Electrolytic coating of titanium onto iron and nickel electrodes in the molten LiF + NaF + KF eutectic

Electrolytic coating of titanium onto iron and nickel electrodes in the molten LiF + NaF + KF eutectic

125 J. Elecrroanal. Chem., 230 (1987) 125-141 Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands ELECTROLYTIC COATING OF TlTANIUM ONTO IR...

2MB Sizes 2 Downloads 83 Views

125

J. Elecrroanal. Chem., 230 (1987) 125-141 Elsevier Sequoia S.A., Lausanne - Printed

in The Netherlands

ELECTROLYTIC COATING OF TlTANIUM ONTO IRON AND NICKEL ELECTRODES IN THE MOLTEN LiF + NaF + KF EUTEmC

A. ROBIN,

J. DE LEPINAY

and M.J. BARBIER

Centre de Recherche en Electrochimie Minkale, E.N.S.E. E.G.-Inshtut Domaine Universitaire, BP 75, 38402 Saint Martin &H&es (France) (Received

18th March

1986; in revised form 4th March

National Polytechnrque de Grenoble,

1987)

ABSTRACT

The electrolytic reduction of Ti(II1) species in LiF+NaF+ KF eutectic on iron and nickel, studied using voltammetric techniques, proceeds reversibly in the 600-700 o C temperature range. No interaction occurs between the iron electrode and the titanium coating, whereas the solubility of titanium in nickel has been proved. Chronoamperometric measurements at constant potentials more positive than the Ti(III)/Ti equilibrium potential indicate that the electrochemical incorporation of titanium in nickel is controlled by intermetallic diffusion. Coatings of pure titanium 20 to 30 pm thick were produced on iron at 700 o C and on nickel at 600 and 700 Q C; they were homogeneous and well-crystallized. X-ray fluorescence analysis showed that the pure titanium coating is joined to the nickel supporting metal by a multilayer zone consisting of Ti,Ni, TiNi and TiNi, definite compounds.

INTRODUCTION

Some attempts to obtain titanium electroplates have been carried out in molten halide media. Thus, Novitskaya et al. [l] used alkali metal chlorides for titanium coatings on steel. Nevertheless, the low solubility of TiCl, [2] and the existence of different low-stability oxidation states [3,4] prevent a precise control of the conditions of the final reduction stage yielding metallic titanium; the coatings are consequently either powder or dendritic. Fluoride melts and mixed fluoride + chloride melts, favourable to the elaboration of coherent coatings of refractory metals in most cases [5,6], seem to ensure better results. Indeed, the complexation of the tetravalent and trivalent titanium ions by fluorides allows the use of concentrated solutions and eliminates the stability domain of the divalent state, which is responsible, in chloride media, for dismuta0022-0728/87/$03.50

0 1987 Elsevier Sequoia

S.A

126

tion equilibria leading to titanium powder formation. Then, the reduction mechanism becomes limited to two reversible steps [7-91: TiFz- + e- + TiFb3and TiF:-

+ 3 e- + Ti + 6 F-

which are well separated owing to the high stability of the TiF:- ions. Besides, fluoride melts, which are good solvents of oxides, may provide an initial cleaning of the electrodes, improving the adherence and the uniformity of the electrolytic coatings. Earlier, titanium electroplating was carried out by Stetson [lo] and Matiasovsky et al. [ll] onto iron, steel and copper in K,TiF, + NaF melts in the 800-1000°C temperature range. More recently, in chloride + fluoride melts, Novitskaya et al. [l] obtained, on iron, coatings consisting of two compounds, Fe,Ti and FeTi. A Texas Instruments [12] patent claims 75 pm thick titanium coatings on iron, nickel and copper, in the LiF + NaF + KF eutectic at 700” C. For the same medium at 700 and 800 o C, a patent has been applied for by Kline [13] on titanium-clad copper electrodes involving the formation of interdiffusion layers. Using the LiF + NaF + KF eutectic, we have formed [14,15] compact coatings on copper consisting of a 30 pm thick external titanium layer and several diffusion layers of definite compounds whose compositions have been determined. We have shown that the growth rate at potentials more positive than the pure titanium equilibrium potential is limited by the intermetallic diffusion of titanium and copper. As an extension of Renaud’s work, we have studied the mechanisms and the kinetics of the electrolytic formation of titanium on iron and nickel, and the possible interactions between titanium and the substrate. We present here our results on the main characteristics of the reduction mechanisms on iron and nickel at 600 and 700 o C, determined using convolution analysis of the voltammograms, and on the structure of the coatings.

