge geometry used in MCNPX. Left: front view.
Right: cross-section along the depth of Stonehedge. Insert: detector and PVC tube.

274

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

relative standard deviation 1.6%). The PVC tube was modelled as

C2H3Cl with a density of 1.38 kg dm3 (Carvalho et al., 2009). Air
was taken as a mixture of N2 (77% mass) and O2 (23% mass) with
a density of 0.001205 kg dm3 (Leal Iga et al., 2008).

600

550

4.2. Radionuclide concentrations

Concentration (Bq kg-1)

450

Th
U

50

40

30

1A
1B
2A
2B
3A
3B
4A
4B
5A
5B
6A
6B
7A
7B
8A
8B
9A
9B
10A
10B

The Stonehedge clay bricks only contain the natural activity that
is incorporated in the clay from which they were manufactured.
The activity concentrations of K, U and Th were measured on
crushed samples (grain size of less than 3 mm) from one of the
Stonehedge bricks. After crushing, these samples were transferred
to six 100 mL pill boxes (diameter 3.5 cm, Rubox BV, Hazerswoude,
the Netherlands). This pill box is one of the calibrated geometries of
the KVI HPGe gamma spectrometer (Shakhashiro et al., 2007). The
pill boxes were measured using this gamma spectrometer and
analysed according to the Dutch protocol for spectrometry
(NEN 5623, 2002). This protocol described the determination of the
activity concentrations based on a peak content analysis, taking
into account measurement time, detector efciency, branching
ratio and sample mass and density. Furthermore, to assess the
variation between the bricks of Stonehedge, 10 bricks were cleaved
in two geometrically identical halves. The resulting set of 20 half
bricks had a mean mass of 1.45 kg with a relative standard deviation of 2.2%. Each half brick was individually measured for at least
1 h (resulting in the uncertainties due to counting statistics to be
smaller than 1%) on the Medusa gamma spectrometer that consist
of the 3  3 inch NaI detector inside a 10 cm thick shield of lowactivity lead. The half bricks were positioned with their square
10  10 cm2 surface centred directly on the detector surface. These
20 measurements were analysed with standard spectra simulated
for the geometry of a half brick on a 3  3 inch NaI detector (Fig. 5).
For these simulations the lead shield was also modelled to account
for back-scattered gamma radiation. The homogeneity of this data
set is depicted in Fig. 6 and summarized in Table 1. For all three
radionuclides, the homogeneity of the activity concentration
dened as the ratio of the standard deviation versus the mean of
the 20 half bricks is better than 3.5%. Moreover, the mean values of
the activities concentrations of the set of half bricks are consistent

500

Sample
Fig. 6. Activity concentrations (in Bq kg1) of Th, U and K in 10  2 half bricks as
measured with the Medusa spectrometer. Solid lines represent means of the presented
data. Dashed lines represent mean values of the six crushed samples as measured with
the KVI HPGe detector.

with the activity concentrations derived from the HPGe measurements of the crushed samples.
4.3. Monte Carlo evaluation of Stonehedge
Two aspects of the Stonehedge facility were assessed using
Monte Carlo simulations. Firstly, the facility is located in a large
industrial storage hall. Radioactivity in the building materials of
this hall may affect measurements in Stonehedge. The inuence of
the radioactivity in the oor and wall on these measurements was
calculated. Secondly, the dimensions of Stonehedge are large but
nite. It was estimated to which degree Stonehedge approximates
an innite geometry.
The Stonehedge calibration facility is situated on a concrete
oor. The back wall of Stonehedge is close to a wall of the storage
hall made of low activity sandstone bricks. Besides the radionuclides in the Stonehedge bricks, this oor and wall are the major
other sources of radionuclides that emit gamma radiation that may
reach the detector being tested. To estimate the magnitude of this
contribution that does not originate from the bricks of Stonehedge,
spectra were calculated for the large sized detector (4  4  16 inch
CsI). In these calculations, a 20 cm thick oor and 15 cm thick wall
element, both with a 200  200 cm2 surface area, were used in the
Table 1
Activity concentrations (Bq kg1) of K, U and Th. Values given are the mean and
standard deviation of the n 20 (half bricks) and n 6 (crushed bricks) data sets.

Fig. 5. Geometry used in MCNPX of Medusa gamma spectrometer consisting of

3  3 inch NaI detector inside a 10 cm thick cylindrical lead shield. The half brick is
positioned on top of the detector.

Radionuclide

Half bricks

Crushed brick

K
U
Th

519  15
36.5  1.2
47.3  1.3

537  8
36.1  0.8
45.6  0.9

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

275

Fig. 7. Schematic view of the Stonehedge geometry extended with a oor (right) and a wall (left), respectively.

