U.S. patent application number 14/660696 was filed with the patent office on 2015-09-24 for electrochemical ion separation in molten salts.
The applicant listed for this patent is Sandia Corporation. Invention is credited to Harlan James Brown-Shaklee, Jon Ihlefeld, Leo J. Small, Erik David Spoerke, Karen Waldrip, David R. Wheeler, Jill S. Wheeler.
Application Number | 20150267316 14/660696 |
Document ID | / |
Family ID | 54141547 |
Filed Date | 2015-09-24 |
United States Patent
Application |
20150267316 |
Kind Code |
A1 |
Spoerke; Erik David ; et
al. |
September 24, 2015 |
Electrochemical Ion Separation in Molten Salts
Abstract
A purification method that uses ion-selective ceramics to
electrochemically filter waste products from a molten salt. The
electrochemical method uses ion-conducting ceramics that are
selective for the molten salt cations desired in the final purified
melt, and selective against any contaminant ions. The method can be
integrated into a slightly modified version of the electrochemical
framework currently used in pyroprocessing of nuclear wastes.
Inventors: |
Spoerke; Erik David;
(Albuquerque, NM) ; Ihlefeld; Jon; (Albuquerque,
NM) ; Waldrip; Karen; (Albuquerque, NM) ;
Wheeler; Jill S.; (Albuquerque, NM) ; Brown-Shaklee;
Harlan James; (Albuquerque, NM) ; Small; Leo J.;
(Albuquerque, NM) ; Wheeler; David R.;
(Albuquerque, NM) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Sandia Corporation |
Albuquerque |
NM |
US |
|
|
Family ID: |
54141547 |
Appl. No.: |
14/660696 |
Filed: |
March 17, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61955347 |
Mar 19, 2014 |
|
|
|
Current U.S.
Class: |
205/770 ;
204/243.1 |
Current CPC
Class: |
B01D 69/02 20130101;
C04B 35/634 20130101; C22B 26/10 20130101; C04B 35/447 20130101;
C04B 2235/3251 20130101; C04B 2235/6562 20130101; C04B 2235/6587
20130101; C22B 26/20 20130101; B01D 61/44 20130101; B01D 2325/14
20130101; C04B 2235/3418 20130101; C04B 2235/761 20130101; C04B
35/495 20130101; C04B 35/6261 20130101; C04B 35/632 20130101; C04B
2111/00801 20130101; C04B 2235/9669 20130101; G21F 9/30 20130101;
C04B 35/638 20130101; C04B 2235/3203 20130101; C22B 15/0026
20130101; C04B 2235/76 20130101; G21F 9/12 20130101; C04B 2235/3201
20130101; C04B 2235/3227 20130101; C04B 2235/764 20130101; C04B
2235/3244 20130101; B01D 71/024 20130101; C04B 2235/3215
20130101 |
International
Class: |
C25F 1/00 20060101
C25F001/00; C25F 7/00 20060101 C25F007/00 |
Goverment Interests
STATEMENT OF GOVERNMENT INTEREST
[0002] This invention was made with Government support under
contract no. DE-AC04-94AL85000 awarded by the U. S. Department of
Energy to Sandia Corporation. The Government has certain rights in
the invention.
Claims
1. An electrochemical cell for electrochemical ion separation,
comprising: an ion-selective ceramic membrane having selectivity
for at least one molten salt cation and selectivity against
transport of at least one contaminant cation, a molten salt
electrolyte, a cathode immersed in a purified molten salt
electrolyte on the cathode side of the ceramic membrane, an anode
immersed in a contaminated molten salt electrolyte comprising the
at least one molten salt cation and the at least one contaminant
cation on the anode side of the ceramic membrane, and an electrical
circuit for applying an electrical bias across the anode and
cathode, thereby causing the at least one molten salt cation to
transport across the ceramic membrane but substantially excluding
the at least one contaminant cation from transporting across the
ceramic membrane and thereby concentrating the at least one
contaminant ion in the contaminated molten salt electrolyte on the
anode side of the ceramic membrane.
2. The electrochemical cell of claim 1, wherein the ion-selective
ceramic membrane is selective by size exclusion or valence-charge
exclusion of the at least one contaminant cation.
3. The electrochemical cell of claim 1, wherein the ion-selective
ceramic membrane comprises
Na.sub.1+xZr.sub.2Si.sub.xP.sub.3-xO.sub.12.
4. The electrochemical cell of claim 1, wherein the
Na.sub.1+xZr.sub.2Si.sub.xP.sub.3-xO.sub.12 comprises
cation-substituted Na.sub.1+xZr.sub.2Si.sub.xP.sub.3-xO.sub.12.
5. The electrochemical cell of claim 4, wherein the
cation-substituted Na.sub.1+x-Zr.sub.2Si.sub.xP.sub.3-xO.sub.12
comprises KZr.sub.2(PO.sub.4).sub.3.
6. The electrochemical cell of claim 1, wherein the ion-selective
ceramic membrane comprises a garnet-structured lithium lanthanum
tantalate ceramic.
7. The electrochemical cell of claim 6, wherein the
garnet-structured lithium lanthanum tantalate ceramic comprises
Li.sub.6BaLa.sub.2Ta.sub.2O.sub.12.
8. The electrochemical cell of claim 1, wherein the molten salt
electrolyte comprises LiCl--KCl.
9. The electrochemical cell of claim 8, wherein the molten salt
electrolyte comprises a eutectic of LiCl--KCl.
10. The electrochemical cell of claim 8, wherein chloride ions in
the molten salt electrolyte are oxidized to form chlorine gas at
the anode and the chlorine gas is reduced to chloride ions at the
cathode.
11. The electrochemical cell of claim 1, wherein the at least one
contaminant cation comprises a fission product.
12. The electrochemical cell of claim 1, wherein the at least one
contaminant cation comprises a monovalent cation.
13. The electrochemical cell of claim 12, wherein the monovalent
cation comprises Cs.sup.+.
14. The electrochemical cell of claim 1, wherein the at least one
contaminant cation comprises a divalent cation.
15. The electrochemical cell of claim 14, wherein the divalent
cation comprises Cu.sup.2+ or Sr.sup.2+.
16. The electrochemical cell of claim 1, wherein the anode
comprises copper.
17. The electrochemical cell of claim 1, wherein the cathode
comprises platinum, carbon, tungsten, copper, or steel.
18. The electrochemical cell of claim 1, wherein the ion-selective
ceramic membrane comprises a planar configuration.
19. The electrochemical cell of claim 1, wherein the ion-selective
ceramic membrane comprises a tubular configuration.
20. A method to purify KCl--LiCl--CuCl.sub.2 molten salt,
comprising: adding lithium formate and/or potassium formate to a
KCl--LiCl--CuCl.sub.2 molten salt to precipitate copper metal
according to the reaction:
2LiCOOH+CuCl.sub.2.fwdarw.2LiCl+Cu(s)++2CO.sub.2 +H.sub.2, and/or
2KCOOH+CuCl.sub.2.fwdarw.2KCl+Cu(s)++2CO.sub.2+H.sub.2.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application claims the benefit of U.S. Provisional
Application No. 61/955,347, filed Mar. 19, 2014, which is
incorporated herein by reference.
FIELD OF THE INVENTION
[0003] The present invention relates to pyroprocessing and, in
particular, to a method and materials for electrochemical ion
separation in molten salts. Although the method can be applied to
molten salt purification generally, described herein is an example
of the electrochemical purification of molten metal chloride salts
used for nuclear waste remediation.
BACKGROUND OF THE INVENTION
[0004] Nuclear reprocessing technology has been developed to
chemically separate and recover plutonium, uranium, and other
actinides from irradiated nuclear fuel. As shown in FIG. 1,
pyroprocessing is a high-temperature method that involves
introducing spent nuclear fuel into a large scale electrochemical
cell, where uranium is first electrochemically dissolved into a
molten salt electrolyte and then reductively plated at the cell
cathode where it can be collected and ultimately recovered as pure
uranium. See H. Lee et al., Sci. and Tech. Nuc. Install. 1 (2013).
Left behind in the molten salt electrolyte (e.g., LiCl--KCl),
however, are residual fission products, transuranics, and other
contaminants in the molten salt electrolyte that can degrade the
electrochemical efficiency of the system and pose significant
challenges to ultimate waste packaging and disposal. Developing a
method to efficiently remove the contaminants would facilitate
molten salt recycling and reduce the volume of hazardous waste to
be packaged for disposal.