EXPERIMENTAL

Melt preparation

The LiF + NaF + KF (46.5 + 11.5 + 42 mol%) eutectic was prepared by mixing pure and finely crushed fluorides (Merck pro analysis), placing them in a nickel crucible and introducing the latter in the electrolysis cell, which has been described previously [8,9]. The cell is maintained under vacuum during heating by 100°C steps until the salt melts, then under argon atmosphere. K,TiF, (Ventron pro analysis) was added via a lock-chamber either as compressed pellets or as blocks of a frozen 0.5 (LiF + NaF + KF) + 0.5 K2TiF6 mixture,

127

then reduced to the trivalent state by connecting the crucible to a titanium rod immersed in the bath. The reduction proceeds according to: 3 TiFi- + Ti + 6 F- +A TiF:In order to eliminate metallic impurities and oxides, the mixture was pre-electrolysed for 16 h using a 15 cm* nickel cathode and a 5 cm* ti~um anode. Electrodes To study the reduction mechanisms, the electrodes used were 5 X 5 X 1 mm sheets or 0.5 mm diameter wires of iron and nickel (Johnson Matthey). For coatings, larger sheets (20 x 5 x 1 mm) were used. Iron electrodes were first degreased by anodic polarization (100 mA cm-‘) in a hot (90 o C) concentrated soda solution (70 g I-‘), then rinsed in a 5% HCl solution and in distilled water. Nickel electrodes were cleaned in a 50-50 vol.% HNO, + CH,COOH mixture at 60° C. These electrodes were ultimately dried by alternate rinsings in alcohol and ether. The counter electrode was a 6 mm diameter titanium rod (99.7% Ventron). The nickel crucible was used as the reference electrode. Nevertheless, all electrode potentials are reported versus the K+/K equilibrium potential [14]. The crucible potential was checked periodically by plotting voltammograms in the potential domain of the solvent negative limit (see Fig. 1 for an example). Electrical device The electrolytic reduction mechanisms of Ti(II1) species have been studied by means of voltammetric techniques and constant potential chronoamperometry. The signals actually applied to the electrodes being too much distorted by the ohmic drop, the data processing was improved by means of convolution analysis [16,17]; with this method, all information included in the whole experimental curves can be considered whereas only some particular points are taken into account when classical techniques are used. The data acquisition system was composed of a PAR 173 potentiostat driven by a PAR 176 signal generator, a Nicolet 2090 digital oscilloscope for recording transients and a Hewlett Packard 9826 calculator for data processing. The convolution treatment of i(t) transients was carried out on the basis of Oldham’s Gl and G2 algorithms [18]. The efficiency of this method of analysis, rarely used in molten salt media f19,20], in spite of the importance of ohmic drops due to the distances between the reference electrode and the working electrode which are involved in these media, has been proved in our laboratory j&9,14,15,21-231. Efectrocoating experiments Deposits were obtained by means of 1 to 3 h constant current polarization at 600 and 700’ C. 25 to 50 mA cm-* current densities were applied.

