Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
Available online 16 April 2021
0584-8547/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Monte Carlo simulation for the analysis of various solid samples using
handheld X-ray fluorescence spectrometer and evaluation of the effect by
environmental interferences
Woojin Kim a,*
, Jaeyeong Jang a
, Do Hyun Kim b
a
Korea Institute of Nuclear Nonproliferation and Control, 1534, Yuseong-daero, Yuseong-gu, Daejeon 34054, Republic of Korea
b
Korea Atomic Energy Research Institute, 111, Deadeok-daero 989 beon-gil, Yuseong-gu, Daejeon, 34057, Republic of Korea
A R T I C L E I N F O
Keywords:
Nuclear Forensics
Nuclear material
Handheld-XRF
Environmental interferences
Monte Carlo simulation
A B S T R A C T
Handheld X-ray fluorescence (HH-XRF) has expanded its utilization areas according to recent technological
developments. Most current applications, though, are still concentrated in traditional areas including mineral
resource analysis and environmental regulation rather than forensic science for the purpose of investigating a
nuclear security event involving nuclear material out of regulatory control. To apply HH-XRF to nuclear material
analysis, it is necessary to first obtain calibration data using standard reference materials. Considering the difficulty
in obtaining such standard reference materials as well as the high costs involved, one well-known
alternative method is to use Monte Carlo simulation code. This study investigated the feasibility of employing
Monte Carlo N-Particle transport 6 (MCNP6) simulation to provide calibration data through comparison with
experimental measurements of pure solid samples of graphite, copper, SiO2, and UO2 using HH-XRF. The results
showed that the MCNP6 simulation results were entirely consistent with the measurement spectra, except for
environmental interferences stemming from interactions with the mechanical components below 10 keV which
varied slightly according to sample type. To quantitatively evaluate the effect of these environmental interferences
on the whole spectrum, the coefficient of determination (R2
) was used. In the case of graphite, the
effect of the environmental interferences was evaluated to be about 20% on the conformity of the measured and
simulated results, while those for copper, SiO2, and UO2 were about 1%, 3%, and less than 1%, respectively.
These results indicate that samples having elements with higher rates of photoelectric absorption followed by
fluorescence compared to scattering tend to decrease the effect of the environmental interferences over the entire
spectrum. The origin of the environmental interferences was estimated to be interference with the detector shield
and/or X-ray tube collimator, which are particular design features of the device used. Their effect on contributing
to the environmental interferences was evaluated by experiment for the detector shield and simulation for the Xray
tube collimator. As the detector shield was found to only contribute to a decrease in overall spectrum intensity,
the major contributor to the environmental interferences was determined to be the collimator. It is
believed that the results of this study will help to confirm that Monte Carlo simulation can properly provide
calibration data for using HH-XRF on nuclear materials for which reference materials are hard to obtain.
1. Introduction
As X-ray tube and detector technologies have become more sophisticated,
handheld X-ray fluorescence (HH-XRF) applications have
expanded [1–5]. For instance, HH-XRF has become a type of forensic
science equipment, with several applications including gunshot residue
analysis, glass fragment analysis, and the matching of samples associated
with suspects or crime scenes [6]. In the field of nuclear security,
while XRF techniques have been proposed for use in nuclear forensics in
cases encountered out of regulatory control [7], however, related applications
have focused on desktop types of analytic tools in the laboratories
and have not considered handheld types of equipment for usage
in the field. Absolutely, desktop equipment in laboratory settings provides
more accurate and therefore confident performance than handheld
types; however, such large equipment involves high costs and is
generally hard to use after moving into the crime scene. Furthermore,
* Corresponding author.
E-mail address: kimwj@kinac.re.kr (W. Kim).
Contents lists available at ScienceDirect
Spectrochimica Acta Part B: Atomic Spectroscopy
journal homepage: www.elsevier.com/locate/sab
https://doi.org/10.1016/j.sab.2021.106203
Received 14 August 2020; Received in revised form 9 April 2021; Accepted 12 April 2021
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
2
samples containing nuclear material or radioactively contaminated
material are often not able to be safely handled in laboratories or loaded
into equipment due to issues with cross-contamination and safety regulations.
