Available online at www.sciencedirect.

com

ScienceDirect
Availableonline
Available onlineatatwww.sciencedirect.com
www.sciencedirect.com
Energy Procedia 00 (2016) 000–000

ScienceDirect
ScienceDirect
www.elsevier.com/locate/procedia

Energy Procedia
Energy Procedia 00
127(2017)
(2017)000–000
76–86
www.elsevier.com/locate/procedia

International Youth Nuclear Congress 2016, IYNC2016, 24-30 July 2016, Hangzhou, China

Rapid Nuclear Forensics Analysis via Laser Based Microphotonic

The 15th International Symposium on District Heating and Cooling
Techniques Coupled with Chemometrics
Assessing
Bobby Bhattathe feasibility
*, Kalambuka Husdon of usingAngeyothe a heat
, Alix demand-outdoor
Dehayem- Kamadjeu a
temperature Department
function for a long-term district heat demand forecast
0F
0 F

of Physics, University of Nairobi, P.O. box 30197-0010, Nairobi, Kenya,

a
a

I. Andrića,b,c*, A. Pinaa, P. Ferrãoa, J. Fournierb., B. Lacarrièrec, O. Le Correc

Abstract
a
IN+ Center for Innovation, Technology and Policy Research - Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001 Lisbon, Portugal
b
Veolia Recherche & Innovation, 291 Avenue Dreyfous Daniel, 78520 Limay, France
c
Département Systèmes Énergétiquestool
Nuclear forensics (NF) is an important et Environnement - IMTattribution
for analysis and Atlantique, of
4 rue Alfred Kastler,
nuclear 44300 Nantes,
and radiological France (NRM) in
materials
support of nuclear security. The critical challenge in NF currently is the lack of suitable microanalytical methodologies for direct,
rapid and minimally-invasive detection and quantification of NF signatures. Microphotonic techniques can achieve this task
particularly
Abstract when the materials are of limited size and under concealed condition. The purpose of this paper is to demonstrate the
combined potential of chemometrics enabled LIBS and laser Raman spectromicroscopy (LRS) for rapid NF analysis and
attribution. Using networks
District heating LIBS, uranium lines at 385.464
are commonly nm,in385.957
addressed nm andas386.592
the literature one ofnm thewere
mostidentified
effectiveassolutions
NF signatures in uranium
for decreasing the
ore surrogates.
greenhouse gasAemissions
multivariate
fromcalibration
the buildingstrategy
sector. using
These artificial neural high
systems require network was developed
investments which areforreturned
quantification
throughofthetrace
heat
sales. Due
uranium. to the component
Principal changed climate
analysis conditions
(PCA) ofand LIBSbuilding
spectrarenovation policies,
achieved source heat demand
attribution of theinores.
the LRS
futurestudies
could ondecrease,
UCl33,
prolonging the investment return period.
UO33(NO33)22.6H22O, UO22SO44.3H22O and UO33 in pellet state identified the bands associated with different uranium molecules as
The main
varying scope
in the of this
range papertois867)
of (840 to assess
± 15the
cmfeasibility of using
-1. Using this
-1 the heat
signature, we demand – outdoor temperature
have demonstrated function
spectral imaging of for heat demand
uranium under
forecast. The district of Alvalade, located in Lisbon (Portugal), was used as a case study. The district is consisted of 665
concealed conditions.
©buildings
2016 Thethat vary in
Authors. both construction
Published by Elsevierperiod
Ltd. and typology. Three weather scenarios (low, medium, high) and three district
©renovation
2017 The Authors.
scenariosPublished by Elsevier
were developed Ltd. intermediate, deep). To estimate the error, obtained heat demand values were
(shallow,
Peer-review
Peer-review under responsibility
under responsibility ofof the
the organizing
organizing committee
committee of of IYNC2016
IYNC2016.
compared with results from a dynamic heat demand model, previously developed and validated by the authors.
The results
Keywords: showed
Nuclear that when
Forensics; Laseronly weather
Induced change
breakdown is considered,
Spectroscopy; Laserthe margin
Raman of error could be
Spectromicroscopy; acceptable Attribution.
Chemometrics; for some applications
(the error in annual demand was lower than 20% for all weather scenarios considered). However, after introducing renovation
scenarios, the error value increased up to 59.5% (depending on the weather and renovation scenarios combination considered).
1.The value of slope coefficient increased on average within the range of 3.8% up to 8% per decade, that corresponds to the
Introduction
decrease in the number of heating hours of 22-139h during the heating season (depending on the combination of weather and
renovation scenarios considered). On the other hand, function intercept increased for 7.8-12.7% per decade (depending on the
The illicit
coupled trafficking
scenarios). of nuclear
The values suggestedandcould
radiological
be used tomaterials
modify the(NRM) ledparameters
function to the emergence of a new
for the scenarios branch and
considered, of
science in the 1990’s known as “nuclear
improve the accuracy of heat demand estimations. forensics” (NF) [1, 2]. NF is the analysis of intercepted illicit NRM to