128 i

I

Acm” A’

Fig. 1. Typical voltammograms showing the behaviour of the iron and silver electrodes solution in molten Flinak at 700 o C. X T,cmj = 1.5 x 10W2; u = 0.1 V s-l. (a) Iron electrode: V -+ 1.66 V -+ 1 V; (b) silver electrode: 1 V + -0.24 V --) 2.96 V--f 1 V.

in a Ti(II1) 1 V + -0.24

The electrolysis being completed, the electrode was lifted 1 cm above the bath level for draining, then removed from the cell by means of the lock chamber. Then it was washed in water, submitted to ultrasound and dried with alcohol and ether. The coatings were observed using a scarming electron microscope and analysed by means of X-ray diffraction and X-ray fluorescence in the electron microscope. REDUCTION

MECHANISM

OF Ti(II1) SPECIES

ON IRON

ELECTRODES

Figure 1 permits a convenient comparison between the electrochemical behaviour of iron and silver electrodes in a Ti(II1) solution in the LiF + NaF + KF eutectic at 700 o C. We notice that the anodic oxidation of iron starts at E = + 1.5 V, i.e. at a potential more negative than peaks B and B’ at E = + 1.94 V and E = + 2.12 V, which are related to the Ti(IV)/Ti(III) redox equilibrium [9]. On both voltammograms, we distinguish reduction peak A of TiF:ions beginning at E = + 0.47 V before the negative limit of the melt due to the reduction of K+ ions [S] during the negative sweeps and, after scan reversal, the associated peak A’ for the anodic oxidation of pure titanium. We note a good proportionality of the peak current i,, on voltammograms recorded using iron electrodes against the square root of the scan rate (Fig. 2) and the solute concentration (Fig. 3). Assuming that the variation of the peak potential

129

Fig. 2. Peak current variation versus fi.

Iron electrode; XT,t,IIj = 5.1X10m3.

(a) 600°C;

(b) 700°C.

with the scan rate (Fig. 4a) results mainly from the ohmic drop, particularly when high sweep rates are used, we report in Fig. 4b the semi-integral curves m(E) relative to the voltammograms of Fig. 4a, where

is the semi-integral of current Z(t) [16]. On these curves, the ohmic drop has been corrected according to E = V- rZ,where V is the potential supplied by the potentiostat and r the electrolyte resistance [8]. Notwithstanding the extent of the ohmic drop as suggested in Fig. 4a, the semi-integral curves (Fig. 4b) present a limiting value m* in the very negative domain, independent of the scan rate; the same goes for the whole cathodic part of the semi-integrals, proving that the electrochemical system behaves reversibly. Moreover, the direct and reverse scan

T

Fig. 3. Peak current variation versus Ti(II1) concentration at 600 o C. Iron electrode; u = 0.1 V s-‘.

130 m A.s “2 I

1

E K:K/”

(b)

Fig. 4. Voltammograms for trivalent titanium reduction on iron at 700 o C. XTi(llI)= 2.7 X 10K3; S = 0.5 cm*. v = (a) 0.1, (b) 0.2, (c) 0.5, (d) 1, (e) 2, (f) 5 V s-l. (b) Semi-integral curves corresponding to the voltammogmms plotted in (a).

plots are almost the semi-integral relevant to the product (Fig. 5,

the same, as is expected for a reversible electron exchange. One of curves of Fig. 4b has been analysed using logarithmic transforms hypothesis of a reversible exchange, giving either an insoluble curve a) where:

E=E,,+(RT/nF)ln[(m*-m)/m*] or a soluble product (Fig. 5, curve b) according to: E=E’+(RT/2nF)ln(D,/D,)+(RT/nF)In[(m*-m)/m]

where E, is the equilibrium potential in the studied solution, E o the standard potential, D, and Do the diffusion coefficients of reduced and oxidized species, respectively. The linearity of curve a is consistent with the insolubility of titanium in iron and in good agreement with the shape of redissolution stripping peak A’ (Fig. 1, curve a). The slope of the linear part of curve a (Fig. 5) yields the number of electrons exchanged during the TiF:- reduction equal to: n = 3 f 0.2. It can therefore be stated that the reduction of TiF:- ions proceeds reversibly through the direct exchange of three electrons without any slow intermediary step

131

In +$

(a)

A

b

Fig. 5. Logarithmic analysis of semi-integral curves plotted in Fig. 4b: (a) hypothesis of insoluble compound formation; (b) hypothesis of soluble compound formation.