It is therefore important to conduct an in-situ preliminary
analysis to minimize the safety risks of first responders and establish
appropriate analytical plans prior to the collection, sampling, and
transport of such materials to forensic laboratories. In this light, HH-XRF
equipment can be an effective analytical tool that can provide an
elemental characteristic of various types of samples can be detected
practically in the field.
To date though, HH-XRF has rarely been used for nuclear material
analysis, particularly when the nuclear material is meant to include
uranium, plutonium, or thorium as major constituents. In order to apply
HH-XRF to nuclear material for elemental analysis, additional optimization
work is requisite because there currently is insufficient calibration
data for nuclear material samples to determine and quantify
elemental composition. Some researchers have produced experimental
data directly through measurements of standard reference materials
(SRMs) or synthetic samples made by themselves [8–10]. Most studies
involved liquid samples dissolved by nitric acid or the content analysis
of uranium and plutonium mixtures. For any such studies, it is not easy
to obtain the relevant SRMs and then conduct the desired experiments,
since all nuclear material is rigorously controlled by international norms
and national regulations.
Considering these difficulties, Monte Carlo simulation has long been
an alternative method to conducting experiments with sensitive nuclear
materials. In terms of X-rays, most simulations focus on the estimation of
X-ray beam profiles produced from X-ray tubes for managing the radiation
exposure of patients and workers; such beam profile estimation has
been known to achieve high accuracy from X-ray sources [11–13].
However, additional design parameters for HH-XRF such as geometry
and material compositions, need to be considered in detail to more
precisely simulate the energy spectrum that results from the detector.
Such data are not opened to the public nor to equipment users, though,
because they include the manufacturers’ know-how.
The purpose of this study is to simulate the X-ray energy spectra of a
range of samples using openly available data and then verify the feasibility
of the simulation by comparing the results to experimental measurements.
To do so, spectra of graphite, copper, SiO2, and UO2 solid
samples were experimentally collected using HH-XRF and compared
with Monte Carlo simulation results. During the investigation, unexpected
fluorescence peaks stemming from interactions with the instrumentation,
which is called environmental interferences [14], were
identified in all measurement spectra but not in simulation; the effects of
these peaks on the measurement performance for each sample were
evaluated using the coefficient of determination (R2
). From this simulation
study, it is expected that the energy spectra of nuclear material
samples, with which experimental measurements are not easily conducted,
can be obtained.
2. Materials and methods
2.1. Samples
The samples used in this study were graphite, copper, SiO2, and UO2,
as shown in Table 1. Graphite and copper were selected as representative
pure single elements, obtained as reference materials from FLUXANA.
In particular, graphite was selected as a blank sample presenting a
background spectrum when other elemental samples are measured. It is
predicted to give a reasonable approximation of the original beam
profile produced by the X-ray tube due to the X-ray scattering properties
of materials with low atomic numbers and the relatively low energy of Xray
fluorescence. Copper belongs to the transitional metals having very
high fluorescence rates, so its use was expected to reveal related phenomena
by interactions in the detector such as sum peaks. SiO2 represents
one of the major matrix elements in geological samples. The
material of interest, UO2, is the major chemical form conventionally
used in the commercial nuclear fuel cycle, constituting the fuel rods in
many nuclear power plants.
2.2. Instruments
A Bruker S1-Titan 600 handheld XRF analyzer (Bruker, Billerica, MA,
USA) was used for the experimental measurements. It consists of a fast
silicon drift detector (SDD), an ultralene beryllium (Be) window, and an
end window transmission type X-ray tube with the following specifications:
Maximum accelerating voltage 50 kV, Maximum power 2 W, and
Rhodium (Rh) target with 0.65 μm of thickness and 1.85 g/cm3
of
density. In this study, the spectrometer mode without any calibration
was used. A desktop kit provided by the manufacturer was also used to
achieve controlled measurement geometry.