© 2017 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the Scientific Committee of The 15th International Symposium on District Heating and
* Corresponding author. Tel.: 254-788503899; fax: +0-000-000-0000 .
Cooling.
E-mail address: bobbybhatt@students.uonbi.ac.ke
Keywords: Heat demand; Forecast; Climate change
1876-6102 © 2016 The Authors. Published by Elsevier Ltd.
Peer-review under responsibility of the organizing committee of IYNC2016.

1876-6102 © 2017 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the Scientific Committee of The 15th International Symposium on District Heating and Cooling.
1876-6102 © 2017 The Authors. Published by Elsevier Ltd.
Peer-review under responsibility of the organizing committee of IYNC2016
10.1016/j.egypro.2017.08.072
 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86 77
2 Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000

provide evidence for attribution. NF and nuclear forensic interpretation are important for monitoring nuclear
safeguards in support of nuclear security [3, 4]. Microphotonic techniques are attractive for NF analysis as they have
the ability to perform rapid, minimally invasive and non-destructive analysis with high sensitivity. In this paper, we
discuss the utility of laser-induced breakdown spectroscopy (LIBS) and laser Raman spectromicroscopy (LRS) as
examples of such techniques in the field of NF. LIBS is an emission spectroscopic technique, in which the radiation
emitted by the micro plasma formed on focusing a powerful laser beam onto the sample is used to fingerprint the
elements associated with the spectral peaks [5, 6, 7]. On the other hand, LRS uniquely identifies and images specific
chemical compounds and microstructure in a given sample based on molecular vibrations following little or no
sample preparation [8, 9]. Although these techniques have potential to identify and quantify at stand-off distance the
elemental, molecular, and isotopic composition of NRM, their practical utility is limited by the complexity of the
spectra in air at atmospheric pressure and the interpretative challenges of the multivariate data [10]. For instance,
colored compounds are difficult to analyze by LRS because of intense fluorescence, laser heating and diversity of
oxidation states and deviations from stoichiometry; and for LIBS the analyzes gets difficult due to the matrix effects
which are driven by the strongly coupled transient laser-matter interactions. The LIBS spectra are optically thick and
therefore trace element spectral peaks are masked and buried by the strong background continuum. These limitations
may to a large extent be overcome by using multivariate chemometrics to reduce the data dimensionality and to
extract and model the subtle NF signatures in aid of NF attribution [5].

1.1. The Problem

Traditional nuclear forensic science makes use of analytical techniques such as gamma-ray spectrometry, alpha
spectrometry, mass spectrometry and microstructural techniques, which were developed for applications related to
the nuclear fuel cycle [1]. Unfortunately, most of these techniques are time-consuming and destroy the integrity of
the sample. The critical challenge in NF currently is the lack of suitable microanalytical methodologies for direct,
rapid and non-invasive detection and trace analysis of nuclear forensic signatures in NRM. The ability to rapidly
detect and measure trace NF signatures in NRM such as uranium (U) and its compounds in samples of limited size,
typical in NF, at stand-off distance and mostly under concealed conditions without destroying their integrity has
significance in nuclear security. Additionally, it could prove useful in combating illicit trafficking of NRM and fight
against nuclear terrorism that may employ improvised nuclear devices (IND) and/ radiological dispersal devices
(RDD).
Doucet et al. (2009) determined the isotopic ratio of partially resolved uranium-235/uranium-238 and
hydrogen/deuterium isotope shift lines in air at atmospheric pressure using partial least squares (PLS1) regression on
LIBS spectra for nuclear safeguard inspection [11]. Cremers et al. (2012) developed a LIBS system with lightweight
and compact high-resolution spectrometers to rapidly detect different ratios of isotopes of H and U and highly
enriched samples of 6Li and 7Li to monitor radiological, nuclear and explosive (RNE) threats [12]. Chinni et al.
(2009) performed a detailed study of uranium detection on surfaces and mixed in bulk materials (e.g., soil) by LIBS
for application to activities associated with environmental surveillance and detecting weapons of mass destruction
(WMD) [13]. These demonstrate the applicability of LIBS in air and atmospheric pressure for comprehensive in-
field NF analysis for a variety of NRM both in bulk and on surfaces (samples of limited size), as well as the benefits
of coupling multivariate chemometrics with the method.
The high collection efficiency of the microscope optics and the multi-scanning capability of the micro-Raman
spectrometer were utilized to characterize uranium oxide particles and a range of uranium compounds [14, 15]. It
was concluded that the use of in situ Raman provides a powerful technique to study the mechanism of
transformation and the stability of uranium oxides under oxidizing conditions [15]. Raman spectroscopy was used
on 95 samples comprising mainly of uranium ore concentrates as well as some UF4 and UO2 samples, in order to
classify uranium compounds for nuclear forensic purposes using PCA for the first time by Ho [16]. The use of LRS
in the study of uranium compounds gets limited due to fluorescence and ignition of the colored samples. However,
these limitations can be overcome to a great extent by the use of low energy laser and chemometrics. While low
laser energy reduces the fluorescence and burning of the colored samples, chemometrics helps to extract the
required information from the weak Raman signal arising due to low energy laser.
78 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86
Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000 3

