Virginia Commonwealth University
VCU Scholars Compass
" %+& )+!1 )".+$&+"".&+$1)& 0&,+/ "-0,#" %+& )+!1 )".+$&+"".&+$
Laser-Induced Breakdown Spectroscopy (LIBS) in
a Novel Molten Salt Aerosol System
Ammon N. Williams
Virginia Commonwealth University3&))&*/+2 1"!1
Supathorn Phongikaroon
Virginia Commonwealth University
,)),30%&/+!!!&0&,+)3,.(/0 %8-/ %,)./ ,*-//2 1"!1"$*+-1/
.0,#0%" " %+& )+$&+"".&+$,**,+/+!0%" 1 )".+$&+"".&+$,**,+/
67"10%,./"-.&+0/+!-".*&//&,+//$"-1 ,1(',1.+)/".*&//&,+/+2
7&/.0& )"&/.,1$%00,4,1#,.#.""+!,-"+ "//40%""-0,#" %+& )+!1 )".+$&+"".&+$0 %,)./,*-//0%/""+
"-0"!#,.&+ )1/&,+&+" %+& )+!1 )".+$&+"".&+$1)& 0&,+/4+10%,.&5"!!*&+&/0.0,.,# %,)./,*-//,.*,."
&+#,.*0&,+-)"/" ,+0 0 )& ,*-//2 1"!1
,3+),!"!#.,*
%8-/ %,)./ ,*-//2 1"!1"$*+-1/
Article
Laser-Induced Breakdown
Spectroscopy (LIBS) in a Novel
Molten Salt Aerosol System
Ammon N. Williams and Supathorn Phongikaroon
Abstract
In the pyrochemical separation of used nuclear fuel (UNF), fission product, rare ear th, and actinide chlorides accumulate in
the molten salt electrolyte over time. Measuring this salt composition in near real-time is advantageous for operational
efficiency, material accountability, and nuclear safeguards. Laser-induced breakdown spectroscopy (LIBS) has been
proposed and demonstrated as a potential analytical approach for molten LiCl–KCl salts. However, all the studies con-
ducted to date have used a static surface approach which can lead to issues with splashing, low repeatability, and poor
sample homogeneity. In this initial study, a novel molten salt aerosol approach has been developed and explored to
measure the composition of the salt via LIBS. The functionality of the system has been demonstrated as well as a basic
optimization of the laser energy and nebulizer gas pressure used. Initial results have shown that this molten salt aerosol–
LIBS system has a great potential as an analytical technique for measuring the molten salt electrolyte used in this UNF
reprocessing technology.
Keywords
Laser-induced breakdown spectroscopy, nuclear safeguards, pyroprocessing, molten salts
Date received: 24 November 2015; accepted: 1 March 2016
Introduction
Pyroprocessing technology is a dry reprocessing alternative
to traditional aqueous methods.
1
In this process, the uran-
ium in the used nuclear fuel (UNF) is electrochemically
dissolved at an anode and transported through a molten
salt electrolyte to a cathode inside an electrorefiner (ER).
As part of the process chemistry, fission products, rare
earth elements, and transuranics (including plutonium)
accumulate in the molten salt. Monitoring the composition
of the molten salt within the ER is important from oper-
ation efficiency, material accountability, and nuclear safe-
guards perspectives.
2
However, analytical methods to
measure the salt have been limited due to a high tempera-
ture operating condition (typically, 723–773 K) and high
radiation environment that exist within the process. As a
result, salt samples are usually drawn and measured via
inductively coupled plasma mass spectrometry (ICP-MS).
3
This process of extraction, material transfer, and later
sample preparation is cumbersome and can impose signifi-
cant delay between sampling and compositional results.
Laser-induced breakdown spectroscopy (LIBS) has been
proposed as an alternative analytical approach to ICP-MS
since it can be done in situ, with little to no sample
preparation, and compositional information of the sample
can be obtained in near real-time.
4
In LIBS, a pulsed laser is
focused onto a material surface to vaporize and ionize the
sample in the creation of a plasma plume. As the plasma
cools, the ionized atoms return to ground states with the
emission of characteristic light which can be collected via
optics. From the spectrum, qualitative and quantitate infor-
mation about the sample composition can be obtained.