[0005] Currently there are several methods under development to
recapture and purify the molten salts including zeolite-based ion
exchange and melt crystallization. See H. Lee et al., Sci. and
Tech. Nuc. Install., 1 (2013); D. Lexa and I. Johnson, Mettalurq.
and Mater. Trans. B 32B, 429, (2001); M. Shaltry et al., Micropor.
Mesopor. Mater. 152, 185 (2012); and A. Williams et al., Chem. Eng.
Sci. 89, 258 (2013). In the ion-exchange process, contaminated
molten salt is run through a column containing an alumino-silicate
zeolite (e.g., Zeolite-4A), which extracts contaminants such as
Sr.sup.2+ or Cs.sup.+ from the salt. These zeolites can then be
sintered into waste forms for disposal. In contrast, melt
crystallization purifies the salts by taking advantage of the
increased solubility of contaminants in lower temperature eutectic
melts. Contaminated salt is melted, and then slowly cooled from the
top down, concentrating the contaminants in the liquid melt phase.
Once the salt has completely resolidified, the impurities are
concentrated in the lowest portion of the solid, which can be
removed and prepared for disposal.
[0006] The method of the present invention can facilitate the rapid
removal of contaminants, and could potentially be integrated into
the existing electrochemical system currently used for
pyroprocessing of nuclear waste.
SUMMARY OF THE INVENTION
[0007] The present invention is directed to a purification method
that uses ion-selective ceramics to electrochemically filter waste
products from a molten salt. This invention can be used in
collaboration with other methods as part of a multi-stage
purification process or it can be used in place of these approaches
as a stand-alone purification process. Further, the electrochemical
method can be integrated into a slightly modified version of the
electrochemical framework currently used in pyroprocessing. Central
to this concept is the use of ion-conducting ceramics that are
selective for the molten salt cations desired in the final purified
melt, and selective against any contaminant ions. Because these
ceramics are typically selective for only cationic transport,
charge compensation of ions transported through the ceramic may be
necessary. This charge compensation can be achieved either through
oxidation or reduction of gases, or addition of sacrificial cation
or anion-donors in anodic or cathodic salts. In the particular
example described below, the ion-selective ceramics of the
invention can be used to electrochemically remove contaminants such
as Cs.sup.+ from LiCl--KCl molten salts. The chemically and
thermally stable ion-conducting ceramics are capable of high rate,
selective ion transport. Described herein are the syntheses of
ion-conducting ceramics KSICON, LLTO, and LBLTO, though the process
may include the use of any ion conducting ceramic (including
glasses) or combinations of ceramics with the chemical stability,
thermal stability, and selective ion conductivity appropriate for
the molten salt and contaminants of a given process. The syntheses
of these materials can be tailored to modify critical properties
such as ionic conductivity, stability, and ionic selectivity, and
these particular ceramics have shown meaningful ionic conductivity
(particularly at elevated temperatures) of Li.sup.+ and/or K.sup.+,
relevant to this example. Further, the ion conducting ceramics show
selectivity against Cs.sup.+ transport, both under passive
conditions and under DC bias at elevated temperatures (up to
500.degree. C.). The ceramics also show selectivity against
divalent cations (e.g., Cu.sup.2+). Both KSICON and LBLTO show
chemical and structural stability against LiCl--KCl molten salts up
to 500.degree. C. Further, the rapid, macroscopic dissolution of Cu
metal anodes, and the cathodic plating of Cu and Li metals during
galvanostatic discharge experiments (at 100 mA/cm.sup.2) indicate
that this method can support the rapid mass transport needed to
provide a scalable salt purification process.
BRIEF DESCRIPTION OF THE DRAWINGS
[0008] The detailed description will refer to the following
drawings, wherein like elements are referred to by like
numbers.
[0009] FIG. 1 is a schematic illustration of pyroprocessing nuclear
waste recycling technology.
[0010] FIG. 2 is a schematic illustration of an electrochemical
cell for electrochemical removal of Cs.sup.+ from LiCl--KCl molten
salt. Chloride ions in the molten salt have been omitted for
clarity.
[0011] FIG. 3A is an illustration of the molecular structure of
NaZr.sub.2P.sub.3O.sub.12 showing zirconia octahedra, phosphate
tetrahedra, oxygen (small spheres), and sodium (large spheres).
FIG. 3B is a schematic view of the NaSICON structure showing a
pathway for ion conduction.
[0012] FIG. 4 is an illustration of the molecular structure of
KSICON (KZr.sub.2(PO.sub.4).sub.3) showing zirconia octahedra,
phosphate tetrahedra, oxygen (small spheres), and potassium (large
spheres). Substitution of K.sup.+ for Na.sup.+ in the lattice
produces an expansion of the c-axis and a contraction of the
a-axis.
[0013] FIGS. 5A and 5B are illustrations of the molecular structure
of Li.sub.5La.sub.3Ta.sub.2O.sub.12 showing Ta octahedra, oxygen,
La, and Li. The structure in FIG. 5A shows Li in octahedral
coordination, while the structure in FIG. 5B shows
tetrahedrally-coordinated Li.
[0014] FIG. 6A is a side-view schematic illustration of an
assembled electrochemical pellet stack. An ion conducting membrane
separates two different molten salt electrolyte binder pellets. A
copper anode and stainless steel current collectors cap the
assembly. FIG. 6B is a photograph of the pellets. FIG. 6C is a
top-view photograph of the assembled device, sandwiched between two
pieces of mica, that is connected to electrical leads before placed
between heated platens and tested electrochemically.
[0015] FIG. 7. is a photograph of a high temperature
electrochemical test platform (inside Ar glove box). Current
collectors attached to the molten salt/ceramic pellet stacks (not
shown) were connected to 4 electrical leads, and the assembly was
inserted between two platens heated to 500.degree. C.
[0016] FIG. 8 is a graph of the X-ray diffraction patterns for
NaSICON (NaZr.sub.2P.sub.3O.sub.12) and KSICON
(KZr.sub.2P.sub.3O.sub.12). (h k l) indices are provided for the
major peaks. Lattice shifts show c-axis expansion and a-axis
contraction in the KSICON material.
[0017] FIG. 9A -9C show EDXS spectral maps of KSICON ceramic,
infiltrated with LiCl--KCl--CsCl molten salt. FIG. 9A is a map
showing overlaid Cs signal from CsCl and Zr signal from KSICON.
FIG. 9B is a map showing Cs signal only. FIG. 9C is a map showing
Zr signal only.
[0018] FIG. 10 is a graph of imaginary versus real impedance
collected from KSICON at room temperature.
[0019] FIG. 11A is a graph of the X-ray diffraction spectrum for
LBLTO compared against the calculated pattern for LLTO
(Li.sub.5La.sub.3Ta.sub.2O.sub.12). FIG. 11B is a graph showing the
variation of lattice parameter with Ba-substitution into LLTO. The
green dot corresponds to composition of LBLTO shown in FIG.
11A.
[0020] FIG. 12A is a photograph comparing LLTO ceramic pellets
before and after exposure to LiCl--KCl molten salt at 500.degree.
C. for 30 minutes. FIG. 12B is a graph of the X-ray diffraction of
exposed pellet, indicating no changes in crystal structure during
30 minute exposure (compared with FIG. 11A above). *There is a
small amount of residual KCl evident in the diffraction
spectrum.
[0021] FIG. 13 show EDXS maps of LBLTO pellets as made (above) and
after immersion in molten LiCl at 650.degree. C. for 4 hours
(variation in sample texture is due to sample preparation). As
made, La (magenta), Ta (green), and Ba (yellow) appear uniformly
distributed throughout the grain structure of the ceramic. After
LiCl treatment, La and Ta remain uniformly distributed, but
segregated regions of high Ba content appear.
[0022] FIG. 14A is an SEM micrograph of LBLTO ceramic
microstructure. FIG. 14B is an EDXS spectrum from LBLTO after
exposure to LiCl--KCl--CsCl molten salt for 60 minutes at
400.degree. C.
[0023] FIG. 15 is a graph of imaginary versus real impedance
collected from LBLTO at room temperature.
[0024] FIG. 16A is a schematic illustration of electroreduction of
copper in a pellet stack using electrolyte binder pellets and an
ion-conducting separator. FIG. 16B is a schematic illustration or
electroreduction of lithium in a pellet stack using electrolyte
binder pellets and an ion-conducting separator.
[0025] FIG. 17 is a graph of real versus imaginary impedance for
KSICON at 500.degree. C. between two LK-EB pellets. Inset is the
equivalent circuit used to fit the data.