and leads to insoluble metallic titanium according to the reaction: TiF:-

+ 3 e- P Ti + 6 F-

Using the following relation: m* =

3Fc,@

where c0 is the concentration of the TiF:- ions in the bulk, we have calculated the values of 0, at 600 and 700 ’ C. These values are respectively (1.2 + 0.5) x 10m5 and (2.7 f 0.5) X lo-'cm2 s-l. REDUCTION MECHANISM OF Ti(II1) SPECIES ON NICKEL ELECTRODES

The electrochemical behaviour of nickel (curve a) and iron (curve b) electrodes in a Ti(II1) solution at 700 o C are compared in Fig. 6. We notice the metal oxidation at E = + 1.9 V and the reduction of K+ ions at the same potential as on iron. We also recognize the reduction peak A of TiFi- ions in pure titanium and the corresponding dissolution peak A’; furthermore, the rise of several peaks B, B’. . . at more positive potentials than the A/A’ couple is noted. The non-existence of intermediate valences during the reduction of TiF:- ions, as shown on iron electrodes and already stated on molybdenum [14], allows us to associate peaks B, B’ . . . , with the formation and oxidation of alloys of titanium and nickel, mainly as definite compounds.

132

A‘

A

Fig. 6. Typical voltammograms showing the behaviour of nickel (a) and iron (b) in a trivalent titanium solution in molten F’linak at 700 o C. X,,,,,, = 1.5 X lo-‘; u = 0.1 V s-l. (a) 1 V + -0.24 V --*1.92 V + 1 V; (b) 1 I’-, -0.24 Vd1.66 V-1 V.

The current associated with peak A varies proportionally with the square root of the scan rate (Fig. 7). The much too low values obtained at higher scan rates might result from ohmic drop and from the overlapping between peaks A and B, which are both related to reactions consuming Ti(II1) species. The proportionality between the current of peak A and the solute concentration has been verified for 0.1 V s-l.

1i pd/A.cm-’

I

Fig. 7. Peak current variation versus \/;. Nickel electrode;

XTi(mj = 2.9 x lo-‘;

T = 700 o C.

133

m I As’/2

Fig. 8. Semi-integral curve of a voltammogram obtained on nickel at 700 o C. XTi(,,Ij = 2.9 V s-‘. (a) Direct scan; (b) reverse scan.

x

lo-*;

u= 5

Figure 8 presents the semi-integral m(E) of a voltammogram recorded at a scan rate of 5 V s- ‘. The proximity of the direct and reverse scan plots in the potential range less positive than +0.45 V proves that the reversibility of the system is still observed for fast sweeps. The semi-integral curves m(E) tend to a well-defined limiting value m*; we verified that m* is independent of the scan rate. The logarithmic transform (Fig. 9) of Fig. 8 presents no linear part either for the hypothesis of an ideal solution formation (curve a) or for the hypothesis of insolubility of reduction products (curve b). The presence of the anodic peak A’ (Fig. 6, curve a) reveals the accomplishment of surface saturation with pure titanium; the insolubility, at least partial, of the coating is ascertained by the positive value taken by the semi-integral during reverse scan for potentials higher than +0.5 V (Fig. 8, curve b). The potential gap between the positive and the negative parts of the semi-integral observed during the reverse scan is larger on nickel than on iron (Fig. 4b, curve f) indicating that the solubility of titanium in nickel is higher than in iron. Due to their proximity, peaks Bl, B2 and B3, occurring at potentials more positive than the A/A’ couple (Fig. lo), are relatively interdependent. Nevertheless, the semi-derivative curve d’121/dt ‘I2 [24], presents (Fig. 11) several well-separated peaks which must correspond to the initiation potential of new phases. In Table 1, the depolarizations AE = E, - EA of the experimental peak potentials E, against the titanium deposition potential EA are compared with the theoretical values AEth. The theoretical values were calculated using the Gibbs energies of formation AGP of the definite compounds Ti,Ni, TiNi and TiNi, [25-271 reported on the Ti-Ni phase diagram [28] according to: AEthTiNi3 = - AGPTiNiJ3F A E,TiNi