model. The oor and wall were directly placed under and adjacent
to the Stonehedge outer boundaries, respectively (Fig. 7).
In Fig. 8, simulated spectra are compared for a 4  4  16 inch CsI
detector resulting from the activity in Stonehedge and from the
activity of the oor and wall. The intensity of the spectra from the oor
and wall is about three orders of magnitude smaller than the intensity
of the spectra from the Stonehedge formation. Consequently, if it is
assumed that radionuclide activity concentrations in the oor and

wall are identical to the concentrations in the Stonehedge bricks, it

can be estimated that the contribution from the oor and wall is
a factor of 1000 smaller than the contribution of Stonehedge. This
assumption is conservative as the bricks have relative high activity
concentrations. The inuence of the oor and wall will, thus, be
overestimated.
Ideally, a calibration facility should be innitely large such that
all radiation that contributes to the count rate of the detector surface
originates from the facility. This section estimates how well Stonehedge approximates such an innite geometry. At this point it is
emphasized that Stonehedge not being innite does not disqualify it
as a calibration facility. By modelling the precise geometry of the
facility and assuring that disturbances from radiation not originating
from the facility are small, the shape and intensity of spectra
resulting from the activity in the Stonehedge bricks can be accurately calculated and compared with measurements.
To estimate how well Stonehedge approximates an innite
geometry, spectra calculated for the Stonehedge geometry versus an
innite geometry are compared. The innite geometry is modelled
as a square (20  20 cm2) 60 cm long air gap at the centre of a solid
sphere with radius of 100 cm of Stonehedge bricks. Inside this gap the
20 cm diameter, 60 cm long PVC tube is placed together with
a 4  4  16 inch CsI detector. The radius of 100 cm of the innite
sphere was estimated from simulations using a source of 2.6 MeV
gamma rays distributed homogeneously throughout a sphere. The
total count rate (TC) in the detector was calculated as a function of the
radius of this sphere. It was found that the ratio of TC at a radius of
100 cm and the saturation value of TC for innite radius is larger than
0.999. This saturation value was calculated from a t of a function
representing an exponential increase towards a saturation value to
the data points from the simulations. Comparing the TC rates over the
energy interval of 0.3e3.0 MeV for the two geometries (Table 2), we
conclude that TC rates in the Stonehedge geometry for all three
radionuclides are larger than 96% of the value for the innite geometry, indicating that Stonehedge is very similar to an innite situation.
The ratio of the volume of Stonehedge (100  100  120

Table 2
Total count rates (cps per Bq kg1) in Stonehedge and innite geometry.
Fig. 8. Spectra (cps per Bq kg1) of Th, U and K simulated for the Stonehedge
formation and for an underlying oor and an adjacent wall (as in Fig. 5). Spectra are for
a 4  4  16 inch CsI detector and are Gaussian energy-broadened to mimic a detector
resolution of 10% at 662 keV.

Radionuclide

Stonehedge

Innite

Ratio

Th
U
K

22.20
19.20
1.596

23.06
19.83
1.659

0.963
0.968
0.962

276

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

(20  20  80) 1.17106 cm3) and the 100 cm radius geometry ((4/3)
p1003(20  20  80) 4.16106 cm3) is 0.28. This implies that for
the 100 cm radius geometry, the outer 72% of its volume only
contributes 4% (100%e96%) of the radiation.
In conclusion, the Stonehedge facility is very similar to an
innite geometry, with negligible contributions of radiation from
radionuclides that are not accounted for in the Monte Carlo simulations. Inuences of the oor and wall of the storage hall, in which
the facility is situated, are estimated to contribute less than 0.1% to
the signal in the detector being tested, from radiation from the
bricks in the Stonehedge facility itself. As a consequence no extra
shielding (e.g. with lead or copper) of the facility is needed.
5. Measured and simulated detector spectra
Applications of the Stonehedge facility to calibrate scintillation
detectors are described in the sections that follow. Spectra were
collected using detectors (described in Section 2.1) in Stonehedge,
in the Medusa gamma spectrometer (example of a nite geometry)
and in a typical in situ situation with a detector on a sand bed. The
latter situation is very similar to the way the method may be used
for eld conditions. For each of these situations also standard
spectra were calculated using Monte Carlo simulations. The validity
of the calibration procedure has been tested by comparing activity
concentrations derived from the FSA analysis with simulated
standard spectra with activity concentrations in certied reference
materials or measured with a well-calibrated HPGe gamma
spectrometer.
5.1. Spectrum acquisition
In all cases, spectra were collected using a photomultiplier tube
(PMT) base with integrated bias supply, preamplier and multichannel analyzer (1024 channels, ORTEC digiBASE) connected to
the PMT of the detector. This spectroscopy system was read out
with a portable computer running the Maestro-32 (ORTEC, 2009)
software package.
5.1.1. Stonehedge
Spectra were collected for 333 s (4  4  16 inch CsI) and 2692 s
(3  3 inch NaI) with the detector positioned in the centre of the air
gap in the Stonehedge facility. The spectra are shown Figs. 9 and 10,
respectively. The Total Counts (TC) rate in the energy interval
0.3e2.75 MeV is approximately a factor of 10 higher for the larger
detector (2530 compared to 262 cps for the CsI and the NaI
detector, respectively). The acquisition times used result in uncertainties (in TC) due to counting statistics of less than 0.2%. For both
detectors the t resulting from the FSA appears to approximate the
measured data very well. The parameters that were found from the
FSA are given in Table 3. The activity concentrations of Stonehedge
as found for the 3  3 inch NaI detector and the 4  4  16 inch CsI
detector are close (within 5e10%) to the values from the
measurements with the HPGe and the Medusa gamma spectrometer. Using the b0-parameter in Eq. (4) the resolution at 661.66 keV
(137Cs) was calculated as 7.4 and 11.5% for the 3  3 inch NaI and the
4  4  16 inch CsI detectors, respectively. These values are typical
 ski et al., 2006; Gong
for these sizes and types of detectors (Moszyn
et al., 1988; Yoshimori et al., 1983).
5.1.2. Medusa gamma spectrometer
The Medusa gamma spectrometer is used to analyse natural
radioactivity in soil samples. The samples are measured in reentrant 1 L (Type 133N) Marinelli beakers (ANSI, 1986). The
Marinelli beakers are polypropylene (density 0.9 g cm3)
containers with a wall thickness of 2 mm. The shape of the sample