[0026] FIG. 18 is a graph of galvanostatic discharge curves for the
electroreduction of copper and lithium using KSICON and LBLTO
separators at 100.degree. C. Curves were obtained at a constant
current density of 100 mA/cm.sup.2.
[0027] FIG. 19A-19D are photographs showing the formation of
electroreduced copper metal (FIG. 19A) in tested pellet stacks
(with KSICON separator). Reduced copper at the cathode and the
dissolved copper anode are also visible in the dissected cell (FIG.
19B). Formation of lithium metal (evident as beads in FIG. 19C)
after lithium electroreduction using LBLTO as a separator. In the
dissected cell (FIG. 19D) reduced lithium is visible on the cathode
and the cathode-side of the LK-EB pellet. Again, the copper anode
has been clearly dissolved.
[0028] FIG. 20A shows EDXS spectra and inset SEM micrographs from
CsLK-EB cross sections and FIG. 20B shows EDXS spectra from CuLK-EB
pellet cross-sections after copper electroreduction at 500.degree.
C. with a KSICON separator. The presence of Cu in the CsLK-EB
pellet indicates solubility of the oxidized copper anode. Absence
of Cs in the CuLK-EB pellet indicates ion transport selectivity
against Cs.sup.+.
[0029] FIG. 21A shows EDXS spectra and inset SEM micrographs from
CsLK-EB cross section and FIG. 21B shows LK-EB pellet
cross-sections after lithium electroreduction at 500.degree. C.
with a KSICON separator. The presence of Cu in the CsLK-EB pellet
indicates solubility of the oxidized copper anode. Absence of Cs
and Cu in the LK-EB pellet indicates ion transport selectivity
against Cs.sup.+ and Cu.sup.2+.
[0030] FIG. 22 shows a process to chemically reduce trace
CuCl.sub.2.
[0031] FIG. 23A is a side-view schematic illustration of
"concentric reactor" for molten salt purification using closed-end
ceramic tubes as the ion-conducting separator. FIG. 23B shows the
concentric reactor after purification of molten salt.
[0032] FIG. 24A is a schematic illustration of a VCR assembly. FIG.
24B shows the VCR assembly installed in a VCR reactor.
[0033] FIG. 25A is a side-view photograph of a VCR-type reactor
with copper anode extending from the reactor internal volume. FIG.
25B is a top-view photograph of a reactor assembled in testing
apparatus after cooling from an electrochemical test.
[0034] FIG. 26 is a graph of the current response over time for a
VCR-type reactor cell configured with an LLTO ceramic. Constant
applied potential was -200 mV across the cell. Charge balance was
maintained using Cu redox chemistry.
[0035] FIG. 27A is a graph of the evolution of Cs:Li cation ratio
in anodic and cathodic salts and FIG. 27B is a graph of the
evolution of Cs:K cation ratio in anodic and cathodic salts,
indicating selective transport of Li.sup.+ and K.sup.+, but not
Cs.sup.+ through LLTO during molten salt purification.
[0036] FIG. 28 is a schematic illustration of a closed-loop
pipe-based reactor design incorporating the VCR-type reactor for
gas-mediated salt purification.
DETAILED DESCRIPTION OF THE INVENTION
[0037] Pyroprocessing has emerged as a promising approach to
electrochemically separate recyclable actinides from waste fission
products. However, further segregation of select fission products
(FPs) is still needed for efficient and cost effective waste
disposal. The present invention uses ion-conducting ceramics as
selective ion filters in the electrochemical segregation of waste
FP constituents, specifically the segregation of cesium from molten
LiCl-KCl eutectic electrolytes. Removal of short-lived, high
heat-generating FPs is an important objective in the consolidation
of radioactive waste and is key to recycling the LiCl--KCl molten
salt electrolyte used in pyroprocessing.
[0038] The method for electrochemical ion separation of the present
invention uses ion-selective ceramics for mixed waste separations.
The method can use modified potassium-substituted NaSICON (KSICON)
and garnet-structured lithium lanthanum tantalate (LLTO) ceramics,
including a barium-modified LLTO (LBLTO). Additional ceramics or
ceramic glasses can also be used, including LiSICON, LIPON, lithium
lanthanum zironates, .beta.''-alumina, and others. In the present
case, both the KSICON and LLTO-based forms of ion-conducting
ceramics exhibit stability against LiCl--KCl eutectic molten salts
at elevated temperatures (e.g., 500.degree. C.). Critically, based
on impedance spectroscopy and galvanostatic electroreduction
experiments, both ceramics also show effective conductivity of
lithium and/or potassium ions, while selecting against the
transport of contaminant cesium ions. Therefore, the ion-selective
ceramics can be used to electrochemically purify LiCl--KCl-based
molten salts. Although the examples described below focus on the
removal of Cs.sup.+ from LiCl--KCl molten salt, the system can be
used to remove other monovalent and multivalent cations from the
system as well.
[0039] FIG. 2 schematically illustrates an exemplary
electrochemical cell 10 for separating the contaminant Cs.sup.+
from a LiCl--KCl molten salt. The cell 10 uses an electric field to
move cations in a molten salt through an ion-selective ceramic
membrane 11. Based on characteristics such as ionic radius or
valence charge, the ceramic will selectively filter out the
contaminants, allowing only purified cations of Li.sup.+ or K.sup.+
to pass through. This example relies primarily on size exclusion to
separate Cs.sup.+ from the Li.sup.+ and K.sup.+. To achieve this
electrochemical separation, Cs.sup.+-contaminated LiCl--KCl salt 12
is placed opposite purified LiCl--KCl electrolyte 13 and separated
therefrom by the ion-selective ceramic membrane 11. Under an
electrical bias applied between an anode 14 immersed in the
contaminated salt 12 and a cathode 15 immersed in the purified salt
13, Li.sup.+ and K.sup.+ ions are transported across the
ion-conducting ceramic membrane 11, while larger Cs.sup.+ cations
are excluded, effectively filtered and ultimately concentrated in
the contaminated salt 12 on the anode-side of the cell 10.
Ultimately, the resulting heavily Cs.sup.+-enriched salt can be
processed for disposal, but the volume of salt waste is
dramatically reduced. This exemplary method takes specific
advantage of the fact that the ionic radius of Cs.sup.+ is
significantly larger than either Li.sup.+ or K.sup.+ (i.e.,
r.sub.Li=0.090 .ANG., r.sub.K=1.52 .ANG., r.sub.Cs=1.81 .ANG.). To
maintain charge balance on the anodic side of the cell, "excess"
chloride ions in the molten salt can be electrochemically oxidized
at the anode to form chlorine gas. Meanwhile, chlorine gas
(Cl.sub.2), ideally that produced at the anode, is bubbled into the
cathode where it is reduced to chloride ions, balancing the influx
of K.sup.+ and Li.sup.+ coming through the ceramic membrane.
[0040] Naturally, the properties of the ceramic separator will
strongly influence the effectiveness of this method. In particular,
the ceramic material preferably serves as a stable physical barrier
between contaminated and purified salts at 500.degree. C., is
resistant to radiation exposure, is electrical insulating, and
enables efficient, selective ion transport of Li.sup.+ and K.sup.+.
The higher the ionic conductivity through the ceramic, the greater
the ionic current through the cell and the faster the salt can be
purified. The greater the selectivity of the ceramic, the higher
the purity of the recycled salts. As described below, NaSICON-type
ceramics and garnet-structured lithium lanthanum tantalates are
especially attractive ceramics that can provide both efficiency and
selectivity.
NaSICON-Type Ceramics
[0041] The term NaSICON (Na Super Ion CONductor) commonly refers to
the ion-conducting ceramic
Na.sub.1+xZr.sub.2Si.sub.xP.sub.3-xO.sub.12, but more generally
belongs to a family of compounds with the basic formula
[M.sup.I.sub.1][M.sup.II.sub.2][A.sup.VI.sub.2][B.sup.IV.sub.3]O.sub.12
(most often designated as M.sub.xZr.sub.2(PO.sub.4).sub.3, where M
is typically an alkali cation). As shown in FIG. 3A, these
materials form a rigid hexagonal framework of corner-linked
ZrO.sub.6 octahedra and PO.sub.4 tetrahedra and containing
so-called M1 and M2 interstitial positions, typically filled with
alkali cations such as sodium. See M. Cretin et al., Sensors and
Actuators B 43, 224 (1997). As shown in FIG. 3B, the conduction of
alkali cations through NaSICON-type materials is based on the
movement of the ion from site to site through the crystal lattice
via channels (bottlenecks) constricted by zirconia octahedra and
phosphate tetrahedra. As a result, ion conductivity through this
matrix depends strongly on both the size of these conduction
channels and the size of the mobile cation. See D. Bykov et al.,
Sol. St. Ionics 182, 47 (2011).
