Вы находитесь на странице: 1из 7

JOURNAL OF RARE EARTHS, Vol. 28, No. 3, Jun. 2010, p.


Electrochemical studies on cerium(III) in molten fluoride mixtures

(Romanian Academy, Institute of Physical Chemistry Ilie Murgulescu, Department of Molten Salts Splaiul Independentei 202, Bucharest-Romania) Received 22 September 2009; revised 9 March 2010

Abstract: This study aims to determine the principal electrochemical characteristics of the electrodeposition of cerium metal from molten fluoride systems. The cathodic process of Ce3+ ions in LiF-NaF and LiF-NaF-CaF2 molten salts was studied using electrochemical techniques as steady state and cyclic voltammetry methods. The decomposition potential (Ed) and the overvoltage( ) were determined for NaCeF4 using current-potential curves under galvanostatic conditions. The Ed was found to be 2.025 V in LiF-NaF and 2.045 V in LiF-NaF-CaF2. It was also found that the ohmic drop potential (E ) was not dependent on NaCeF4 concentration and it rose as the current intensity increased. The overvoltage ( ) was determined from the polarization curves and the Tafel coefficients and kinetic parameters were calculated on the assumption that the process constitutes of direct discharge of Ce3+, with no solvent-solute interaction. In order to elucidate the cathodic process the investigation by cyclic voltammetry technique was finally used. From the evolution of the voltammograms we concluded that the electrochemical reduction of Ce3+ ion was actually a reversible process on the molybdenum electrode and cathodic reduction of Ce3+ took place in one single step involving three electron exchanges. Keywords: cerium; molten salts; lanthanides; NaCeF4; electrolysis; cyclic voltammetry

Cerium is the most abundant and the most widely used one of the rare earth metals (RE), making up about 0.0046% of the Earth's crust by weight. It is found in a number of minerals and its commercial applications are numerous. They include catalysts, additives to fuel to reduce emissions and to glass and enamels to change their color. The relatively high prices of the RE elements results not from their rarity but from the extreme difficulty of extracting them from their ores and, more significantly, purifying them. From the late 1950s a considerable volume of work has been expended on the issue of obtaining higher purity lanthanide metals. Molten salts have been used in industrial metallurgical processes for many years and in the last ~15 years many attempt was made in order to obtain purity refractory and rare earth metal by electrolysis in molten salts[1]. The most efficient method for the production of high purity rare earth metals was found to be the electrowinning in molten salts and there are many reports on the electrochemical studies on rare earth ions and some of them are presented in the references[27]. A molten fluoride bath has a great advantage compared with the chloride bath as high purity metal is obtained with improved current efficiency[8]. However, the molten fluoride bath has the disadvantage of higher melting point and aggressive corrosion. Cerium is the most abundant one of the rare earth metals. Rare earth metals are the elements in row 6 of the periodic table. The rare earth elements are not really rare. In fact, cerium ranks about number 26 in abundance
Foundation item: Project supported by the a NATO grant (nr.004-005/2002)

among elements found in the Earth's crust. Cerium was the first rare earth element to be discovered and is an iron-gray metal with a melting point of 1068.15 K that easily reacts with other elements. This study is a part of a systematic investigation on the alternative method for obtaining metallic cerium by using molten fluorides (in this case eutectic compositions of LiF-NaF and LiF-NaF-CaF2) as a solvent while the solute was the complex of cerium(III) fluoride and sodium fluoride, NaCeF4 (50 mol.% NaF and 50 mol.% CeF3, Te=999.15 K). Previous papers described the synthesis of this complex and its crystallographic parameters[9,10]. The purpose of this study was to determine the principal electrochemical characteristics of the electrodeposition of cerium metal from molten fluoride systems.