= ( AGP TiNi 3 - 3AGP TiNi)/6F

AEtiTi2Ni = ( AGPTiNi

- AGPTi,Ni)/3F

Fig. 9. Logarithmic analysis of the semi-integral curve plotted in Fig. 8. (a) Hypothesis compound formation; (b) hypothesis of insoluble compound formation.

of soluble

We note that the experimental and calculated values are in good agreement. The incorporation process of titanium in nickel is corroborated by the current variations recorded during constant potential polarizations at more positive poten-

Fig. 10. Voltammogram for Ti(III) reduction on nickel. S = 0.25 cm2; 1.05 V + -0.20 V + 2.20 V + 1.05 V.

T = 700 ’ C; X,,,,,,,

= 2 x lo-*;

v = 0.1 V s-‘;

135

A

Fig. 11. Semi-differentiation of the voltammogram plotted in Fig. 10.

tials than the pure titanium deposition potential. We observe that the current is constantly decreasing during polarizations lasting more than several minutes, as is expected for diffusion-controlled kinetics in solid phases. Figure 12 shows that the current density varies proportionally to t-l/* in the < 0.135 V. The existing methods for the potential domain 0.075 < E - EthTi/Ti(lII) determination of interdiffusion coefficients b [29,30] are restricted to the formation of solid solutions. We suppose that titanium and nickel atoms diffuse through an homogeneous layer of TiNi, since the equilibrium potential of the nickel electrode in our solution stays in the domain 0.20 < E - E*Ti/Ti(III)< 0.25 V. We use Oldham’s relation in the case of solid solution formation, taking electrode growth into account: i(t) = (noF/v,)fi where V, is the molar volume of the alloy and u a function of the surface atomic fraction of titanium X,, which is fixed by constant potential applied to the electrode.

TABLE 1 Thermodynamic values relative to Ti,Ni, Compound

AH,/J

Ti,Ni TiNi, TiNi

-80,400*6,000~ -67,800~4,000 -138,800~8,000 -171,600*4,000

a Ref. 25. b Ref. 26. ’ Ref. 27.

A&/J

mol-’ ’ ’ =

TiNi and TiNi, formation at 700 o C mol-’

-11.145b -12.444 b -26.008 b

K-’

AGp/J mol-’ - 69,556 f 6,000 - 55,692 f 4,000 -113,494~88,000 - 146,294 f 4,000

s*b ab a.b b.c

AE,/mV

AE/mV

+47*34 +921t35 +392&27 + 505 f 14

+30 +90 +450

136

Fig. 12. I - t- V2 curves for different E - ETlflL(Il~) = 73.5 my 135.5 mV.

constant

potential

(b) E - ET,,T~(III) = 90 my

polarizations.

X,,(,,,) = 5 X 10w3; T = 550 o C. (a)

(C) ~5 - ETi/Tr(III)

=I08

mv;

(4

E -

ETlp(l,,)

=

The composition ranges of TiNi, and TiNi are very narrow and no data are available about activity coefficients in the transition domain between these phases. Assuming an exponential increase of X, with potential up to the equilibrium potential between the TiNi phase and the solution (= EthrI,rioIIJ + 0.090 V), we obtain: X,=0.25

+0.25

exp(3F

[0.090-(E-E,i/riclll))]/RT)

(Iwascalculated according to ref. 29 and the values of b deduced from data of Fig. 12 are reported in Table 2. We obtain almost constant fi values for the diffusion in the TiNi, phase of the same magnitude as the literature values [31]. TABLE