We report in this paper progress in the development of chemometrics based microphotonic methodologies for
rapid nuclear forensic analysis and attribution in support of nuclear security exploiting the multi-modal capabilities
of LIBS and LRS. The current effort is to develop NF signatures, which are capable of identifying seized materials
(uranium compounds mixed in various binders such as cellulose and PVC, uranium-bearing mineral ores namely
high background radiation area (HBRA) soils and carbonatite rocks) with the greatest degree of confidence and
determine their origin.

2. Materials & Methods

2.1. Laser Induced Breakdown Spectroscopy (LIBS)

A Q-Switched Nd-YAG (Big Sky Laser; Ocean Optics, Inc.) laser, delivering maximum energy of 50 mJ at a
fundamental wavelength of 1064 nm and 10 Hz fixed pulse repetition frequency was optimized based on signal
(uranium line 386.592 nm) to noise ratio for analysis of uranium in uranium trioxide, uranium ore surrogates and
uranium-bearing minerals from Kenya pelletized in cellulose. The emission from the plasma plume was recorded by
seven HR2000+ High-resolution Miniature Fiber Optic Spectrometer, with a fused silica optical fiber (0.22
numerical aperture) of 101 mm focal length, placed at right angle to the direction of plasma expansion. About 50
spectra from the different points of the pellets were averaged to get a single LIBS spectrum of the sample.
Uranium lines were identified using a pellet with 108517 ppm of uranium (prepared by mixing UO3 in cellulose)
and later these lines were verified using pellets of 19 ppm, 142 ppm and 237 ppm of uranium. Pellets of uranium
ranging from 215 ppm to 2025 ppm were also similarly prepared to mimic typical scenarios of illicit NRM. These
pellets were used to build a multivariate calibration model for quantitative analysis of uranium using ANN.
Certified reference material (CRM), IAEA-RGU-1, Uranium Ore (IAEA/RL/148) was used for model validation.
Uranium-bearing carbonatite rocks from two regions of Kenya namely South Ruri and Lake Magadi and soils from
Coast, Lake Magadi and North Ruri were similarly prepared and analyzed.

2.2. Laser Raman Spectromicroscopy (LRS)

Pellets of uranium chloride (899 ppm of U), uranyl sulphate (875 ppm of U), uranyl nitrate (832 ppm of U) and
uranium trioxide (150 ppm, 800 ppm and 900 ppm of U) were prepared by mixing cellulose with the respective
uranium compounds to study the uranium bands associated with the different uranium compounds. Another pellet
was prepared by mixing all the four uranium compounds to study the variability of the uranium bands associated
with the different compounds in a mixture. In addition to these pellets, a pellet was prepared with cellulose only to
identify the Raman bands associated with cellulose. The pellet was mounted on microscope slide and focused using
50X short-range microscope objective. Raman spectra were collected with a confocal laser Raman spectrometer
(STR Raman Spectrum System, Seki Technotron Corp) equipped with a 300 mm imaging triple grating
monochromatic spectrograph and two lasers emitting at 532 nm and 785 nm. A 785 nm laser was used for all the
measurements. The laser 785 nm was operated at a power of 4.71 mW to minimize the fluorescence arising from
these colored uranium compounds and the burning of the sample due to laser heating. Measurement with silicon was
done prior to the measurements to ensure that the sensitivity of the instrument does not change significantly at the
time of measurement. Calibration of the wavenumber was done automatically by the instrument [16]. About 20
spectral data were acquired with exposure time 10s and 10 accumulations from each pellet and averaged to get a
single spectrum. An extended scan from 98 cm-1 to 1800 cm-1 was acquired with a center grating of 1050 cm-1.