Fig. 9. Spectrum of 4  4  16 inch CsI detector in Stonehedge. Also shown is the

individual contribution of K, U and Th. The solid line represents the t from the Full
Spectrum Analysis with the parameter values from Table 3.

in the Marinelli beaker can be visualised as a hollow cylinder

(height 7.7 cm, inner radius 4.7 cm and outer radius 7.0 cm) with
a solid cylinder (height 2.085 cm and radius 7.3 cm) on top. The
volume of such a sample is 1000 cm3. The 3  3 inch NaI detector
ts inside the hollow cavity of the Marinelli beaker (Fig. 11).
Five Marinelli beakers were lled to 1 L with quartz powder or
sand spiked with certied reference materials (van der Graaf and
ten Have, 2002). A sixth Marinelli beaker was lled with demineralized water and was used for an assessment of the background.
The quartz powder (Mallinckrodt Baker BV, Deventer, the
Netherlands) was used as matrix for the beakers containing U, Th
and K reference material and for a beaker considered as a low
activity sample. The sand was silver sand obtained from Zilverzand
exploitatie Beaujean B.V. in Heerlen, the Netherlands. This sand
(99.9% SiO2) had a mean grain size of approximately 250 mm and
had very low natural radioactivity concentrations. The silver sand
was used as matrix for the beaker spiked with 137Cs. As the 137Cs

Fig. 10. As Fig. 7 except for a 3  3 inch NaI detector.

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282
Table 3
Activity concentrations (Bq kg1) and detector parameters resulting from the Full
Spectrum Analysis of spectra measured with two detectors in Stonehedge.
Detector
Radionuclide

3  3 inch

4  4  16 inch

K
U
Th
Detector parameter
ao
a1
a2
b0

528  3
39.6  0.3
50.2  0.3

498  3
38.8  0.3
47.1  0.3

2.67
0.574
0
0.61
14

3.91
0.782
0
0.94
25

c2red

spike was in liquid form, sand was preferred over quartz powder
because the coarser grain size allows for better non-coagulating
mixing of the sample.
The uranium and thorium spikes originated from the IAEA,
Vienna, Austria (IAEA, 1987). These spikes were prepared and certied to be in secular equilibrium by CCRMP-CANMET, Ottawa, Canada.
The uranium spike is coded as IAEA/RGU-1 and the certied uranium,
thorium and potassium mass contents are 400  2 ppm, less than
1 ppm and less than 20 ppm, respectively. For the thorium spike
(coded IAEA/RGT-1) these values are 6.3  0.4 ppm (U), 800  16 ppm
(Th) and 0.02  0.01% (K), respectively. The potassium spike (Merck
company) is coded IAEA/RGK-1 and has certied (by IAEA, Vienna,
Austria) uranium, thorium and potassium mass contents of less than
0.001 ppm, less than 0.01 ppm and 44.8  0.3%, respectively. The
caesium (137Cs) spike was purchased from Canberra Eurisys Benelux
in a 5 mL solution of 0.1 M HCl (Product code: CDZ24, activity
25.90 kBq, Reference time: 12.00 GMT, September 1, 2000). Three
batches of about 1.4 L of quartz powder and one batch of 1.4 L of sand
were spiked by adding (U, Th, K) or dripping (Cs) weighed amounts of
spike, and thereafter homogenized with a rotational mixer. Each
batch contained only one of the spikes. From each batch ten 100 mL
fractions were taken and the homogeneity of the batch was checked
by measuring the count rate of one of the characteristics gamma lines
of the spike radionuclide in each fraction in a 100 mL pill box on an
HPGe gamma detector. For all batches the relative standard deviation
of the measurements on the ten fractions was less than 3%. These
procedures for U, Th and K are conformed with the methods
described in the Dutch norm for determination of the natural
radioactivity in stony building materials (NEN 5697, 2001; Blaauw
et al., 2001) and for Cs conform with the methods used for the validation of the Dutch norm for the determination of the activity of
gamma ray emitting nuclides in a counting sample (NEN 5623, 2002;
van der Graaf and de Meijer, 2002).

Fig. 11. Geometry of the Medusa gamma spectrometer with 1 L Marinelli beaker.