1 Experimental
1.1 Reagents The used fluorides (NaF, LiF, CaF2 and CeF3) were p.a. grade reagents from Merck and Flucka. Every fluorinated compound was dehydrated under vacuum heating. The NaCeF4 solute was selected from all the compounds formed by CeF3 with alkali fluorides because it was easily synthesized and was not as hygroscopic as the compounds formed with potassium. 1.2 Apparatus and procedure

Corresponding author: VIRGIL CONSTANTIN (E-mail: virgilconstantin@yahoo.com; Tel.: +40-0213121147) DOI: 10.1016/S1002-0721(09)60127-7

VIRGIL CONSTANTIN et al., Electrochemical studies on cerium(III) in molten fluoride mixtures


The experimental setup used in our electrochemical studies consists of three parts: reaction system (electrical furnace-electrochemical cell), atmosphere control (vacuum and argon gas flushing system) and electronic equipment for electrochemical investigation. The details for the assembly and working procedure were described previously[11,12]. The overpotential was determined by current-potential measurements in a stationary regime under galvanostatic conditions. Variable current steps between 2050 mA were applied at a rate of 23 steps/min. These steps were higher than those usually applied for this type of determinations as NaCeF4 electrolysis proceeds of very high current densities and potentials (specific for lanthanide compound electrolysis). The cyclic voltammetry was performed by using a VoltaLab 050-Radiometer potentiostat with Volta Master-4 packaged software. The potential scan rate was between 0.050.3 V/s. A SU-2000 type glassy carbon crucible served both as a container for the melt and counter electrode. This was placed in a hermetically sealed retort which could be filled with purified argon (produced by Air Liquid-France with impurities 1 ppm). The melt temperature was maintained 2 C with an Alcatel thermoregulator. The voltammograms were registered on a Mo electrode ( =2 mm) by a Pt quasi-reference electrode (QRE). In order to accurately delimit the surface area exposed to the melt the cathode was fastened in a boron nitride sheath.

of NaCeF4 in both electrolytes was determined. Figs. 1 and 2 represent the current-potential curves in the systems LiF-NaF-5% NaCeF4 and LiF-NaF-CaF2-5% NaCeF4 when working temperature was 1003.15 K; the shapes of the U-I and E-I curves are similar in both systems. The resistance of the electrolyte was measured with a B641Wayne-Kerr auto-balance bridge (with a 0.1% precision) and found to be 0.28 for the first solvent and 0.36 for the second one, respectively. Starting from the classical relation: E=UIR (1) where: E(V) is counter electro-motive force, U(V) cell voltage, I(A) current intensity and R( ) cell resistance, current-voltage curves (I-U) were transformed into I-E curves. The decomposition potential (Ed) was evaluated from the extrapolation to zero current of the linear parts of these plots. The mean value of the decomposition potential was found to be Ed=2.025 V for the system LiF-NaF-5% NaCeF4 and Ed=2.045 V for LiF-NaF-CaF2-5% NaCeF4, respectively where Ed is an absolute value. The Ed was also determined for 10%NaCeF4 in both studied electrolytes and, as we expected, concentration of the cerium complex does not influence the value of the decomposition potential. Some complementary information was obtained from the ohmic drop and polarization curves. From the evolution E =f(I), presented in Fig. 3, where E represents the sum of the values E in the electrolytes with 5% and 10% NaCeF4, it was found that the ohmic drop potential (E =IR) is not

2 Results and discussion

The electrolytes were selected to have low eutectic temperature and good conductivity. The physico-chemical and thermodynamic properties of the electrolytes selected for this study are presented in Tables 1 and 2[13,14]. 2.1 Decomposition potential and ohmic drop In order to know if the chosen electrolytes (LiF-NaF and LiF-NaF-CaF2) are adequate for the electrodeposition of the electroactive species (NaCeF4), the decomposition potential
Table 1 Thermodynamic properties of the employed fluorides[13]
Compound (solid state) LiF NaF CaF2 CeF3 Gf, 1000/ (kJ/mol ) 523 468.608 1058.552 1443.48 Gf,298/ ( kJ/mol ) 585.76 543.92 1163.152 1615.024 E0298 [Me F2]/ V 6.06 5.60 6.03 5.58

Fig. 1 Experimental Ed for LiF-NaF-5% NaCeF4 system

Table 2 Physico-chemical constants of the solvents

No. 1 2 System LiF-NaF LiF-NaF-CaF2 Composition/ mol.% 49.250.8 Te/ K 923 d/

K/ 494 508 2.90 1.98

Ed*/ 3.97 4.02

(kg/m ) ( /m) (mPas) V 1980 3730

28.964.56.6 888

*Ed of those systems was additively calculated from individual decomposition potentials of the components[12]

