Reduction process of uranium(IV) and uranium(III) in molten fluorides

Article abstract of DOI:10.1016/j.electacta.2006.11.018

This study focused on the electroreduction process of uranium cations in molten fluorides. It involved cyclic voltammetry, chronopotentiometry with and without current reversal, and square wave voltammetry. The results indicate a two-step reduction process for uranium(IV). The first step U(IV)/U(III) exchanging one electron corresponds to a soluble/soluble system and is limited by U(IV) diffusion with D_U(IV) = 1.25 ± 0.35 × 10^-5 cm² s^-1 in LiF-NaF at 720 °C. In order to perform a thorough study of the second step U(III)/U(0) in the reduction process, the melt was chemically reduced in U(III) with U metal as reducing agent. Alternatively to the use of LiF-NaF where U metal is unstable at 720 °C, the chemical reduction of U(IV) in U(III) was performed in a LiF-CaF₂-UF₄ solution containing U metal at 810 °C. It has been confirmed that the reduction of U(III) proceeds in one step exchanging three electrons and by a diffusion controlled process with D_U(III) = 2.2 ± 0.7 × 10^-5 cm² s^-1 in LiF-CaF₂ at 810 °C.

Full text of DOI:10.1016/j.electacta.2006.11.018

Electrochimica Acta 52 (2007) 3995–4003
Reduction process of uranium(IV) and uranium(III) in molten fluorides
a
b
a
c
d
b,∗
C. Hamel , P. Chamelot , A. Laplace , E. Walle , O. Dugne , P. Taxil
a
CEA/DEN/VRH/DRCP/SCPS/LPP, site de Marcoule, B aˆ timent 399, BP 17171, 30207 Bagnols sur C e` ze, Cedex, France
Laboratoire de G e´ nie Chimique UMR 5503, Universit e´ Paul Sabatier, 118 route de Narbonne, 31062 Toulouse, Cedex 04, France
b
c
Laboratoire EDF R&D, Groupe chimie et Corrosion, Site des Renardi e` res, Ecuelles, 77818 Moret sur Loing, France
d
CEA/DEN/DTEC/STCF/LMAC, BP 111, 26702 Pierrelatte, Cedex, France
Received 25 August 2006; received in revised form 17 November 2006; accepted 18 November 2006
Available online 3 January 2007
Abstract
This study focused on the electroreduction process of uranium cations in molten fluorides. It involved cyclic voltammetry, chronopotentiometry
with and without current reversal, and square wave voltammetry.
The results indicate a two-step reduction process for uranium(IV). The first step U(IV)/U(III) exchanging one electron corresponds to a
−
5
2
−1
◦
soluble/soluble system and is limited by U(IV) diffusion with DU(IV) = 1.25 ± 0.35 × 10 cm s in LiF–NaF at 720 C.
In order to perform a thorough study of the second step U(III)/U(0) in the reduction process, the melt was chemically reduced in U(III) with
U metal as reducing agent. Alternatively to the use of LiF–NaF where U metal is unstable at 720 C, the chemical reduction of U(IV) in U(III)
solution containing U metal at 810 C. It has been confirmed that the reduction of U(III) proceeds in one step
exchanging three electrons and by a diffusion controlled process with DU(III) = 2.2 ± 0.7 × 10 cm s in LiF–CaF
2006 Elsevier Ltd. All rights reserved.
◦
◦
was performed in a LiF–CaF
2
–UF
4
−
5
2
−1
◦
2
at 810 C.
©
Keywords: Uranium; Molten fluorides; Voltammetry; Chronopotentiometry; Electrowinning
1
. Introduction
formances being of primary importance, a major objective is to
acquire a greater basic knowledge of the chemistry of actinides
and fission products in molten salts and on process engineering
considerations in order to answer to the following question: is
it possible to find the right combination – media and separation
technique – that allows a quantitative recovery of all the actinides
with sufficient separation efficiency from fission products. Elec-
trolytic recovery in molten fluorides is one of these separation
techniques and it has to be assessed.
This project focused on the following elements: neodymium
[2] and uranium. The experimental work first consisted in the
electrochemical study of those elements in molten fluorides rel-
ative to valences, exchanged electrons and diffusion coefficient.
In relationship with these electrochemical studies, we are inter-
ested in the actinides–lanthanides separation. The aim of this
project is to determine the capability of the An–Ln separation
(i) by measuring their potential difference [3] and (ii) by testing
actinides electrolytic extraction.
In France, hydrometallurgical processes are the reference
route either for the reprocessing of spent nuclear fuel (PUREX
process) or for the partitioning and transmutation strategies of
long-lived radionuclides.
However, in the long-term strategies on nuclear fuel cycle,
they could be not well appropriate for the reprocessing of future
nuclear fuels, which will contain large amount of transuranium
elements and refractory matrix. In that case, alternative pro-
cesses such as the pyrochemical processes could be revealed as
promising options thanks to presumed intrinsic properties: large
fissile material loading (criticality constraints are less restrictive
than in aqueous media), no sensitivity to radiolysis, ability to
treat refractory material.
