Journal of Nuclear Materials 444 (2014) 261–269

Contents lists available at ScienceDirect

Journal of Nuclear Materials

journal homepage: www.elsevier .com/ locate / jnucmat

Electrochemical reduction of UO2 in LiCl–Li2O molten salt using porousand nonporous anode shrouds

0022-3115/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.jnucmat.2013.09.061

⇑ Corresponding author. Tel.: +82 42 868 8968; fax: +82 42 868 8317.E-mail address: eychoi@kaeri.re.kr (E.-Y. Choi).

Eun-Young Choi ⇑, Chan Yeon Won, Ju-Sun Cha, Wooshin Park, Hun Suk Im, Sun-Seok Hong, Jin-Mok HurKorea Atomic Energy Research Institute, Daedoek-daero 989-111, Yuseong-gu, Daejeon 305-353, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 July 2013Accepted 30 September 2013Available online 8 October 2013

Electrochemical reductions of uranium oxide in a molten LiCl–Li2O electrolyte were carried out using por-ous and nonporous anode shrouds. The study focused on the effect of the type of anode shroud on thecurrent density by running experiments with six anode shrouds. Dense ceramics, MgO, and MgO(3 wt%) stabilized ZrO2 (ZrO2–MgO) were used as nonporous shrouds. STS 20, 100, and 300 meshesand ZrO2–MgO coated STS 40 mesh were used as porous shrouds. The current densities(0.34–0.40 A cm�2) of the electrolysis runs using the nonporous anode shrouds were much lower thanthose (0.76–0.79 A cm�2) of the runs using the porous shrouds. The ZrO2–MgO shroud (600–700 MPaat 25 �C) showed better bending strength than that of MgO (170 MPa at 25 �C). The high current densitiesachieved in the electrolysis runs using the porous anode shrouds were attributed to the transport ofO2� ions through the pores in meshes of the shroud wall. ZrO2–MgO coating on STS mesh was chemicallyunstable in a molten LiCl–Li2O electrolyte containing Li metal. The electrochemical reduction runs usingSTS 20, 100, and 300 meshes showed similar current densities in spite of their different opening sizes. TheSTS mesh shrouds which were immersed in a LiCl–Li2O electrolyte were stable without any damage orcorrosion.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Despite the wide use of nuclear energy, the accumulating spentfuel from current nuclear power plants, which fuel is mainly com-posed of uranium oxides, remains a formidable challenge. One ofthe most practical solutions is to reduce the spent oxide fuel andrecycle it as metal fuel for fast neutron reactors. The nature ofmetal fuel enables us to employ pyrometallurgical reprocessing(pyroprocessing), which has several benefits such as its inherentproliferation resistance, the compactness of the process equip-ment, and the relatively low cost. Pyroprocessing involves thereduction of spent oxide fuel to a metal through an electrochemicalreduction process and the recovery of the fuel components bymeans of an electro-refining process [1–13].

In the electrochemical reduction process, oxides fuels areloaded at the cathode in molten LiCl–Li2O at 650 �C. The cathodereactions are as follows [14]:

Liþ þ e� ! Li ð1Þ

MO2 þ 4Li!Mþ 2Li2O ðsalt phaseÞ ð2Þ

MO2 þ 4e� !M ðactinideÞ þ 2O2� ðsalt phaseÞ ð3Þ

Li2O produced by reaction (2) in molten LiCl is dissociated into Li+

and O2�:

Li2O! Liþ þ O2� ðsalt phaseÞ ð4Þ

When a platinum (Pt) anode is employed, oxygen ions (O2�) are oxi-dized to give oxygen (O2) gas on the anode surface, as follows:

2O2�ðsalt phaseÞ ! O2 ðgasÞ þ 4e� ð5Þ

When an electrical potential is applied, the actinide metal oxide isreduced to metal and remains at the cathode. The O2� ions pro-duced at the cathode are transported through the salt and dischargeat the anode to form O2 gas [15–21].

The anode is generally surrounded by a nonporous ceramicshroud such as one of MgO with an open bottom to offer a pathfor O2 gas produced on the anode surface [17,22–24]. However,the O2� ions are transported only in a limited fashion throughthe open bottom of the anode shroud because the nonporous cera-mic shroud hinders the transport of the oxygen ions to the anodesurface, which leads to a decrease in the current density and anincrease in the operation time of the process. In the present study,electrochemical reductions of UO2 were carried out in a moltenLiCl–Li2O salt using various porous and nonporous shrouds toinvestigate how effectively the porous structures of the anodeshrouds can contribute to the increase in the current density. InFig. 1, the transport of O2� ions between the cathode and the anodewith a nonporous shroud is schematically compared with that with

Fig. 1. Schematic diagram of comparison between the nonporous and porous anode shrouds for the electrochemical reduction process.