2.3. Experimental setup
A number of parameters can influence the beam profile irradiated
into samples from the X-ray tube as well as the result energy spectrum
measured by the SDD. These parameters include accelerating electron
voltage, current, measured time, applied filters, and mechanical components.
The voltage and filters directly influence the incident X-ray
beam in both energy distribution and intensity, while current and
measuring time are proportional to the intensity at each energy interval.
This study selected the no filter option because it is more convenient
to minimize any interference factors for ease of comparison with the
Monte Carlo simulation results. Accelerating electron voltage was set at
50 kV, while current and operating time were fixed at 35 μA and 30 s in
the real-time setting except for the UO2 sample, for which 60 s was set.
The HH-XRF device used in this study does not provide a function to
control live-time; rather, the real-time setting is applied as an experimental
setup parameter and then normalized for comparison with simulations.
Here, real time means the actual time which has elapsed
between starting and stopping a measurement, while live time represents
an estimate of the total effective time that the digital pulse processor
could have been acquiring valid counts. All experimental
parameters were maintained in the same conditions.
2.4. Monte Carlo simulation
By analyzing the energy spectrum obtained from simulations of the
experimental measurements, the feasibility of applying simulation to the
measurement of nuclear material was evaluated. HH-XRF measurements
were simulated using Monte Carlo N-Particle Transport Code System
version 6.2.0 (MCNP6, Los Alamos National Lab, NM, USA). The crosssection
library used to compute the transport of electrons and photons
was the Electron Photon Relaxation data 14 (EPRDATA14) library.
Table 1
Specifications of the experimental samples.
Sample Type Density Certifying
Agency
Serial
Number
Graphite Solid/Cylinder 2.26 g/
cm3
FLUXANA FI001953
Copper Solid/Cylinder 8.96 g/
cm3
FLUXANA FM0011942
SiO2 Solid/Sample cup
(powder)
2.65 g/
cm3
Sigma-Aldrich MB060-50G
UO2 Solid/Cylinder 10.5 g/
cm3
KEPCO Nuclear
Fuel
8F2H364
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
3
2.4.1. Profiling of the X-ray source beam
Fig. 1 presents the variation of the X-ray beam geometric size
generated by the HH-XRF device according to the increasing distance
from the beam port [15]. Generally, manufacturers provide this kind of
radiation profile to let users know about the radiological conditions
during instrument operation. This information is obtained from radiation
measurements using a radiation survey instrument in accordance
with the IEC62495 which have been used for their licensing.
Considering the X-ray tube specifications and parameters, X-ray
beam generation was simulated using MCNP6 code. The thickness and
density of the Rh target were set to 0.65 μm and 12.41 g/cm3
, respectively,
and the thickness and density of the beryllium (Be) window were
125 μm and 1.85 g/cm3
, respectively. The accelerating voltage of electrons
was set to 50 kV, same as in the experiment. The electron beam was
assumed to be incident vertically to the center of the Rh target in a
vacuum.
To calculate the energy distribution of the primary X-ray beam
emitted from the Rh target, which is often called the primary X-ray beam
source profile or open beam profile, the number of X-rays passing
through the surface corresponding to the incidence angle range of the Xray
beam was obtained using F1 tally in MCNP6. The tally card is a
function to specify what type of information the user wants to gain from
the Monte Carlo calculation, and the F1 tally is a card that calculated the
current integrated over a surface. The incidence angle of the X-ray beam
was set to 20 degrees as shown in Fig. 1, by referring to the manual of
HH-XRF. The relative intensity by X-ray energy on the surface was
calculated. Based on these calculation results, Fig. 2 shows the primary
X-ray beam source profile calculated for both vacuum and air media.
The former represents the X-rays immediately after emission from the
Rh target, while the latter represents reflection by air of the X-rays between
the HH-XRF and the contact surface of the sample.
2.4.2. Material and geometry setup
The simulation was also performed by MCNP6 using an F8 Tally
(Pulse Height Tally) card which derives the output of the energy distribution
of pulses created by radiation in a detector. MCNP6 was used
to simulate the energy spectrum generated by the detector. It was
derived from the specifications provided by the manufacturer [16,17]
such as the incidence angle of the X-ray, the placement of the detector,
and the distance between the major components in the HH-XRF device.
