The oxygen electrode in fused lithium chloride-potassium chloride eutectic containing oxide ion

Electroanalytical Chemistry and Interfacial Electrochemistry, 58 (1975) 339-348 ,.~-~Elsevier Sequoia S.A., Lausanne

339

Printed in The Netherlands

THE OXYGEN ELECTRODE IN FUSED LITHIUM CHLORIDEPOTASSIUM CHLORIDE EUTECTIC CONTAINING OXIDE ION

YASUSHI KANZAKI* and MASAO TAKAHASHI

Institute of Materials Science and Engineerin9, Faculty of Enyineering, Yokohama National University, Ohoka-machi, Minami-ku, Yokohama (Japan) (Received 1st February 1974; in revised form 17th September 1974)

INTRODUCTION

The oxide ion is known as a strong Lewis base. It plays an important role in the passivation phenomenon of metals 1'2 and in the acid-base equilibrium in fused salts 3'4. The log of the concentration of the oxide ion, p O 2 - - - - l o g [O2-], in fused salts corresponds to the concept of pH in the aqueous solution 3. Realizing these points the authors have dealt with the oxygen electrode in fused lithium chloride-potassium chloride eutectic containing oxide ions through the oxygen evolution reaction. Oxygen gas is evolved at inert electrodes in fused salts. The electrode reaction is often revealed to have a far simpler mechanism than those in the aqueous solution as will be contended in this article. The overall oxygen evolution reaction in the fused lithium chloride-potassium chloride eutectic can be assumed to be 20 z- (melt) - 4e ~ 02 (gas)

(1)

Reaction mechanisms for this electrode process have already been proposed by several authors 5-9. For instance, Inman et al. considered the rate determining step of the reaction 1 as the formation of adsorbed oxygen atoms on the gold electrode from the analysis of the Tafel slope of the polarization curve 6. They reconsidered the reaction later and regarded the potential determining process as the 0 2 - / 0 2 equilibrium from the analysis of the Nernst plot of the 0 2 - / 0 2 system v. These previous works have suggested that the oxygen electrode in fused lithium chloride-potassium chloride eutectic was influenced by the presence of the peroxide ion, 0 2 - , or the superoxide ion, O2. However, important experimental defects seem to be included in these works. One consists in the interpretation of the polarization curve, and the other in the determination of the oxide ion concentration in the melt. This paper deals with the oxygen evolution reaction in an attempt to correct these defects. Such a revision may suggest the possibility of the reversible oxygen electrode in fused lithium chloride-potassium chloride. * Present address: Department of Chemistry, College of Science and Engineering, Aoyama Gakuin University. 6-16-1 Chitosedai, Setagaya-ku, Tokyo, Japan.

340

Y. KANZAKI, M. TAKAHASHI

EXPERIMENTAL

Reagents Reagent grade lithium chloride, potassium chloride and silver chloride (Wako Pure Chemical Ind., Ltd.) were used without further purifications. Lithium oxide (99~o, Mitsuwa Pure Chemicals) and barium oxide were used as the source of oxide ions. The concentration is shown as the ionic fraction unless otherwise stated.

Preparation of the fused salt Lithium chloride and potassium chloride were dried separately at 400°C for 4 h. They were mixed together and dried again at 250°C for 6 h in vacuo before melting. Anhydrous hydrogen chloride (99.7~o, Tsurumi Soda Co. Ltd.) was bubbled through melted salt to remove oxide ions, hydroxide ions and the residual moisture 1o. After exhaustion of hydrogen chloride the cell was saturated by dried argon gas during measurements or by oxygen gas when the oxygen electrode was constructed. Atmospheric pressure correction was not made in the oxygen electrode.

Cells The cell assembly is illustrated in Fig. 1. The electrolytic cell was enveloped by a Pyrex glass tube. The tube was plugged by a silicone-rubber stopper. It supported electrodes and a glass tube for the thermocouple. The cell was thermostatted by a high-low thermo-controller with a precision of 2°C. Measurements were made at 400°C unless otherwise stated. The glass cell was made of Pyrex glass. A platinum cell was used for comparison, because the glass cell consumes oxide ions. C RW

Ag

AgCl(0.l) sbestos i

i

7ram

lOcm Fig. 1. Electrolytic cell. (W) working electrode, (R) reference electrode, (C) counter electrode, (E) ground. (I) steel shield, (Thl, Th2). PR-thermocouples. Detailed structure of reference electrode is shown on the right-hand side of the cell.