262 E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269

a porous shroud. We used MgO and MgO (3 wt%) stabilized ZrO2

(ZrO2–MgO) as the nonporous anode shrouds and ZrO2–MgOcoated stainless steel wire mesh (STS mesh) and uncoated STSmeshes with three different opening sizes (20, 100, 300 meshes)as the porous shrouds. Photographs of these anode shrouds beforeand after the electrochemical reduction are shown in Table 1.

2. Experimental

All of the electrochemical reduction experiments were per-formed in a high-purity argon atmosphere glove box, as LiCl (99%purity, Alfa Aesar) is very hygroscopic and the metallic uranium,reduced from the uranium oxide, is easily reoxidized, even witha low concentration of oxygen.

Fig. 2 shows a schematic of the cell used for the electrochemicalreduction process in the present study. The electrolysis cell con-sists of a crucible, a salt, and electrodes. The salt makes contactwith the electrodes, which are placed on a top flange containinga heat shield and ports. Fig. 3(a) and (b) show the top, the side,and the bottom of the flange before use. To minimize the temper-ature gradient and the salt vaporization, the flange is composed often heat shields; there are also four ports in which to place theelectrodes and the sampling rod. The ports, r, s, t, and u, indi-cated in Fig. 3(b), are for the anode, cathode, reference electrode,and sampling rod, respectively. First, 700 g of LiCl was put in anSTS crucible (diameter of 95 mm and height of 160 mm) at roomtemperature. The reactor was heated to 650 �C for 4 h and main-tained at that temperature. Then, 7.0 g of Li2O (99.5% purity, AlfaAesar) was fed into the reactor to reach the desired concentration.After the complete dissolution of Li2O was confirmed via titration,the electrodes were lowered into the molten salt. A liquid Li–Pballoy (32 mol% Li, 13 in Fig. 2) was used as a reference electrode; thiswas similar to the reference electrode used by Sakamura et al. [16].This reference electrode was prepared by placing 1 g of the Li–Pb al-loy in an MgO tube (14 in Fig. 2) with an inner diameter of 5 mm anda porous bottom. A tantalum wire (with a diameter of 0.5 mm, 15 inFig. 2) was then put into the tube and immersed in the liquid alloy foruse as an electrical lead. A cylindrical cathode basket (diameter of20 mm, 10 in Fig. 2) was surrounded with an STS wire mesh sheet(STS 316L) to contain the UO2 particles (9 in Fig. 2). The STS wiremesh sheet for the cathode was fabricated from a three-ply layerof 20-mesh STS wire cloth (outer surface), 100-mesh STS wire cloth(middle layer), and 325-mesh STS wire cloth (inner surface). Thewall of the cathode basket (salt immersion = 20 mm in height) wasinsulated by applying the electrical potential only to the center rod(salt immersion = 10 mm in height) of the cathode (11 in Fig. 2)and by surrounding it with UO2 particles. A photograph of the uti-lized UO2 particles is shown in Fig. 4. The size of the UO2 particles,

with a density of 10.59 g/cm3, ranged from 1 mm to 4 mm. In addi-tion, a 10 mm wide plate-type Pt anode (2 in Fig. 2) with cup-shapedPt (3 in Fig. 2) as its top was surrounded with a shroud (1 in Fig. 2) toprovide a path for the O2 gas produced on the anode surface. The uti-lized anode shrouds (salt immersion = 55 mm in height) are listed inTable 1. Also, the specifications of the utilized meshes for the porousanode shrouds are listed in Table 2. The Pt plate anode (salt immer-sion = 45 mm in height) was connected to the STS rod (8 in Fig. 2). Acylindrical alumina tube (5 in Fig. 2) was placed on the cup-shaped Ptto protect the STS rod from corrosion by the O2 gas. During the elec-trolysis, O2 gas was continuously exhausted using a pump from theupper end of the shroud to the O2 gas outlet (7 in Fig. 2). An AgilentE3633A power supply was used for voltage control electrolysis in or-der to reduce the UO2. The cathode potential during the electro-chemical reduction was monitored with a digital multimeter(Agilent, 34405A). After the electrolysis run, the variation of theLi2O concentration in the molten salt was checked using an autoti-trator (G20, Mettler Toledo). The conversion rate of metallic U fromUO2 was determined by means of a thermogravimetric analyzer (TG/DSC 1, Mettler Toledo). The surface morphology of the products wasobserved with a scanning electron microscope (SEM, Hitachi, SU-8010). The elements of the ZrO2–MgO coating on the STS mesh afteran immersion test in the molten salt were identified with an energydispersive X-ray (EDX) spectrometer (Horiba, EX-250 X-max).