The detector was assumed to consist of pure silicon in a cylindrical shape
of size 0.51 cm × 0.05 cm (diameter × length), as referenced from the
specifications of the SDD in the HH-XRF equipment. The energy spectra
measured via HH-XRF for all samples were obtained through the simulation
using the primary X-ray beam source profile from Section 2.4.1.
Fig. 1. Schematic of the emission angle and diameter of the X-ray beam at
various distances from the beam port [15].
Fig. 2. Primary X-ray beam source profile through MCNP6 simulation for
vacuum and air media. The accelerating voltage is 50 kV.
Fig. 3. Measured energy and MCNP6 simulation spectra for the graphite
sample at energy ranges (a) 0–40 keV and (b) 0–10 keV. The accelerating
voltage was 50 kV and the filter option was not selected.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
4
3. Results
3.1. Energy spectra of experiment and MCNP6 simulation
3.1.1. Graphite
Fig. 3 shows a comparison between the experimental and simulated
spectra of the graphite sample. The main difference is the existence of
environmental interferences stemming from interactions with the
instrumentation, which are existed in the experiment but absent in
simulation. This is because, in the MCNP6 simulation, only the results of
scattering and fluorescence by the primary X-ray incident on the
graphite sample were calculated. The conformity of the spectra was
evaluated through the coefficient of determination (R2
). For the entire
energy range (0–40 keV), the R2
was 0.744. For the energy range
excluding up to 10 keV, where the environmental interferences were
mainly generated, the R2
was 0.946. The result shows that the environmental
interferences had about a 20.2% effect on the conformity of
the energy spectra in the case of the graphite sample.
3.1.2. Copper
Fig. 4 shows a comparison between the experimental and simulated
spectra of the copper sample. The sum peaks indicating that the peaks
occurred experimentally due to limitations in detector resolution,
present in the experimental results but did not appear in the MCNP6
simulation. For the entire energy range, the R2
was 0.977, showing a
high similarity of the spectra. In the range from 0 keV to 10 keV, without
the sum peaks, the R2
was 0.985. This result indicates that the environmental
interferences under 10 keV and the sum peaks had relatively
little effect on conformity, and accordingly, the MCNP6 simulation
shows high accuracy for the copper sample.
3.1.3. SiO2
Fig. 5 shows a comparison of the simulated and experimental spectra
of the SiO2 sample. The main difference between the results was caused
by the environmental interferences, like in the previous cases. Since
normalization was performed based on the peak of silicon Kα1 fluorescence
at 1.74 keV, a difference occurred at the Compton peak of the Rh K
line. For the entire energy range, the R2
was 0.900. Excluding the range
of 3.1 keV to 9 keV, where the environmental interferences were mainly
generated, analysis of conformity gave an R2
of 0.930, meaning that the
effect of the environmental interferences on the conformity of the SiO2
spectra is about 3%.
Fig. 4. Measured energy and MCNP6 simulation spectra for the copper sample
at energy ranges (a) 0–40 keV and (b) 5–20 keV. The accelerating voltage was
50 kV and the filter option was not selected.
Fig. 5. Measured energy and MCNP6 simulation spectra for the SiO2 sample at
energy ranges (a) 0–40 keV and (b) 0–10 keV. The accelerating voltage was 50
kV and the filter option was not selected.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
5
3.1.4. UO2
Fig. 6 shows a comparison of the simulated and experimental spectra
of the UO2 sample. Similar to the copper sample, the difference due to
the environmental interferences was not large. The peaks of the uranium
L and M lines showed high agreement between the experiment and
MCNP6 simulation. Over the entire energy range, the R2
was 0.906.
Analyzing the conformity excluding the range 3.5–9 keV where the
environmental interferences were mainly generated, the R2
was 0.906.
This means that the environmental interferences have little effect on
spectral conformity in the case of the UO2 sample.
All spectra of each sample’s measurements are shown in Fig. S1-S4
(Appendix). Additional efforts to find out their origin are discussed in
Section 4.