Fig. 2 Experimental Ed for system LiF-NaF-CaF2-5% NaCeF4 system


JOURNAL OF RARE EARTHS, Vol. 28, No. 3, Jun. 2010

dependent on NaCeF4 concentration and it rises as the current intensity increases. On the other hand because E is also determined by a direct measurement of the electrolyte resistance, it is obvious that the evolution in Fig. 3 is conditioned only by the electrolyte resistance variation. One can see that E increases faster with current in LiF-NaF-CaF2 solvent where the energy losses are subsequently higher. A comparison between anode-cathode polarization curves in the studied systems shows that the electrodes reach faster a stationary regime and intensities rise appreciably only in LiF-NaF solvent, while a stationary regime is not practically reached in LiF-NaF-CaF2. Comparing the experimental decomposition data with deposition data from Tables 1 and 2 and with evolution of the ohmic drops, it is obvious that cerium deposition by dissolution of NaCeF4 in the two chosen fluoride solvents can be done as there is no interference between solute and solvent during electrolysis. All those data are important for the cathodic overvoltage calculation. 2.2 Overpotential A study of the electrode processes is useful as it provides information on the sequence of partial electrode reactions as well as electrode reaction rates which, according to Faradays law, are proportional to current density (i). The dependence of current density on both potential R and concentration allows for an evolution of the following important parameters: exchange current (i0), transfer coefficients ( ) electrochemical reaction orders, limiting reaction currents, valence of the electrode reaction (n), etc. The potential of an electrode (Ep) through which current flows differs from the equilibrium potential (Erev) established when no current passes through the electrode. The difference between these potentials was defined as the overpotential[15]: (2) c=EpErev

The overpotential offers direct information on several of the above mentioned parameters. Therefore, electrode kinetic studies consist always in establishing the relationship between electrode overpotential ( ) and electrochemical reaction rate expressed as current density (i), that is, the =f(i) dependence. The study of the cathodic overvoltage at the electrolysis of NaCeF4 dissolved in the fluoride mixture electrolytes above presented aims at gathering some information about the mechanism of cerium electrocrystalization process. Fig. 4 presents the evolution of the polarization potential (Ep) with the current density (i) after the correction with the E for the studied systems. The cathode polarization curves Ep=f(i) for the system LiF-NaF-NaCeF4 indicate the rapid increase in the first part up to a value of ~2.0 V, after which the slope increases slowly till it reaches a flat region. As for the second system, LiF-NaF-CaF2-NaCeF4, the polarization curves have a slowly ascending line within the 01.5 V interval and then has an ascendant evolution with a constant increase. This representation suggests a little too high energetic consumption in the second system in comparison with the first one. In order to obtain the cathodic overvoltage ( c) in the studied systems, the polarization data were improved with the value of the equilibrium potential at zero current. The typical =f(i) curves depicted in Fig. 5 as a semilogarithmic dependence vs. logI are based on Tafel equation: =a+b lgi (3)

Fig. 4 Cathodic polarization curves for LiF-NaF-NaCeF4 and LiFNaF-CaF2-NaCeF4

Fig. 3 Ohmic drop as a function of current obtained for NaCeF4 in both selected solvents

Fig. 5 Cathodic overvoltage vs. current density (logarithmic scale) and the correspondent Tafel lines obtained at 1003 K in molten systems with 5% NaCeF4

VIRGIL CONSTANTIN et al., Electrochemical studies on cerium(III) in molten fluoride mixtures


The relationship between current density and chargetransfer overpotential can be expressed in terms of two quantities, i.e., the exchange current density (io) and transfer coefficient ( ). Table 3 presents the kinetic parameter (io), Tafel coefficients (a and b, calculated from the Tafel line as the intersection and respectively slope) and Tafel range of validity for the Tafel equation ( =a+blgi) as obtained from the Tafel plots.
Table 3 Kinetic parameters, Tafel coefficients and Tafel ranges
System LiF-NaF-NaCeF4 a/ V 1.1578 1.3410 LiF-NaF-CaF2-NaCeF4 0.6314 0.7038 b/ V 0.4877 0.700 0.1943 0.4040 io/ (A/cm2) 1.5422 2.8773 1.2337 2.4360 Tafel range/ (A/cm2) 0.313.5 4.7747.13 0.311.91 1.9136.3