2.3. Multivariate Chemometrics

1) Principal Component Analysis (PCA) - In our study PCA was utilized for data reduction and to find
combinations of variables (Principal Components) to describe any significant patterns within the entire data set. The
scores of the individual principal components obtained from each PCA model described the variation in the samples
comprising the data set, while loadings describe the correlations among the variables [17].
 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86 79
4 Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000

2) Artificial Neural Networks (ANN) - Due to the enhanced background, matrix effects as well as the non-
linearity of the noisy LIBS spectra, a multivariate calibration strategy using artificial neural network (ANN) coded
in Matlab was developed for quantitative analysis. A two layer (in-built with back propagation training function)
neural network (NN) was generated to quantify uranium concentrations. The network prediction ability is finally
evaluated by the validation set [18]. Back propagation training functions with Levenberg–Marquardt (LM)
algorithm was used to train the feed forward networks for non-linear regression [19]. The model was validated using
independent simulate samples of known concentration of uranium and CRM (IAEA/RL/148).

3. Results and Discussion

3.1. Qualitative analysis for identification of uranium lines using LIBS

The LIBS spectra from a pellet of 108517 ppm of uranium were collected to explore the uranium lines, which
could be detected in air and atmospheric pressure, using our LIBS set-up (shown in Fig (a)). The uranium lines
identified have been listed in Table 1. We observe that most of the uranium lines in the LIBS spectra lie in the
visible and UV region.

Table 1. Detectable U lines from the LIBS spectra of UO3 samples.

λ (nm) (UV Region) λ (nm) (Visible Region) λ (nm) (Visible Region) λ (nm) (Visible Region)

U I 348.937 UII 383.146 U II 393.202 U II 436.205

U I 356.659 U I 383.963 U II 405.004 U II 454.363
U I 358.488 U II 385.464 U I 415.397 U I 591.539
U I 365.915 U II 385.957 U II 417.159
U II 367.007 U II 386.592 U II 424.166
U II 378.284 U I 387.104 U II 434.169
U II 378.284 U I 387.104 U II 434.169

a b.

Fig.1. (a) LIBS spectra from a pellet of 108517 ppm of uranium bound in cellulose; (b) LIBS spectra from pellets of 19 ppm, 142 ppm
and 237 ppm of uranium bound in cellulose.

The uranium lines listed in Table 1 can be considered as NF signatures for detection of uranium concealed in
organic binder like cellulose, in air and atmospheric pressure. Although at 19 ppm of uranium most of the uranium
lines are masked by the strong continuum radiation the uranium lines at 385.464 nm, 385.957 nm and 386.592 nm
could be clearly identified as shown in Fig 1(b). These uranium lines were also identified in the LIBS spectra of soil
samples collected from high background radiation areas (HBRA) of Kenya namely: Kiruku (KRK-32-SL),
Msambeni (3-SL), Menzamwenye (K-6-SL) and Jombo Hills (DZHS-19) as shown in Fig 2(a) as well as in those
80 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86
Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000 5

collected from HBRA carbonatite rocks sampled from South Ruri (SRR-14, SRR-13, SRR-09) and Lake Magadi
(LM-10) as seen in Fig 2(b). The gamma radiation dose in most parts of Kenya’s HBRA is five times the world
average [20]. NF laboratories needs to deal with trace level of uranium in limited size or even swipes collected from
explosive or incident sites. The swipes or the samples may even be so little that the particle could be made visible
with the aid of a microscope after swiping it on glass slides. Under such circumstances, laser ablation in LIBS needs
to be performed with very low laser energy to refrain the sample from damage. Low energy leads to non-optimal
excitation of plasma, therefore weak spectral lines. Under such condition the uranium lines at 385.464 nm, 385.957
nm and 386.592 nm can be utilized to detect uranium concealed in cellulose or buried in soil and rock.
a b

Fig. 2. (a) LIBS spectra from soil (HBRA) pellets; (b) LIBS spectra from rock (HBRA) pellets.