3. Results and discussion

The electrochemical reduction runs conducted for this studyand their results are listed in Table 3. They were conducted with20 g of UO2 in a molten LiCl–Li2O (1 wt%) salt at 650 �C. The elec-trolysis runs were conducted at a constant voltage and were inter-rupted at appropriate intervals. The lithium deposition on thecathode surface could be indicated by the cathode potential atthe open circuit potential [16,17]. The cathode potential was low-ered to a potential more negative than �0.6 V by applying a cellvoltage of (3.0 or 3.3 V); this was done because a cyclic voltamme-try test in our previous study [24] showed that UO2 is reduced atpotentials more negative than �0.6 V. The Li/Li+ potential mea-sured during the interruption of the electrolysis run was �0.58 Vagainst the Li–Pb reference electrode. Each run was finally finishedafter 150% (10.7 A h) of the theoretical electric charge had passed.One set of the electrolysis results (Run 7) is presented in Fig. 5.These results show the electrolysis curves of (a) the cell voltage,(b) the current, and (c) the cathode potential as a function of time.The response current was more than 8 A during Run 7 (Fig. 5b); thecalculated average of the current density during Run 7 was0.76 A cm�2 for 11 cm2 of the anode surface area. Electrochemicalreductions using the other anode shrouds were also conducted in

Table 1List of six types of the anode shrouds used for the present study.

Nonporous shroud Porous shroud

Run 2 Run 3 Run 4 Run 5 Run 6 Run 7

Dense MgO MgO (3 wt%) –ZrO2 MgO (3 wt%) –ZrO2

coated STS 40 meshSTS 20 mesh STS 100 mesh STS 300 mesh

Before electrochemical reduction

After electrochemical reduction

Fig. 2. Schematic diagram of the electrochemical cell containing the salt, thecathode, the anode and the reference electrode.

E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269 263

a manner similar to that used for Run 7. The obtained metallicproducts after Run 1–Run 9 were shown in Table 3. Theirconversion rates of metallic U from UO2 which were determinedby means of thermogravimetric analyzer were higher than 95%.

In Run 1, we attempted the electrochemical reduction of UO2

without the use of any anode shroud. After the electrolysis, thecathode basket was cut along the radial direction in a high-purityargon glove box. A photograph of a cross-section of the product(Table 3) shows that U was metallic grey after a charge supplyof 150%, unlike the brown UO2. The change in the microstructureof UO2 before and after the electrochemical reduction was con-firmed by means of SEM. While pores in the fresh UO2 can scarcelybe observed in Fig. 6a, the structure became dramatically porousafter the electrochemical reduction (Fig. 6b). The generation ofpores in UO2, caused by reactions (2), (3), and (5) is accompaniedby cohesion of the U metal, which results in volume shrinkage anda heterogeneous structure [16,21]. The averaged current density ofRun 1 was the highest among the performed electrolysis runs.However, it was revealed that the bottom heat shield of the flangewas severely corroded because O2 gas was not effectively collectedand exhausted to the gas outlet (Fig. 3d). Unlike the case of thebottom of the flange used for Run 7, which had a porous anodeshroud (Fig. 3c), brown corrosion products around the round portsin Fig. 3d for Run 1 were observed. Moreover, the pipe for the an-ode (6 in Fig. 2) was stuck in the flange port due to the presence ofcorrosion products, as Fig. 3d shows. These results indicate thatthe anode shroud plays an effective role as a path for outgassingO2 and that the anode shroud is necessary to protect the materialof the electrochemical cell. After Run 1, the flange was cleaned forfurther experiments.

