Prompt Gamma Neutron Activation Analysis of toxic elements in
radioactive waste packages
J.-L. Ma1, C. Carasco*1, B. Perot1, E. Mauerhofer2, J. Kettler2, A. Havenith2
1
CEA, DEN, Cadarache, Nuclear Measurement Laboratory, F-13108 St Paul-lez-Durance
2
Institute of Energy and Climate Research – Nuclear Waste Management and Reactor Safety,
Forschungszentrum Jülich GmbH
Abstract
The French Alternative Energies and Atomic Energy Commission (CEA) and National
Radioactive Waste Management Agency (ANDRA) are conducting an R&D program to
improve the characterization of long-lived and medium activity (LL-MA) radioactive waste
packages. In particular, the amount of toxic elements present in radioactive waste packages
must be assessed before they can be accepted in repository facilities in order to avoid
pollution of underground water reserves. To this aim, the Nuclear Measurement Laboratory of
CEA Cadarache has started to study the performances of Prompt Gamma Neutron Activation
Analysis (PGNAA) for elements showing large capture cross sections such as mercury,
cadmium, boron, and chromium. This paper reports a comparison between Monte Carlo
calculations performed with the MCNPX computer code using the ENDF/B-VII.0 library and
experimental gamma rays measured in the REGAIN PGNAA cell with small samples of Ni,
Pb, Cd, As, Sb, Cr, Mg, Zn, B, and Li in view to verify the validity of the numerical model
and gamma-ray production data. The measurement of a ~ 20 kg test sample of concrete
containing toxic elements has also been performed, in collaboration with Forschungszentrum
Jülich, to validate the model in view of future performance studies for dense and large LLMA waste packages.
Pacs: 28.41.Kw, 28.20.Np, 24.10.Lx, 89.20.Bb
Keywords: Radioactive waste characterization, PGNAA, Monte Carlo simulation.
*Corresponding author: Tel: +33 442256130, Fax: +33 442252367, cedric.carasco@cea.fr
1. Introduction
To cope with European regulation [80/68/EEC], the long term disposal of radioactive waste
in dedicated European repositories must prevent any pollution of ground water from toxic
chemicals, even long after the radioactivity of the waste has reached the level of the natural
radioactivity [EUPractices]. Additionally, the waste must contain low amounts of reactive
chemicals, in order to avoid gas production that would impact the integrity of the waste
matrix [GasProduction]. Accordingly, the French and German national radioactive waste
management agencies, ANDRA (Agence Nationale pour la gestion des Déchets RAdioactifs)
and BFS (Bundesamt für Strahlenschutz) respectively, have built a list of chemicals that
should be avoided for disposal of LL-MA waste. In principle, traceability and quality
controls in the waste production and conditioning process allow to quantify these chemicals,
but information is often missing in historical waste. Non destructive analysis based on Prompt
Gamma Neutron Activation Analysis (PGNAA) can provide information concerning the
chemical composition of the radioactive waste [Westinghouse]. In this paper, we present a
comparison between PGNAA signatures acquired with small samples of toxic elements in the
graphite measurement cell REGAIN of CEA Cadarache [Jallu], and MCNPX calculations
[MCNPX] in view to qualify the numerical model for coming R&D studies on the PGNAA
performances for French LL-MA waste packages. Further, a 20 kg concrete sample with a
1.53 g.cm-3 density, containing toxic elements homogenously distributed like cadmium and
mercury, has also been measured in collaboration with Forschungszentrum Jülich
2. Measurements with small samples
50 cm3 samples containing liquid solutions or powders with the Ni, Pb, Cd, Hg, As, Sb, Cr,
Mg, Zn, B, and Li elements have been measured in REGAIN with a GENIE 16 pulsed
neutron generator from SODERN. For each sample, capture gamma rays have been recorded
with a 30% relative efficiency High Purity Germanium detector (HPGe) from CANBERRA
during one hour with an average neutron emission of about 108 neutrons per second. The
neutron pulse duration was 100 µs and the capture gamma rays were collected within the 2 ms
following the pulse, before the next pulse is produced. Due to the electronics upper energy
threshold, only gamma rays with energy lower than 8 MeV have been detected. The samples
net energy spectra have been obtained after subtraction of the spectrum recorded with the
empty measurement cell. The present study focuses on the detection of boron, cadmium,