OXYGEN ELECTRODE IN FUSED CHLORIDE SALTS

341

Electrodes

Platinum, palladium, rhodium and gold wires were used as the working electrode. Each had a 1 mm diameter and 16 mm length. They were fitted with glass tubes. The thermoelectric power of these with the copper lead wire was ignored. The electrode used as the working electrode was platinum unless otherwise stated. Spectroscopic grade graphite was used as the counter electrode. The silver silver chloride system (ag/agCl(0.1), [LiCl(0.59~KCl(0.41)](0.9) (asbestos)) was used as the reference electrode (Fig. 1). The electrode potential of the system was - 1.010 V at 400°C vs. the standard chlorine electrode ( - 1.002 and - 0 . 9 9 5 V at 450°C and 500°C, respectively11). The reproducibility of the electrode potential was about + 2 mV. Penetration of silver ions through the asbestos was almost negligible during measurements. The silver chloride melt was prepared separately. Voltammetr y

A potentiostat (PS-1000, H o k u t o Denko Co. Ltd.) was used for the measurement of c u r r e n t ~ o t e n t i a l characteristics. Current-potential curves were recorded on a XY recorder. The electrode potential was measured with a chopper type pH-meter. The input impedance was sufficiently high compared with the resistance of the junction of the reference electrode. RESULTS AND DISCUSSION The potential sweep method was adopted for the voltammetric measurement. Figure 2 shows typical current-potential curves for the oxidation of oxide ions at the platinum electrode. When the electrode was polarized anodically in the melt containing oxide ions, a'rising current was observed at about 0 V vs. Ag/AgCI(0.1). The current was attributable to the generation of oxygen. Bubble evolution was observed simultaneously at high oxide ion concentration. The potential is about 0.5 V lower than that of the anodic dissolution of platinum 1,2,12. When the concentration of oxide ions was low enough, the polarization curve was similar to the theoretical one for the simple diffusion-controlled two-electron transfer reaction (curve 1). The shape was turned into that for the steady (hydrodynamical) currentpotential curve at high oxide ion concentration (curves 2 and 3). This change must be due to the convection around the electrode caused by bubbles of oxygen. The maximum current in curve 1 should be proportional to the concentration of the reactant. Plots of the maximum current, iv, vs. the amount of added oxides (lithium oxide or barium oxide) are shown in Fig. 3. Both oxides gave a quite similar result. Degradation of oxide ions was observed in the glass cell as shown by curve 1. The degradation degree corresponded to the area of glass surface. The ip value, however, was almost independent of the time elapsed. On the other hand no degradation of oxide ions was observed in the platinum cell. The ip value was closely proportional to the amount of added oxides (curve 2). Since it was found from the polarization curve that the reaction product of oxide ions with glass did not distort the polarization curve, qualitative experiments were made by using the glass cell. The real concentration of oxide ions in case of the glass cell can be deducted on comparing two curves in Fig. 3.

342

Y. KANZAKI, M. TAKAHASHI

3

25

/

4C 2O

2C

E]5 u < E 10

1E ~. E u 12

2

,

,

2

3

2

E'8 \

1

-0.1

0 0.1 0.2 E / V vs. Ag/AgCI(0,1)

.o. 0.3

4 0

1

4

5

6

""~2"

×10-3

[ 0 2-3

Fig. 2. Current-potential characteristics for oxidation of O 2- at platinum electrode in LiC1 KC1 eutectic. Scan rate 0.5 V min- 1: concns, of O2-: (1) 5 x 10 -3, (2) l'x 10 -2, (3) 2 x 10 -2. Cell, Pyrex glass. Fig. 3. Relations between maximum current, ip, and amount of added oxides, in which X-axis is transformed into conch, of O 2-. (1) Plots obtained in glass cell at scan rate 0.5 V min -I. (2) plots in platinum cell at scan rate 1 V min-1.