In Run 2, the electrochemical reduction of UO2 was performedusing an MgO anode shroud, which shrouds have typically been

used in molten LiCl–Li2O electrolytes. As listed Table 3, the currentdensity was low (0.34 A cm�2) compared to those (>0.76 A cm�2)of the electrolysis runs using porous anode shrouds; this low cur-

Fig. 3. Photographs of the flanges with a set of heat shields and ports for the electrodes: (a) the side and top before use, (b) the bottom heat shield before use (the ports forr-the anode and its shroud, s-the cathode, u-reference electrode, t-sampling), (c) the bottom heat shield after Run 7, and (d) the bottom heat shield after Run 1.

Fig. 4. Photograph of the fresh UO2 particles.

Table 2Specifications of the porous STS meshes.

Mesh Wire diameter (mm) Opening (mm) Open area (%)

20 0.229 1.04 67.240 0.254 0.380 36.0

100 0.114 0.140 30.3300 0.040 0.045 28.0

264 E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269

rent density was the result of the limited transport of O2� ionsthrough only the open bottom of the MgO anode shroud. In gen-eral, Li2O is added to molten LiCl in order to speed up the electro-chemical reaction and reduce the electrolysis time. Thus, thecommon range of O2� concentration in molten LiCl is 0.5–1.0 wt%because a low O2� concentration (<0.5 wt%) leads to a rapid in-crease of the anode potential and dissolution of the Pt material.The O2� concentration in the salt is almost constant during theelectrochemical reduction process thanks to the diffusion of O2�

between the cathode and the anode [16]. However, a slow diffusionor dissociation rate of O2� from metal oxide particles to the anode

surface via the bulk salt may lead to a slight decrease in O2� con-centration. Thus, it is important to observe the O2� concentrationnear the surface of the Pt anode [21,25–28]. After the terminationof the electrolysis runs, the local O2� concentrations inside of theanode shrouds were measured, as follows. First, after the Pt anodeplate (2 and 3 in Fig. 2) and its connected rod (5 and 8 in Fig. 2)were removed from the port of the flange at one time right afterthe termination of the electrolysis run, an STS sampling rod was in-serted into the molten Li2O–LiCl and then quickly taken out. Then,the salt on the sampling rod was collected and its O2� concentra-tion was measured using the titration method. The O2� concentra-tion inside of the anode shrouds after Run 2 was 0.915 wt%, whichis relatively low compared to the O2� concentrations (>0.940 wt%)after the electrolysis runs using the porous shrouds. Also, it shouldbe noted that the O2� concentration in the bulk electrolyte slightlydecreased from 1.09 wt% to 1.06 wt%. This difference of O2� con-centration between inside of the shroud and the bulk (outside ofthe shroud) electrolytes implies that the side wall of the nonporousMgO shroud plays the role of a barrier to the flow of O2� ions fromthe bulk electrolyte to the Pt anode surface.

Table 3Results of the electrochemical reduction of UO2 using various anode shrouds.

Shroud type Shroud RunNo

Photographafter reduction

Initial Li2Oconc. (wt%)

Li2O conc. in thebulk electrolyteafter O R (wt%)

Li2O conc.within theshroud afterOR (wt%)

Averagecurrentdensity(A cm�2)

Electro-lysis time(h)

Electricchargepassed (%)

Cellvoltage(V)

No shroud 1 1.05 1.01 – 0.81 1.3 150 3.3

Non-porous MgO 2 1.09 1.06 0.915 0.34 3.5 150 3.3

ZrO2–MgO 3 1.06 1.03 0.922 0.40 2.8 150 3.3

Porous ZrO2–MgOcoated STS 40mesh

4 1.03 0.98 0.940 0.76 1.5 150 3.3

STS 20 mesh 5 1.03 0.99 0.947 0.79 1.4 150 3.3

STS 100 mesh 6 0.98 0.96 0.945 0.77 1.5 150 3.3

STS 300 mesh 7 1.05 1.00 0.941 0.76 1.5 150 3.3

Porous STS 100 mesh 8a 1.01 0.970 0.950 0.77 1.5 150 3.3

STS 20 mesh 9b 0.980 0.950 0.940 0.46 2.0 150 3.0

a Run 8 was performed after the STS pipe was replaced into MgO for the connected pipe with the STS 300 mesh shroud (see Fig. 9c) The other experimental conditions weresame as those of Run 7.

b Run 9 was performed to monitor the change in the potential of the STS mesh shrouds.