[0042] As shown in FIG. 4, the present invention provides as an
example material a potassium-substituted NaSICON variant, KSICON
(KZr.sub.2(PO.sub.4).sub.3). Because K.sup.+ is larger than
Na.sup.+ (r.sub.K=1.52 .ANG., r.sub.Na=1.16 .ANG.), forcing
transport of K.sup.+ ions through a sodium-based NaSICON structure
would have required a lattice expansion "in situ," a process that
could be detrimental to the structure and integrity of the ceramic.
By synthesizing the lattice with potassium already in place, the
lattice is created "ready" to accommodate K.sup.+ for ion
transport. Transport of the much smaller Li.sup.+ does not pose any
steric problems in this lattice. In contrast, steric hindrance can
severely limit transport of the larger Cs.sup.+ cation
(r.sub.Cs=1.81 .ANG.) through the lattice. Ultimately, this ceramic
ion conductor allows facile ion transport of Li.sup.+ and K.sup.+,
but prevents transport of Cs.sup.+, providing the ion selectivity
desired to realize the electrochemical ion separation scheme shown
in FIG. 2.
[0043] In addition to alkali cation substitution to create KSICON,
ion selectivity can be further tuned through modifications to the
zirconia octahedra and the phosphorus tetrahedra to vary the shape
of the ion transport pathways. For example, partial substitution of
silicon for phosphorus in NaSICON produces a notable increase in
sodium conductivity, in large part by changing the shape of the
tetrahedral side of the ion channel bottleneck. See O. Bohnke et
al., Sol. St. Ionics 122, 127 (1999).
LLTO-Type Ceramics
[0044] Garnet-structured lithium lanthanum tantalate ceramics
(Lithium Lanthanum Tantalate or LLTO), shown in FIGS. 5A and 5B,
have recently been shown to possess high ionic conductivities with
stability against facile reduction of the constituent ions. See J.
Awaka et al., Sol. St. Ionics 180, 602 (2009); J. Ihlefeld et al.,
Adv. Mater. 23, 5663 (2011); and V. Thangadurai and W. Weppner,
Adv. Funct. Mater. 15, 107 (2005). Lithium ions in
Li.sub.5La.sub.3Ta.sub.2O.sub.12 and related materials move within
a three-dimensional network of energetically equivalent partially
occupied sites. Lithium is both octahedrally and tetrahedrally
coordinated in this lattice, and ion conduction is possible along
both sites, though it is believed that transport along the
octahedral sites is more favorable.
[0045] As a known lithium ion conductor, this material is naturally
suitable for Li.sup.+ transport. According to the present
invention, the crystal growth can be modified to accommodate
K.sup.+ transport as well (while excluding Cs.sup.+ transport). As
in the NaSICON modification described above, this involves
expansion of the crystal lattice and, consequently, ion channel
size via substitution of lanthanum with combinations of alkali and
alkali-earth ions. It has been shown that by expanding the lattice
parameter (e.g., through substitution of Ba.sup.2+ (r=1.49 .ANG.)
for La.sup.3+ (r=1.174 .ANG.)) the polyhedra expand and conduction
via tetragonal sites as well as octahedral sites becomes more
favorable. See J. Ihlefeld et al., Adv. Mater. 23, 5663 (2011); and
V. Thangadurai and W. Weppner, Adv. Funct. Mater. 15, 107 (2005).
The low occupation of lithium in the polyhedra means that a hop to
a neighboring site has greater than 50% probability of
occurring.
Material Syntheses and Processing
[0046] To demonstrate the invention, ion-conducting ceramics were
synthesized utilizing primarily solid-state synthetic techniques,
tailoring the chemistry to accommodate variations in material
composition and crystal structure. All chemistries described below
utilize surrogate ions.
NaSICON-Type Ceramics
[0047] The reagents used for solid state synthesis of a variety of
NaSICON-type powders included sodium phosphates
(NaH.sub.2PO.sub.4or Na.sub.2HPO.sub.4) potassium phosphate salts
(KH.sub.2PO.sub.4or K.sub.2HPO.sub.4) zirconium oxide (ZrO.sub.2),
and phosphorus pentoxide (P.sub.2O.sub.5). Stoichiometric
quantities of target reagents were combined in a mortar and pestle
with hexanes, limiting or eliminating exposure to moisture. Solvent
was allowed to evaporate and the resulting powder was calcined at
500-800.degree. C. in an alumina crucible. The resulting hard,
glassy solid was removed from the crucible and ball-milled with
ZrO.sub.2 media in hexanes overnight. The milled powder was
collected, dried at 60.degree. C., and then combined with 2.5 wt %
propylene carbonate binder in acetone before uniaxial pressing at
10,000 psi in a 0.25 inch diameter steel die. Pellets were placed
on a bed of mother powder and fired to 1000-1100.degree. C. for 12
hours in static air (ramp rate of 5.degree. C./min). During
heating, a 4 hour hold at 400.degree. C. was used to ensure
effective binder burnout before reaching the elevated sintering
temperatures. Pellets were slowly cooled to room temperature.
Powder samples were similarly prepared, but were not mixed with
binder or pressed into pellets.
LLTO-Type Ceramics
[0048] Ceramic test specimens of LLTO-based garnet phases were
prepared using solid-state ceramic synthesis techniques. Starting
materials were lithium carbonate, lanthanum oxide, tantalum oxide,
and barium carbonate. The powders were combined and mixed by ball
milling with ZrO.sub.2 media in hexane for 24 hours. The hexane was
then evaporated from the combined powders using a rotary evaporator
to minimize segregation. Powders were then calcined in closed
crucibles at 1000.degree. C. for 6 hours to decompose the carbonate
precursors and to react the oxides to form the LBLTO compounds. The
calcined powders were subsequently milled in hexane for 24 hours to
break-up sintered and agglomerated particles. After evaporating the
hexane liquid, the powders were combined with 3 wt % poly vinyl
butyrol and ethanol binder and pressed into 0.5 inch diameter
pellets under a uniaxial pressure of 10,000 psi. The pellets were
then isostatically pressed under a pressure of 22,000 psi. Pellets
were sintered into ceramic monoliths by burying in the mother
powder to minimize lithium volatilization and firing to
1050.degree. C. in air for 6 hours using 5.degree. C./min
temperature ramps. A 500.degree. C., 3 hour rest was utilized on
the increasing temperature ramp to remove the binder from the
samples prior to densification.
Materials Characterization
[0049] The morphology, chemistry, and crystal structure of
synthesized materials, both in powder form and as sintered pellets,
were characterized before evaluating the ion-transport and molten
salt stability of these compounds.
Structural Characterization
[0050] To investigate ceramic microstructure, morphology and
qualitative chemical composition, platinum-coated samples were
examined in a Zeiss Supra.TM. 55VP scanning electron microscope
(SEM) at 10 keV, fitted with an Oxford X-Max energy dispersive
x-ray spectroscopy (EDXS) detector and AZtec.RTM. software.
Crystallography of ion-conducting materials was evaluated by x-ray
diffraction (XRD), performed on both powder samples and ceramic
pellets using a Panalytical X'Pert Pro powder diffractometer or a
Siemens D500 theta/theta diffractometer (Cu Ka radiation).
Molten Salt Stability
[0051] Sintered ceramic pellets of NaSICON, KSICON and LBLTO were
prewarmed and immersed in eutectic molten salts of KCl--LiCl (56-44
wt %) and KCl--LiCl--CsCl (18-25-57 wt %) to evaluate ceramic
stability. The binary KCl--LiCl melt temperature is
.about.355.degree. C., but that temperature decreases with
increasing CsCl content, becoming as low as 260.degree. C. with 29
mol % (57 wt %) CsCl.
[0052] Initial stability tests were performed in an Ar glove box,
immersing test pellets into eutectic KCl--LiCl molten salt for 30
minutes at 500.degree. C. The eutectic salts were melted in a glass
test tube and heated in a custom resistive heater. Temperature in
the molten salt was monitored using a pyrex-coated type K
thermocouple. After 30 minutes of exposure, the samples were cooled
and the molten salt was removed either manually or with aqueous
washing.