The evolution of the Tafel plots for both studied systems have the same evolution, but for LiF-NaF-CaF2-NaCeF4 the obtained values are more favorable. One can notice that Ce electrodeposition on a Mo cathode occurs in both studied systems at a high cathodic overpotential when potentials and current densities also reach high values, a fact usually encountered in lanthanide electrowinning. Moreover, the exchange currents, i0, are high and Tafel behavior evidences the presence of two linear sections with different slopes due to various causes such as change of the rate determining step, reaction mechanisms or oxidic state of the electrode surface. In order to discuss all these aspects we start from the specificity of the metal electrodeposition processes which differ from most other electrode processes by the occurrence of a new solid phase at the electrode-electrolyte interface due to a complex series of either successive or simultaneous steps[16]. According to all these steps, two successive electrochemical phases are evidenced. The first important for metal deposition, one referees to the rate of change transfer reaction, while the second to the crystal growth. Electrodeposition process thus transferring the load will be the first stage in which kinetic ion complex, such as ion (CeF4) in this study, electrical double layer loses its shell ion (F) and passes as decomplexed ion on the electrode surface (in the form of adion), where by continuing the charge transfer reaction turns into an adatom. During this step, charge transfer is rate determining and the overpotential corresponding to the process is the charge transfer one ( t). Subsequently, the second kinetic step consists in including the adatom within the crystalline network and a crystal growth. This process is accompanied by crystallization overpotential ( cr) which is determined by a delay of the atom inclusion into the ordered network of the solid metal electrode. Taking into account Butler-Volmer equation and Vetter theory[14] and coming back to Tafel curve evaluation in Fig. 5, it becomes obvious that the two different slopes are

caused by a change of the rate determining step along with the increase of current density. As is well known that the electrolysis of the rare earth is performed at high voltage and current density[17,18] and taking into account the above obtained data, we can conclude that the obtained Tafel plots suggest these particularities: - both systems have two straight lines with different slopes; this evolution indicates the change of the rate determining step once the current density is increased; - when the C 150 mV, the rate determining step is controlled by the charge transfer. It is important to mention that the cathodic area exposed to the melt was 0.0314 cm2, which explains once again high current densities obtained by us. Moreover within this domain, Tafel slope values (b= 2.303RT/nF) calculated for n=3 lead to a theoretical value of 0.437 V under the specific conditions of our experiments, which is in a quite good agreement with the experimental b=0.479 obtained for LiF-NaF-CaF2-NaCeF4 system. The transfer overpotential contribution to the total cathodic overpotential and mechanism of a direct discharge of Ce3+ ions by a three electron exchange process are once again confirmed. - When the overvoltage has a high value ( 0.600 V), the rate-determining step is controlled by the nucleation; - Residual overvoltage has a very low value and is placed in the domain of the highest value of current densities; this overvoltage represents the diffusive component of the total cathodic overvoltage, which is controlled by the mass transfer from the inside of the electrolyte towards the cathode surface. So, we can say that the cathodic overvoltage at electrodeposition of metallic cerium is composed of three types of overvoltages (charge transfer, crystallization and diffusion) and also that total cathodic reaction is composed by the following partial reactions: charge transfer, crystallization and mass transport. From the data obtained so far, one can state that NaCeF4 electrolysis in the studied fluorinated electrolytes can be conducted in promising conditions, that is: - without solvent-solute interactions because EdNaCeF4 values are much lower than those for solvent; - it is done by direct discharge of Ce3+ ions on the cathode surface; - crystallization overpotential is decided by the nucleation rate which forms microcrystalline powder deposits; - the process proceeds with a reduced energy consumption because of the low temperature electrolysis of 1003 K (approximately 423 K lower than classical procedures). 2.3 Cyclic voltammetry In order to elucidate the cathodic process the investigation by cyclic voltammetry technique of the electrodeposition of metallic lanthanides (in particular cerium) is very important.