Therefore, a R&D program has been launched to assess the
potentialities of high temperature molten salt technology for the
selected applications for long-term options [1]. Separation per-
In this paper, results are given on uranium(IV) and ura-
nium(III) electrochemical behaviour in molten fluorides.
The first part of this paper discusses our experimental condi-
tions such as the solvent and the cathode material. The second
∗
Corresponding author. Tel.: +33 561556828; fax: +33 561556139.
E-mail address: taxil@chimie.ups-tlse.fr (P. Taxil).
0
013-4686/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2006.11.018

3
996
C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
part details our electrochemical investigations on the uranium
ions reduction process.
Up to day several studies have been realized on uranium(IV)
ions in molten fluoride melts. They can be distinguished in two
types: with BeF2 [4–6] and without BeF2 [7–9].
for analysis, ACS, ISO 99% minimum); CaF2 (Prolabo norma-
pur, 98% minimum). The composition and the melting points of
these solvents in eutectic compositions are presented in Table 1.
Before the experiments, they are previously dehydrated by heat-
ing under vacuum up to their melting point for 48 h.
Uranium tetrafluoride (Comhurex) was used as solute. This
compound is introduced into the melt through an air lock under
argon.
Uranium metal (CEA Pierrelatte) was cut in plates
(10 mm × 35 mm × 8 mm). They were placed on a steel connec-
tor with a silver basket and plunged into the molten salt during
the pre-reduction step.
All these publications deal with results on the reduction or
oxidation process of UF4. Only the first step of the reduc-
tion process (U(IV) → U(III)) has been studied. The number
of exchanged electrons, the diffusion coefficient at several tem-
peratures and the kinetics limitation of this reaction have been
determined.
All previous studies show that the reaction is reversible
and limited by diffusion. The results given for the number of
exchanged electrons are 1 for this first step [4,5,9] except for
Clayton and Mamantov [8] in LiF–NaF–KF who obtained a
range of 1.4–1.7. These authors explain this difference by a
possible disproportionation reaction; 4UF3 = U + 3UF4 and a
reaction with the solvent; U + 4K(I) = U(IV) + 4K: these reac-
tions generating U(IV) disturb the calculation of n.
2.2. The electrodes
The working electrode was a 1-mm diameter silver wire
(Goodfellow 99.95%) or a 1-mm diameter tantalum wire (Good-
fellow 99.9%); the contact surface was determined after each
experiment by measuring the immersion depth in the bath.
The auxiliary electrode was a 7-mm diameter graphite rod
The second step (U(III) → U(0)) is not accurately studied in
molten fluorides. Chardard [4] is the first author who identified
the U(III)/U(0) system by electrochemical techniques. Cartier
(Graphitec, carbon for spectrography).
The potentials are referred to a 0.5-mm diameter platinum
[
9] showed that a initial U(III) solution could be the favourable
wire (Goodfellow 99.9%) immersed into the molten electrolyte,
condition to study the formation of the metal. In molten chlo-
rides, a pre-reduction step (3UCl4 + U → 4UCl3) has already
been studied [10].
2−
acting as a quasi-reference electrode Pt/PtOx/O , reliable in
molten media where the acidity is invariable [2,11].
In this paper, the first step of the reduction process is studied
from UF4 and the second step from a U(III) molten solution
which will be obtained by the chemical reduction of U(IV) in
the molten bath.
2
.3. Electrochemical techniques
Cyclic voltammetry, chronopotentiometry and square wave
voltammetry were used for the investigation of the uranium
reduction process.
2
. Experimental
Cyclic voltammetry and chronopotentiometry were per-
formed using an Autolab PGSTAT21 controlled by a computer
using the specific software GPES version 4.9. The square wave
voltammograms were performed using an EG&G m 273-A
potentiostat/galvanostat connected with a computer equipped
with the research software Headstart.
2
.1. The electrochemical cell and chemicals
A vitreous carbon crucible (Carbone Lorraine) containing
the electrolyte is introduced into a refractory steel cylinder
where the inner wall is protected by a graphite liner. During the
experimentation, the cell, closed by a stainless steel lid cooled
by circulating water, is placed under inert argon atmosphere
2.4. Characterization techniques
(
LINDE U quality), previously dehydrated and deoxygenated
The deposits and salts were characterized using (i) XRD
SEIFERT) to analyse the crystallized compounds and (ii) scan-
ning electron microscopy (LEO 435 VP) coupled with an EDS
probe (INCA 200) for the observation and the chemical analysis
of the deposits, respectively.
using a purification cartridge (Air Liquide). The cell is heated
using a programmable furnace, the temperatures being measured
using a chromel–alumel thermocouple.