Effenberger was among the first to demonstrate LIBS ana-
lysis in a molten LiCl–KCl salt by using a static surface
approach,
5
meaning that the laser light was focused down-
wards onto the top surface of the melted salt in a crucible.
Here, CrCl
3
, CoCl
2
, and MnCl
2
were studied in the molten
salt matrix at 773 K. This work showed that LIBS analysis in
the molten salt was feasible. Using the static surface con-
figuration similar to Effenberger,
5
Hanson et al. performed
Department of Mechanical and Nuclear Engineering, Virginia
Commonwealth University, Richmond, VA, USA
Corresponding author:
Ammon N. Williams, Department of Mechanical and Nuclear Engineering,
Virginia Commonwealth University, 401 West Main Street, Richmond, VA
23284, USA.
Email: williamsan25@vcu.edu
Applied Spectroscopy
2017, Vol. 71(4) 744–749
! The Author(s) 2016
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0003702816648965
journals.sagepub.com/home/asp
an in-depth LIBS study of CeCl
3
and MnCl
2
in LiCl–KCl
salts at temperatures ranging from 523 K (solid) to 773 K
(liquid).
6
Results of the study showed that, in general, the
molten samples had less self-absorption and lower relative
standard deviations (%RSDs) than the solid samples.
Weisberg et al. also studied a molten LiCl–KCl salt with
additives of EuCl
3
and PrCl
3
using the static surface
approach.
7
In this work, a thin film was observed on the
top surface of the salts containing Eu and Pr but not on
blank LiCl–KCl samples. The composition of the film with
respect to the bulk was not reported; however, the pres-
ence of the film influenced the shot-to-shot results since
plasma formed on the liquid surface differed from those
formed on the solid film surface. To address this issue,
the authors used conditional analysis to eliminate large
shot-to-shot fluctuations within the collected spectra,
thus effectively grouping shots that fell only on the liquid
surface. In addition to the conditional analysis, a partial least
squares (PLS) method was used to obtain qualitative and
quantitative results. More recently, Smith et al. from the
Argonne National Laboratory have designed a probe that
can be used to remotely analyze the static surface of the
molten salt within a glovebox environment.
8
To date, feasi-
bility studies have been done to measure U, Pu, and Np in
the LiCl–KCl salt.
These aforementioned LIBS studies conducted in a
molten LiCl–KCl salt have used a static surface approach.
Whereas this approach is feasible and can yield compos-
itional information of the salt, many potential issues exist.
For example, in the formation of the plasma on a liquid
sample, a shock wave is created which tends to splash the
sample—potentially causing a quenching of the plasma and/
or lens damage and degradation. In addition, liquid quenching
of the plasma can further reduce signal and the plasma life-
time.
9
Other potential issues are the film formation/dross
layer and bubble formation on the top surface of the salt,
which can lead to homogeneity concerns. These aspects
provide a motivation for the development of a new innova-
tive LIBS sampling configuration for molten salt analysis.
One potential method is to use a LIBS–aerosol meas-
urement technique.
10
In this approach, the laser is focused
within an aerosol gas stream to create the LIBS plasma. This
technique is often used for online monitoring of environ-
mental air samples for toxic metals.
11–13
In these cases, the
aerosol concentration within the gas stream is low, leading
to high shot-to-shot variation as a result of low probability
of particle interaction with the plasma. Diwakar et al. have
shown that the plasma-particle interaction can be defined
by Poisson modeling statistics.
14
As a result, this supports
the conditional analysis technique for excluding shots below
a certain threshold.
In many cases, the aerosol is generated from a liquid
sample via an aerosol generator. Here, the aerosol concen-
tration within the gas stream can be sufficiently high as to
reduce the probability of null shots; however, significant
shot-to-shot variation is still observed. Huang et al.
observed that the shot-to-shot variation could be large
due to fluctuations in the number of aerosol particles con-
sumed per plasma volume.