E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269 265

In our previous study [29], we reported that ZrO2 can chemi-cally react with Li2O to form Li2ZrO3 in molten Li2O–LiCl, asfollows:

Li2Oþ ZrO2 ! Li2ZrO3 ð6Þ

This reaction can lead to a decrease in the Li2O concentration in LiCl.Nonetheless, before using this material as an anode shroud, thestability of ZrO2–MgO was tested, as follows. First, 450 g of LiClcontaining Li2O was put in a ZrO2–MgO crucible (9.1 cm OD � 8.1cm ID � 11 cm H) at room temperature. After the salt was

completely melted at 650 �C, the change in the Li2O concentrationin LiCl was monitored for 25 h by sampling the salt and measuringthe concentration. As Fig. 7 shows, the Li2O concentration decreasesfor the initial 8.5 h, which may be caused by reaction (6). However,this concentration leveled off afterwards, even though a continuousdecrease in Li2O concentration was expected according to the re-sults of our previous study [29]. Such an apparent discrepancycan be explained by the initially formed Li2ZrO3 on the ZrO2–MgOcrucible surface. The formed Li2ZrO3 seems to play a role as a pro-tective layer and prevents the continuous reaction between the

Time [min]

Cel

l vol

tage

[V]

2.6

2.8

3.0

3.2

3.4

Time [min]

Cur

rent

[A]

0

2

4

6

8

10

12

Time [min]

0 20 40 60 80 100 120

0 20 40 60 80 100 120

0 20 40 60 80 100 120

Cat

hode

pot

entia

ll [V

]

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

(a)

(b)

(c)

Fig. 5. Plots from electrolysis Run 7: (a) the cell voltage–time, (b) the current–time,and (c) the cathode potential-time.

Fig. 6. Microstructures of UO2: (a) before and (b) after the electrochemicalreduction.

266 E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269

ZrO2 of the crucible and the Li2O in the molten LiCl. Thus, in ourprevious study [30], we have demonstrated a successful electro-chemical reduction using the ZrO2–MgO crucible. In Run 3, the elec-trochemical reduction of UO2 was carried out using a nonporousZrO2–MgO shroud. As listed in Table 3, the electrolysis results aresimilar to those for MgO (Run 2) due to the dense structure of thenonporous shrouds. The bulk Li2O concentration after Run 3 didnot show a dramatic decrease (1.06 wt% ? 1.03 wt%), as it did inthe case of the tested ZrO2–MgO crucible. Moreover, ZrO2–MgO isvery not easily broken because the bending strength of ZrO2–MgOis much higher (600–700 MPa at 25 �C) than that of MgO(170 MPa at 25 �C). This advantage is very useful for large scale cellsbecause the equipment and components of the process should behandled by a mechanical manipulator.

In another previous study [31], we tested the chemical stabil-ity of a plasma spray ZrO2–MgO coating on Inconel 713 LC in anLiCl–Li2O molten salt. The ZrO2–MgO coatings on Inconel 713 LCexhibited a superior resistance to hot corrosion in the presenceof LiCl–Li2O molten salt for 216 h under oxidizing atmosphere.Inspired by these results, we used a ZrO2–MgO coated STS meshas a porous anode shroud. The coating method was described indetail in our previous report [31]. Before using this material asan anode shroud, we conducted a stability test of the ZrO2–MgO coated STS mesh by immersing the mesh (15 mm � 15 mm)in LiCl–Li2O at 650 �C for �21 days and observing the changes.

As can be seen in Fig. 8(b–d), the ZrO2–MgO coating on theSTS meshes exhibited resistance in the presence of the salt, com-pared to the case of the initial surface (Fig. 8a). Even thoughsome partial spallings were observed after 14 days (Fig. 8c andd), EDX analysis revealed that the remaining coating containsZr even after spalling, which indicates that the ZrO2–MgO coat-ing had not been completely lost (Fig. 8f). In Run 4, electrochem-ical reduction was performed using the ZrO2–MgO coated STS 40mesh as a porous anode shroud. The wire diameter, opening size,and open area are listed in Table 2. High current density wasachieved compared to that of Runs 2 and 3 due to the use ofthe porous shrouds; hence, U metal was successfully obtainedwithin a relatively short time. Moreover, the Li2O concentrationwithin the ZrO2–MgO coated STS mesh after the termination ofthe electrolysis run was higher than those of Runs 2 and 3 be-cause the O2� ions were transported through the side of the por-ous shroud. However, and unfortunately, SEM analysis revealedthat the ZrO2–MgO coating on the STS meshes was severelydamaged after Run 4, as shown in Fig. 8e. The reason for thedamage lies in the presence of Li metal in the molten LiCl–Li2O electrolyte. During the electrochemical reduction process,Li metal is generated by reaction (1) and exists in the electrolyte.We carried out an additional test via a 1 day-immersion of theZrO2–MgO coated STS mesh in molten LiCl–Li2O electrolytecontaining 0.3 wt% of Li metal. The damage to the ZrO2–MgO