Reversibility of the electrode reaction It is well known that platinum gives a high oxygen overpotential in aqueous solution. On the other hand the oxygen evolution reaction seems to be rapid in the chloride melt (Fig. 2). It is useful to introduce the cyclic voltammetry at the solid electrode to examine reversibility of this electrode reaction 15. Figure 4 shows an experimental curve obtained by the multi-sweep method. It resembles to the theoretical curve though it is not symmetrical about the point of origin which is determined by zero current and the average potential of the anodic and the cathodic peak. A small solubility of oxygen might cause such an inconsistency. Such an inconsistency is supposed not to disturb the comparison between the experimental aqd the theoretical curve seriously. The potential difference between the cathodic and the anodic peak was about 80 mV, which is similar to the theoretically presumed g (n=2) 2 0 -2 -4

4 3

u

1

-] -2

0

0.1

O.2

E / V vs.Ag/AgCI(Q1)

Fig. 4. Cyclic voltammogram of oxide ion obtained after ten scans. Scan rate 0.5 V min-t. Concn. of 0 2 : 5 x 10 -3. Cell Pyrex glass. Upper axis is indicated in ~-units is.

O X Y G E N E L E C T R O D E IN F U S E D C H L O R I D E SALTS

343

value of the two-electron transfer reaction, 2.2RT/2F (65 mV at 400°C; the upper scale of the Figure gives the 4-unit) a 5. Reversibility of the oxidation and the reduction reaction of the oxygen electrode is obvious from this curve. To examine this reaction more quantitatively, the single potential sweep method was applied. The overall reaction was assumed to consist of the following elementary steps: 0 2- - 2 e ~ O

(2)

OJl-O

(3)

~

0 2

For convenience of analysis the subsequent dimerization process was neglected at first, i.e. the dimerization reaction was assumed to have an infinitely slow reaction rate. Then the Randles-Sevcik type current-potential curve can be given as

follows13 15: i( 4) = nFc* D~( nF/R T) ~ v ½71(4)

(4)

where i(4) is the current density, n the number of electrons transferred in the electron transfer step, c* the concentration of the reactant in the solution bulk, D the diffusion coefficient and v the scan rate. F, R and T have their usual significance. ~(4) determines the shape of the current-potential curve and is a function of the electrode potential. ~ is given as - n F / R T ( E - E~), where E~ is the half wave potential and E the applied potential. 7' is independent of the rate of the electron transfer step and the transfer coefficient if the rate constant is sufficiently large. This is valid if the 15

10

E t) E

-Q4

-Q2

0

i

i

0.2

0.4

i

0.6 0,8 ,

E / V vs. Ag/AgCI(0.1) Fig. 5. Dependence of polarization curve on scan rate. Concn. of 0 2 - 5 x 10 3 in glass cell. Scan rates: (1) 0.5, (2) 1.0. (3) 2.0, (4) 6.0, (5) 12.0 V min -1.

standard rate constant is larger than 10-1 cm s -a in our present case. Figure 5 shows_the dependence of the polarization curve on the scan rate. The potential at which the maximum current occurs, Ep, was independent of the scan rate. The maximum current density, ip, was correctly proportional to the square root of the scan rate. These two observations are consistent with polarographic reversibility. In Fig. 6 the experimental current-potential curve and the theoretical ones are compared. For convenience, the potential at which the maximum occurs is chosen as a base line. The abscissa is shown in the volt-unit. Abscissae shown in the 4-unit are indicated for

344

Y. KANZAKI, M. TAKAHASH1 E I V vs. Ag/AgCl(0.1) -0.3 -0.2 T

-0.I

0

0.I

0.2

Q3 12

0.5 0.477

/7' II I]

0.4 Q3

o 8 8~

~ 0.2

E 4 ...

0.1

/

2

0 _......-/ n=l £ o=z

n:4

.1.1__ o '

'

46 2/~

'!-' -!~ oJ2/~

0

-~

( = -n F/RT(E i - E!~+ vt))