3.2. Quantitative analysis of uranium using ANN on LIBS spectra

The area under the peak of a spectral line is a measure of the concentration of the respective element in the
sample. However, this line intensity depends on the laser pulse energy, plasma temperature, plasma size, atomic
parameters of the transition line, sample surface and detector response function [21]. Multivariate calibration
strategy using ANN was used to overcome this non-linear relationship and quantify trace uranium. The regression
model was built based on feature selection so as to completely capture the variables influencing the concentration of
uranium. In addition to this, the data size is reduced and the noisy spectral region that affects results is avoided [17].
Thus, the model was developed utilizing the variables (wavelengths) associated with the uranium lines as listed in
Table 1. The ANN algorithm was trained using known concentration of uranium in the range 215 ppm to 2025 ppm.
The training phase involved finding the best set of weights and bias values that would minimize the network output
errors. The regression model developed for predicting the unknown concentration of uranium is shown in Fig 3. The
model predicted the uranium concentration with R2 ≈ 0.85.
 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86 81
6 Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000

Fig. 3. Regression plot utilizing the uranium lines using ANN (back-propagation algorithm).

3.3. Exploratory analysis of the HBRA samples using Principal Component Analysis (PCA)

LIBS spectrum provides a unique chemical signature of a material, its geochemical fingerprint, and this
information can be used to discriminate geological specimens originating in one place from samples of the same
kind from other locations. The idea of geochemical fingerprinting is based upon the fact that the Earth's crust is
compositionally heterogeneous, both horizontally and vertically, and that minerals forming within the crust will
reflect that inherent geographic heterogeneity [22]. The PCA was performed on the entire LIBS spectra (200 nm to
980 nm) to avoid eliminating any factor, which may be responsible for establishing the pattern formation based on
its geological locations. The score of the principal components obtained from the PCA model for the complete LIBS
spectra is shown in Fig 4(a). The PCA score plot displayed the coordinates of the spectra in the plane of the
principal components (PCs) i.e., PC1 (57%) and PC2 (20%) to contain 76 % while PC3 and PC4 together contribute
to 21% of the total spectral information. We are still exploring the utility of PC3 and PC4 in attribution. The HBRA
soil samples collected from the different regions of Kenya – North Ruri, Lake Magadi and Coast were seen to form
three clusters attributing to their sources. The PC1 and PC2 loading spectrum as shown in Fig 4(b) reveals that the
source attribution of the HBRA soil samples is closely associated to the Rare-Earth Elements (REE) namely
dysprosium (Dy), holium (Ho), praseodymium (Pr), promethium (Pm), samarium (Sm) and titanium (Ti) besides
elements like scandium (Sc), thorium (Th), uranium (U), cobalt (Co) and strontium (Sr).
b.
a.

Fig. 4. (a) Principal component analysis score plot of HBRA soil samples analyzed using LIBS spectra (200-980) nm; (b) Loadings plot of the
PCA score of the LIBS spectra of HBRA soil samples using the complete spectra (200-980) nm.
 82 BobbyetBhatt
Bobby Bhatt et al. /Procedia
al./ Energy Energy Procedia 127
00 (2016) (2017) 76–86
000–000 7