15,16
Poulain and Alexander,
17
as
well as Schechter,
18
observed the large shot-to-shot vari-
ation in liquid aerosols and concluded that the variation can
be attributed to (1) the number of particles within the
plasma volume, (2) the location and size of the particles
within the plasma, and (3) liquid droplet interference with
the incoming laser light. As a result, Schechter developed a
basic algorithm to perform conditional analysis on the col-
lected spectra to reduce variation prior to ensemble aver-
aging.
18
Using the above conditional analysis scheme,
Schechter was able to significantly improve the sensitivity
of the system.
Despite challenges with particle–plasma interactions and
interference, LIBS analysis from liquid aerosols has been
shown to perform well. Kumar et al. used a Meinhard nebu-
lizer to generate an aerosol and found that the aerosol
approach yielded better results than a liquid jet method.
19
They concluded that this in part was due to better utiliza-
tion of the laser energy to ionize the sample in the aerosol
stream, whereas much of the laser energy was used to first
vaporize the sample in the liquid jet case. Martin and Cheng
measured chromium in water and reported a limit of detec-
tion for Na down to 400 ng per dry standard cubic meter
(dscm).
20
Cahoon and Almirall measured the concentra-
tions of Sr, Ba, Mg, and Ca in water using both an aerosol
and micro-droplet approach.
9
Here, both methods resulted
in ultralow detection limits with a %RSD of the order of 1%.
Alvarez-Trujillo et al. performed stand-off LIBS in a liquid
aerosol and determined that they could detect Na down to
55 ppm in water at a distance of 10 m.
21
In addition to a good performance, an aerosol approach
has several advantages to a static surface configuration with
regards to an online measuring system for molten salts. In
an aerosol sampling approach, the liquid sample can be
drawn continuously from the bulk of the salt, thus reducing
variation due to inhomogeneity. In addition, since splashing,
bubble formation, and surface perturbations are not an
issue with the aerosol approach, the shot-to-shot frequency
of the sampling system is less limited. Lastly, the laser
energy is better utilized to ionize the sample. Due to
these advantages, it has been proposed to use this approach
to measure molten salt. To date, the high temperature
aerosol generation of molten salts has never been
reported. As a result, a novel molten salt aerosol and
LIBS system is being developed (patent pending) and has
been shown to have a promising potential for online mea-
suring of the molten salt composition. In this system, argon
gas is used to create aerosol in high temperature (773 K)
salt using a Collison nebulizer. The aerosol is introduced
into a sampling chamber where the laser light and plasma
emissions are transmitted in and out via quartz ports.
Following the sampling chamber, the aerosol sample is
Williams and Phongikaroon 745
passed through a coalesce filter to remove the molten salt
droplets prior to cooling the gas stream to room tempera-
ture. Currently, the system has been tested and optimiza-
tion of the laser energy and gas pressure will be initially
reported here—illustrating its promising potential in com-
position detection in a molten salt system for pyroproces-
sing technology.
Setup and Method
An illustration of the molten salt aerosol–LIBS configur-
ation using a Collison nebulizer is shown in Figure 1. In
the first step of the process, cold argon gas is heated to
773 K before entering the Collison nebulizer to prevent salt
freezing. This is done using 4.3 m of coiled 0.25 inch diam-
eter stainless steel tubing within a Kerr Lab furnace. The
hot gas then enters a three-jet Collison nebulizer pur-
chased from BGI, Inc., to generate the molten salt aerosol.
The nebulizer was modified in order to operate safely up to
773 K. The flow of argon and molten salt droplets then
passes into a sampling chamber with three quartz ports
from Rayotek, Inc. Through the top port, the laser light is
focused into the aerosol stream to create the plasma.
Through the back port, the plasma light is collected via a
fiber optic cable and optics positioned outside the sampling
chamber. The front port is used for visualization of the
plasma. The sampling chamber was designed and built in
house to accommodate up to 100 psi. The aerosol stream
then passes into a coalesce filter (with high temperature
seals) from United Filtration Systems, Inc., which removes
the aerosol from the gas stream. The entire system was
fitted together using Swagelok and NPT fittings with a high
temperature thread sealant design. The sampling chamber
and coalesce filter were heated via two 125 W cartridge
heaters. The hot gas exiting the coalesce filter then passes
through a heat exchanger to cool it down to ambient
temperatures before being passed through a HEPA filter
and vented into the exhaust system.