2

Interdiffusion

coefficients

in the TiNi,

phase (V, = 7 cm3 mol-‘)

cE - '%IT,,T,(I,I,)/~

XS

0

it’/2

0.1355 0.108 0.090

0.286 0.3668 0.5

0.0275 0.1 0.23

0.018 0.058 0.154

/

A

a-2

s1/2

10” d/cm2 2.5 2.0 2.6

s-l

137 TITANIUM

COATINGS

Titanium coatings on iron electrodes The pure titanium coatings we obtained at 600°C were porous and irregularly distributed at the surface. X-ray fluorescence analysis of the electrode surface, revealing the presence of both titanium and iron elements, confirmed that the coatings did not cover the iron electrode completely. On the other hand, the coatings carried out at 700 o C were continuous (Fig. 13) and surface analysis did not reveal any other element than titanium. Figure 13 shows there is no diffusion layer between the titanium layer and the iron substrate. This is corroborated by X-ray fluorescence analysis which revealed no contamination of titanium by iron and of iron by titanium in the close proximity (i.e. about 1 pm) of the electrode/coating interface. Titanium coatings on nickel electrodes The coatings carried out on nickel at 600 and 700° C were more homogeneous than those on iron. They covered the whole electrode surface and were not too dendritic as can be seen in Fig. 14. They were composed of well-crystallized joined grams whose size was 20 to 40 pm (Fig. 15). X-ray fluorescence analysis did not reveal the presence of any nickel at the surface. Figure 16 represents a cross section of a deposit carried out at 700 o C. Between the 20 to 30 pm thick external layer of pure titanium and the nickel substrate, one can distinguish three underlayers whose compositions, determined by X-ray fluores-

Fig. 13. Scanning electron micrograph of a cross section of a titanium deposit on an iron electrode. X,i(,,rj = 5.1 x 10e3; T = 700 o C; i = 34.5 mA cme2. Polarization duration = 1 h 40 min.

138

Fig. 14. Titanium deposit obtained on nickel electrode. cmT2. Polarization duration = 1 h 30 min.

XTICm,= 1.5 X lo- -2. T=,jOO=-C;

Fig. 15. Scanning electron micrograph of the titanium deposit of Fig. 14.

j=50

mA

139

Fig. 16. Scanning electron’dcmgraph of a cross section of a titanium deposit on a nickel electrode. X,,(,,,) = 1.5 x lo-‘; T = 700 o C, i = 37 mA cm-‘. Polarization duration = 1 h 30 min.

cence analysis, correspond to the definite compounds Ti,Ni, TiNi and TiNi,. Underlayers of the same structure but 3.5 times less thick than at 700 QC, were also obtained at 600” C. The position of the Kirkendall interface which was directly contiguous to the nickel substrate, showed that nickel diffuses much faster in titanium than the other way round. CONCLUSION

We have described the main features of the electrochemical reduction of TiF:ions on iron and nickel electrodes in the LiF + NaF + KF eutectic in the 600-700 o C temperature range by means of convolution linear scan voltammetry. At iron electrodes, the three-electron exchange proceeds reversibly following a single step according to: TiF:-

+ 3 e- + Ti + 6 F-

and the diffusion coefficients for TiFi- ions are 1.2 X lo-’ cm2 s-l at 600°C and 2.7 x 10e5 cm2 s-l at 700 OC. The pure titanium deposit is very stable and does not interact with the supporting metal. At nickel electrodes, the overall process is also reversible but the deposition of pure titanium is preceeded by several steps leading to the formation of titaniumnickel alloys. Since the voltammetric transients presented peaks in this potential

140

domain which were too much overlapped, the curves were roughly deconvoluted using semi-derivative analysis which presents three well-separated peaks corresponding to the formation of definite compounds according to: TiF:2 TiF:-