The solvents used as electrolytic bath were pure or mixed
alkaline fluorides: LiF (Merck suprapur 99.99%); NaF (Merck
(
Table 1
◦
Comparison between the reduction potential of the solvent cations and the reduction potential of U(IV) (UF4) and U(III) (UF3), at 720 C with
−
U(IV)] = [U(III)] = 0.05 mol kg
1
[
LiF–CaF2
LiF
LiF–NaF
UF3
UF4
Molar percent composition
Melting point ( C)
77–23
760
100
846
60–40
652
◦
Reduction potential of the least stable cation of the
solvent and uranium cations (V vs. ref F2/F )
Li(I)/Li (−5.44)
Li(I)/Li (−5.42)
Na(I)/Na (−4.99)
U(III)/U (−4.53)
U(IV)/U(III) (−3.81)
−
−
Origin of the potential scale: the standard potential of the F2/F system.

C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
3997
3
. Results and discussion
.1. Selection of the solvent and the working electrode
3
material
3
.1.1. Selection of the solvent
Table 1 helps to predict the reduction potential of the cations
of each solvent compared to the reduction potential of a solu-
◦
tion of uranium cations at 720 C. The standard potentials
displayed here are converted from Gibbs energy for the pure
solid compounds provided by Ref. [12] and referred to the flu-
−
orine electrode (F2/F ). In the case of binary mixtures, which
are considered as ideal, the activity of each compound is equal
to its molar fraction; and the value of the potential of the least
stable element is calculated.
According to these assumptions, all melts are suitable to
observe both U(IV)/U(III) and U(III)/U(0) system.
We began our experiments on UF4 in the LiF–NaF eutectic
mixture which has the lowest melting point.
In order to perform a thorough study of the second reduction
step, a first attempt was made in LiF–NaF to chemically reduce
UF4 into UF3 with uranium metal immersed in the electrolyte
Fig. 1. Cyclic voltammogram of LiF–NaF–UF (0.12 mol kg 1) solution at
−
4
◦
2
720 C. Working electrode: Ag (surface area = 0.42 cm ); counter electrode:
−
1
graphite; electrode for potential reference: Pt; scan rate = 0.1 V s
.
which are, respectively, oxidized at −0.36 V (Ia) and −0.77 V
(IIa) in the anodic run.
This first part of this paper will be dedicated to the first reduc-
tion process Ic/Ia while the IIc/IIa system will be studied in the
second part (cf. Section 3.3).
The first cathodic peak (Ic) can be obviously attributed to the
reduction of U(IV) in an intermediate species (possibly U(III))
which is reduced in U metal in the step IIc.
Indeed, we observe a linear relationship between the peak
current density and the uranium(IV) ion concentration (Fig. 2).
For a reversible soluble/soluble system, the intensity of the
cathodic peak is correlated with the potential scanning rate by
the following relationship [21]:
(
U + 3UF4 ↔ 4UF3).
This reaction was found to be unsuccessful in LiF–NaF at
◦
7
20 C; U metal being unstable in this solvent. However, conclu-
sive results were obtained in a LiF–CaF2 melt (cf. Section 3.3).
The investigation on the U(III) reduction was thus performed in
this latter media.
3
.1.2. Selection of the working electrode material
ꢀ
ꢁ
1/2
Uranium can form, in the temperature range of our exper-
iments, alloys or stoichiometric compounds with a series of
metals, usually used as working electrodes (e.g. Cu, Ni, Fe, Pt)
nF
RT
0
1/2 1/2
Ip = 0.446nFSC
D
v
(1)
2
0
[
13,14]. Such metals have not been selected as cathode material
where S is the electrode area in cm , C the solute concentra-
− −1
3
2
because their use would obviously generate a shift of the elec-
trodeposition potential of uranium in the anodic direction due to
the lower activity of the electrodeposited U in the alloyed form
tion in mol cm , D the diffusion coefficient in cm s , F the
−
1
Faraday constant (96,500 C mol ), n the number of exchanged
−1
electrons, v the potential scanning rate in V s and T is the
temperature in K.
[
15–18].
Taking into account these considerations, we only considered
The linearity of Ip versus v1/2 shows that the electrochemi-
cal reduction process is controlled by the diffusion of uranium
for this study silver, tungsten or tantalum as working electrode
materials which are inert with uranium.
3
.2. Study of the first step of the reduction process:
reduction of uranium(IV) in LiF–NaF
The reduction process of U(IV) was first studied in LiF–NaF,
a molten fluoride melt often used in our laboratory [19,20] and
which exhibits a low melting point.
3
.2.1. Cyclic voltammetry
Cyclic voltammetry was carried out on a silver electrode in
◦
a LiF–NaF–UF4 melt at 720 C. The cathodic and anodic limits
of the electrochemical window correspond to the reduction of
Na(I) and the oxidation of silver, respectively.
Fig. 2. Linear relationship between the cathodic peak current density Ic and the
U(IV) concentration in the LiF–NaF melt (eutectic composition 60:40 mol/mol).
Working electrode: Ag; counter electrode: graphite; electrode for potential ref-
−1 ◦
erence: Pt; scan rate = 0.1 V s ; T = 720 C.