[0053] The ternary immersion tests were performed in a dry N.sub.2
purged glove box. Inside this glove box, glass vials containing the
salts of a desired salt composition were fitted into a chambered
aluminum block that was heated on a hot plate. A type J
thermocouple, inserted into the block, was used to monitor
temperature and maintain a temperature of 400.degree. C. inside the
aluminum block (400.degree. C. was the maximum temperature
achievable with this heating configuration).
[0054] At specific intervals, samples were allowed to cool and the
pellets were removed from the solidified molten salt either
manually or with aqueous washing. Select pellets were either
examined by XRD to determine any changes in the crystal structure
during molten salt exposure. Other pellets were fractured after
exposure to reveal cross-sections of the ceramic interior for SEM
and EDXS analysis of ceramic morphology and chemistry.
[0055] Higher temperature testing for stability in LiCl molten
salts was performed by burying LBLTO ceramics in LiCl powder inside
a glass container (vial). The sample was heated to 650.degree. C.
under flowing N.sub.2 and held for 4 hours. After slowly cooling to
room temperature, the pellet was removed from the molten salt and
examined by SEM/EDXS.
Impedance Spectroscopy
[0056] Gold blocking electrodes were sputter deposited onto the
parallel faces of the 0.25-inch diameter KSICON pellets to enable
impedance measurements. Impedance data were collected using an
Agilent 4284A instrument at room temperature. An oscillator voltage
of 20 mV rms was applied over a frequency range from 1 kHz to 1 MHz
and the absolute impedance and phase angle measured.
High Temperature Electrochemical Characterization
[0057] The high temperature electrochemical behavior of the
ceramics was determined using "pellet stacks," configured as shown
schematically in FIG. 6A. These stacks were composed of molten salt
pellets, an ion-conducting ceramic, a copper anode, and stainless
steel current collectors. FIGS. 6B and 6C show the molten salt
pellets, also known as electrolyte binder (EB) pellets, and an
assembled device structure, respectively. Molten salt pellets were
11 mm diameter, while KSICON pellets were typically 13.3 mm in
diameter and 2.9 mm thick, and LBLTO pellets were typically 15.6 mm
in diameter and 3.2 mm thick. The assembled structures were
sandwiched between carefully stapled sheets of mica to help hold
the stacks together (until heated) and to electrically insulate the
structure from the platens used to heat the stacks.
[0058] To protect the integrity of the molten salt pellets, stack
assembly was performed in a dry room and placed in an Ar-glove box
for electrochemical evaluation. Inside the glove box, the four
current-collector tabs were connected to a Solartron SI 1287
electrochemical interface for 4-point electrochemical measurements,
and the pellet stack was placed between two platens heated to
500.degree. C., as shown in FIG. 7. Platen temperature was
monitored by thermocouples built into the platens. Once the EB
pellets were molten, ensuring good contact between the layers of
the stacks, high temperature impedance measurements and
galvanostatic discharge experiments were performed to evaluate the
preliminary ion-transport properties of these systems.
[0059] Table 1 identifies several phases that were targeted for
synthesis and briefly summarizes the results of the synthesis. In
particular, KZr.sub.2P.sub.3O.sub.12 (KSICON) and
Li.sub.6BaLa.sub.2Ta.sub.2O.sub.12 (LBLTO) ceramics were readily
synthesized and exhibited the crystal chemistry preferred for
removal of Cs.sup.+ from LiCl--KCl molten salts. Refinement of the
chemistry and processing of the other phases, however, is possible
and may provide improved ceramic properties. In addition,
commercially-available NaSICON-based ceramics, such as those
produced by Ceramatec, Inc., may be suitable for this
application.
TABLE-US-00001 TABLE 1 Compositions of targeted ion conducting
ceramics and the results of attempted syntheses. Target Compound
Synthetic Outcome* NaZr.sub.2P.sub.3O.sub.12 Single phase NaSICON
Na.sub.3Zr.sub.2Si.sub.2PO.sub.12 NaSICON with minor ZrO.sub.2
contaminant KZr.sub.2P.sub.3O.sub.12 Single phase KSICON
Li.sub.5La.sub.3Ta.sub.2O.sub.12 Single phase LLTO
Li.sub.6BaLa.sub.2Ta.sub.2O.sub.12 LBLTO with minor
Ba.sub.2LaTaO.sub.4 phase K.sub.5La.sub.3Ta.sub.2O.sub.12 Multiple
phases Na.sub.5La.sub.3Ta.sub.2O.sub.12 Multiple phases
K.sub.3Zr.sub.2Si.sub.2PO.sub.12 Multiple phases
K.sub.3Zr.sub.2Si.sub.2PO.sub.12 Multiple phases *"Single phase"
indicates a single phase evident by x-ray diffraction
KSICON
Crystallography
[0060] Solid-state synthesis of both KSICON and NaSICON at
1100.degree. C. produced white ceramic powders (or pellets) that
appeared as single phase by x-ray diffraction, as shown in FIG. 8.
Comparing the diffraction pattern for KSICON with that of the
sodium-version of this crystal structure reveals a significant
difference in lattice parameters between these two materials. There
is a notable expansion of the crystallographic c-axis, coincident
with a slightly less significant contraction of the a-axis (i.e.,
C.sub.NaSICON=22.746 .ANG., C.sub.KSICON=23.890 .ANG.;
a.sub.NaSICON=8.815 .ANG., a.sub.KSICON=8.710 .ANG.). If the
standard sodium-filled NaSICON were used to conduct potassium ions
from the molten salt, these data indicate that the lattice would
have to undergo considerable rearrangement in situ to accommodate
the K.sup.+. Either resistance to this rearrangement would prevent
K.sup.+ transport, or the rearrangement itself would degrade the
integrity of the bulk ceramic separator. Furthermore, it is
reasonable to infer that a similar in situ lattice expansion to
accommodate the still larger Cs.sup.+ ions would provide an even
greater energetic deterrent to Cs.sup.+ transport, thus providing
the desired ion selectivity.
KSICON Stability in Molten Salt
[0061] The use of these ceramics in electrochemically purifying
molten salts requires ceramic stability against molten salts at
elevated temperatures. To verify this stability, KSICON ceramic
pellets (preheated) were immersed in a LiCl--KCl--CsCl eutectic
molten salt at 400.degree. C. for 30 and 60 minute exposures.
Relatively low density pellets (.about.60% theoretical density)
were used to most effectively probe the interactions between the
salt and the ceramic matrix (i.e., by providing higher ceramic
surface area exposure to molten salt). Visually, the pellets were
unchanged during their immersion (with some bubbling as the molten
salt infused the pellet), and they remained physically intact after
cooling and removal of the molten salt caked around the pellet.
X-ray diffraction data confirmed retention of the KSICON phase,
though there was significant evidence of chloride salts present in
the material. In agreement with these data, EDXS data collected
from pellet cross-sections in the SEM qualitatively indicated that
the crystal chemistry remained unchanged, and further revealed
significant molten salt infiltration as suggested by the XRD data.
The ceramic microstructure appeared otherwise unaffected by the
interaction with the molten salt. The effective infiltration of the
molten salt also indicates good interfacial wetting at the
ceramic-salt interface, an important consideration for efficient
ion transport in the purification cell.
Passive KSICON Ion Selectivity
[0062] The EDXS data also provides insight into the selectivity of
these ion conductors against Cs.sup.+ ions. FIGS. 9A-9C show
elemental mappings of a local region of salt-infiltrated KSICON.
The composite map in FIG. 9A overlays the spectral map for Cs in
the molten salt and Zr in the KSICON while FIGS. 9B and 9C show
these spectra separately. The maps show KSICON domains surrounded
by domains of salt, but the interfaces between these different
phases remain sharp; there is no indication of Cs diffusion into
the KSICON domains. These data represent evidence that the KSICON
ceramic is stable in the LiCl--KCl--CsCl molten salt at elevated
temperatures, and that the ion conductor is selective against
significant Cs transport.