+ 3 Ni + 3 e- + TiNi, + 6 F+ TiNi, + 6 e- + 3 TiNi + 12 F-

and TiF:-

I- TiNi + 3 e- -+ Ti,Ni + 6 F-

These conclusions are corroborated by the analysis of underlayers resulting from titanium coating on nickel cathodes. The rate of these processes is controlled by intermetallic diffusion whereas the pure titanium deposition is limited only by the diffusion of titanium ions in the melt. The interdiffusion coefficient in the TiNi, phase is 2.5 x lo-” cm* s-r at 550°C. Titanium electroplating was attempted on iron and nickel and resulted in 20-30 pm thick coatings of pure titanium on iron at 700°C and on nickel at 600 and 700°C. The deposits covered the whole surface and were well crystallized and the average size of the grains amounted to 20-40 pm. Whereas any solubility of iron into titanium was not observed, we obtained a multilayer interdiffusion zone joining the titanium coatings to the nickel substrate. X-ray fluorescence analysis has given the composition of the successive layers which are, starting from the nickel based metal: TiNi,, TiNi, Ti,Ni and pure titanium external coating. Work is now in progress in our laboratory on titanium electroplating on nickel and iron materials at higher temperatures with the possible interference of the reduction of K+ to potassium metal and on the electrochemical study of the interdiffusion parameters during the growth of layers of intermetallic compounds. Acknowledgement

The authors wish to express their indebtedness to the C.N.R.S board for financial assistance (A.T.P. “Applications de l’electricitt a la chimie” No. 2064). REFERENCES 1 G.N. Novitskaya, L.I. Zarubitskaya and V.I. Shapoval, Khim. Tekhnol. (Kiev), 5 (1981) 57.

2 3 4 5 6 7 8 9 10 11 12

M.V. Smimov and U.S. Maksimov, Electrochem. Molten Solid Electrolytes, 6 (1968) 30. R. Baboian, D.L. Hill and R.A. Bailey, Can. J. Chem., 43 (1965) 197. M. Nardin and G. Lorthioir, J. Less Common Metals, 56 (1977) 269. G.W. Mellors and S. Senderoff, J. Elctrochem. Sot., 114 (1967) 586. G.W. Mellors and S. Senderoff, Can. Pat., 688. 546, 1964. F.R. Clayton, G. Mamantov and D.L. Manning, J. Electrochem. Sot., 120 (1973) 1193. P. Paillere, Thbse., Grenoble, 1982. J. de Lepinay and P. Paillere, Electrochim. Acta, 29 (1984) 1243. A.R. Stetson, Mater. Des. Eng., 57 (1963) 81. K. Matiasovsky, Z. Lubyova and V. Danek, Electrodep. Surf. Treat., 1 (1973) 43. Texas Instruments Inc., Fr. Demande, 2. 075. 857, 1971.

141

13 G.A. Kline, Eur. Pat. Appl., 79055, 1983. 14 D. Renaud, These, Grenoble, 1985. 15 J. de Lepinay, J. BouteiIIon, S. Traore, D. Renaud and M.J. Barbier, J. Appl. Electrochem., 17 (1987) 294. 16 K.B. Oldham, Anal Chem., 44 (1972) 1%. 17 J.C. Imbeaux and J.M. Sav&ant, J. EIectroanaI. Chem., 44 (1973) 169. 18 K.B. Oldham, J. Electroanal Chem., 121 (1981) 341. 19 F. Seon, G. Picard and B. TremiIlon, Electrochim. Acta, 27 (1982) 1357. 20 D. Ferry, These, Paris 6,1985. 21 M. Taoumi, These, Grenoble, 1985. 22 M. Taoumi and J. BouteiIlon, EIectrochim. Acta, 31 (1986) 837. 23 J.P. Gamier, These, Grenoble, 1985. 24 P. Dahymple-AIford, M. Goto and K.B. Oldham, J. Electroanal. Chem., 85 (1977) 1. 25 0. Kubaschewsky, Trans. Faraday Sot., 54 (1958) 814. 26 L. Kaufman and H. Nesor, Calphad, 2 (1978) 81. 27 J.C. Gachon, M. Notin and J. Hertz, Thermochim. Acta, 48 (1981) 155. 28 D.M. Poole and W. Hume-Rothery, J. Inst. Met., 83 (1954-55) 473. 29 K.B. Oldham and D.O. Raleigh, J. Electrochem. Sot., 118 (1971) 252. 30 F. LanteIme and M. ChemIa, Z. Naturforsch., 38a (1983) 1%. 31 G.F. Bastin and G.D. Rieck, Met. Trans., 5 (1974) 1827.