The cyclic voltammogram in Fig. 1, exhibits two reduction
peaks in the cathodic run at −0.64 V (Ic) and −0.92 V (IIc) ver-
sus the quasi reference electrode Pt/PtOx/O2 , giving products
−

3
998
C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
Fig. 5. Series of chronopotentiograms of the LiF–NaF (60:40 mol/mol)–UF4
Fig. 3. Linear relationship of U(IV) reduction peak current density vs. the
square root of the potential scanning rate. Working electrode: Ag (surface
−
1
(
0.09 mol kg ) solution with various currents. Working electrode: Ag (surface
2
2
area = 0.35 cm ); counter electrode: graphite; electrode for potential reference:
area = 0.35 cm ); counter electrode: graphite; electrode for potential reference:
◦
Pt; T = 720 C.
◦
−1
Pt; T = 720 C; electrolyte: LiF–NaF (60:40 mol/mol) [U(IV)] = 0.09 mol kg
.
3
.2.2. Chronopotentiometry
Fig. 5 shows the evolution of the chronopotentiograms of
ions in the solution. The measurement of the slope of the line
◦
shown in Fig. 3 yields the following relation at T = 720 C and
UF4 in LiF–NaF with the applied current density on a silver
electrode. These curves exhibit a single wave in the same poten-
tial range as that observed on the cyclic voltammogram and
therefore associated to the reduction of uranium(IV) ions into
uranium(III).
0
−2
−1
C = 9.0 × 10 mol kg
:
Ip
1/2
−1/2
−2
cm
=
−0.089 A s
V
(2)
1/2
Sv
The so-called transition time τ indicates the time necessary to
observe the complete depletion of the electroactive species (here
the U ion), resulting from the diffusion in the layer of elec-
trolyte at the electrode surface, when a current pulse is applied
to the cathode; τ is measured by the duration of the first part of
the chronopotentiogram; as developed in Refs. [21–23], the data
plotted in Fig. 6 follow the Sand’s law [23], expressed below,
proving that the electrode process is diffusion controlled:
The slope of the straight line, assuming that n = 1, allows to
calculate the diffusion coefficient of U(IV) ions: DU(IV) = 1.1 ×
IV
−
5
2
−1
1
0
cm s
.
◦
At 720 C, from 10 attempts and six different con-
centrations the diffusion coefficient of U(IV) ions is
−
5
2
−1
.
1
.5 ± 0.5 × 10 cm s
The influence of the temperature on the value of the diffusion
coefficient was investigated by performing a series of cyclic
voltammograms at several temperatures. The results plotted in
Fig. 4 obey Arrhenius’ law expressed as:
1/2
0
1/2
Iτ
= 0.5nFSC (πD) = constant
(4)
The data of Fig. 6 allow the following equation to be established
at T = 720 C and C = 9.0 × 10 mol kg
ꢀ
ꢁ
Ea
◦
0
−2
−1
:
D = D0 exp −
(3)
RT
1
/2
Iτ
1/2
−2
cm
=
−0.049 ± 0.003 A s
(5)
where Ea is the activation energy, is found from the data of Fig. 4
to be 49.2 kJ mol and D0 = 5.66 × 10 cm s
S
−
1
−3
2
−1
.
Fig. 6. Values of Iτ1/2 vs. the cathodic current for the U(IV)/U(III) system.
2
Fig. 4. Linear relationship of logarithm of U(IV) diffusion coefficient vs. the
Working electrode: Ag (surface area = 0.35 cm ); counter electrode: graphite;
◦
reverse of the absolute temperature. Working electrode: Ag; T ranging from 680
electrode for potential reference: Pt; T = 720 C; electrolyte: LiF–NaF (60:40
◦
−1
−1
to 760 C; electrolyte: LiF–NaF (60:40 mol/mol) [U(IV)] = 0.09 mol kg
.
mol/mol); [U(IV)] = 0.09 mol kg
.

C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
3999
Eqs. (1) and (4) [21,23]:
ꢀ
ꢁ
2
Ip
1
Rπ
n =
T
(7)
1/2
1/2
2
v
Iτ
4 × 0.446 F
According to the results provided by Fig. 3 and Fig. 6, the
number of exchanged electrons is 1.1.
◦
From seven attempts at 720 C, and four different concen-
trations, an average number of 1.1 ± 0.1 was found. Thus, we
concluded that one electron was exchanged.
3
.2.3.2. Square wave voltammetry. In this technique, derived
Fig. 7. Plotting of Iτ1/2/C vs. the UF4 concentration in the LiF–NaF (60:40
from cyclic voltammetry, the scanning of the working electrode
potential proceeds by staircase with superimposition, on each
step of the staircase, of two potential pulses, direct and reverse,
with the same intensity [25]. When the differential current mea-
sured on each step between the successive pulses, is plotted
versus the potential, each electrochemical reaction is associated
to a Gaussian shape peak [26]. The mathematical analysis of the
peak yields a simple and useful equation associating the width of
the half peak (W1/2 in V) and the number of exchanged electrons
[25,27]:
mol/mol) mixture. Working electrode: Ag; counter electrode: graphite; electrode
◦
for potential reference: Pt; T = 720 C.