Room Temperature Impedance Spectroscopy of KSICON
[0063] KSICON pellets were coated with gold electrodes to determine
room temperature ionic conductivity using electrochemical impedance
spectroscopy (EIS). FIG. 10 is a plot of the imaginary versus real
impedance for KSICON collected at room temperature. The high
frequency (1 MHz) real portion of impedance was attributed to
potassium-ion conductivity within the grains of the material. See
R. W. Powers and S. P. Mitoff, J. Electrochem. Soc. 122, 226
(1975). This value was used to calculate an approximate grain
conductivity of 2.times.10.sup.-4 S/cm (three samples tested), a
value comparable to lattice sodium conductivity values in
comparable NaSICON materials. See O. Bohnke et al., Sol. St. Ionics
122, 127 (1999). Ultimately, however, these room temperature
blocking electrode measurements provide only limited insight into
the ion transport properties desired for electrochemical waste
separation. They account for only potassium conductivity, and they
cannot account for changes in ionic mobility of the system at
higher temperature. The high temperature EIS and galvanic discharge
experiments described below provide a better picture of the ionic
conductivity.
LBLTO
Crystallography
[0064] Solid-state synthesis of the LBLTO garnet phase at
1050.degree. C. produced white ceramic pellets with geometric
densities of .about.84% of theoretical density. X-ray diffraction
revealed the expected LBLTO garnet phase with a minor
Ba.sub.2LaTaO.sub.4 second phase, likely produced because of
lithium deficiency (either through volatility during sintering or
batching deficiency of hygroscopic precursors). Thermogravimetric
analysis of the starting powders was used to identify the actual
starting masses of the constituent powders to correct this problem
in subsequent batches. Comparison of the lattice parameters of the
Ba-containing and non-Ba-containing compositions revealed that the
inclusion of Ba.sup.2+ did result in an increased lattice
parameter.
[0065] FIG. 11A shows the x-ray diffraction pattern for LBLTO,
compared with the calculated pattern for the non-barium containing
Li.sub.5La.sub.3Ta.sub.2O.sub.12 phase. These data show that the
partial substitution of the larger Ba.sup.2+ for La.sup.2+ did
increase the lattice parameter from 12.829 .ANG. to 12.936 .ANG..
FIG. 11B shows the predicted variation of lattice parameter as a
function of Ba.sup.2+ substitution, with the partially
Ba-substituted compound
(Li.sub.5.6Ba.sub.0.6La.sub.2.4Ta.sub.2O.sub.12) indicated by the
green data point. The evident expansion of the crystal lattice not
only stands to increase the conductivity of Li.sup.+ in these
materials, but may accommodate the transport of K.sup.+ ions as
well. Importantly, FIG. 11B shows that the lattice can be tuned as
a function of Ba.sup.2+ content, allowing for refinement and
optimization of ion selectivity.
LBLTO Behavior in Molten Salt
[0066] The stability of LBLTO ceramic pellets was evaluated by
immersing preheated pellets in LiCl--KCl eutectic molten salt at
500.degree. C. for 30 minutes, LiCl--KCl--CsCl eutectic molten salt
at 400.degree. C. for 60 minutes, and molten LiCl at 650.degree. C.
for 4 hours. Visually, the pellets were unchanged both during their
immersion and after removal from each of these salt treatments, as
shown in FIG. 12A.
[0067] Comparison of x-ray diffraction patterns collected from the
pellets before and after exposure to the eutectic salt compositions
revealed no measureable change in the lattice parameters or crystal
phase, as shown in FIG. 12B. In addition, EDXS data collected from
pellet cross-sections in the SEM qualitatively indicate that the
crystal chemistry remained unchanged. These tests indicate
stability of the LBLTO ceramic against LiCl--KCl molten salts at
500.degree. C. or LiCl--KCl--CsCl eutectic molten salt at
400.degree. C.
[0068] When cross-sections of samples exposed to LiCl at
650.degree. C. for 4 hours were examined in the SEM, EDXS data
indicated the formation of minor BaCl.sub.2 inclusions within the
structure. As seen in the elemental maps in FIG. 13, LBLTO examined
before high temperature exposure to LiCl showed uniform
distribution of La, Ta, and Ba. After exposure, however, La and Ta
remained uniformly distributed, but there was evidence of Ba
segregation (likely as BaCl.sub.2), indicated by the emergence of
bright yellow spots in the lower right map of FIG. 13. These data
suggest that at this elevated temperature, a fraction of Ba.sup.2+
cations become sufficiently mobile to escape the LLTO lattice,
forming secondary inclusions in the material. The remainder of the
ceramic composition appeared unchanged.
[0069] For purification of LiCl--KCl molten salts, the addition of
Ba.sup.2+ to the LLTO lattice is intended to keep the crystal
structure sufficiently expanded to allow for K.sup.+ transport as
well as Li.sup.+ transport. At the lower temperatures (e.g.,
500.degree. C.), required for this process, the Ba.sup.2+ appears
to be much more stable, based on XRD data (above) and EDXS analysis
(not shown). For purification of LiCl, however, the mobility of
Ba.sup.2+ at elevated temperatures (650.degree. C.) suggests that
unmodified LLTO (no Ba.sup.2+) may be a more appropriate ion
conductor. Since there is no need to transport K.sup.+ in such a
system, the absence of the lattice expanding Ba.sup.2+is not
expected to pose a problem.
Passive Ion Selectivity in LBLTO
[0070] LBLTO samples were also immersed in LiCl--KCl--CsCl eutectic
molten salt for 60 minutes at 400.degree. C. to probe both
stability and passive ion selectivity (against Cs.sup.+). FIG. 14A
shows an electron micrograph and FIG. 14B shows an EDXS spectrum
from a cross-section of a pellet after salt exposure. The
microstructure appears intact and the EDXS data show,
qualitatively, that the LBLTO chemical composition was preserved.
Importantly, there is no evidence of Cs.sup.+ in the EDXS spectrum,
suggesting that not only is the material stable against the molten
salt, but it is selective against Cs.sup.+ transport.
Room Temperature Impedance of LBLTO
[0071] As with the KSICON pellets above, LBLTO pellets were coated
with gold electrodes for room temperature EIS. FIG. 15 plots the
imaginary versus real impedance for LBLTO collected at room
temperature. The higher density of these ceramic pellets allowed
for a more complete impedance measurement. After fitting the data
to a circuit model (Z-plot), the room temperature Li-ion
conductivity in these materials was determined to be
.about.3.times.10.sup.-5 S/cm, a value consistent with literature
reports of similar materials. See V. Thangadurai and W. Weppner,
Adv. Funct. Mater. 15, 107 (2005).
High Temperature Electrochemical Behavior
[0072] High temperature electrochemical testing of the KSICON and
LBLTO ceramics in contact with molten salts provided information
about the stability, ionic conductivity, and ionic selectivity of
these ceramic materials under more relevant conditions. Both KSICON
and LBLTO ceramics were tested under identical conditions, and for
the sake of simplicity, the results from both materials types are
described together below.
[0073] Ideally, these tests would have been conducted according to
the scheme shown in FIG. 2, in which charge balance is maintained
through the oxidation and reduction of chlorine. However, an
alternative scheme was devised to more safely and immediately
assess the selective ion-transport properties of the KSICON and
LBLTO ceramics at 500.degree. C. This alternative scheme utilized
stacks of molten salt pellets, separated by ceramic ion conductors,
all sandwiched by a copper anode and a stainless steel cathode (a
stainless steel current collector was also placed in contact with
the copper anode). In this system, shown in FIGS. 16A and 16B,
charge balance is maintained not through the oxidation and
reduction of chlorine, but through the oxidation of copper metal at
the anode and the reduction of either copper or lithium at the
cathode.
[0074] Key to this process was the development of molten salt
pellets (as shown in FIG. 6B), which are similar to those used in
thermal battery systems. These pellets were pressed from mixtures
of target salts, as shown in Table 2, and an electrochemically
inert, inorganic electrolyte binder that allowed the pellets to
maintain their shape and integrity within the assembled testing
stacks at 500.degree. C., even after melting of the molten
salt.
TABLE-US-00002 TABLE 2 Compositions of electrolyte binder pellets
used for electroreduction studies. All values are weight percents.
LiCl KCl CsCl CuCl.sub.2 LK-EB 44 56 0 0 CsLK-EB 40.1 48.7 11.2 0
CuLK-EB 32.7 41.6 0 25.7
Electrochemical Impedance
[0075] Impedance measurements were conducted prior to any galvanic
discharge studies, which tended to produce irreversible changes to
the stacks. FIG. 17 plots the imaginary versus real impedance for
KSICON measured at 500.degree. C. between two LK-EB pellets.