Fig. 6. Analysis of polarization curve obtained by potential sweep method for oxidation of oxide ions. All curves are shown in volt-unit for abscissa. Origin for abscissae shown in ~-unit is half wave potential ( . . . . . . ) Experimental curve. Theoretical curves are shown by (-- - - -) for n = 1, ( ) for n = 2 and ( - - - ) for n = 4. Experimental curve was obtained at the following conditions. Temp. 400°C; conch, of 02- 5 x 10 3; electrode Pt; scan rate 6 V rain- 1. comparison. The experimental curve agrees well with the theoretical one for n = 2. The half wave potential was calculated to be + 0.15 V vs. the reference electrode. According to Saveant and Vianello 16, the shape of the curve scarcely depends on the rate of the following dimerization reaction. The rate of the dimerization reaction, therefore, must be examined in some other way. Fortunately, the E v value depends o n the scan rate when the dimerization reaction is fast 16, AEp = - ( R T / 3 F ) In Av. Since E v did not depend on the scan rate in our present reaction, the dimerization reaction must be slow. The upper limit of the rate could not be determined because the dissociation constant of eqn. 3 was not known. Analyses of Figs. 4 and 5 show that (1) the oxygen evolution reaction is polarographically reversible, (2) the n u m b e r of electrons involved in the electron transfer step is two and (3) the reduction reaction of oxygen is reversible. It could not be p r o v e d from these experiments whether the reactant in the cathodic reaction, the reactant which gave a cathodic peak in Fig. 4, was an oxygen molecule or an oxygen atom. The Arrhenius plot of the ip value was quite linear in the temperature range 400-500°C. The activation energy of diffusion of the oxide ion was calculated to be 10.6 kcal mol -~. The diffusion coefficient of the ion was 2 . 4 x 10 -6 cm 2 s -1 at 400°C.

Standard electrode potential

The standard electrode potential of the oxygen electrode, 0 2 -/02, in lithium c h l o r i d e - p o t a s s i u m chloride eutectic has been determined by several authors. M a s u k o et al. 4 calculated the value from the solubility of refractory metal oxides and the standard chemical potential of their parent cations. The standard electrode potential

345

OXYGEN ELECTRODE IN FUSED CHLORIDE SALTS

was - 0 . 6 6 V at 450°C vs. the standard chlorine electrode, in which the concentration was given in ionic fraction. Experimentally, it was reported that the reversible oxygen electrode could not be constructed in this melt as in the aqueous solution v8. However, some ambiguities are supposed to be included in these experiments v. For example, these authors considered the added amount of oxide (Li20) as the concentration of oxide ions and they did not analyse the concentration properly. 0.12

>

-O.88

0

0.10

-q90

0.08

-0.92

0.06

-0.94

0.04

-0.06

0.02

-0.98

0.00

-1.oo >

-0.02

-1.02 hJ

(.9

],I

-1.04

-0.04 , . .

'

10_4 2 3 5 10_3 [ 0 2- ]

~

'

~

10_2

Fig. 7. Nernstian plots of OZ-/O2(Pt, Po2= 1 atm) system in LiC1 KC1 eutectic. X-axis is determined by the added amount of oxide. Cell platinum, Temp. 450°C. (O) Li20, ( A ) BaO.

In accordance with the results stated in the preceding section the possibility of the direct measurement of the O2-/O2(Pt) system seems to be quite probable. Figure 7 shows the Nernstian plots of the oxygen electrode. The slope of the plots is fairly consistent with the theoretical one (solid line). The discrepancy in the dilute region was caused by the effect of glasses which were used for the electrode support. When the concentration of oxide ions was corrected from the height of the polarographic diffusion current (Fig. 5), the plot showed a good linearity. The standard electrode potential was estimated as E = - 1.173 + ( R T / 4 F ) In Po~_- ( R T / 4 F )

In [0 2-]2

at 450°C vs. the standard chlorine electrode. This result as well as the results obtained by the voltammetric measurement show that the reversible oxygen electrode in lithium chloride-potassium chloride can be constructed. Although values of the potential obtained here were similar to those reported by Wrench and Inman v, the slope of the Nernst plot in Fig. 7 was twice as large as that presented by the latter authors. Comparing these two results, our result is supposed to be more reliable since the concentration of oxide ions was calibrated polarographically in our work, while in the latter it was not. The reaction mechanisms proposed by Wrench e t a l. 7"8 and the conclusion that the reversible oxygen electrode could not be constructed seem, therefore, rather questionable.

346

Y. K A N Z A K I , M. T A K A H A S H I

Other electrodes

Current-potential characteristics of rhodium and palladium electrodes were similar to the results obtained at the platinum electrode in the oxygen evolution region, though they were different from the latter in some minor details 17. 25 20

k5

S ~0

1

5

0

-0,4 -0.2

0

I

I

0.2

0.4

I

I

0.6 0.8

E I V vs. AglAgCKO.I)

Fig. 8. Current-potential characteristic of Au for oxidation of O 2-. Concn. of 0 2- : 1 x 10-2. Cell Pyrex glass, scan rates (curve numbers) can be referred to those in Fig. 5.

