ELECTROCHEMICAL BEHAVIOR OF LANTHANUM AND YTTRIUM IONS IN TWO MOLTEN CHLORIDES WITH DIFFERENT OXOACIDIC PROPERTIES : THE EUTECTIC LiCl-KCl AND THE EQUIMOLAR MIXTURE CaCl 2-NaCl

The electrochemical behavior of LaCl3 and YCl3 was studied in two molten chloride mixtures with different oxoacidic properties, the eutectic LiCl-KCl and the equimolar CaCl2-NaCl melt at different temperatures. The stable oxidation states of both elements have been found to be (III) and (0) in both melts, and it was found that both La(III) and Y(III) cations were less solvated by the chloride ions in the calcium-based melt, which was explained by the stability of CaCl4 ions in that melt. Transient electrochemical techniques, such as cyclic voltammetry, chronopotentiometry and chronoamperometry were used in order to study the reaction mechanism and the transport parameters of electroactive species at a tungsten electrode. The results showed that in the eutectic LiCl-KCl, the electrocrystallization of lanthanum and yttrium seems to be the controlling electrochemical step, while in CaCl2-NaCl this phenomenon has not been observed. That was explained in terms of the differences in the physicochemical properties of the systems, especially interfacial tensions. Castrillejo.indd 7/7/03, 12:26 PM 1 Y. Castrillejo et al J. Min. Met. 39 (1 ‡ 2) B (2003) 110 In the eutectic LiCl-KCl, chronoamperometric studies indicated instantaneous and three dimensional nucleation and crystal growth of lanthanum and yttrium whatever the applied overpotential of the rare earth metal is, whereas in the equimolar mixture CaCl2-NaCl, the corresponding electrochemical exchanges were found to be quasi-reversible, and the values of the kinetic parameters, K0 and α, were obtained for both reactions. Mass transport towards the electrode is a simple diffusion process, and the diffusion coefficients have been calculated. The validity of the Arrhenius law was also verified by plotting the variation of the logarithm of the diffusion coefficient versus 1/T.