3.4. Qualitative analysis for determination of molecular uranium bands using LRS

Raman spectra from the pellets prepared as described above were first subjected to pre-processing (detrending,
baseline offset, smoothing (Savitzky-Golay)) using Unscrambler software (Version 10.5) to remove the background
noise and fluorescence. These spectra were analyzed to study the Raman band associated with each of these
compounds as shown in Fig 5(a). Raman spectrum in the range 100 cm-1 to 1250 cm-1 is presented as the band of
interest lies in this range. The Raman spectra of cellulose have bands at 380 cm-1, 458 cm-1, 477 cm-1, 1090 cm-1 and
1120 cm-1 [23]. The Raman bands associated with each uranium compound and the assignments for the Raman
features are summarized in Table 3. Raman spectra from uranium trioxide pellet reveal the U-O stretching bond at
840 cm-1. The Raman spectra from uranium chloride reveal bands at 848 cm-1and 860 cm-1, which have been
attributed to UO2Cl+ [24, 25]. The observation of the UO2Cl+ band in uranium chloride pellet maybe due to the
hydroscopic nature of uranium chloride or due to the oxidation of the uranium chloride pellet under laser irradiation.
The Raman band associated with UO3NO3+ in uranyl nitrate was observed at 862 cm-1, while the band associated
with the nitrate group was observed at 1034 cm-1 and 1049 cm-1. In addition to these, we observed few more Raman
bands at 194 cm-1, 1010 cm-1 and 1072 cm-1, which are likely to be associated with the nitrate ion. In uranyl
sulphate, the Raman band associated with UO2(SO4) was observed at 867 cm-1 and that associated with the sulphate
ion at 1050 cm-1. The Raman spectra from the pellet prepared by mixing the four uranium compounds – uranium
trioxide, uranium chloride, uranyl nitrate and uranyl sulphate is shown in Fig 5(b). In this spectrum, the band
associated with the uranium molecule appears to be broadened. The broadening of the Raman band may be
accounted to the overlapping of the different bands associated with each uranium molecule (U-O at 846 cm-1,
UO3NO3+ at 876 cm-1, UO2SO4 cm-1 at (860±2) cm-1, and UO2Cl+ at (866±2) cm-1 and 841 cm-1) in the mixed pellet.
The Raman bands associated with the different anions (NO3- at 1047 cm-1 and SO42- at 1048 cm-1) in the mixed pellet
are not easy to identify as they lie very close to each other. It may be noted that Raman band associated with
uranium was distinctly observed in this pellet consisting of four uranium compounds. We therefore conclude that
using a laser with =785 nm the Raman band associated with uranium varies in the range of (840 to 867)  ± 15 cm-1.
This band may be used to detect the presence of uranium in uranium ore concentrates hidden in an organic matrix
like cellulose.

a b

Fig. 5. (a). Raman spectra from the uranium compounds using 785 laser; (b) Raman spectra from a pellet prepared by mixing the four uranium
compounds (uranium trioxide, uranium chloride, uranyl sulphate and uranyl nitrate) using 785 laser.
 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86 83
8 Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000

Table 3. Assignment of vibrational frequency to the uranium compounds.

Speciation Band assignment This work Literature Ref
(cm-1) (cm-1)
UO3 U-O stretching band 840 846 [15]
UO3(NO3)2.6H2O
UO3NO3+ UO22+ symmetric stretch 862 876 [15]
NO3 -
NO3 symmetric stretch
-
1034 1034 [15]
Free NO3 (ν1)
-
1049 1047 [24]
UO2SO4.3H2O
UO2SO4 867 (860±2) [24]
SO42- 1050 1048 [24]
UCl3
UO2Cl+ 860 (866±2) [24, 25]
848 841 [25]

3.5. Detection of uranium in concealed condition using LRS

Uranium trioxide bound in cellulose mimics an ideal scenario of trafficking uranium in concealed condition.
Therefore, LRS was used to study the detectability of 150 ppm uranium in a pellet. On analyzing the pellet surface
using the inbuilt microscope of the spectrometer, some regions appeared slightly yellowish in color while most of
the surface appeared to be colorless. Raman spectra from the yellowish region reveal the presence of U-O band at
840 cm-1. It was also observed that the intensity of uranium band (U-O) from the yellowish region of the pellet
increased linearly with increase in the total exposure time (number of accumulations times exposure time) as shown
in Fig 6. The presence of yellow traces on certain regions of the pellet surface and its disappearance in the rest of the
surface may be accounted to the fact that the uranium compounds do not mix uniformly with cellulose.

Fig.6. Raman spectra from the yellow traces on the UO3 pellet (150 ppm of uranium) using 785 laser.

To ease the process of detection of uranium, we randomly performed spectral imaging at 50 spots with exposure
time 10s and 10 accumulations on the 150 ppm of uranium pellet as well as on the pellet with four uranium
compounds (uranium trioxide, uranyl sulphate, uranyl nitrate and uranium chloride). Raman spectra collected from
these 50 shots is shown in Fig. 7 (a) As our investigation is related to the detection of uranium on the pellet we
assigned the range (821.277 to 870.812) cm-1 to the Red-band (R-band) for the mixture pellet. The distribution of the
uranium molecule was obtained using the AutoMap Software of the Raman set-up. The dark region in Fig. 7 (b)
represents the absence of the assigned R-band while the red region represents the presence of uranium molecule.
This method of spectral imaging can serve to detect uranium molecule by assigning a color to the Raman band (840
to 867) ± 15 cm-1. It is however not easy to identify the anion part of the molecule as the bands associated with the
sulphate and nitrate are very close to each other and may even overlap at certain time.
84 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86
Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000 9

a. b.