Fig. 7. Time vs. the Li2O concentration in ZrO2–MgO crucible containing moltenLiCl.

Fig. 9. (a) The measured temperatures at the vertical positions on the STS meshshroud and its connected pipe, (b) the photograph used for the anode shroudconnected STS pipe Run 5 and (c) the photograph of the anode shroud connectedMgO tube modified for anti-corrosion (Run 8).

E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269 267

coating, similar to that shown in Fig. 8e, was also observed (datanot shown). These results imply that the use of a ZrO2–MgOcoating is not appropriate in a molten LiCl–Li2O electrolyte con-taining Li metal.

In Runs 5, 6, and 7, electrochemical reductions of UO2 were per-formed using uncoated STS meshes with different opening sizes asporous shrouds. The opening sizes of the 20, 100, and 300 meshesare 1.04, 0.140, and 0.045 mm, respectively. As listed in Table 3,

Fig. 8. SEM images of the immersed ZrO2–MgO coated meshes in LiCl–Li2O at 650 �C (elemental identification of the rectangular area in (d) using EDX.

their resultant current densities were much higher than those inthe runs using the nonporous shroud. These samples did not showsignificant differences in spite of the different opening sizes of theSTS meshes. These results mean that 0.045 mm of the 300 mesh of-fers sufficient passage size for the transport of the O2� ions. AfterRuns 5, 6, and 7 the O2� concentrations within the shrouds werehigher (>0.940 wt%) than those in the nonporous shrouds. Afterthe use of the STS anode shroud in Run 5, the area that had beenimmersed in the molten LiCl–Li2O electrolyte at 650 �C was

a) original, (b) for 7 days, (c) for 14 days, (d) for 21 days, (e) during Run 2 and (f)

Fig. 10. SEM images of STS 20 mesh: (a) original (the indicated thickness of themesh – 290 and 298 lm), (b) the area immersed in LiCl–Li2O during Run 5 (theindicated thickness of the mesh – 289 and 295 lm) and (c) the corroded area on theupper position of the immersion area during Run 5.

Time [min]0 20 40 60 80 100 120 140

Pote

ntia

l of t

he a

node

shr

ouds

[V]

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

MgO anode shroud

STS anode shroud

Fig. 11. Time vs the change in the potential of the MgO (Run 2) and STS anodeshrouds (Run 9).

268 E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269

observed, as can be seen in Fig. 9b. While corrosion products in thisarea could scarcely be observed, significant amounts of corrosionproducts from the released O2 gas from the anode surface were ob-served on the upper position of the immersion area. These resultswere also confirmed by means of SEM analysis. As can be seen inFig. 10b, the mesh area that had been immersed in the moltenLiCl–Li2O electrolyte does not show any significant difference withthe original mesh (Fig. 10a) while spallings are observed on themesh area that had been on the upper position of the immersionarea (Fig. 10c).