277

Table 4
Activity concentrations (Bq kg1) in ve Marinelli beakers calculated from added
spike activities and HPGe measurements, and from measurements with the Medusa
gamma spectrometer in combination with FSA and Stonehedge Monte Carlo calibration. Values for caesium refer to time of measurement.
Radionuclide
Marinelli
Cs
K
U
Th
Marinelli
K
U
Th
Marinelli
K
U
Th
Marinelli
K
U
Th
Marinelli
K
U
Th

From spikes

with caesium spike

133.2  0.6
1.17  0.06
1.72  0.04
1.74  0.06
with potassium spike
570  5
1.50  0.09
2.05  0.08
with uranium spike
12.4  0.5
144.4  0.4
1.64  0.09
with thorium spike
12.4  0.5
5.49  0.13
145.1  1.4
with quartz powder only
12.7  0.5
1.57  0.08
2.14  0.09

From FSA
133.5
4
0.4
1.1






c2red

1.2
3
0.4
0.4

11

619  15
1.6  0.4
1.3  0.4

70  20
136  4
0.7  1.0

32

47  11
12.2  1.5
134.3  1.9

12.7  1.2
1.39  0.12
0.43  0.12

39

To account for the activity concentrations of U, Th and K in the

matrix material both the unspiked quartz powder and the sand
were measured (for 60,000 s) in a 1 L Marinelli beaker using an
HPGe gamma detector and density dependent calibration curves
(linear relation) determined earlier (van der Graaf et al., 2000) were
used to calculate the activity concentrations of 238U, 232Th and 40K
from the gamma spectra. For calculations the procedures described
in NEN 5697 were followed. More specically, this implies that the
gamma rays of 214Pb (352 keV), 208Tl (583 keV), 228Ac (911 keV) and
40
K (1461 keV) were used to obtain the activity concentrations of
238
U, 232Th and 40K, respectively. Using the weighed amounts of the
added certied spikes and correcting for the low activities
concentrations of the matrix materials, the activity concentrations
in the four spiked and the low activity Marinelli beakers were
calculated (Table 4).
Spectra were collected for all six Marinelli beakers using the
Medusa gamma spectrometer. Acquisition times were between 1
and 2 h, except for the spectrum of the water-lled Marinelli which
had a live time of 16 h. Fig. 12 presents the raw spectra for the

Fig. 12. Example of raw data measured with the Medusa gamma spectrometer. Spectra
are shown for 1 L Marinelli beakers lled with water, unspiked and uranium-spiked
quartz powder, respectively.

278

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

beaker spiked with uranium, for the water-lled (background)

beaker and the beaker with unspiked quartz powder. The spectrum
of the uranium-spiked Marinelli beaker has clearly higher count
rates than the other two spectra. As expected, the spectrum for the
water-lled Marinelli beaker had the lowest count rate. The raw
spectra are overlapping for channel numbers larger than 450, corresponding to the energy region larger than 3 MeV where no
gamma rays from K, U, or Th are present and the only events are due
to cosmic radiation.
The spectrum of the water-lled Marinelli beaker was used as
a background spectrum in the Full Spectrum Analysis of the spectra
for other ve Marinelli beakers. Clearly, this background correction
is small for the spectra of the spiked Marinelli beakers. However, for
the low-activity Marinelli beaker, this correction is relatively large.
Standard spectra of K, U, Th and Cs that were used in the FSA were
created with MCNPX for the geometry shown in Fig. 11. The Cs
standard spectrum was only used in the analysis of the spectrum
for the Cs-spiked Marinelli beaker. In the FSA, the a parameters
(see Eq. (2)) were adjustable to allow for (in all cases, very small)
corrections for temperature drift and the b0-parameter (dening
detector resolution) was xed and taken from the Stonehedge
calibration values (Table 3). The energy-calibrated spectra from the
FSA are shown in Figs. 13 and 14, together with the best FSA t
(solid lines) and the spectrum resulting from using the known
activity concentrations and the simulated standard spectra (dashed
lines). In general, the agreement between the FSA t and the

measurements is within 2e3%. Especially, for the radionuclides that

emit mono-energetic gamma rays (40K and 137Cs) the agreement up
to the primary full energy peaks was excellent (within 1%). In the
energy region after these peaks the statistics of the spectra are low
but within the uncertainties FSA t and measurements agree. For
both the U and Th spectra there were some differences between the
FSA t and the measurements (inserts in Fig. 13). For the U spectrum, the two peaks directly after the 1.12 MeV (214Bi) peak were
underestimated and FSA compensated for this by increasing the 40K
concentration so that the 1.461 MeV full energy peak contribution
lifted the tted spectrum. For the Th spectrum a similar situation
was encountered directly after the 0.911 MeV (228Ac) peak. In this
case, FSA compensated for the underestimate in this region by
increasing both the U and 40K concentrations. These differences
between the FSA t and measurement are reected in the activity
concentrations from the FSA (Table 4). All of the FSA concentrations
were consistent with those calculated directly from the added
spikes except for the 40K concentration in the Marinelli beakers
with the U and Th spikes and the U concentration in the Marinelli
beaker with the Th spike.
5.1.3. Detector on sand bed
A large eld that was recently prepared for building was
selected for the assessment of the Monte Carlo based calibration
procedure in a typical in situ application situation. At the site over
an area of approximately 50  50 m2 a layer of 35 cm of sand was

Fig. 13. Spectra measured with the Medusa gamma spectrometer of l L Marinelli beakers lled with quartz powder spiked with Cs, K, U and Th, respectively. Solid line is FSA t to
the data points; dashed line is calculated from standard spectra and known activity concentrations in the Marinelli beakers. Inserts show details of spectra where deviations
between FSA t and measurement are largest.