Fig. 7. (a) Raman spectra collected from 50 spots using 785 nm laser with R-band assigned to (821.277 to 870.812) cm-1; (b) Distribution of
uranium compound on the pellet surface.

4. Conclusion

The atomic uranium lines at 383.164 nm, 385.464 nm, 385.957 nm and 386.592 nm were identified as NF
signatures of uranium for preliminary investigations of trace uranium in cellulose, soil and rock samples using LIBS.
These NF signatures were used to detect the presence of uranium in the uranium ore surrogates. A multivariate
calibration strategy for the quantification of trace uranium was developed using ANN (back-propagation
algorithm). PCA performed on the complete LIBS spectra of the various soil samples collected from HBRA of
Kenya clustered into three groups, which related to their origins. The PCA loading spectrum revealed that the source
attribution of the HBRA soil samples is closely associated to the REE namely Dy, Ho, Pr, Pm, Sm and Ti besides
elements like Sc, Th, U, Co and Sr.
The uranium compounds (uranium trioxide, uranium chloride, uranyl sulphate and uranyl nitrate) studied using
LRS reveals that the Raman band associated with these uranium compounds ranges between (840 cm-1 - 867cm-1) ±
15 cm-1. We concluded that spectral mapping of the pellet can be utlized to detect uranium molecule when uranium
compounds are trafficked under concealed condition in organic binder like cellulose.
The primary objective in NF analysis is detection of NRM (uranium in our case). The NF signatures (elemental
and molecular) identified in the above study can be utlitized to develop a library for detection of uranium. Handheld
LIBS and LRS with this inbuilt library of elemental (for LIBS) and molecular (for LRS) NF signatures of uranium
can be developed for detection of uranium (concealed in oraganic compunds like cellulose) at the airports and sea
ports. The algorithm developed for quantification of uranium can also be embedded in the handheld LIBS to provide
preliminary information on uranium level in the seized material. On positive detection of uranium the material can
be taken to the NF laboratory for further analysis to throw light on its attribution. NF analysis cannot be successfully
achieved without International collaboration among different countries. The fingerprints of uranium ores collected
from different geographical locations with the help of international collaboration can be used to create a library.
LIBS coupled with principal component analysis (PCA) can be used for exploratory analysis to attribute an
unknown NRM to its source (mine) with the aid of this library. Thus, with the support of the airport and seaport
authorities these instruments can be set-up at the sensitive airports and seaports to restrict trafficking of uranium
concealed in organic compunds.
Thus, laser based microphotonic techniques (LIBS and LRS) coupled with chemometrics have the potential to
detect uranium in NRM rapidly under concealed condition without destroying the integrity of the sample. Such
analysis will enable to relate the seized material to its origin, thereby strengthening the nuclear security of the
country. These techniques have potential to be developed for in-field nuclear forensic analysis.
 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86 85
10 Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000

Acknowledgements

The first author would like to thank her husband and children for the constant support and encouragement. She
would also like to thank International Science Programme (ISP), Uppsala, Sweden for donating LIBS and LRS.
Finally, she would like to thank the Lord for giving her the strength to achieve her goal.