4. Discussion
4.1. Origins of the environmental interferences
In all experimental spectra, the unexpected fluorescence peaks by
environmental interferences generally depend on the sample type. In
particular, these environmental interferences have a large effect on
conformity in the graphite sample. So, the graphite sample was selected
at first, then other samples are included for further simulation in this
discussion chapter. As shown in section 3.1.1, the energy spectrum
measured by graphite sample is best to let us know what is difference
between the experiment and simulation because of its simple interaction
with primary X-ray.
It is derived below that they are produced by interference with the
Fig. 6. Measured energy and MCNP6 simulation spectra for the UO2 sample at
energy ranges (a) 0–40 keV, (b) 0–10 keV. The accelerating voltage was 50 kV
and the filter option was not selected.
Fig. 7. Images of the detector shield. Scale bars are 3 mm (left) and 300 μm (right).
Fig. 8. Comparison of energy spectra with and without the detector shield.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
6
instrumentation; considering the design features of the HH-XRF equipment,
two possible origins are investigated as the major origins of the
environmental interferences.
The first possible origin was estimated to be interference from the
detector shield, which protects the detector window from being punctured
by sharp objects. To confirm this estimation, the detector shield
was extracted from the detector, as shown in Fig. 7, for elemental
composition analysis by a Bruker M4 Tornado Micro-XRF analyzer
(Bruker, Billerica, MA, USA). The elemental composition by weight
percent was calculated as Fe (51.19 wt%), Cr (9.65 wt%), Ni (6.84 wt%),
and In (25.94 wt%).
Using the MCNP6 code, the detector shield was applied as shown in
Fig. 10(a) to simulate the measurement of the graphite sample. As
previously calculated, an energy spectrum was obtained by specifying
SDD as F8 tally as shown in Fig. 8. As a result of the simulation applying
the detector shield, it is confirmed that the shield affects a decrease in
intensity across the entire spectrum. In addition, fluorescence peaks by
Fe, Cr, Ni, and In are obviously observed. These peaks are due to the
fluorescent X-ray generated from the detector shield in that they were
not present in the case without the shield. Especially, it is determined
that Fe and In mainly affect due to their high content. Therefore, it could
be confirmed that some of the environmental interferences are caused by
the detector shield.
The second possible origin was estimated to be interference from the
X-ray tube collimator that sharpens the beam and makes it closer to the
sample in the employed device than in traditional designs. To confirm
this estimation, an additional MCNP6 simulation was performed to
identify the extent that the collimator may produce environmental interferences.
The element composition of the collimator was analyzed
using the M4 Tornado Micro-XRF analyzer for the simulation. The
element composition of the inner collimator by weight percent was
calculated as Al (97.12 wt%), Fe (2.16 wt%), and traces of other elements
such as Ca, Cu, Ni. The composition of the outer collimator was
calculated as Cu (60.82 wt%), Zn (36.4 wt%), Pb (2.59 wt%), and Fe
(0.19 wt%). The element composition data of the collimator was applied
and the simulation was performed. The energy spectrum calculated by
specifying SDD as F8 tally as shown in Fig. 9 presents that there are
fluorescent X-ray peaks by Fe, Cu, Ni, and Zn below 10 keV. Especially, it
is determined Al and Fe mainly affected due to their high content. The
result of the simulation shows that the environmental interferences can
arise from interference with the collimator in response to the primary Xray
source beam.
Additional simulation was applied using both tentative two origins
as shown in Fig. 10 in order to confirm the effcts of both the detector
shield and collimator at the same time. Fig. 11 shows the energy spectra
of both experiment and simulation, normalized assuming they have the
same total count rate. Compared to the simulated spectrum shown in
Fig. 3, it could be seen that it is quite similar to the energy spectrum
obtained by the experiment. Environmental interferences, which mainly
occur in the 0–10 keV energy range, were also simulated fairly well, and
the continuous spectrum in the energy region of 25 keV or more showed
good agreement. As a result, the coefficient of relation (R2
) is 0.930,
which is 19% higher than the previous R2
of 0.744.