chromium, nickel, mercury and arsenic, the capture signal related to lead, antimony,
magnesium, zinc and lithium being too weak for a clear identification. Antimony has been
however clearly identified via the 564.6 keV delayed gamma-ray from activation product
122
Sb (T1/2 = 2.72 d), whereas lead and zinc would be better detected with delayed gamma
rays induced by fast neutrons activation. Despite a significant capture cross section, small
amounts of aluminium cannot be detected in REGAIN because the frame of the cell is made
of aluminium. Finally, magnesium and lithium could not be detected due to their small
capture cross section. The experimental setup has been modelled with the MCNPX computer
code, which already proved to correctly reproduce PGNAA measurements of chlorine
concentration in concrete samples [Naqvi]. In a first neutron-photon calculation, a timeenergy dependent point detector (F5 tally of MCNP) estimates the capture gamma ray flux
reaching the HPGe after a neutron pulse. For cadmium, capture gamma-ray production data
being wrong in the ENDF/B-VII.0 library, a laboratory-made data file has been used. The
photon energy deposited in the HPGe crystal has then been calculated with the MCNP pulse
height tally (F8). This two step calculation was used for chromium, nickel, boron, and
cadmium. However, for mercury and arsenic, capture gamma-ray data being (again)
unreliable and no corrected file being available, we have used a different approach. The
neutron capture rate in the samples has been estimated with the MCNP F4 tally (average flux
in a volume) and the FM4 multiplier card (capture reaction) in a only neutron calculation.
Then, capture gamma rays have been produced with proper intensities [Lone] as the source of
a photon calculation to determine the pulse height spectrum in the HPGe detector with the F8
tally. The ratios between the net peak areas of the measured spectrum and MCNP
expectations are shown in Fig. 1 for the peaks that have been clearly identified experimentally.
MCNP reproduces the order of magnitude of the measured data. For mercury, calculation
significantly overestimates the measured data, which probably comes from a systematic bias
(only statistical uncertainties are shown in Fig. 1). The major uncertainties in the MCNP
model concern the amount of boron in graphite and the emission of the neutron generator.
Indeed, REGAIN has been built with graphite blocks loaded with boron impurities. A 16 ppm
boron concentration has been modelled, based on chemical analysis of one graphite sample,
but this data might be not representative of the entire cell because REGAIN has been built
with graphite parts stemming from different origins. MCNP calculations indicate that the
capture rate in the small samples drops by 2.5% per ppm of boron impurity. On the other hand,
the neutron generator control unit does not allow to determine the emission with precision.
We estimate this uncertainty to about 20% and in future experiments it will be necessary to
monitor this emission, for instance with gamma rays produced in a calibration sample placed
close to the neutron source and not influenced by the measured objects. Additionally, the
precision of the ENDF/BVII.0 gamma-ray production data for the elements of interest must
be considered as a source of uncertainty. The validity of the data will be checked by
comparative measurements with a well-known sample of chloride (e.g. sodium chloride)
because a good agreement between calculation and experiment has already been observed for
this element [Naqvi]. These sources of uncertainty may explain the large fluctuations of the
observed calculation-to-experiment ratios. The agreement is sufficient at this step of our study,
the objective of which being to assess the feasibility of the PGNAA technique to detect toxic
elements in radioactive waste packages. Calculations will be used to estimate the order of
magnitude of the detection limits and to design a dedicated system which will be qualified
experimentally if the performances are promising.
3. Measurement with a concrete sample
To be more representative of the densities and volumes of waste packages, the numerical
model has been tested with the measurement of a 20 kg concrete block (see Fig. 2, in
REGAIN) of density 1.53 g.cm-3 prepared at Forschungszentrum Jülich with toxic elements
additives, especially 15 g of cadmium and 20 g of mercury leading to detectable gamma rays.