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

Fig. 14. Spectrum measured with the Medusa gamma spectrometer of l L Marinelli
beakers unspiked quartz powder. Solid line is the FSA t to the data points; dashed line
is calculated from standard spectra and known activity concentrations in the Marinelli
beakers.

deposited on an old agricultural eld mainly composed of clayey

soil. To assess the radionuclide concentrations in the sand bed, six
samples were taken within a radius of 2 m around the detector.
Although the rst 35 cm is expected to give the major part of the
signal in the detector also one sample was taken from the original
eld underneath the sand bed to be able to assess the contribution
from this layer. These samples were analysed using the KVI HPGe
gamma spectrometer for the well-calibrated 1 L Marinelli geometry
(Shakhashiro et al., 2007). The activity concentrations in the six
sand samples (Table 5) showed that the sand bed around the
detector was homogeneous (relative standard deviation) within 5%
(K and U) and 10% (Th). It is also evident that the underlayer is
approximately a factor of two more radioactive than the sand bed.
At the centre of the site, a spectrum (live time: 2093 s) was taken
with the 4  4  16 inch CsI detector. A background spectrum (live
time 1662 s) was collected in the middle of a lake (Schildmeer,
Groningen, the Netherlands). This latter measurement was performed with the system in an unmanned plastic canoe at a lake
depth of approximately 1.2 m. The depth of the lake, and the large

279

distance from the shores and operators assures negligible contributions of environmental radionuclides and of 40K present in the
operators. Both raw spectra are presented in Fig. 15. The spectrum
measured on the sand bed had 1e2 orders of magnitude higher
count rate than the background spectrum. For the background
spectrum some structure could be observed; rstly, the 511 keV
annihilation peak is clearly visible, and secondly, there appeared to
be some small contributions from both uranium and potassium. The
latter contributions might have been due to some gamma rays
originating from the lake bottom shining through the water or even
from natural radionuclides in the water itself. The uranium contribution in the background spectrum might also have been caused
partly by radon decay products present in the atmosphere. Although
this background spectrum contained some contributions of environmental radionuclides, it was used for background correction in
the FSA of the spectrum measured on the sand bed. As the count rate
of the background was much lower than the intensity of the sand
bed spectrum, the inuence of the background was small.
In Fig. 16, the background-corrected spectrum is presented
together with its FSA decomposition in standard spectra simulated
with MCNPX for a atbed geometry, approximated by a 100 cm
thick disk of SiO2 with a radius of 240 cm and the detector positioned in the centre. No attempt was made to include the information about the real geometry (35 cm of sand on a clay layer), as
this information would not have been available during a typical in
situ measurement. To include the effect of possible backscatter by
the air on top and surrounding the detector, this stack of air should
be included in the model geometry. Ideally, a much larger atbed,
together with a high air stack on top should be used in the model.
Unfortunately, modelling of such a large geometry requires
computing power beyond the present state of the art. As
a compromise, the situation was investigated in which backscatter
effects were approximated using the same truncated model
geometry with a 1 m high stack of air, but assuming a density of the
air around the detector of 0.001205 (t, no backscatter in Fig. 16)
and 1 kg dm3 (t, backscatter in Fig. 16). Thus, in the latter
geometry, an air stack of 800 m (corresponding to approximately 5
times the half thickness for absorption in air for the 2.6 MeV
gamma ray of 208Tl; Billings, 1998) has been compacted in a 1 m air
layer on top of the atbed.
The FSA t using standard spectra without backscatter underestimated the measured spectrum at energies below 0.5 MeV and
slightly overestimated the measured spectrum at energies above
approximately 1.5 MeV (Fig. 16). The FSA t with standard spectra

Table 5
Activity concentrations (Bq kg1) in samples of the sand bed and a sample of clayey
underlayer from HPGe measurements and from an in situ measurement with the
4  4  16 inch CsI gamma detector in combination with FSA and Stonehedge Monte
Carlo calibration for standard spectra without and with backscatter correction.
Sample

Sand 1
Sand 2
Sand 3
Sand 4
Sand 5
Sand 6
Mean  STD
Underlayer
In situ (FSA, no backscatter)
In situ (FSA, backscatter)

172
180
183
185
187
188
183
343
153
197

U
3
3
3
3
3
3
6
6
6
4

4.95
5.36
5.30
5.54
5.59
5.51
5.4
11.2
10.5
10.7

Th
 0.15
 0.16
 0.10
 0.14
 0.15
 0.18
 0.2
 0.3
 0.6
 0.3

5.12
5.85
6.05
6.22
6.62
6.46
6.1
19.0
6.3
7.4

Cs
 0.15
 0.17
 0.15
 0.18
 0.16
 0.15
 0.5
 0.4
 0.5
 0.4

<0.2
<0.2
<0.2
<0.2
<0.2
<0.2
<0.2
0.76  0.08
2.1  1.1
1.7  0.7

Fig. 15. Raw data measured with the 4  4  16 inch CsI detector at the centre of
a large sand bed and in the middle of a lake.