Complete interpretation of these data is complicated by numerous
interfaces, multiple transport species, and several
ion-transporting phases in these systems. However, by fitting the
data to the equivalent circuit, shown as an inset in FIG. 17,
analysis suggests a total ionic conductivity through the ceramic
membrane on the order of .about.3.times.10.sup.-3 S/cm, an increase
of several orders of magnitude relative to the room temperature
results described earlier. Similar analysis of the LBLTO samples
indicated conductivities on the order of .about.2.times.10.sup.-2
S/cm, again significantly higher than the room temperature values
above.
Galvanostatic Electroreduction
[0076] The practical value of the conductivity in these materials,
however, is better illustrated by the galvanostatic
electroreduction experiments. Discharge curves generated during the
cathodic electroreduction of both Cu.sup.2+ and Li.sup.+ using both
LBLTO and KSICON ion-conductors are shown in FIG. 18. Discharges
were performed at a relatively high constant current density of 100
mA/cm.sup.2. Electroreduction of copper was tested according to the
scheme shown in FIG. 16A, in which the ionic conductors were
sandwiched between CsLK-EB and CuLK-EB pellets with a copper anode
and a stainless steel cathode. Current flowing through the device
during discharge was expected to oxidize the copper at the anode
and reduce Cu.sup.2+ at the cathode, theoretically a 0-volt net
reaction. The discharge potential for the electroreduction of
copper was a consistent 0.3V, and was similar for both LBLTO and
KSICON. This small overpotential (relative to 0V) likely represents
a small activation energy required to drive ionic transport across
interfaces or through the ion-conducting ceramics.
[0077] Similar discharge experiments were performed in which LBLTO
or KSICON ceramics were sandwiched between CsLK-EB and LK-EB
pellets, again using a copper anode and a stainless steel cathode,
as shown in FIG. 16B. In this case, the anodic half-reaction
involved dissolving the Cu metal into the CsLK-EB pellet, and the
cathodic reaction involved reduction of Li.sup.+ in the LK-EB
pellet to Li metal. The discharge curves seen in FIG. 18 show an
increase in the discharge potential to slightly greater than 6V.
This value is considerably higher voltage than the .about.3.4V
expected for the sum of the half reactions for lithium reduction
and copper oxidation, indicating a significant overpotential,
likely associated with the formation of lithium metal at the
LK-EB/Stainless steel interface. There is also a notable
intermediate voltage plateau in the LBLTO case. This plateau may be
due to secondary reactions in the LBLTO pellets. This observation
suggests that in the development of a larger-scale system, the
reduced voltages of an electrochemically symmetrical system (such
as the reduction/oxidation of chlorine) may offer benefits to the
stability of the ceramics.
[0078] For both of these electroreduction demonstrations, and with
both ceramic conductors, the electroreductions resulted in
macroscopic changes in the device structure. FIGS. 19A and 19B show
the copper reduction using KSICON, while FIGS. 19C and 19D show
lithium reduction using LBLTO (the same results were observed for
copper plating with LBLTO and lithium plating with KSICON, but only
one example of each is shown for conciseness). At the cathode,
reduced copper or lithium metals plated visibly at the interfaces
between the stainless steel cathodes and the CuLK-EB or LK-EB
pellets, respectively, as shown in FIGS. 19A and 19C. When these
cells were disassembled after testing, it was evident that not only
were these reduced metals visible at the cathode, but the copper
anodes, originally disks, had been etched into rings with the
electrochemical dissolution of the anode into CsLK-EB pellets, as
shown in FIGS. 19B and 19D. Clearly, the anodic oxidation of Cu and
the cathodic reduction of Cu.sup.2+ or Li.sup.+, supported by ionic
transport through the ceramic membranes, were remarkable efficient.
With both the KSICON and LBLTO tests, the ceramic separators
remained stable and intact. These macroscopic changes to the device
structure during the discharge provide compelling evidence of the
highly effective ion transport properties of these ceramics at
500.degree. C.
Ionic Selectivity During Electroreduction
[0079] These ion conductivity experiments also inform the
selectivity of the ion transport. Energy dispersive x-ray
spectroscopy (EDXS) was used to examine the chemical composition of
molten salt pellet cross sections after testing. FIGS. 20 and 21
show a series of electron micrographs of gold-coated (for SEM
imaging) EB pellet cross-sections and the corresponding EDXS
spectra collected from those cross-sections.
[0080] The spectrum in FIG. 20A was collected from the CsLK-EB
pellet after discharge through a KSICON separator and shows
evidence of the KCl salt (Li is not visible in EDXS), the CsCl
salt, the inorganic binder, and copper, confirming the dissolution
of anodically oxidized copper into the pellet during discharge. The
data in FIG. 20B were collected from the CuLK-EB pellet in the same
experiment. This spectrum shows expected peaks from KCl, residual
CuCl.sub.2, the inorganic binder, and a minor Fe contaminant
believed to come from the stainless steel current collector.
Critically, though, there is no evidence of Cs present, indicating
that the Cs did NOT cross the ceramic separator in any measureable
quantities. This important evidence of ionic selectivity during the
electrochemical discharge was consistent when both KSICON and LBLTO
membranes were used. Moreover, this high temperature selectivity
under DC bias is in agreement with the passive selectivity
described in FIGS. 9 and 14 for KSICON and LBLTO, respectively.
[0081] FIGS. 21A and 21B present similar data collected from EB
pellets (platinum coated for SEM imaging in this case) used in the
electroreduction of lithium using a KSICON separator. Once again,
FIG. 21A confirms the electrochemical dissolution of copper into
the CsLK-EB pellet, while the spectrum in FIG. 21B again shows the
absence of any Cs in the LK-EB pellet.
[0082] Interestingly, CuCl.sub.2 was not originally present in the
either the anodic CsLK-EB pellet or the cathodic LK-EB pellet,
which means that this system can be used to track the transport of
Cu.sup.2+ through the cell. Copper was clearly electrochemically
dissolved at significant levels in the CsLK-EB pellet, but the
persistent absence of copper in the anodic LK-EB pellet suggests
that Cu.sup.2+ did not cross the ceramic separator. This suggests
that these ceramics are also selective divalent ion transport,
which may be useful in more general molten salt purification
experiments. As with the copper electroreduction, the observations
for the Li reduction were the same for both KSICON and LBLTO
conductors.
[0083] Subsequent analysis of the molten salt pellets using
quantitative ion concentration analysis has confirmed the results
provided in the EDXS experiments. Molten salt pellets from both the
anodes and cathodes of the galvanostatically-discharged test cells
described above were dissolved in deionized water. The samples were
centrifuged to remove any insoluble materials and the supernatant
was analyzed for ion content using inductively coupled plasma.
Samples of these data are provided in Table 3 below:
TABLE-US-00003 TABLE 3 Concentrations of target ions in molten salt
pellets after galvanostatic discharge through LBLTO and KSICON.
CuLK-EB pellets were on the cathodic side of the cell, while
CsLK-EB pellets were on the anodic side. All values are in parts
per million, normalized by sample size and molecular weight. Salt
lithium potassium cesium LBLTO: CuLK-EB 107 69.1 0.008 LBLTO:
CsLK-EB 90.8 66.5 4.97 KSICON: CuLK-EB 63.4 51.15 0.023 KSICON:
CsLK-EB 138 97.2 6.47
[0084] The stark differences in Cs content between the CuLK-EB
pellets and the Cs-LK-EB pellets provides compelling evidence for
the selectivity against Cs.sup.+ transport in both the LBLTO and
KSICON ceramics. Moreover, analysis of the ratios of Li:K in each
of the pellets, when compared with the initial ratios of these
elements (Table 2), shows that both ceramics were capable of
K.sup.+ and Li.sup.+ transport, though not surprisingly, Li.sup.+
transport was slightly favored in the LBLTO case.
Process to Chemically Reduce Trace CuCl.sub.2
[0085] The above description provides a process in which the
electrochemical reactions involved in moving Li.sup.+ and K.sup.+
ions across ion-selective membranes to purify LiCl--KCl molten
salts are balanced by the reduction of CuCl.sub.2 and the oxidation
of Cu metal. This process was proved to be very effective in
facilitating Li.sup.+ and K.sup.+ transport, but stands to leave
behind trace CuCl.sub.2 in the purified salt. Therefore, FIG. 22
shows a process to chemically reduce any residual CuCl.sub.2 using
lithium formate or potassium formate. This process not only removes
the residual CuCl.sub.2, but the byproducts of the reaction are
LiCl, KCl, CO.sub.2 and H.sub.2. In the example described below, a
dark brown molten salt of CuCl.sub.2--LiCl--KCl was transformed at
400.degree. C. into a mixture of white LiCl and Cu metal (although
only Li-formate was used for this example, K-formate will also
work).