Introduction
During the last decades, molten salts and molten chlorides particularly have extensively proved to be suitable reaction media for performing selective solubilization or precipitation in chemical reactions, and have been already proposed as promising route for the treatment of raw materials.Moreover, electrochemical processes with metals (electrowinning, electrorefining, electroplating and electroforming) in molten salts media have extensively been proved to be more advantageous than those carried out in aqueous solutions.Higher efficiency of the electrolysis, lower energy consumption, often high electrodeposition rates and much better characteristics of the deposits [1] can be pointed out as some of the main advantages.The accumulated knowledge concerning their high-temperature electrochemistry allows the deposition of metals and their alloys.Those possibilities lie in the fact that, because of their variety, one can always find a solvent whose chemical and electrochemical characteristics and melting point are suitable to carry out the given process.
In the last years a new field has been developed -the use of molten salts media for pyrochemical separation as a promising option in the nuclear fuel technology in the future [2][3][4][5][6][7][8].The interest is due to the progress in the assessment of new concepts for transmutation and the corresponding fuel cycles [9].Today's main emphasis is put on the maximal cost reduction for cycle technology.In order to assess the feasibility of such pyrochemical separation, several processes have been developed for the recovery of minor actinides from spent metallic, nitride, oxide nuclear fuels, and high level radioactive liquid wastes [10,11].
One of the most important steps in the pyrometallurgical reprocessing is the electrorefining from molten chlorides.In this step, spent metal fuel is anodically dissolved into molten chlorides, and the minor actinides are selectively recovered at the cathode due to the differences among the redox potentials of the elements, while fission products remain in the anode and/or in the electrolyte salt.The determination of thermodynamic data of solutions as well as the electrochemical behavior of the elements is of crucial importance for the understanding of the process and the design of the separation cell.
The work presented here is a part of a wider project focused on the separation of actinides from rare earths -the most difficult fission product to separate due to their similar properties -from a simulated nuclear fuel (SIMFUEL).Hence, obtaining basic data of fission products in molten halogenide salts is the major concern.The work presents a study on the electrochemical properties of two rare earths, lanthanum and yttrium, trichlorides and has been carried out in two molten chlorides with different oxoacidity properties, the eutectic LiCl-KCl and the equimolar mixture CaCl 2 -NaCl.
A few studies relating to the chemical behavior of lanthanum and yttrium in molten chlorides have been conducted.The eutectic LiCl-KCl and the equimolar mixture NaCl-KCl have been used for this kind of study [12][13][14][15][16][17][18].It was Plambeck who systematized the standard potential values for a number of rare earths in the eutectic LiCl-KCl at 723 K [12], while the temperature dependence of these values were determined experimentally in [13][14][15][16][17][18].
The fundamental studies of the mechanism of electrode reactions in molten chlorides are, as yet, sparse.In the eutectic LiCl-KCl and LiCl-KCl-NaCl the electroreduction of lanthanum and yttrium trichlorides, at low concentration at solid cathodes such as W, seems to consist of only one electrochemical step, whose potential is close to the potential of the alkali metal electrodeposition [12,17,19].The diffusion coefficients have been determined in works [17,[19][20][21], and some differences have been found.
It is known that rare earth metals are soluble in their molten chlorides [22][23][24][25] being responsible for the low current efficiency of the electrolysis, moreover, mixtures of rare earth chlorides and rare earth metals give rise to electronic conductivity, which may influence electrochemical measurements.The work of Keneshea and Cubbicciotti [25] indicates that the solubility of lanthanum in its liquid trichloride is 9% at 826 0 C and 11% at 914 0 C. Therefore, dilute solutions of RECl 3 in the LiCl-KCl or CaCl 2 -NaCl melts are expected to show very low metal solubility.
The purpose of our investigation was to determine the electrochemical behavior of lanthanum and yttrium ions in two molten chlorides with different intrinsic acidities, the eutectic LiCl-KCl melt at 723 K and the equimolar mixture CaCl 2 -NaCl at 823 K, added as lanthanum and yttrium trichloride.Tungsten wires were utilized as working electrodes.Transient electrochemical techniques, such as cyclic voltammetry, chronopotentiometry and chronoamperometry, were used in order to study the reaction mechanism and the transport parameters of electroactive species.Moreover, nucleation studies of the rare earth metals in the eutectic mixture LiCl-KCl were carried out, something that, to our knowledge, is rarely found in the literature published so far.The results show under the experimental conditions the deposition of yttrium and lanthanum at a tungsten electrode can be explained in terms of a model involving instantaneous and three-dimensional nucleation, where the growth of the crystals is controlled by hemispherical or linear diffusion over the whole cathodic potential range studied.

Preparation and purification of the melt
The experimental cell was carefully prepared.Therefore all handling of the salts was carried out in a glove box mBraun Labstar 50 under argon atmosphere.Moreover, purification of the cell was very important in order to obtain consistent results.
The chloride mixtures (CaCl 2 -NaCl or LiCl-KCl, analytical grade) were melted in alumina or glassy carbon (GC) crucibles placed in a quartz cell inside a Taner furnace.A West 3300 programmable device controlled the temperature of the furnace to a precision of ±2 o C. The working temperature was measured with a thermocouple protected by an alumina tube inserted into the melt.
The mixture was fused under vacuum, then the pressure was raised to the atmospheric value using dry argon; afterwards it was purified by bubbling HCl through the melt for at least 30 min, and then kept under argon which removed the residual HCl and maintained inert atmosphere during experiments.This procedure has been used previously [26][27][28][29][30][31].

General features
Solutions of the electroactive species were prepared by direct additions of RECl 3 .The experimental problems related to the low solubility of LaOCl and Y 2 O 3 prevented the preparation of stable solutions of RE(III) for longer periods, more than one day, and it was hard to know the exact amount of salt introduced into the melt; for that we bubbled HCl each day prior to determinations.The total concentration of dissolved rare earth was calculated by ICP-AES analysis of frequent melt samples.