280

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

Fig. 16. Background corrected spectrum measured with the 4  4  16 inch detector at the centre of a large sand bed. Also shown is the individual contribution of K, U, Th and C. The
solid line represents the t from FSA with the detector parameter values from Table 3 and using the backscatter corrected standard spectra. The grey dashed line results from the
FSA t with standard spectra that do not include backscatter.

that included backscatter was found to correspond much better to

the measured spectrum. Not only had the discrepancy at energies
lower than 0.5 MeV disappeared, the t to the two peaks between
the 1.461 MeV 40K peak and the 2.6 MeV 208Tl peak had slightly
improved. In comparing the HPGe based activity concentrations of
the sand bed with those derived from the in situ measurement and
FSA analysis (bottom row, Table 5) it is apparent that for K and Th,
the FSA values are not signicantly different from the HPGe values.
For U the comparison is less favourable, the FSA value is a factor 2
higher than the HPGe value.
6. Discussion
6.1. Stonehedge
Comparison between a measured spectrum in Stonehedge and
the t resulting from the FSA decomposition for the two scintillation detectors indicated that values of these spectra agreed within
a few percent. The activity concentrations resulting from the FSA t
were within 5e10% of independently determined values. It should
be noted that the MCNPX model does not mimic light transport in
the scintillator or photomultiplier efciency. The fact that we nd
agreement between Monte Carlo model and measurements implies
that virtually every gamma interaction that deposits energy in the
scintillation material is registered by the data acquisition chain. A
priori, this cannot be taken for granted, and indeed, some detectors
were tested in Stonehedge in which a constant scaling factor
between the measured and calibrated concentration was found.
This scaling factor should then subsequently be used to multiply
the simulated standard spectra in transferring the Stonehedge
calibration to other geometries.
6.2. Medusa gamma spectrometer
Small discrepancies in shape between FSA t and the
measurements were found for the Marinelli beakers spiked with U
and Th. The most likely explanation for these discrepancies seems
to be coincident summing effects, as these are not included in the
simulations of the standard spectra. Moreover, it is well-known in
close-to-the detector geometries, such as the Marinelli beaker
setup summing effects are enhanced (Dcombaz et al., 1992; Lee
et al., 2008) and that measurements of gamma emissions from

the natural decay series are prone to summing interference. The

overall effect will be that due to summing out, some intensity in the
spectrum will be shifted from the peaks towards the continuum.
For the FSA procedure, this will result in concentrations being
underestimated, as the peak channels have the highest intensity,
and thus, have the largest weights in the least-squares procedure.
For the concentrations from the FSA t for the Marinelli beakers
with the U and Th spikes (Table 4) underestimates of 6 and 8% are
found, respectively. This is consistent with results derived for HPGe
detectors for Marinelli beaker geometries where summing out
effects of this order are reported (Garca-Talavera et al., 2001). In
case of the Th-spectrum, a possible candidate to cause the unexplained peak at approximately 1.1 MeV is the coincident summing
in of the 0.511 (8.1%) and 0.583 (30%) gamma emissions of 208Tl.

6.3. Detector on sand bed

It is evident that transferring the Stonehedge calibration to in
situ situations is not trivial. For large size geometries, it is crucial to
take the backscatter effects of the surrounding materials (air, in this
case) into account, otherwise inconsistencies may occur in the
spectral shape. Gammas that are emitted from the sand bed that
initially miss the detector may be scattered back towards the
detector through interaction with the air around the detector. This
results in mainly lower-energy events in the detector. In reality,
there is a large stack of air (instead of the 100 cm in the model) that
provides this backscatter, and moreover, also the amount of soil
that contributes to this part of the signal is likely much larger in
reality than is modelled. To include the effect of backscatter, spectra
were simulated with the density of air articially increased.
Another complication for this in situ situation is that the actual
geometry in which the measurement was performed (35 cm sand on
top of a clayey layer) did not perfectly match the model geometry
(uniform atbed). As the activity concentrations in the clayey layer
were rather different from the concentrations in the sand, the
possible inuence of the clayey layer was assessed by Monte Carlo
simulations. Standard spectra were simulated for a 35 cm top layer
of sand and for a semi-innite layer of clay with a layer of 35 cm of
(no activity containing) sand on top. The count rates of the standard
spectra for the underlayer were only 1.8% (U), 2.2% (Th) and 3.2% (K)
of those of the top layer. Considering the factor 2e3 higher activities
in the bottom layer, the total effect on the concentrations is expected