The inert atmosphere glovebox and the experimental
setup are shown in Figure 2a and 2b, respectively. Here, a
Q-smart 450 Nd:YAG laser (Quantel USA, Inc.) was
mounted vertically on the outside of the glovebox and
the laser light was directed via optics into the glovebox
through a quartz window located in the glovebox wall. A
laser mirror within the glovebox was used to guide the
laser into the sampling chamber through a lens with a
75 mm focal length (f). The focused laser light created a
plasma in the center of the sampling chamber. Light from
the plasma was collected via an f ¼ 75 mm lens and then
focused (f ¼ 100 mm lens) into a 4 m long fiber optic cable
from Ocean Optics, Inc. The fiber optic cable carried the
light to a Michelle 5000 spectrometer and iStar ICCD
detector located outside of the glovebox. The Michelle
spectrometer and ICCD detector were purchased from
Andor Technologies, Inc.
Salt samples were made using CeCl
3
(99.99%), LiCl
(99.99%), and KCl (99.95%) purchased from Alfa Aesar.
The LiCl and KCl salts were mixed at the eutectic ratio
(44 wt % LiCl and 56 wt % KCl) and the CeCl
3
composition
was set at 5 wt % for this study. The salts were formed into
ingots by mixing them in an inert atmosphere glovebox,
drying at 573 K for 24 h, and then pre-melting for 12 h
before rapidly freezing them into solid slugs. Each ingot
contained approximately 41.5 g of total salt. For each
experiment, a single salt ingot was loaded into the
Collison nebulizer jar and then the system was ramped
up to 773 K at approximately 4 K/min. The system was
held at temperature for 8 h prior to operating the system
to ensure that the sample was fully melted and homoge-
neous. Following melting, argon gas at the desired pressure
was introduced into the nebulizer to create the aerosol.
The Collison nebulizer uses the Venturi effect to draw up
Figure 1. Illustration of the proposed aerosol–LIBS setup.
746 Applied Spectroscopy 71(4)
and cycle the sample within the nebulizer jar. As a result of
the long melting period and forced convection during oper-
ation of the nebulizer, it is assumed that the salt compos-
ition in the aerosol stream was the same as the bulk salt
composition.
Initial experiments were done to explore the effect of
the nebulizer pressure and laser energy on the signal-to-
noise ratio (SNR). Here, the SNR is the peak intensity
divided by the background intensity near the peak of inter-
est. The laser energy was varied between 40 mJ and 140 mJ
and the nebulizer pressure was varied between 10 psi and
60 psi. For each variation, three repetitions were com-
pleted, each comprising of 250 laser shots. The 250 spectra
from each repetition were averaged to provide a single
representative spectrum.
Results and Discussion
For each element in the sample, the strongest expected
peaks were determined using the NIST database for
atomic spectral lines.
22
Three cerium peaks were selected:
418.660 nm (the dominant peak), 428.994 nm and 457.228
lines. These lines were picked due to high intensity and/or
low interference with other peaks. For Li, there are a
number of strong peaks; most notably are the lines at
460.289 nm, 610.354 nm, and 670.776 nm. For K, there
are two strong lines at 766.489 nm and 769.896 nm. A rep-
resentative spectrum for the 5 wt% CeCl
3
–LiCl–KCl salt
obtained via the aerosol–LIBS system is shown in Figure 3.
Figure 4 shows the SNR for the 418.660 nm,
428.999 nm, and 457.228 nm lines as a function of the
Figure 3. A representative LIBS spectrum obtained from the 5 wt % CeCl
3
–LiCl–KCl samples.
Figure 2. (a) Picture of the inert atmosphere glovebox with laser mounting and safety curtains. (b) Close up of the aerosol–LIBS
system mounted within the glovebox.
Williams and Phongikaroon 747