Electrochemical apparatus and electrodes
Cyclic voltammetry and pulse techniques were performed with a PAR EG&G Model 273A potentiostat/galvanostat controlled by the PAR EG&G M270 software package.
The reference electrode consisted of a silver wire (1 mm diameter) dipped into a silver chloride solution (0.75 mol kg -1 ) in the CaCl 2 -NaCl or LiCl-KCl molten mixture placed in a Pyrex tube.Potentials were measured against the potential of the AgAgCl couple and translated into potentials versus Cl 2 /Cl -to make the comparison clear.The working electrode was a 1mm tungsten wire.Another tungsten wire was used as the counter electrode.The lower end of the tungsten electrode was polished thoroughly by using SiC paper.Then they were cleaned in ethanol using ultrasound followed by heating under vacuum.The active surface area of the working electrode was determined by measuring the depth of immersion.
Auxiliary techniques as SEM, X-ray diffraction analysis and ICP-MS analysis were also used.

Voltammetric characterization of the electrochemical systems
The stable oxidation states of the studied rare earth elements, lanthanum and yttrium, were identified by different electrochemical techniques, i.e. cyclic voltammetry, chronopotentiometry and square wave voltammetry.
Figures 1(a-b) show some voltammograms obtained in the eutectic LiCl-KCl melt containing RE(III) ions at a tungsten electrode.The shape of the wave A, in the potential range close to the cathodic limit of the melt (electrodeposition of liquid lithium or sodium respectively) is characteristic of the formation of a new phase, steep rise and slow decay.The anodic peak A' has the expected characteristics for a stripping peak -the decay is steeper than the rise due to the depletion of the metal deposited during the forward scan.Moreover, the ratio of the forward to reverse current peaks Ip a /Ip c is higher than unity.
In the case of the equimolar CaCl 2 -NaCl, the voltammograms recorded show similar peaks (Fig. 2(a-b)).In agreement with the standard potential values obtained by us previously [18], the peak potentials for the deposition and dissolution of the metals are shifted towards less negative values.It can be explained in terms of lower cation complexation by the bath, probably due to the formation of the CaCl 4 2-complex [33, 34], which leads to the decrease in free chloride ion concentration in the calcium melt in comparison with the molten LiCl-KCl mixture.In addition, a small oxidation wave B' (related to an inflection of the cathodic wave, B) is observed, which can be explained by the formation of a RE-Na alloy [35].
Square wave voltammograms were also obtained (Fig. 3).This technique was described in details by Osteryoung and Osteryoung [36], and Ramaley and Krasue [37].The potential-time function consists of the sum of a synchronized square wave and a staircase potential ramp.The current is sampled at the end of every half wave and then differentiated.This allows capacitive and residual currents to be eliminated and makes the method highly sensitive.For a simple reversible reaction, the net current-potential curve is bell-shaped and symmetrical about the half-wave potential, and the peak height is proportional to the concentration of the electroactive species.The width of the halfpeak, W 1/2 , depends on the number of electrons exchanged and the temperature as follows: (1) A single signal was obtained when sweeping in the negative direction, W 1/2 giving values of 3.0±0.2electrons for lanthanum and yttrium.Moreover, chronopotentiograms in both melts were obtained (see Figures 4(a-b)).The graphs indicate the existence of a potential plateau at approximately -3.3 and -3.2 V (vs.Cl 2 /Cl -).After this plateau there is a rapid decrease in the potential, and when the constant current was maintained for times longer than the transition time, the electrode potential reaches a limiting value corresponding to the deposition of the alkali metal (sodium or lithium).In the case of the calcium-based melts it is possible to observe the formation of the RE-Na alloy (zone B in the Figure 4b).
Analysis of the electrodeposits obtained by coulometry at controlled potential confirmed the formation of the metal.All these results show that the only stable oxidation states of lanthanum and yttrium in the melts studied are 0 and the dissolved La(III) and Y(III).