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

to be an overestimate (from FSA) for the concentrations of the sand

layer of less than 4e9%. The activity concentrations for K
(197  4 Bq kg1) and Th (7.4  0.4 Bq kg1) calculated using the
standard spectra with backscatter corresponded well with the
activity concentration expected from the HPGe measurements
including the extra contribution of the clayey underlayer, namely
183 0.032  343 194  6 Bq kg1 and 6.1 0.022 
19.0 6.5  0.5 Bq kg1 for K and Th, respectively. For U the expected
concentration is 5.4 0.018  11.2 5.6  0.2 Bq kg1, and consequently, the FSA result was 5 Bq kg1 too high. A likely explanation
for this overestimate of U by FSA is the inuence of radon. Radon that
emanates into the pore space of the clay will diffuse upwards and
end up in the pore space of the sand layer and also partly exhale in
the air surrounding the detector. In the simulations it was assumed
that the U-series was in secular equilibrium. In the case of the sand
bed this was most likely not true because extra radon was introduced from the clay, both in the pore space of the sand and in the air
around the detector. As almost 97% of the total gamma-ray intensity,
with energies in the FSA-analysis range of 0.25e2.75 MeV, originate
from radon decay products the spectrum produced by radon
strongly resembles a uranium spectrum. Radon emanation coefcients of clay are typically in the order of 25% (Baixeras et al., 2001;
Damkjaer and Korsbech, 1985; Greeman and Rose, 1996). This
implies that due to radon from the underlayer, an extra 2.8 Bq kg1
might be added to the signal from uranium. Moreover, the radon
concentration in the air around the detector may be increased
resulting in a further enhancement of the uranium signal in the
detector.
For in situ applications, one needs to be cautious of possible
differences between the actual geometry and the geometry used to
derive the standard spectra for FSA. In the present case with
a detector on a sand bed, the premise that the sand bed was
homogeneous and semi-innite, was not entirely satised. Such
situations are likely often encountered during in situ surveys. This
does not disqualify the use of standard spectra for homogeneous
semi-innite media. However, it should be realized that the activity
concentrations obtained are averages over the eld of view of the
detector, which is for a situation with direct contact between
detector and soil surface typically 1e2 m around the detector.
7. Conclusions
Monte Carlo simulations are a convenient and powerful tool to
transfer calibration information for gamma-ray detectors from
a calibration setup to other geometries. Using a measurement in the
calibration geometry and analysing the obtained spectrum with the
Monte Carlo simulated response for the radionuclides involved,
results in a complete set of detector-specic calibration parameters.
Monte Carlo simulations of detector responses in any other
geometry in which the detector will be deployed can then be
combined with the detector-specic parameters to obtain a calibration for the other geometry. A simple calibration facility built of
bricks was used to test the validity of this calibration transfer for
both a situation with a Marinelli beaker on a detector in a lead
shield and a situation with a detector on top of a large sand bed. It
was concluded that with the exception of some issues concerning
coincidence summing and radon interference this transfer results
in activity concentrations that are in agreement (within 5e10%)
with activity concentrations that are derived by independent
measurements.
Acknowledgment
This project was conducted with support of the Casimir programme of the Netherlands Organization for Scientic Research

281

that aims at increasing the mobility of research staff in the public

and private sectors.