JOURNAL OF RARE EARTHS, Vol. 28, No. 3, Jun. 2010

Cerium is one of the elements whose electrochemistry is not yet clear due to a very high reduction potential. Herein we studied the electrochemical behavior of trivalent cerium in LiF-NaF and LiF-NaF-CaF2. Cerium is the second most active lanthanide after europium. Since at elevated temperature europium interacts with traditional electrode materials yielding alloys/carbides, a negative opinion about platinum, copper, silver, glassy carbon working electrodes is expressed[16]. It is obvious that cerium will have the same behavior and so, in order to avoid mistakes in electrode reaction interpretation, molybdenum, which does not form intermetallic compounds and has a low interdiffusion coefficient (with europium[19,20]), was used for this part of the work. The cyclic voltammetric curves in LiF-NaF and LiF-NaFCaF2 melts obtained with Mo cathode and vs. a Pt (quasi- reference electrode, q.r.e) electrode are presented in Fig. 6, where typical voltammograms for the Ce3+ reduction are depicted. This shape of the voltammogram is typical for the product adherent to the cathode, which was also the case of Ce. There is no clear peak of cerium reduction on the ascending section of the cathodic part of voltammograms. Cerium discharges at high negative potentials together with alkali metal ions, which is in agreement with the behavior found for europium[21], but sharp anodic peak of Ce dissolution is observed. That peak grows up when cerium concentration is higher in the electrolyte. The cyclic voltammetric curves in LiF-NaF and LiF-NaF-

CaF2 melts obtained with Mo cathode and vs. a Pt (q.r.e) electrode are presented in Figs. 7 and 8, where typical voltammograms for the Ce3+ reduction are depicted. Those figures show the evolution of the obtained voltammograms with the scan rates and potential variation. This shape of the voltammogram is typical for the product adherent to the cathode, which was also the case of Ce. Cathodic current started flowing at 2 V versus platinum, which corresponded to the cerium electrodeposition and the anodic peak, and corresponded to the dissolution of cerium. However, the top of the reduction wave of cerium ion was not clearly observed.

Fig. 6 Cyclic voltammograms of cerium in molten LiF-NaF-NaCeF4 and LiF-NaF-CaF2-NaCeF4 (c(NaCeF4)=1 wt.%; WE=Mo, RE=Pt (quasi reference), AE=GC, T=1003.15 K, v=100 mV/s)

Fig. 7 Cyclic voltammograms for LiF-NaF-NaCeF4 system; influence of scan rates (a) and potential variation (b)

Fig. 8 Cyclic voltammograms for LiF-NaF-CaF2-NaCeF4 system; influence of scan rates (a) and potential variation (b)

VIRGIL CONSTANTIN et al., Electrochemical studies on cerium(III) in molten fluoride mixtures


From the cyclic voltammograms with different scan rates it was found that both cathodic and anodic peak potentials hardly change with the scanning rate and potential variation. It means that the electrochemical reduction of Ce3+ ion is actually a reversible process on the molybdenum electrode. Also, the ic/ia~1 ratio not depending on the scan rate and cathodic reduction of Ce3+ take place in one single step involving three electron exchanges. Taking into account all those obtained data we propose for the fluorinated systems containing (CeF4) complex ions, that reduction of Ce3+ occurs according to one stage mechanism with a change of three electrons: NaCeF4 Na++CeF4 (3) CeF4 Ce3++4F (4) Ce3++3e Ce0 (5) 4F 4e 2F2 (6) The emission of fluorine ions at the anode opens the perspective for employing this procedure in metallic cerium electrorefining process. However, we consider that a more complex electrochemical investigation is needed for elucidating completely this process.

is also grateful to Prof.dr.Stefania Zuca for the fructuous discussions and help in developing the study of cerium electrodeposition.