The Sand’s law is verified and can be normalized to the U(IV)
concentration (Fig. 7):
1
/2
Iτ
1/2
−1
mol
=
−303 ± 35.0 A cm s
(6)
0
SC
This electrochemical reduction process is confirmed to be lim-
ited by the diffusion of uranium(IV) ions in solution. The
diffusion coefficient is calculated using Eq. (6) and assuming
RT
= 3.52 nF
W
1/2
(8)
−
5
2
−1
◦
that n = 1, DU(IV) = 1.25 ± 0.35 × 10 cm s at 720 C. This
range of D is similar as the one calculated by cyclic voltammetry.
The reversal chronopotentiogram (Fig. 8) also evidences
the formation of a soluble phase at the electrode during the
cathodic run. Indeed, the transition time of the oxidation wave
is three times smaller than the transition time in reduction
As a rule, the technique is valid when the electrochemical
reaction fulfils the reversibility criteria that can be verified if
the current peak is linear with the square root of the potential
signal frequency. Thus, we considered in previous papers that
it is possible to apply this method as long as this linearity is
verified [28–30].
(τox = (1/3)τred) [24].
In the present work, the linearity is observed between 0 and
80 Hz. Fig. 9 shows the square wave voltammogram of UF4 in
3
.2.3. Calculation of the number of exchanged electrons
LiF–NaF obtained in the reliable range of frequencies. It exhibits
a single peak in the same potential range as the one of the previ-
ous cyclic voltammograms. This peak is symmetric as predicted
by theory.
By the measurement of the width of the half peak, we can
perform the calculation of n. For six attempts, the average value
We assumed in the previous paragraphs that the number of
exchanged electrons in the U(IV)/U(III) process was 1. By com-
bining cyclic voltammetry and chronopotentiometry results and
by square wave voltammetry, this hypothesis will be confirmed.
3
.2.3.1. Cyclic voltammetry and chronopotentiometry. The
number of exchanged electrons can be calculated combining
Fig. 8. Reversal chronopotentiogram of LiF–NaF–UF4 (0.09 mol kg 1),
−
Fig. 9. Square wave voltammogram of the LiF–NaF–UF4 (0.09 mol kg
−1
)
−2
2
2
I = ± 0.15 A cm . Working electrode: Ag (surface area = 0.33 cm ); counter
solution at 25 Hz. Working electrode: Ag (surface area = 0.35 cm ); counter
◦
◦
electrode: graphite; electrode for potential reference: Pt; T = 720 C.
electrode: graphite; electrode for potential reference: Pt; T = 720 C.

4
000
C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
of W1/2 is 304 (± 20) mV and the number of exchanged electron
for the U(IV)/U(III) system is 1 (± 0, 1).
reverse reaction would be slightly favoured on a thermochemical
basis. Then, U(III) would be slowly oxidized by a dispro-
portionation reaction, also driven to right by U metal further
consumption.
3
.3. Study of the second step of the reduction process:
reduction of uranium(III) in LiF–CaF2
In our experiment, the same reaction certainly occurred with
sodium vaporization. At 720 C, sodium vapour pressure is
◦
Prior to the U(III)/U(0) system study, a chemical reduction
of UF4 by U metal as reducing agent in the molten melt was
achieved; the Gibbs free energy of the reaction (9) in pure com-
pounds being at 720 C, ꢀrG = −261.3 kJ mol [12]:
0.17 bar which allows its progressive volatilization and depo-
sition on water cooled part of the top of the cell. The cell
atmosphere being never saturated in sodium vapour, this leads
to a full uranium metal oxidation.
◦
0
−1
UF4 + U ↔ UF3
(9)
3
.3.1. Cyclic voltammetry
Cyclic voltammetry was carried out on a silver electrode in
◦
Conclusive results were obtained in LiF–CaF2 at 810 C for
thepre-reductionstep:thebathwasinitiallygreen(typicalcolour
of UF4) and became red (characteristic of UF3) after the addi-
tion of U metal. A XRD analyse proved the presence of UF3 in
LiF–CaF2 (Fig. 10).
◦
LiF–CaF –UF melt at 810 C. The cyclic voltammogram in
Fig. 11, exhibits two peaks Ic and IIc as in LiF–NaF–UF that
were attributed, respectively, to U(IV)/U(III) and U(III)/U(0)
(Fig. 1).
2
3
4
◦
Inversely in LiF–NaF, at 720 C, the metal was found to be
unstable: uranium was oxidized in U(IV) in the bath.
However, the major difference between those two former
voltammograms consists in the relative position of the work-
ing electrode rest potential before the measurement. In Fig. 1
this potential is more anodic than U(IV)/U(III) system, whereas
in Fig. 11 it is more cathodic, showing then the sole presence
of UF species in LiF–CaF . Thus, the peak Ic results from the
oxidized species U(IV) produced in Ia during the anodic run.