Fig. 12 shows the energy spectra of both experiment and simulation,
normalized assuming they have the same total count rate. As in the case
of the graphite sample, environmental interferences were well simulated.
For the entire energy range, R2
was 0.974, 0.969, and 0.904.
In the case of graphite and SiO2 samples, the difference of R2
between
the improved simulation and existing simulation was noticeable at 15%
and 4%. On the other hand, in the case of copper and UO2 sample, since
the intensity of the secondary X-ray generated by the sample was much
higher than that of X-ray from environmental interferences, there was
little difference of R2
between the improved simulation and existing
simulation. In other words, the influence of environmental interferences
had less effect.
4.2. Variation of environmental interferences by sample type
To observe the variation in the environmental interferences by the
type of sample, comparisons were made to distinguish the excitation
spectra between elements in the samples and elements from instrument
contribution, as shown in Fig. 13. All samples contain common spectra
from Cr, Mn, Fe, Ni, and Cu. By normalizing the spectra to the intensity
of the graphite spectrum at 6.4 keV, the heights of the Fe Kα lines, the
Fig. 9. Comparison of energy spectra with and without the collimator.
Fig. 10. Cross-sectional view of the MCNP6 geometry model of (a) the detector shield and (b) the entire simulation.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
7
highest can be compared more easily. From the normalized spectra, it is
observed that the environmental interferences do not significantly
deviate by sample, except for the UO2 spectrum which suggests contributions
from not only elements in the sample but also the instrument
itself.
Section 3.1 above presented the results of conformity evaluations of
the measured and simulated energy spectra. The former represents the
real case with environmental interferences while the latter is the ideal
case without those. In conclusion, it is found their effects to range from 1
to 3% in three samples but up to 18% in the case of graphite. This is
because that graphite has rare fluorescence peaks due to low energy and
relatively strong scattering peaks from a high yield of the primary Xrays,
while the other three samples have strong fluorescence peaks and
relatively fewer scattering peaks. This means that their effect depends on
the fluorescence peaks of the sample elements having significantly
different intensities, as shown in Fig. 14. In addition, for samples containing
elements having high fluorescence peaks such as copper and
UO2, the effect of environmental interference is not significant. Thus,
high consistency between the measured spectrum and the Monte Carlo
simulation is achieved for these samples.
Fig. 11. Comparison of measured and simulated spectra of graphite to check
the effect of collimator and detector shield at energy ranges (a) 0–40 keV, (b)
0–10 keV.
Fig. 12. Comparison of measured and simulated spectra to check the effect of
the collimator and detector shield in energy ranges 0–10 keV (a) at the copper
sample, (b) SiO2 sample, and (c) UO2 sample.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
8
5. Conclusion
This study investigated the feasibility of applying Monte Carlo
simulation to obtain the energy spectra of nuclear samples for use as
calibration data by comparing simulation results to experimental measurements
using HH-XRF for various pure solid samples.
A Conformity comparison between the normalized energy spectra
from simulations and measurements was conducted. The effect of the
observed environmental interference on the whole spectrum was evaluated
using the coefficient of determination (R2
). It was found that
Monte Carlo simulation excluding the low energy range until 10 keV
yielded 95% conformity in the graphite case and 99, 93, and 91% conformity
for the copper, SiO2, and UO2 samples, respectively. The effect
of environmental interferences was evaluated to be about 20% in the
graphite sample while correspondingly about 1, 3, and less than 1% in
the copper, SiO2, and UO2 samples. This result indicates that samples
having elements with higher rates of photoelectric absorption followed
by fluorescence tend to decrease the effect of the environmental interferences
over the entire spectrum. Through this investigation, it was
determined that the assumptions and design parameters employed in the
MCNP6 simulation are feasible to provide energy spectra for use with
HH-XRF.