At the positions at which the corrosion products occurred, thelocal temperatures at the vertical position of the anode shroudand its connected pipe were measured by inserting a thermocoupleinto the anode port without any performance of electrolysis(Fig. 9a). It was revealed that the corrosion products were mainlygenerated in the temperature range of 480–630 �C. Hence, in orderto prevent corrosion in the given temperature range, electrochem-ical reduction of UO2 was additionally performed after the STS pipehad been replaced into MgO for the connected pipe for the STS 100mesh shroud (Run 8). The electrolysis results were very similar tothose of Run 7 as listed in Table 3. As can be seen in Fig. 9c, corro-sion products were scarcely observed after Run 8, which impliesthat an anti-corrosion material such as a ceramic needs to be usedfor the material above the interface of the electrolyte-gas with atemperature range of 480–630 �C. Due to the high temperatureand corrosion environment of the electrochemical reduction pro-cess, it is currently impossible to visually observe the generationof O2 gas bubbles in a molten LiCl and to know their sizes and dis-tribution near the anode surface [21,23,32–35]. Thus, it is hard toexperimentally determine an optimal opening size for the STSmesh shroud that will not allow the passage of the O2 gas bubbles.However, Phongikarroon et al. [32] experimentally observed thesize and distribution of O2 gas bubbles in aqueous glycerol solu-tions of an electrolyte, NaCl, with properties (liquid viscosity andsurface tension) similar to those of LiCl at 650 �C using dispersionfrom a cylinder of pure O2 gas. The diameter of O2 gas bubbles inthe solution ranged from 0.4 to 1.3 mm. Those results suggest thatthe use of a mesh with an opening size smaller than 0.4 mm is re-quired to prevent the escape of O2 gas bubbles from the anodeshroud during the electrolysis run, unless O2 gas bubbles are incontact with the shroud surface and are broken. As shown inFig. 3c, no severe corrosion on the bottom heat shield was ob-served, which implies that O2 gas was exhausted into the gas outletwithout leaking through the open area of the porous STS meshshroud. Nonetheless, a sufficient distance between the anode andits shroud surfaces is required in order to prevent corrosion ofthe material due to the leakage of O2 gas through the open areaof the porous shroud. The effective control of the O2 gas in the por-ous shroud will be the subject of further examination.

Fig. 11 shows the change in the potential of the anode shroudsduring Run 9 using STS 20 mesh shroud; these results were ob-tained by measuring against the reference electrode. The resultsusing the MgO anode shroud (Run 2) were also compared. Theexperimental conditions and electrolysis results shroud were sim-ilar to those of Run 5; however, the applied cell voltage waslowered to 3.0 V in order to monitor changes in the potential ofthe anode shroud for a longer time. The potential of the MgO

E.-Y. Choi et al. / Journal of Nuclear Materials 444 (2014) 261–269 269

shroud showed an almost constant value close to 0 V because theutilized material, MgO, was not conductive. On the other hand,the potential of the STS mesh shroud slightly decreased from�0.76 V to�1.15 V due to its polarization. This decrease mainly oc-curred within the initial 60 min and the potential leveled off after-ward. This observation of the anode revealed that the change in thepotential due to polarization did not affect the anode.

4. Conclusions

The effect of the type of anode shroud (porous or nonporous) onelectrochemical reduction rate was investigated in a molten LiCl–Li2O electrolyte. Six anode shrouds were tested in this study. Denseceramics, MgO and ZrO2–MgO, were used as nonporous shrouds.ZrO2–MgO coated STS 40 mesh, STS 20, 100, and 300 meshes wereused as porous shrouds. Systematic comparisons of the reductionrates among the tested anode shrouds suggests that a porous struc-ture of the anode shroud, due to the active transport of O2� ionsthrough the pores in meshes of the shroud wall, is advantageousto achieve high current density and to speed up the electrochemi-cal reduction process than is a nonporous structure. Moreover, thedecrease in Li2O concentration in the porous shrouds after the elec-trochemical reduction was less significant than that in the nonpo-rous shrouds. This approach may potentially be useful in thedesign of large scale cells with high efficiency because the highcurrent density achieved by the use of the porous shroud can speedup the electrochemical reduction process.

Acknowledgements

This work was supported by the Nuclear Research & Develop-ment Program of the National Research Foundation (NRF), in agrant funded by the Korean Government.

References

[1] J.J. Laidler, J.E. Battles, W.E. Miller, J.P. Ackerman, E.L. Carls, Prog. Nucl. Energy31 (1997) 131.

[2] R.W. Benedict, H.F. McFarlane, Radwaste Magaz. 5 (1998) 23.[3] M. Iizuka, Y. Sakamura, T. Inoue, Nucl. Eng. Technol. 40 (2008) 183.[4] M.F. Simpson, S.D. Herrmann, Nucl. Technol. 162 (2008) 179.

[5] J.-H. Yoo, C.-S. Seo, E.-H. Kim, H. Lee, Nucl. Eng. Technol. 40 (2008) 581.[6] S. Kitawaki, T. Shinozaki, M. Fukushima, T. Usami, N. Yahagi, M. Kurata, Nucl.

Technol. 162 (2008) 118.[7] J. Serp, R.J.M. Konings, R. Malmbeck, J. Rebizant, C. Scheppler, J.-P. Glatz, J.