The sharp anodic peak IIa is associated with the cathodic one
IIc observed on the reverse scan. The shape of the anodic peak
is typical of the dissolution of a metal deposited in the cathodic
run (stripping peak). Accordingly, we can assume that uranium
metal is deposited in a single step by reduction of U(III) ions
into U(0).
To study the U(III)/U(0) system, the same methodology as
for the U(IV)/U(III) system was used.
The height of the peak IIc increases with the uranium ion
content and its intensity is correlated with the scanning rate of
potentials by the following relationship, reliable for a reversible
soluble/insoluble system [32]:
Fifteen grams of U metal were introduced in 200 g of molten
◦
LiF–NaF at 720 C. Increase of U(III) and U(IV) concentra-
tions was then observed by cyclic voltammetry. After 2 days,
the experiment was stopped. The molten bath was fully green
and UF4 presence was confirmed by XRD. Powder inflammation
was also observed when the cell was open.
3
2
Several authors have already reported U metal instability in
LiF–NaF–KF melts (46.5–11.5–42.0 mol%) [8,31]. Young [31]
◦
worked at 525 C and studied uranium by spectrophotometry.
Under the conditions of his experiment, he showed that uranium
metal and U(III) were slowly oxidized into U(IV) and alkali
metal vapor was also observed.
Clayton and Mamantov [8] studied uranium electrochemical
behaviour in the same bath at 500 C. They also observed U(0)
and U(III) oxidation. As the previous author, they attributed it
to a reaction of U metal with potassium: U + K(I) = U(IV) + 4K.
The volatilization of potassium metal from the melt to a cooler
region of the cell (water cooled parts) would be a driving force
shifting the previous equilibrium to the right even through the
◦
ꢀ
ꢁ
1
/2
nF
RT
0
1/2 1/2
Ip = −0.61nFSC
D
v
(10)
Fig. 11. Cyclic voltammogram of the LiF–CaF2 (77:23 mol/mol)–UF3
−
1
◦
Fig. 10. XRD spectrum of LiF–CaF2–UF3 salt. XRD conditions: 2Θ
scale—start 10.000 ; end 74.380 ; step 0.020 ; step time 6 s, with Ta as standard
reference.
(0.03 mol kg
)
solution at 810 C Working electrode: Ag (surface
◦
◦
◦
2
area = 0.42 cm ); counter electrode: graphite; electrode for potential reference:
Pt; scan rate = 0.1 V s
−
1
.

C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
4001
Fig. 14. Reversal chronopotentiogram of LiF–CaF2–UF3 (0.03 mol kg 1),
−
Fig. 12. Series of chronopotentiograms of the LiF–CaF2–UF3 (0.03 mol kg 1)
−
−2
2
I = ± 0.37 A cm . Working electrode: Ta (surface area = 0.36 cm ); counter
2
solution with various currents. Working electrode: Ag (surface area = 0.42 cm );
counter electrode: graphite; electrode for potential reference: Pt; T = 810 C.
◦
electrode: graphite; electrode for potential reference: Pt; T = 810 C.
◦
The Sand law is verified (Eq. (12) and Fig. 13): this elec-
trochemical reduction process is limited by the diffusion of
uranium(III) ions in solution:
1
/2
◦
0
The linearity of Ip versus v
at T = 810 C and C = 6.6 ×
−5
−3
−1
1
0
mol cm (0.03 mol kg ) proves that the electrochemical
reduction process is controlled by the diffusion of uranium(III)
ions in the solution:
1
/2
Iτ
1/2
−2
cm
=
−0.088 A s
(12)
S
Ip
1/2
−1/2
−2
cm
=
−0.3393 A s
V
(11)
The reversal chronopotentiogram (Fig. 14) evidences the forma-
tion of a solid phase on the Ta electrode during the cathodic run,
since the transition time of the oxidation wave was found to be
equal to the one of the reduction (τox = τred) [33].
1/2
Sv
With the measurement of the slope of the straight line given
in Eq. (11) and assuming that n = 3, the diffusion coefficient
−
5
2
−1
of U(III) ions was calculated: DU(III) = 2.9 × 10 cm s at
◦
8
10 C.
3
.3.3. Calculation of the number of exchanged electrons
We use the same methodology as for the U(IV)/U(III) system
◦
From 10 attempts at 810 C and three different concentra-
tions, the diffusion coefficient of U(III) ions was found to be
to calculate the number of exchanged electrons (Section 3.2.3).
By combining the results from Eqs. (4) and (10) [21,32], and
according to the results provided by Figs. 13 and 15, the number
of exchanged electrons is calculated to be 2.9 ± 0.2.
−
5
2
−1
.