This study applied some assumptions for the Monte Carlo simulation
due to a lack of detailed design parameters of the HH-XRF device. These
assumptions can also contribute to the incompleteness of simulation
results. For more precise prediction by simulation, additional simulations
to identify the possible origins of environmental interferences were
conducted with more instrument details including the collimator and
detector shield with adding in our MCNP6 input geometry. The major
differences between simulation and measurement spectra were identified
to derive from the environmental interferences below 10 keV. It was
estimated that these originated from the detector shield and the X-ray
tube collimator, which are particular design features that were not
simulated in MCNP6. As a result of the simulation for the detector shield,
it could be confirmed that some of the environmental interferences are
caused by the detector shield. In the case of the X-ray tube collimator, it
was observed that it can also arise from interference with the collimator
in response to the primary X-ray source beam. Based on this additional
discussion work, it will help for the MCNP simulation to have more
accuracy.
The MCNP6 simulations here were conducted using limited data
such as user manuals and open information. Nonetheless, they were
successful in simulating the primary beam source profile after interaction
with Rh targets and electrons, as well as the detector response after
interaction with samples except for the environmental interferences
stemming from interactions with the instrumentation. However, it is
well-known that the primary X-ray beam profile is sensitive and variable
as a small design parameter of X-ray tubes [18–20]. Especially, the focal
spot size and the distribution of incident electrons to the target may
affect the primary beam profile. In addition, the distances among the Rh
target, sample, and detector can affect the resulting energy spectrum via
detector response. Therefore, it is recommended that further efforts to
estimate a more precise source beam profile can contribute to having
more accuracy with the Monte Carlo simulation results.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Fig. 13. Excitation spectra from the graphite, SiO2, and UO2 samples (a) in the
5–10 keV range, and (b) normalized at about 6.4 keV.
Fig. 14. Excitation spectra to distinguish the elements in the samples.
W. Kim et al.
Spectrochimica Acta Part B: Atomic Spectroscopy 180 (2021) 106203
9
Acknowledgements
This work was supported by the Nuclear Safety Research Program
through the Korea Foundation of Nuclear Safety (KOFONS) using the
financial resource granted by the Nuclear Safety and Security Commission
(NSSC) of the Republic of Korea (Grant No. 1804026).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.sab.2021.106203.
References
[1] M.C. Corbeil, J. Philip, Potts and Margaret west (Eds.): portable X-ray fluorescence
spectrometry. Capa for in situ analysis, Anal. Bioanal. Chem. 393 (2009)
1385–1386, https://doi.org/10.1007/s00216-008-2587-6.
[2] Eva Marguí, Ren´e Van Grieken, State-of-the-art of X-ray fluorescence
instrumentation for chemical analysis, Petro Industry News (2013) 16–18. June/
July.
[3] Analytical Methods Committee AMCTB No. 89 (1), hand-held X-ray fluorescence
spectrometry, anal, Methods 11 (2019) 2498–2501, https://doi.org/10.1039/
C9AY90060H.
[4] International Atomic Energy Agency, In Situ Applications of X Ray Fluorescence
Techniques, IAEA-TECDOC-1456, IAEA, Vienna, 2005.
[5] Kelsey E. Young, Cynthia A. Evans, Kip V. Hodges, Jacob E. Bleacher, Trevor
G. Graff, A review of the handheld X-ray fluorescence spectrometer as a tool for
field geologic investigations on earth and in planetary surface exploration, Appl.
Geochem. 72 (2016) 77–87. ISSN 0883-2927, https://doi.org/10.1016/j.apgeoch
em.2016.07.003.
[6] Criminal Forensics, Bruker, website, https://www.bruker.com/ko/applications/de
tection-and-environmental/forensics/criminal-forensics.html. (Accessed 9 April
2021).
[7] INTERNATIONAL ATOMIC ENERGY AGENCY, Nuclear Forensics in Support of
Investigations, IAEA Nuclear Security Series No. 2-G (Rev. 1), IAEA, Vienna, 2015.
[8] D. Gallhofer, B.G. Lottermoser, The influence of spectral interferences on critical
element determination with portable X-ray fluorescence (pXRF), Minerals 8 (2018)
320, https://doi.org/10.3390/min8080320.