Results obtained by cyclic voltammetry
A typical cyclic voltammogram of LiCl-KCl-RECl 3 at a tungsten electrode with the evidence of nucleation and crystal growth are shown in Fig. 5(a-b).A "crossover" of the direct and the reverse curve is visible in these voltammograms.The deposition of

Results obtained by chronoamperometry
Chronoamperometry is a technique particularly sensitive to nucleation and growth phenomena [38][39][40][41][42].If the metal is deposited on a foreign substrate, the initial response of the current gives information about the nucleation behavior of the metal on the substrate.
I-t transients of Y(III) and La(III) for various potential pulses at a tungsten electrode are shown in Figures 6(a-b).The appearance of these curves indicated that nucleation and growth phenomena play a part in the overall deposition process.The initial regime of each transient is characterized by a decrease in current, which corresponds, after charging of the double layer, to the formation of the first nuclei.This is followed by an increase in the current associated with crystal growth on the electrode.Finally, the current decays in the usual way with time.The rising part of the current culminates in a maximum, i m , as the individual diffusion zones of the growing nuclei merge, and we can see that the higher the overpotential, the greater the value of i m .The position of this maximum on the time axis, t m , depends on the magnitude of the potential step, and decreases as the applied potential becomes more negative.
After each run the deposited metal was removed from the surface by polarizing the working electrode anodically.The most interesting part of the cathodic i-t transients is the rising portion of the curve, which corresponds to the current before overlapping of the first monolayer of the growing nuclei and therefore can be used to determine the kinetics of nuclei growth.Then, the rising part of the chronoamperograms was analyzed and compared to models developed to describe instantaneous nucleation in which all the RE germs are created at the same moment at the beginning of the electrolysis; and progressive nucleation in which new crystals are continuously created throughout the electrolysis.In order to identify the lanthanum and yttrium nucleation mode, we have analyzed: i) the relationship between current, i, and time, t; and ii) the non-dimensional plots of the chronoamperometric curves according to Scharifker et al [38].
According to the literature [38,42], the relationship between current, i, and time t, at the beginning of the potential pulse is given by an equation of the following type: i=αt x .The exponent x depends on the type of nucleation, the geometry of the nuclei, and the growth conditions.Various models with corresponding values of α and x were presented by Allongue and Souteyrand [42].
The proportionality between i and t 1/2 (Figure 7) for various overvoltages demonstrates that initial stages of the electrochemical deposition of yttrium and lanthanum at a tungsten electrode can be explained in terms of a model involving instantaneous and three-dimensional nucleation.The growth of the crystals is controlled by hemispherical or linear diffusion [42].
Scharifker and Hills [38] developed a non-dimensional model adapted for this situation.Since the entire transient curve is analyzed, the non-dimensional model was found to be more accurate.The entire dimensionless experimental current-time transients obtained at different applied cathodic potentials were compared to the appropriate theoretical transients reported by the authors for instantaneous and progressive nucleation (Equations ( 2) and (3) respectively). (2) (3) The results (see Fig. 8) confirmed the above statement: initial stages of the electrochemical deposition of lanthanum and yttrium on a tungsten electrode in the eutectic 123 LiCl-KCl mixture can be explained in terms of a model involving instantaneous nucleation with three-dimensional growth of the nuclei.Electrochemical behavior of lanthanum and yttrium ions in two molten chlorides .....