DU(III) = 2.2 ± 0.7 × 10 cm s
3
.3.2. Chronopotentiometry
Fig. 12 shows the evolution of the chronopotentiograms
The square wave voltammogramm of UF3 exhibits a peak
IIc which is not exactly symmetric as predicted by theory. A
similar shape was reported in molten chlorides for the reduction
of U(III) ions to uranium metal [34] and attributed to a nucle-
ation effect, since, as observed in cyclic voltammetry, the rise of
the current is delayed by the overpotential caused by the solid
◦
of UF3 with the applied current density at T = 810 C and
0
−5
−3
−1
for C = 6.6 × 10 mol cm
(0.03 mol kg ). These curves
exhibit a single wave associated to the reduction of U(III) ions
into metal in the same potential range as that observed on the
cyclic voltammogram.
Fig. 13. Plotting of Iτ1/2 vs. the cathodic current density for the
Fig. 15. Linear relationship of U(III) reduction peak current density vs. the
square root of the scanning potential rate. Working electrode: Ag (surface
area = 0.42 cm ); counter electrode: graphite; electrode for potential reference:
◦ −1
Pt; T = 810 C; [U(III)] = 0.03 mol kg . Solvent: LiF–CaF2 (77:23 mol/mol).
2
U(III)/U(0) system. Working electrode: Ag (surface area = 0.42 cm ); counter
◦
2
electrode: graphite; electrode for potential reference: Pt; T = 810 C,
−1
[
U(III)] = 0.03 mol kg
.

4
002
C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
tion on a molybdenum cathode. After this electrolysis a thin
layer of uranium metal was deposited on the electrode (Fig. 17).
The nature of this white deposit analysed by EDS confirmed the
formation of uranium metal.
4. Conclusion
This study was focused on the electroreduction process of
uranium cations in molten fluorides. It involved cyclic voltam-
metry, chronopotentiometry with and without current reversal,
and square wave voltammetry.
The molten Li–NaF and LiF–CaF2 eutectics were used suc-
cessfully as solvent. Results demonstrated that uranium(IV) is
reduced in two successive steps:
Fig. 16. Square wave voltammogram of the LiF–CaF2–UF3 (0.03 mol kg 1)
−
2
solution at 16 Hz. Working electrode: Ag (surface area = 0.36 cm ); counter
◦
electrode: graphite; electrode for potential reference: Pt; T = 810 C.
−
−
U(IV) + 1e = U(III) and U(III) + 3e = U(0)
(13)
phase formation. This explains that the increasing part of the
differential current is sharper than the decreasing one. Thus, the
disturbance of the signal due to the nucleation effect should be
resolved if the decreasing part of the peak is only taken into
account for measurement of the half peak width. Accordingly,
measuring the half-width of the half peak, as indicated in Fig. 16
can be performed for the calculation of n.
Using this methodology and after having verified the linear-
ity of the current peak IIc with the square root of the potential
signal frequency, the average value of W1/2 was calculated to be
The first step U(IV)/U(III) exchanging one electron corre-
sponds to a soluble/soluble system and is limited by U(IV)
−5
2
−1
◦
diffusion with DU(IV) = 1.25 ± 0.35 × 10 cm s at 720 C in
LiF–NaF.
In order to perform a thorough study of the second step,
uranium(IV) was chemically reduced in U(III) with U metal
◦
as reducing agent in the molten bath LiF–CaF2 at 810 C.
This reaction was found to be unsuccessful in LiF–NaF at
◦
7
20 C because U metal is unstable in this solvent. The sec-
ond step U(III)/U(0) exchanging three electrons behaves as
a soluble/insoluble system limited by U(III) diffusion with
1
02 mV (three attempts). According to Eq. (8), the result for n
is (3.2 ± 0.3) electrons, so obviously 3.
−5
2
−1
◦
DU(III) = 2.2 ± 0.7 × 10 cm s at 810 C in LiF–CaF2.
In the frame of actinides/lanthanides separation and actinides
recovery by electrolytic extraction, we can compare the nor-
mal potentials of U(III)/U(0) system and previously studied
Nd(III)/Nd(0) system [2] in the same molten bath and evaluate
the capability of the U–Nd separation. The theoretical efficiency
of U–Nd separation has been evaluated and is very good [3].
Besides, our work also showed that LiF–CaF2 eutectic mix-
ture is more appropriate than LiF–NaF one to extract uranium,
because in our experimental conditions uranium metal is not
stable in LiF–NaF. An alternative route for this extraction could
consist of reducing the uranium ions as alloy on non-inert cath-
ode material in order to stabilize the metal inside the alloy.
3
.4. Uranium electrowinning
All the electrochemical results obtained with cyclic voltam-
metry, chronopotentiometry, square wave voltammetry proved
that the reduction of U(III) in molten fluorides proceeds in one
step exchanging three electrons and yields on the inert Ag cath-
ode an insoluble product which is certainly uranium metal. To
◦
confirm this, an electrolysis of UF3 in LiF–CaF2 at 810 C was
carried out during 15 min at the potential of the U(III) reduc-
Acknowledgements
Financial support for this project was funded by CEA Valrho-
Marcoule and EDF Clamart. The authors express their thanks
to Laurent Massot and to Fabienne Leguyadec and Catherine
Fucili DEN/DTE/STME/LMAC (CEA Valrho-Pierrelatte).