References
ANSI, 1986. IEEE Standard Test Procedures for Germanium Gamma-ray Detectors.
ANSI/IEEE Std 325e1986, American National Standards Institute. The Institute
of Electrical and Electronics Engineers, Inc, New York, USA.
Baixeras, C., Erlandsson, B., Font, Ll., Jnsson, G., 2001. Radon emanation from soil
samples. Radiation Measurements 34, 441e443.
Bennett, B., Lager, A., Potter, D.K., Buckman, J.O., Larter, S.R., 2007. Petroleum
geochemical proxies for reservoir engineering parameters. Journal of Petroleum
Science and Engineering 58, 355e366.
Billings S. D., 1998. Geophysical aspects of soil mapping using airborne gamma-ray
spectrometry. Thesis, Department of Crop Sciences, Faculty of Agriculture,
University of Sydney, Australia.
Blaauw, M., Glass, H.J., de Jong, P., van Dijk, W., 2001. The Dutch norm for the
determination of natural radioactivity levels in construction materials applied
to concrete samples. Journal of Radioanalytical and Nuclear Chemistry 249,
177e180.
Carvalho, M.T., Ferreira, C., Portela, A., Santos, J.T., 2009. Application of uidization
to separate packaging waste plastics. Waste Management 29, 1138e1143.
Damkjaer, A., Korsbech, U., 1985. Measurement of the emanation of radon-222 from
Danish soils. The Science of the Total Environment 45, 343e350.
Debertin, K., Helmer, R., 1988. Gamma- and X-ray Spectrometry with Semiconductor Detectors. North-Holland, Amsterdam.
Dcombaz, M., Gostely, J.-J., Laedermann, J.-P., 1992. Coincidence-summing
corrections for extended sources in gamma-ray spectrometry using MonteCarlo simulation. Nuclear Instruments and Methods in Physics Research A312,
152e159.
De Groot, A.V., van der Graaf, E.R., De Meijer, R.J., Mau
cec, M., 2009. Sensitivity of insitu g-ray spectra to soil density and water content. Nuclear Instruments and
Methods in Physics Research A600, 519e523.
Firestone, R.B., 1996. Tables of Isotopes, eighth ed. Wiley.
Garca-Talavera, M., Laedermann, J.-P., Dcombaz, M., Daza, M.J., Quintana, B., 2001.
Coincidence summing corrections for the natural decay series in g-ray spectrometry. Applied Radiation and Isotopes 54, 769e776.
Greeman, D.J., Rose, A.W., 1996. Factors controlling the emanation of radon and
thoron in soils of the eastern U.S.A. Chemical Geology 129, 1e14.
Godin, M., Galiano-Riveros, E., 2001. The subsoil gamma-ray intensity distribution
in the Oriental Region of the Republic of Paraguay. Applied Radiation and
Isotopes 54, 527e533.
Gong, W.G., Kim, Y.D., Poggi, G., Chen, Z., Gelbke, C.K., Lynch, W.G., Maier, M.R.,
Murakami, T., Tsang, M.B., Xu, H.M., Kwiatkowski, K., 1988. Resolution tests of
CsI (Tl) scintillators read out by pin diodes. Nuclear Instruments and Methods in
Physics Research A268, 190e199.
Hendriks, P.H.G.M., 2001. Full-spectrum analysis of natural g-ray spectra. Journal of
Environmental Radioactivity 53, 365e380.
Hendriks, P.H.G.M., 2003. In-depth g-ray studies. Borehole measurements. Thesis,
University of Groningen, The Netherlands.
IAEA, 1987. Preparation and Certication of IAEA Gamma Spectrometry Reference
Materials RGU-1, RGTh-1 and RGK-1, International Atomic Energy Agency, IAEA
Report RL148, Vienna, Austria.
Jones, D.G., 2001. Development and application of marine gamma-ray measurements: a review. Journal of Environmental Radioactivity 53, 313e333.
Leal Iga, J., Leal Iga, J., Leal Iga, C., Ayala Flores, R., 2008. Effect of air density variations on greenhouse temperature model. Mathematical and Computer
Modelling 47, 855e867.
Lee, M., Park, T.S., Woo, J.-K., 2008. Coincidence summing effects in gamma-ray
spectrometry using a Marinelli beaker. Applied Radiation and Isotopes 66,
799e803.
Melcher, C.L., 1989. Scintillators for well logging applications. Nuclear Instruments
and Methods in Physics Research B40/41, 1214e1218.
 ski, M., Nassalski, A., Syntfeld-Ka
Moszyn
zuch, A., Szcze sniak, T., Czarnacki, W.,
Wolski, D., Pausch, G., Stein, J., 2006. Temperature dependences of LaBr3(Ce),
LaCl3(Ce) and NaI(Tl) scintillators. Nuclear Instruments and Methods in Physics
Research A568, 739e751.
NEN 5623, 2002. Radioactivity Measurements e Determination of the Activity of
Gamma Ray Emitting Nuclides in a Counting Sample by Semiconductor Gamma
Spectrometry. Nederlands Normalisatie Instituut, Delft, The Netherlands. NEN
5623:2002 nl.
NEN 5697, 2001. Radioactivity Measurements e Determination of the Natural
Radioactivity in Stony Building Materials by Means of Semiconductor Gammaray Spectrometry. Nederlands Normalisatie Instituut, Delft, The Netherlands.
NEN 5697:2001 en.
ORTEC, 2009. http://www.ortec-online.com/pdf/digibase.pdf.
Pelowitz, D.B., 2005. MCNPX Users Manual Version 2.50, LA-CP-05e0369.
Shakhashiro, A., Sansone, U., Trinkl, A., Makarewicz, M., Yonezawa, C., Kim, C.K., KisBenedek, G., Benesch, T., Schorn, R., 2007. Report on the IAEA-CU-2006-03
World-wide Open Prociency Test on the Determination of Gamma Emitting
Radionuclides. IAEA Report IAEA/AL/171. International Atomic Energy Agency
(IAEA), Seibersdorf, Austria (KVI laboratory code: 280).

282

E.R. van der Graaf et al. / Journal of Environmental Radioactivity 102 (2011) 270e282

Stromswold, D.C., 1995. Calibration facilities for borehole and surface environmental radiation measurements. Journal of Radioanalytical and Nuclear
Chemistry 194, 393e401.
van der Graaf, E.R., Venema, L.B., ten Have, R., 2000. NGD-KVI Participation in the
Interlaboratory Test Programme of NEN 5697 Radioactivity Measurements:
Determination of the Natural Radioactivity in Dense Building Materials by
Means of Semiconductor Gamma-ray Spectrometry. KVI Internal Report S64.
van der Graaf, E.R., de Meijer, R.J., 2002. Gamma-ray spectrometry intra- and interlaboratory test of the Dutch protocol. Applied Radiation and Isotopes 57, 73e83.
van der Graaf, E.R., ten Have, R., 2002. Preparation Procedures and Certicates of
Calibration for the Medusa U, Th-, K-, Cs and Bg-standards for Gamma-ray
Spectrometry. KVI Internal Report S105.

van der Graaf, E.R., Koomans, R.L., Limburg, J., de Vries, K., 2006. In situ radiometric
mapping as a proxy of sediment contamination: assessment of the underlying
geochemical and -physical principles. Applied Radiation and Isotopes 65,
619e633.
van Ditzhuijzen, P., 1994. Petrophysics e in Touch with the Reservoir. Shell Selected
Paper 121147/15m/7.94.
Venema, L.B., de Meijer, R.J., 2001. Natural radionuclides as tracers of the dispersal
of dredge spoil dumped at sea. Journal of Environmental Radioactivity 55,
221e239.
Yoshimori, M., Okudaira, K., Hirasima, Y., Kondo, I., 1983. Gamma-ray spectrometer
aboard solar are observation satellite Hinotori. Nuclear Instruments and
Methods 215, 255e259.