[9] J.P. Mondia, F. Goh, P.A. Bryngelson, J.M. MacPhee, A.S. Ali, A. Weiskopfa,
M. Lanana, Using X-ray fluorescence to measure inorganics in biopharmaceutical
raw materials, Anal. Methods 7 (2015) 3545–3550, https://doi.org/10.1039/
C4AY02936D.
[10] Morita Kazuo, Ogawa Yukiharu, Chi N. Thai, Fumihiko Tanaka, Etsuji Ishiguro,
Spectral analysis of reflected soft x-ray for detecting foreign materials in foods,
Food Sci. Technol. Res. 9 (3) (2003) 231–236. Released March 03, 2007, Online
ISSN 1881–3984, Print ISSN 1344-6606, https://doi.org/10.3136/fstr.9.231.
[11] P. Martinelli, J.L. Boutaine, G. Gousseau, J.C. Tanguy, C. Tellechea, Determination
of uranium and/or plutonium using X-ray fluorescence analysis excited by 192Ir
sealed sources, Nucl. Instruments Meth. Phys. Res. Sec. Accel. Spect. Detect. Assoc.
Equip. 242 (3) (1986) 569–573. ISSN 0168-9002, https://doi.
org/10.1016/0168-9002(86)90468-7.
[12] J. Py, J.-E. Groetz, J.-C. Hubinois, D. Cardona, Determination of Plutonium in
Nitric Acid Solutions Using Energy Dispersive L X-Ray Fluorescence with a Low
Power X-Ray Generator. Nuclear Instruments and Methods in Physics Research
Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, Apr
2015, pp. 131–137. https://doi:101016/jnima201501073.
[13] A. Pandey, F.A. Khan, A. Kelkar, et al., Determination of trace amounts of uranium
in plutonium oxide by wavelength dispersive X-ray fluorescence spectrometry,
J. Radioanal. Nucl. Chem. 324 (2020) 731–736, https://doi.org/10.1007/s10967-
020-07116-6.
[14] R. Redus, Amptek Inc, Amptek Application Note XRF-1: XRF Spectra and Spectra
Analysis Software. https://www.amptek.com/-/media/ametekamptek/document
s/resources/application-notes/xrf_2.pdf?dmc=1&la=en&revisi
on=6c68e243-0759-4a6a-abe7-88e489aa2f63. (Accessed 9 April 2021).
[15] Bruker Corporation, S1 Titan Overview. https://www.bruker.com/content/bruker
/int/en/products-and-solutions/elemental-analyzers/handheld-xrf-spectrometers
/S1-TITAN.html. (Accessed 9 April 2021).
[16] Bruker Corporation, Technical note #1303 Sharp Beam. https://www.technolog
ynetworks.com/analysis/product-news/bruker-introduces-2nd-generation-of-s1-tit
an-handheld-xrf-analyzer-216583. (Accessed 9 April 2021).
[17] Robert F. Shannon, Jr., Compact collimating device, US patent 8,223,925 B2, filled
April 15, 2010, and issued July 17, 2012, https://patents.google.com/pa
tent/US8223925B2/en (accessed 09 April 2021).
[18] M.R. Ay, S. Sarkar, M. Shahriari, D. Sardari, H. Zaidi, Assessment of different
computational models for generation of x-ray spectra in diagnostic radiology and
mammography, Med. Phys. 32 (6) (2005 Jun) 1660–1675, https://doi.org/
10.1118/1.1906126. 16013725.
[19] M. Yücel, E. Emirhan, A. Bayrak, C.S. Ozben, E. Barlas Yücel, Comparison of
simulated and measured spectra from an X-ray tube for the energies between 20
and 35keV, Nucl. Instruments Meth. Phys. Res. Sec. Accel. Spect. Detect. Assoc.
Equip. 799 (2015) 50–53. ISSN 0168-9002, https://doi.org/10.1016/j.nima.20
15.07.055.
[20] Mohammad M. Nasseri, Determination of tungsten target parameters for
transmission X-ray tube: a simulation study using Geant4, Nucl. Eng. Technol. 48
(3) (2016) 795–798.
W. Kim et al.