References
[
1] CEA, Report PG-DRRV/DIR/00-92, Assessment of Pyrochemical Pro-
cesses for Separation & Transmutation Strategies—Proposed Areas of
Research, 2000.
[
[
2] C. Hamel, P. Chamelot, P. Taxil, Electrochim. Acta 49 (2004) 4467.
3] C. Hamel, P. Chamelot, P. Taxil, A. Laplace, J. Lacquement, E. Walle,
Proceedings of the Atalante 2004, Nimes, France, 2004, p. O06.
Fig. 17. Optical photography of a cross-section of an uranium deposit on a
◦
molybdenum electrode in LiF–CaF2 obtained at 810 C by electrolysis at con-
[4] P. Chardard, Note CEA-N-2090, Etude de Quelques Propri e´ t e´ s Elec-
trochimiques de l’Uranium en Milieux Fluorures Fondus, 1979.
trolled potential (U(III)/U(0) potential) during 15 min (Q = 193C).

C. Hamel et al. / Electrochimica Acta 52 (2007) 3995–4003
4003
[
[
[
5] G. Mamantov, D.L. Manning, Anal. Chem. 38 (1966) 1494.
6] D.L. Manning, G. Mamantov, J. Electroanal. Chem. 17 (1968) 137.
7] P. Soucek, F. Lisy, R. Zvejskova, Proceeding of the Global 2003, New
Orleans, USA, 2003, p. 1582.
[20] L. Massot, P. Chamelot, F. Bouyer, P. Taxil, Electrochim. Acta 47 (2002)
1949.
[21] A.J. Bard, L.R. Faulkner, Electrochimie: Principes, M e´ thodes et Applica-
tions, Masson, 1983.
[
[
8] F.R. Clayton, G. Mamantov, J. Electrochem. Soc. 121 (1974) 86.
9] R. Cartier, M e´ moire CNAM, Propri e´ t e´ s Electrochimiques de l’Uranium,
du C e´ rium et du Zirconium Dans l’Eutectique LiF–BaF2, Paris, 1968.
[22] R.W. Laity, D.E. McIntyre, J. Am. Chem. Soc. 87 (1965) 3806.
[23] R.K. Jain, H.C. Gaur, B.J. Welch, J. Electroanal. Chem. 79 (1977) 211.
[24] T. Berzins, P. Delahay, J. Am. Chem. Soc. 75 (1953) 4205.
[25] L.E. Ramalay, M.S. Kraus, Anal. Chem. 41 (1969) 1362.
[26] J.G. Osteryoung, J.J. O’Dea, Electroanal. Chem. 14 (1986) 209.
[27] L.E. Ramalay, M.S. Kraus, Anal. Chem. 41 (1969) 1365.
[28] P. Chamelot, B. Lafage, P. Taxil, Electrochim. Acta 39 (1994) 2571.
[29] P. Chamelot, B. Lafage, P. Taxil, Electrochim. Acta 43 (1997) 607.
[30] P. Chamelot, P. Palau, L. Massot, A. Savall, P. Taxil, Electrochim. Acta 47
(2002) 3423.
[
[
10] K. Serrano, P. Taxil, J. Appl. Electrochem. 29 (1999) 497.
11] Y. Berghoute, A. Salmi, F. Lantelme, J. Electroanal. Chem. 365 (1994)
1
71.
[
12] HSC Chemistry 4.1, Chemical Reaction and Equilibrium Software with
Extensive Thermochemical Database, Outokumpu, 1974–2000.
13] Binary Alloy Phase Diagrams, S.E., ASM International, 1996.
14] R.P. Elliot, Phase Diagramm, McGraw-Hill Book Company, 1965.
15] P. Taxil, J. Mahenc, Corros. Sci. 21 (1981) 31.
[
[
[
[
[
[
[31] J.P. Young, Inorg. Chem. 6 (1967) 1486.
16] P. Taxil, J. Less-Common Met. 113 (1985) 89.
17] P. Taxil, Z.Y. Qiao, J. Appl. Electrochem. 15 (1985) 947.
18] P. Taxil, P. Chamelot, L. Massot, C. Hamel, J. Min. Metall. Sect. B: Metall.
[32] T. Store, Thesis, Norwegian University of Science and Technology, Depart-
ment of Electrochemistry, 1999, p. 22.
[33] H.B. Herman, A.J. Bard, Anal. Chem. 35 (1963) 1121.
[34] K. Serrano, P. Taxil, O. Dugne, S. Bouvet, E. Puech, J. Nucl. Mater. 282
(2000) 137.
3
9 (2003) 177.
[
19] P. Chamelot, B. Lafage, P. Taxil, Electrochim. Acta 53 (1997) 607.