Electroanal. Chem. 561 (2004) 143.[8] J.L. Willit, W.E. Miller, J.E. Battles, J. Nucl. Mater. 195 (1992) 229.[9] S.M. Jeong, S.-B. Park, S.-S. Hong, C.-S. Seo, S.-W. Park, J. Radioanal. Nucl. Chem.

268 (2006) 349.[10] S.B. Park, B.H. Park, S.M. Jeong, J.M. Hur, C.-S. Seo, S.-H. Choi, S.W. Park, J.

Radioanal. Nucl. Chem. 268 (2006) 489.[11] K.M. Goff, J.C. Wass, K.C. Marsden, G.M. Teske, Nucl. Eng. Technol. 43 (2011)

317.[12] T. Koyama, Y. Sakamura, T. Ogata, H. Kobayashi, in: Proc. Global 2009, p. 9161,

Paris, France, 2009.[13] J.-M. Hur, T.-J. Kim, I.-K. Choi, J.B. Do, S.-S. Hong, C.-S. Seo, Nucl. Technol. 162

(2008) 192.[14] J.-M. Hur, C.S. Seo, S.S. Hong, D.S. Kang, S.W. Park, React. Kinet. Catal. Lett. 80

(2003) 217.[15] G.Z. Chen, D.J. Fray, T.W. Farthing, Nature 407 (2000) 361.[16] Y. Sakamura, M. Kurata, T. Inoue, J. Electrochem. Soc. 153 (2006) D31.[17] Y. Sakamura, T. Omori, T. Inoue, Nucl. Technol. 162 (2008) 169.[18] J.-H. Hur, S.M. Jeong, H. Lee, Electrochem. Commun. 12 (2010) 706.[19] K.V. Gourishankar, L. Redey, M. Williamson, Light Metals, 2002, in: W.A.

Schneider (Ed.), The Minerals, Metals, and Materials Society, 2002, p. 1075.[20] L. Redey, K.L. Gourishankar, US Patent 6540902, April 1, 2003.[21] S.D. Herrmann, S.X. Li, M.F. Simpson, S. Phongikarroon, Sep. Sci. Technol. 41

(2006) 1965.[22] M. Iizuka, Y. Sakamura, T. Inoue, J. Nucl. Mater. 359 (2006) 102.[23] S.M. Jeong, H.S. Shin, S.H. Cho, J.M. Hur, H.S. Lee, Electrochim. Acta 54 (2009)

6335.[24] S.M. Jeong, H.S. Shin, S.S. Hong, J.M. Hur, J.B. Do, H.S. Lee, Electrochim. Acta 55

(2010) 1749.[25] S.D. Herrmann, S.X. Li, B.E. Serrano-Rodriguez, in: Proc. Global 2009, Paris,

France, p. 9059, 2009.[26] E.-Y. Choi, I.-K. Choi, J.-M. Hur, D.-S. Kang, H.-S. Shin, S.M. Jeong, Electrochem.

Solid-State Lett. 15 (2012) E11.[27] T.-J. Kim, Y.-H. Cho, I.-K. Choi, J.-G. Kang, K. Song, K.-Y. Jee, J. Nucl. Mater. 375

(2008) 275.[28] S.X. Li, S.D. Herrmann, J. Electrochem. Soc. 149 (2002) H39.[29] W. Park, J.M. Hur, E.-Y. Choi, J.-K. Kim, J. Kor, Rad. Waste Soc. 8 (2010) 19 (in

Korean).[30] W. Park, J.M. Hur, S.-S. Hong, E.-Y. Choi, H.S. Im, S.-C. Oh, J.-W. Lee, J. Nucl.

Mater. 441 (2013) 232.[31] S.H. Cho, S.B. Park, J.H. Lee, J.M. Hur, H.S. Lee, Mater. Chem. Phys. 131 (2012)

743.[32] S. Phongikarroon, S.D. Herrmann, S. Li, M.F. Simpson, in: Proc. AlChE Annual

Meeting, Cincinnati, 2005, p. 7912.[33] S. Phongikarroon, S.D. Herrmann, S.X. Li, Ind. Eng. Chem. Res. 45 (2006) 7679.[34] S. Phongikarroon, R.O. Hoover, E.C. Barker, Ind. Eng. Chem. Res. 49 (2010) 2926.[35] S. Phongikarroon, R.W. Bezzant, M.F. Simpson, Chem. Eng. Res. Des. 91 (2013)

418.