To better account for moderation and attenuation effects in the F4-F8 approach, the concrete
block has been split into concentric layers, in which the neutron capture rate has been
evaluated. The ratios between experiment and calculation presented in Fig. 3 show that the
intensities of the capture gamma rays associated to mercury and cadmium, but also to silicon
and calcium concrete elements, are roughly reproduced by MCNP, within the limitations
mentioned above about measurement and calculation uncertainties. Previous works have
shown that a good agreement can be reached between PGNAA experiments and Monte Carlo
simulations [WestingHouse] [Naqvi], and thus that calculation can be used for preliminary
design and performance assessment before building an optimised system.
4. Conclusion and perspectives
Feasibility PGNAA measurements performed in the REGAIN graphite cell at CEACadarache with small samples of Cd, Hg, As, Cr, B and Ni show capture gamma rays useable
for toxic element characterisation in LL-MA radioactive waste. Small amounts of Cd and Hg
have been detected in a 20 kg concrete block in collaboration with Forschungszentrum Jülich,
showing that toxic elements with a large capture cross section still can be detected in spite of
neutron and gamma attenuation in concrete. Measurements have also been simulated with
MCNPX, using different approaches since capture gamma-ray production data are either
missing in the ENDF/BVII.0 library or unreliable for some elements. Given additional
uncertainties on the boron contamination of the graphite material constituting REGAIN, and
on the imprecise knowledge of the neutron generator emission, the two calculation approaches
reproduce the measured gamma rays with a sufficient precision for the objective of this study,
which is to estimate by numerical simulation the order of magnitude of the toxic element
detection limits in LL-MA radioactive waste. Neutron and photon attenuation will be
particularly studied in the large and dense matrixes of these packages, as well as the
background due to waste radiations. Additionally, the influence of the nuclear waste
radiations on the detection of the neutrons induced prompt and delayed gamma-rays will be
investigated.
Acknowledgments
This work has been performed in the frame of the R&D program conducted by CEA and
ANDRA for the characterization of long-lived and medium activity radioactive waste.
References
[80/68/EEC] The council of the European communities, Council Directive 80/68/EEC of 17
December 1979 on the protection of groundwater against pollution caused by certain
dangerous substances.
[EUPractices] F. Decamps, L. Dujacquier, Overview of European practices and facilities for
waste management and disposal, Nuclear Engineering and Design 176 (1997) 1–7.
[GasProduction] S. Freedman, Corrosion of Nonferrous Metals in Contact with Concrete,
Modern Concrete 36 (1970).
[westinghouse] Abdul R. Dulloo, PGNAA System for the Assay of RCRA Metals in Mixed
Waste: A Review, proceedings of Industry Partnerships for Environmental Science and
Technology Conference (2001)
[Jallu] F. Jallu, F. loche, Improvement of non-destructive fissile mass assays in a low-level
waste drums: A matrix correction method based on neutron capture gamma-rays and a
neutron generator, Nucl. Instr. and Meth. B 266 (2008) 3674–3690.
[Naqvi] A.A Naqvi, M.M. Nagadi, Omar S. Baghabra Al-Amoudi, Elemental analysis of
concrete samples using an accelerator-based PGNAA setup, Nucl. Instr. And Meth. In Phys.
Res. B 225 (2004) 331-338.
[Lone] M. A. Lone, R. A. Leavitt and D. A. Harrison, At. Data Nucl. Data Tables 26, 511
(1981), also available at http://www-nds.iaea.org/pgaa/PGAAdatabase/lone/lone.htm
Figure 1 Comparison between small samples measurements and MCNP/F5-F8 (top) and MCNP/F4-F8
(bottom) calculations.
Figure 2 Concrete block (on top of the red carton folder) in the REGAIN graphite cell.
Figure 3 Comparison between concrete measurements and MCNP/F5-F8 (top) and MCNP/F4-F8 (bottom)
calculations.