[1] Kuznetsov S A. Some aspects of electrochemical behaviour of refractory metal complexes, in molten salts: from fundamentals to applications, ed. Gaune-Escard M, Kluver Academic Publ., Nederland, 2002. 283. [2] Kaneko A, Yamamoto Y, Okada C. Electrochemistry of rare earth fluoride molten salts. J. Alloys Compd., 1993, 193: 44. [3] Kononov A, Kuznetsov S, Polyakov E. Electrorefining in molten salts-an effective method of high purity tantalum, hafnium and scandium metal production. J. Alloys Compd., 1995, 218: 173. [4] Hamela C, Chamelot P, Taxil P. Neodymium (III) cathodic processes in molten fluorides. Electrochim. Acta, 2004, 49: 4467. [5] Massot L, Chamelot P, Taxil P. Cathodic behaviour of samarium(III) in LiF-CaF2 media on molybdenum and nickel electrodes. Electrochim. Acta, 2005, 50: 5510. [6] Gibilaroa M, Massota L, Chamelot P, Cassayrea L, Taxil P. Electrochemical extraction of europium from molten fluoride media. Electrochim. Acta, 2009, 55: 281. [7] Gao F, Wang Ch, Liu L, Guo J, Chang S, Chang Li, Li R, Ouyang Y. Electrode process of La(III) in molten LiCl-KCl. J. Rare Earth, 2009, 27: 986. [8] Constantin V, Popescu A M, Zuca S, Gaune-Escard M, Olteanu M. The cathodic behavior of cerium compound in fluoro-chloro melt. Ann. West Univ. Timisoara, Series Chemistry, 2003, 12: 649. [9] Constantin V, Popescu A M. Tipes of complexes formed in the binary molten systems Alk-LnF3 (in Romanian). Rev. Chim. (Bucuresti), 2000, 51: 751. [10] Popescu A M, Constantin V, Birjega R. Binary systems NaFLnF3 and their molten complexes (in Romanian). Rev. Chim. (Bucuresti), 2000, 51: 428. [11] Constantin V, Popescu A M, Zuca S. Preliminary studies of the obtaining solid metallic cerium from fluoride melts. Z. Naturforsch., 2003, 58a: 57. [12] Popescu A M, Constantin V. The decomposition potential in molten salts-III. NaCeF4 in molten fluorides and chlorides (in Romanian). Rev. Chim.(Bucuresti), 2007, 58: 1198. [13] Janz G J. Molten Salts Handbook, Academic Press, New York, 1967. [14] Ignatiev V, Zakirov R, Grebenkine K. Molten salts as possible fuel fluids for TRU fuelled systems: ISTC#1606 Approach, Proc. of the Global 99 International Conference, Jackson Hole, USA, 1999. 841. [15] Vetter K J. Electrochemical Kinetics-Theoretical and Experimental Aspects, Academic Press, New York, 1967. [16] Constantin V. Electrochemistry of some transitional metals in molten electrolytes, PhD thesis, Bucharest, Romania, 2002. 168.

3 Conclusions
Decomposition potentials for NaCeF4 in the systems LiF-NaF and LiF-NaF-CaF2 at 1003K were found to be Ed = 2.025 V and Ed =2.045 V, respectively. The E f (NaCeF4) proved that no solvent-solute interaction place during electrolysis. The polarization curves showed that the electrodes became rapidly stationary and the current intensity rose noticeable, but for the system Lif-NaF-CaF2- NaCeF4 the stationary stage was not reached. The cathodic overvoltage was high and Tafel ranged large in the case of the studied process: for n=3 (electrons) and using the classically formula for calculating the b, (b=2.303RT/nF), the experimental values of b were in good agreement with the calculated ones. The Tafel equation was verified on a large current density range; applying the Butler-Volmer equation one could conclude that the two different slopes at the Tafel plots were caused by the charge-transfer overvoltage of ( t) for low and by the crystallization overvoltage ( cr) at high . The major influence was of the cr controlled by the nucleation rate so that Ce will be deposited as a microcrystalline powder. The shape of the voltammograms was characteristic for formation of the product that remains adherent to the electrode, confirmed by the examination of the Mo electrode. The cathodic reduction of Ce3+ occurred in one single step involving three electron exchanges and metallic cerium was deposited as fine powder.
Acknowledgements: The authors A.M.P. and V.C. are grateful for support of the voltammetric study under a NATO grant (nr.004-005/2002) and express all their gratitude to Dr. Marcelle Gaune-Escard and the Polytechnical University of MarseillesI.U.S.T.I. for the wonderful working conditions. The author V.C.

434 [17] Takamaru K. Process for producing rare earth metals. USA Patent, 5.932.084/1999. [18] Speeding F H, Daane A H. The Rare Earth, John Willey & Sons Inc., New York, 1961. [19] Kuznetsov S, Rycerz L, Gaune-Escard M. Electrochemical and thermodynamic properties of EuCl3 and EuCl2 in an equimolar NaCl-KCl Melt. Z. Naturforsch., 2001, 56: 741.

JOURNAL OF RARE EARTHS, Vol. 28, No. 3, Jun. 2010 [20] Massot L, Chamelot P, Cassayre L, Taxil P. Electrochemical study of the Eu(III)/Eu(II) system in molten fluoride media. Electrochim. Acta, 2009, 54: 6361. [21] Bukatova G A, Kuznetsov S A, Gaune-Escard M. The electrochemical synthesis of europium boride. J. Mining & Metall., 2003, 39: 251.