Kinetics of the electrode reaction in the equimolar CaCl 2 -NaCl. Reversibility studies
The characteristic crossover in the cathodic branch, whose presence indicates a large overpotential, was never observed in La and Y voltammograms in the calcium-based melt.Overall, the voltammetric peaks of the deposition and striping of these metals at tungsten are unremarkable in appearance.However, the voltammetric curves recorded at different sweep rates (Fig. 9(a-b)) clearly show that the peak potential Ep c shifts negatively, and the peak potential difference Ep c -E p/2 increases when the sweep rate is increased.This fact suggests under these conditions and after correction of the ohmic drop, the electron transfer rate is significantly lower than that of the mass transport [43][44][45].
An alternative approach to the interpretation of the mechanism and the estimation of kinetic parameters is the convolutional analysis of the voltammetric curves [46][47][48][49][50]. Figure 10 shows some voltammograms of RE(III) reduction and their corresponding semi-integral curves.Analyzing the convoluted curves, we can observe that the direct and reverse scans are not identical and hysteresis behavior occurs between the upper and downer sweeps.All these characteristics support the fact that the reduction of La(III) and Y(III) at a W electrode is not reversible.
Logarithmic analysis of the convoluted curves, according to the model of a quasireversible exchange with the formation of an insoluble product, was carried out applying the equation [51][52][53]: (4) (5) where m is the convoluted current, m * its limiting value, A the active surface area of the electrode, and E O 1 the standard potential of the RE(III)/RE(0) couple previously obtained by us by potentiometry [18].
Plots of E vs. log B (see Fig. 10) were a straight line, from whose slope and intercept one can extract the value of the transfer coefficient, α, and the charge-transfer rate constant, k o , respectively.The average values obtained in this way are given in Table 1.
Taking into account the Matsuda and Ayabe criteria [54], in which the charge-transfer rate constant and the sweep rate are related, it can be confirmed that the La(III)/La(0) and Y(III)/Y(0) reactions are quasi-reversible at a tungsten electrode.Rare earth log k o , cm s -1 α La -3.9 ± 0.1 0.48 ± 0.03 Y -3.7 ± 0.2 0.39 ± 0.05 Under the given experimental conditions (diluted RE(III) solutions), we have not observed by the I-t current transients that the nucleation and growth phenomena step controls the overall RE electrodeposition process in this melt.Thus, one can say that it is probably the charge transfer step, which controls the process, rather than nucleation and growth.

Determination of the RE(III) diffusion coefficient. Verification of the Arrhenius law
The diffusion coefficients, D RE(III) , given in Table 2, were calculated from different electrochemical techniques by applying the appropriate equations.It has to be indicated that, although a tungsten wire was used as working electrodes, all the formulas used are relevant to plane semi-infinite diffusion because, under the experimental conditions, the corrections related to cylindrical geometry can be neglected [55][56][57][58].
Chronopotentiometric studies of the reduction of RE(III) ions obeyed the Sand's law [44]: (6)  Transition times for several current densities were measured and the resulting I versus 1/τ 1/2 plot yielded a straight line, indicating that the fluxes of RE(III) species were diffusion controlled (see Fig. 11).From the slope of the plot it was possible to determine the diffusion coefficient of RE(III) ions.
The convoluted curves of the voltammograms exhibited relatively well defined limiting currents (see Figs. 10 and 12), indicating that there is no gross change in the electrode surface area during the scan, to the point that it affects the limiting current for the RE(III) reduction wave.Then the RE(III) diffusion coefficient was computed from the boundary semi-integral values by means of the relationship [44]: It could also be interesting to study the variation of the transport properties of the RE(III) ions in the melts with the temperature.Chronopotentiometric curves of RE(III) solutions were obtained when the working temperature was varied from 673 to 823 K and from 823 to 923 K in the eutectic LiCl-KCl melt and the equimolar CaCl 2 -NaCl respectively.The diffusion coefficients were determined from the chronopotentiometric curves by applying Eq. ( 6).Fig. 13 shows the variation of the logarithm of the diffusion coefficient versus 1/T.Straight lines obtained show the validity of the Arrhenius law.From the equations of the plots obtained (Table 3) the temperature dependence of the diffusion coefficient was established, and the activation energy value for the diffusion process could be extracted.
Although, as stated above, RE(III) cations are less solvated by the chloride ions in the calcium-based melt, the diffusion coefficient estimated shows that RE(III) diffuse more slowly in the equimolar CaCl 2 -NaCl than in the eutectic LiCl-KCl.This behavior can be explained by different viscosity of the two media, which significantly reduce the mobility of RE(III) in the CaCl 2 -NaCl melt.