References

[1] Mayer K, Wallenius M and Ray I. Nuclear forensics-a methodology providing clues on the origin of illicitly trafficked nuclear materials.
Analyst 2005; 130: 433-441.
[2] Hinrichsen Y. Fingerprinting of nuclear material for nuclear forensics. University of Hamburg, Hamburg, German Member State Support
Programme for the IAEA Task; 2011.
[3] Nuclear Forensics Support. The IAEA Nuclear Security Series, International Atomic Energy Agency, Vienna; 2006.
[4] Robel M, Kristo MJ, and Heller MA. Nuclear forensic inferences using iterative multidimensional statistics. United States, Retrieved from
http://www.osti.gov/scitech/servlets/purl/966580; 2009.
[5] Labbé N, Swamidoss IM, André N, Martin MZ, Young TM and Rials TG. Extraction of information from laser-induced breakdown
spectroscopy spectral data by multivariate analysis. Appl. Opt. 2008; 47: G158-G165.
[6] Siu-Lung Lui, and Koujelev A. Accurate identification of geological samples using artificial neural network processing of laser-induced
breakdown spectroscopy data. J. Anal. At. Spectrom, 2011; 26: 2419-2427.
[7] Dwivedi Y, Thakur SN and Rai S. B. Laser induced breakdown spectroscopy diagnosis of rare earth doped optical glasses. Applied Optics
2010; 49: C42-C48.
[8] Lin H, Marjanović O, Lennox B and Clegg IM. Multivariate statistical analysis of Raman images of a pharmaceutical tablet. Applied
Spectroscopy 2012; 66(3):272-281.
[9] Moore DS and Scharff RJ. Portable Raman explosives detection. Analytical And Bioanalytical Chemistry 2009; 393(6): 1571–1578.
[10] Angeyo HK. A conceptual framework towards developing chemometrics and machine learning assisted spectrometries for rapid nuclear
forensic analysis. IAEA Nuclear Security Conference, Vienna, Austria; 2013.
[11] Doucet FR, Lithgow G, Kosierb R, Bouchard P, Sabsabi M. Determination of isotope ratios using laser-induced breakdown spectroscopy in
ambient air at atmospheric pressure for nuclear forensics. Journal of Analytical Atomic Spectrometry. 2011;26(3):536-41.
[12] Cremers DA, Beddingfield A, Smithwick R, Chinni RC, Jones CR, Beardsley B, Karch L. Monitoring uranium, hydrogen, and lithium and
their isotopes using a compact laser-induced breakdown spectroscopy (LIBS) probe and high-resolution spectrometer. Applied spectroscopy
2012;66(3):250-61.
[13] Chinni RC, Cremers DA, Radziemski LJ, Bostian M, Navarro-Northrup C. Detection of uranium using laser-induced breakdown
spectroscopy. Applied spectroscopy 2009 ;63(11):1238-50.
[14] Allen GC and Butler IS, Tuan NA. Characterisation of uranium oxides by micro-Raman spectroscopy. Journal of Nuclear Materials. 1987
Jan 31;144(1):17-9.
[15] Palacios ML and Taylor SH. Characterization of uranium oxides using in situ micro-Raman spectroscopy. Applied Spectroscopy 2000; 54
(9): 1372-1378.
[16] Ho DML, Jones AE, Goulermas JY, Turner P, Varga Z, Fongaro L, Fanghänel T, Mayer K. Raman spectroscopy of uranium compounds
and the use of multivariate analysis for visualization and classification. Forensic Science International 2015; 251: 61-68.
[17] Rai VN. Laser-induced breakdown spectroscopy: A versatile technique of elemental analysis and its applications. arXiv preprint
arXiv:1407.0132. 2014.
[18] Ying Liu, Belle R. Upadhyaya, and Masoud Naghedolfeizi. Chemometric data analysis using artificial neural networks. Applied
Spectroscopy1993; 47 (1): 12-23.
[19] Mukhono PM, Angeyo KH, Dehayem-Kamadjeu A, Kaduki KA. Laser induced breakdown spectroscopy and characterization of
environmental matrices utilizing multivariate chemometrics. Spectrochimica Acta Part B 2013; 87: 81-85.
[20] Patel JP. Environmental radiation survey of the area of high natural radioactivity of Mrima hill of Kenya. Discovery and Innovation 1991.
3: 31-35.
[21] Noll R. Laser-Induced Breakdown Spectroscopy; Springer-Verlag, Berlin Heidelberg (2012).
[22] Hark RR, Remus JJ, East LJ, Harmon RS, Wise MA, Tansi BM, Shughrue KM, Dunsin KS, Liu C. Geographical analysis of “conflict
minerals” utilizing laser-induced breakdown spectroscopy. Spectrochimica Acta Part B: Atomic Spectroscopy. 2012;74:131-6.
[23] Agarwal UP, Reiner RS and Ralph SA. Cellulose I crystallinity determination using FT–Raman spectroscopy: univariate and multivariate
methods. Cellulose. 2010;17(4):721-33.
[24] Lu G, Forbes TZ, Haes AJ. Evaluating best practices in raman spectral analysis for uranium speciation and relative abundance in aqueous
solutions. Analytical chemistry. 2015; 88(1): 773-80.
86 Bobby Bhatt et al. / Energy Procedia 127 (2017) 76–86
Bobby Bhatt et al./ Energy Procedia 00 (2016) 000–000 11

[25] Dargent M, Dubessy J, Truche L, Bazarkina EF, Nguyen-Trung C, Robert P. Experimental study of uranyl (VI) chloride complex formation
in acidic LiCl aqueous solutions under hydrothermal conditions (T= 21 C–350 C, Psat) using Raman spectroscopy. European Journal of
Mineralogy. 2013; 25(5): 765-75.