Concentric Reactor for Molten Salt Purification
[0086] FIG. 23A shows a schematic cross-sectional illustration of a
"concentric reactor" 20 for efficient bulk salt purification.
Contaminated waste 22 can be placed in a stainless steel or boron
nitride-coated steel vessel, heated to 500.degree. C. (or
700.degree. C. for LiCl purification). "Purified" KCL/LiCl salts
23, doped with CuCl.sub.2 can be placed in a closed-end tube 21, or
ceramic cup, formed from an ion-selective ceramic (e.g. LLTO or
KSICON). This tube 21 can be lowered into the contaminated salts 22
along with a high surface area copper anode 24. A suitable (e.g.,
Pt, C, W, Cu, or steel) cathode 25 can be immersed into the
"purified" salt 23 inside the closed-end ceramic tube 21. Under
electrochemical bias, K.sup.+ and Li.sup.+ ions are transported
across the ion-selective ceramic membrane into the ceramic tube 21,
while Cs.sup.+ contaminants are blocked from transport and left
behind, concentrated in the contaminated salt 22. As seen in
previous "pellet-based" studies, charge balance in this concentric
reactor system can be maintained by oxidation of the copper anode
and reduction of Cu.sup.2+ on the cathode. Over time, the ceramic
tube fills up with purified salt, while the contaminated salt is
reduced in volume and concentrated, as shown in FIG. 23B. The
alkali formate treatment described previously can be used to remove
any residual Cu.sup.2+ left behind in the purified salt or to
recover Cu.sup.2+ dissolved in the residual waste (ultimately, the
Cu can be a recycled component of this process). This process
obviates the need for high temperature stable sealants (e.g.,
sealing ceramic separators against steel parts), allows for higher
ceramic surface area (higher transport efficiency), and can be
adapted to existing pyroprocessing waste containment systems.
VCR.RTM.-Type Reactor for Molten Salt Purification
[0087] An alternative configuration can be used for volumetric
purification processes. In this alternative configuration, a
suitable plumbing fixture can be used to seal the ion-conducting
ceramic, separating contaminated and purified salts. As an example
of this configuration, Swagelok.RTM. VCR (variable compression
ratio) male and female fittings 36M and 36F with a copper gasket 37
were used to seal a planar LLTO pellet 31 in a VCR assembly, as
shown in FIG. 24A. This VCR assembly can be used in an
electrochemical cell 30, referred to herein as a "VCR-type"
reactor, that is conceptually similar to the concentric reactor
scheme in FIG. 23A. As shown in FIG. 24B, once a flat, dense,
uniformly thick ion-conducting ceramic pellet 31 has been sealed in
the VCR assembly, the assembly and a copper anode 34 can be placed
in an alumina crucible prefilled with contaminated anodic salt 32.
Purified cathodic salt 33 can be placed inside the top port of the
cathodic volume of the VCR assembly and a copper wire cathode 35
can be placed inside this volume. Key to this process is the
electrical isolation of both the cathode 35 and the anode 34 from
the stainless steel VCR assembly. It is preferable that the only
pathway for charge transport in this system runs through the ionic
ceramic. Therefore, an insulating alumina sheath 38 can be placed
inside the cathodic "chamber" to prevent shorting of the cathode 35
to the VCR fitting 36M, and insulating alumina beads 39 can be
threaded on the copper anode 34. These beads 39 prevent electrical
shorting, but still allow facile contact between the copper anode
34 and the anodic salt 32. This entire reactor 30 can be placed in
an insulated furnace within an Ar-atmosphere glove box, the
electrical leads (not shown) can be connected to anode 34 and
cathode 35, and the reactor 30 can be heated until the salt reaches
500.degree. C. At this point both anodic and cathodic salts 32 and
33 are molten and electrochemical ion separation can occur in the
cell 30.
[0088] FIGS. 25A and 25B show both an individual VCR-type reactor
and a VCR-reactor integrated into a functional purification
assembly following an electrochemical experiment and cooling to
room temperature.
[0089] Electrochemical tests using this volumetric configuration
revealed selective ion-transport behavior consistent with those
observed using the pellet stack configuration described above. FIG.
26 shows the potentiostatic discharge profile for a representative
test run on an LLTO pellet using copper-mediated charge balance.
These tests were typically run using 30-50 g of Cs-contaminated
anodic salt and .about.5-7 g of cathodic salt. The plot shows the
current produced over time when a constant voltage of -200 mV was
applied across the two cell electrodes. Although previous studies
on the "pellet stack" configuration had included galvanostatic
discharge experiments, the volumetric studies were conducted by
applying a specific, constant voltage to drive ion transport.
Ultimately, it would be expected that by monitoring the current in
these potentiostatic experiments it may be possible to gauge the
degree of impurity concentration in situ, based on the amount of
ionic charge (Li.sup.+ and K.sup.+) passed through the ceramic
separator (correlated with the measured current measured in the
test cell). The data in FIG. 26 show a reasonably smooth current
response to the applied voltage over the course of 10 minutes of
discharge (typical tests included multiple 10-20 minute discharges
with similar responses).
[0090] Control experiments run using stainless steel coupons in
place of the ion-conducting ceramics showed a current response
(believed to be due largely to capacitive charging of cell
interfaces) more than an order of magnitude smaller than that shown
here, indicating that the current measured is the result of
ion-transport through the ceramic separator. These data demonstrate
volumetric electrochemical reactions using the VCR-reactor
configuration.
[0091] In addition to performing volumetric electrochemical
reactions, an objective of these reactions is to segregate and
concentrate Cs.sup.+ in the anodic salt. To determine whether or
not this concentration took place during the electrochemical
experiments described above, molten salt samples were taken from
the cathode and anode after tests performed with LLTO pellets.
These salt samples were dissolved and the elemental compositions
determined. FIGS. 27A and 27B show the elemental ratios (based on
atomic concentration) of Cs:K and Cs:Li for "as-made salts" as well
as salts harvested from LLTO-mediated separations. Clearly, the
ratios of Cs:K and Cs:Li increased substantially at the anode with
no significant Cs measured at the cathode. The increases in the
ratios of Cs:K and Cs:Li at the anode demonstrate that both
Li.sup.+ and K.sup.+ were transported from the anode, while the
Cs.sup.+ was not, effectively concentrating the Cs.sup.+ in the
anodic molten salt. The notable absence of any significant Cs at
the cathode is further indication of the excellent selectivity of
the LLTO against Cs.sup.+ transport.
[0092] Although this configuration does not provide as much surface
area for ionic separation as the tube-based concentric reactor
design described above, and it does require sealing of the ceramic
parts within the reactor, this system does offer a number of
advantages over the concentric design. First, it can be readily
used to study a variety of different ceramic systems without having
to developed closed-end ceramic tubes from each material--a
potentially significant technical obstacle. In addition, the system
is scalable. By increasing the size of the vessel containing the
salt, multiple reactors, potentially containing a variety of
different ceramic separators can be simultaneously used in
parallel. Although this can also be achieved with closed-end tubes,
again the relative ease of incorporating different materials into
the same reactor scheme is an advantage of the VCR-type reactor
system.
[0093] Finally, the VCR fitting apparatus is amenable to
integration into a closed system suitable for gas-mediated
electrochemical purification. FIG. 28 shows a schematic
illustration how the VCR-type assembly can be integrated into such
a closed-loop electrochemical cell 40 using stainless steel tubing
and gas-tight Swagelok.RTM. unions. The cell comprises an
ion-selective ceramic membrane 41 in a VCR-type assembly, an anode
44 immersed in the contaminated molten salt 42 on anode side of the
membrane 41, and a cathode 45 immersed in the purified molten salt
43 on the cathode side of the membrane 41. In this closed-loop
reactor, contaminated salt can be continuously feed into the anode
side and purified salt can be removed from the cathode side through
an extraction port 46. The modular, connectable nature of this
design makes incorporation of ports, valves, vents, and other
reactor access points relatively straightforward. Moreover, because
the size of the system can be varied by changing the tubing
diameter, or even more easily, the tubing length, the scale of the
system can also be readily varied.
[0094] The present invention has been described as a method for
electrochemical ion separation. It will be understood that the
above description is merely illustrative of the applications of the
principles of the present invention, the scope of which is to be
determined by the claims viewed in light of the specification.
Other variants and modifications of the invention will be apparent
to those of skill in the art.
* * * * *