Conclusions
The electrochemical properties of lanthanum and yttrium have been studied using tungsten as working electrode in two molten chloride mixtures with different oxoacidic properties -the eutectic LiCl-KCl at 723 K and the equimolar CaCl 2 -NaCl melt at 823 K -founding some differences in behavior of the two chloride mixtures.
The cathodic peak potential values of the RE(III)/RE electrochemical exchange were less negative in the case of the calcium-based melt, which could be explained by a lower cation complexation in such melts due to the formation of the CaCl 4 2-complex, which leads to the decrease in free chloride ions concentration than in the case of the molten eutectic LiCl-KCl mixture.The standard potential values of the different electrochemical exchanges and the activity coefficient values of the RECl 3 species obtained in both melts and reported elsewhere also support this fact.
By combining different electrochemical techniques (i.e.voltammetry, chronopotentiometry and square wave voltammetry), it was possible to determine the stable oxidation states of lanthanum and yttrium in the molten chloride mixtures.Then, the stability of the rare earth metals was demonstrated in both melts, and their solubility in the molten chloride containing RE(III) ions was considered to be negligible.No UPD of metallic lanthanum and yttrium on tungsten substrates was found, and possible formation of RE-Na alloys was evidenced in the calcium-based melts.
Voltammetric and chronoamperometric techniques showed that the nucleation and growth of the metallic lanthanum and yttrium deposit show an important role in the overall electrodeposition process in the case of the molten eutectic LiCl-KCl mixture.The analysis of the transient curves (both of the rising part and the complete curve according to a non-dimensional model) showed that the initial stages of electrochemical deposition of lanthanum and yttrium at a tungsten electrode can be explained in terms of a model involving instantaneous nucleation (i.e.all the nuclei are formed "immediately" after applying the potential step) with three-dimensional growth of the nuclei.Identical results were found when varying the working temperature from 673 to 823 K.
However, in the case of the equimolar CaCl 2 -NaCl melt at 823 K it was found that it is the charge transfer step, which controls the electrodeposition process, rather than nucleation and growth.The values of the charge transfer constant, k o , calculated by logarithmic analysis of the convoluted voltammetric curves, showed a quasi-reversible behavior of the systems.
Moreover, the diffusion coefficients of the RE(III) ions were obtained by different electrochemical techniques (i.e.convolution and chronopotentiometry), and they showed a temperature dependence according to the Arrhenius law in both melts.It could be expected that some of the values obtained are somehow influenced by the alkali metal co-deposition, especially in the CaCl 2 -NaCl melt.
In addition, the obtained diffusion coefficient values show that, although RE(III) cations are less solvated by the chloride ions in the CaCl 2 -NaCl melt, they are found to diffuse slower than in the eutectic LiCl-KCl at 723 K.That is probably due to higher viscosity of the calcium-based melt, which significantly reduce the mobility of RE(III) chlorocomplexes.

Figure 3 .
Figure 3. a) Net-current square wave voltammogram for the reduction of Y(III) at a tungsten electrode in the eutectic LiCl-KCl melt at 723 K; pulse height: 25 mV; potential step: 1 mV; frequency: 50 Hz; A=0.44 cm 2 ; b) Net-current square wave voltammogram for the reduction of La(III) at a tungsten electrode in the eutectic CaCl 2 -NaCl melt at 823 K; pulse height: 25 mV; potential step: 1 mV; frequency: 30 Hz; A=0.23 cm 2 .

Figure 5
Figure 5 Cyclic voltammograms illustrating the "nucleation crossover effect" on the return sweep for the deposition of: (a) lanthanum onto tungsten electrode; (b) yttrium onto tungsten electrode.

Table 1 .
-Values of the kinetic parameters, k o and α, corresponding to the La(III)/La(0) and Y(III)/Y(0) electrochemical systems in the equimolar mixture CaCl 2 -NaCl at 823K

Table 2 .
-Diffusion coefficients of La(III) and Y(III) calculated by different electro chemical techniques

Table 3 .
Variation of the diffusion coefficient of La(III) and Y(III) ions with the temperature and activation energy of the diffusion process in the eutectic LiCl-KCl