The gold electrode showed a different behaviour from the above three electrodes. The current-potential curves for the oxygen evolution reaction at the gold electrode are shown in Fig. 8. It clearly gives two peaks which depend on the oxide ion concentration. The analysis of these curves revealed that the first wave is a one-electron transfer reaction. In addition, the first wave and the second have the same height. It can be concluded that the oxygen evolution reaction at this electrode consists of the following elementary steps: O2---e

---} O -

first wave

O--e ~ O 0 + 0 ~ O2

second wave

0 2--e

}

(5) (6 / (7)

or

~ O-

(5) first wave

O-+O-

~ O2-3

02--2e

~ 02

(8) second wave

(9)

Since the number of electrons involved in the second wave seems to be one judging from its wave form, the first reaction mechanism looks more appropriate. The electrode potential of the 02 -/Oz(Au ) system was several tens of millivolts lower than that of the platinum electrode. The potential measured might be a mixed potential of the reactions 5 and 6. Although the phenomena at the gold electrode were quite remarkable, further experiments are required. Tafers relation has already been examined by Inman et al. 6 for the oxidation of oxide ion at the gold electrode. They concluded from the Tafel slope that the number of electrons involved in the electron transfer step was two and the transfer coefficient 0.5. Since the oxygen evolution reaction was shown by us to have a very

347

OXYGEN ELECTRODE IN F U S E D C H L O R I D E SALTS

high exchange current density at both platinum and gold, the electrochemical kinetic parameters must be difficult to obtain unless the very high frequency or a superior pulse technique were used. The polarization curve obtained by Inman et al. must be a diffusion controlled one because it was measured by means of the usual stationary method. From such a curve one can determine only the number of electrons involved.

Oxygen evolution reaction from O H - and H20 Figure 9 shows current-potential curves for the evolution of oxygen from melts of different composition. Curve 3 shows the curve in the melt containing hydroxide ions. The potential of the maximum current shifts about 0.9 V more positive than that of the oxide ion (curve 1). The shallow slope of the rising current might be caused by the fact that the activity of products, 02 and H +, changed together as the current became large. The electrode potential of the OH-/O2(Pt) system was not measured since the potential region corresponds to the passivation region of platinum*. Curve 2 shows the polarization curve in the melt containing equivalent amount of oxide ions and hydroxide ions. In the melt containing water, its concentration being known from the current height of the wave for the hydrogen evolution reaction 1°, no anodic current was observed until chlorine gas was evolved (curve 4). In this case the glassy-carbon electrode was used because platinum dissolves at about +0.5 V*. Further quantitative treatments for 0 2 - , OH and H20 in this melt will be presented in the following paper of this series 1°. 10

L)

4 E \

2

-

0

i

i

-0.5

•

jj.. 0 0.5 E / V vs.Ag/AgCII01)

1.0

Fig. 9. Polarization curves for evolution o f O 2 in melts containing (1) O 2 (5 x 10- 3), (2) 02 (2.5 x 10-3) plus O H - (2.5x10-3), (3) O H - ( 5 x 10 a), and (4) H 2 0 (satd.). Electrode: (1)-(3) platinum; (4) glassy-carbon. Scan rate: 1 V min 1; temp. 450°C. Current rise at 1.0 V was attributed to evolution of C12 in curve 4. -

SUMMARY

The oxygen evolution reaction in a fused lithium chloride~otassium chloride eutectic containing oxide ions was examined by means of the voltammetric method. The electrode reaction was polarographically reversible at platinum, rhodium and palladium electrodes. The cyclic voltammogram obtained showed that the oxidationreduction reaction of the oxygen electrode was also reversible. The analysis of the * The anodic dissolution of platinum is prevented by the formation of a protective oxide film when oxide ions or hydroxide ions are added I'z.

348

Y. KANZAKI, M. TAKAHASHI

current-potential curve revealed that the number of electrons involved in the electron transfer step was two. The gold electrode gave a different behaviour from the above electrodes. It gave two separate waves, both depending upon the oxide ion concentration. The electrode potential of the oxygen electrode, O2-/O2(Pt), was measured on the basis of the voltammetric measurement. The Nernstian plot of this system was linear and the slope was in good agreement with the one theoretically presumed. The standard electrode potential was -1.173 V at 450°C vs. the standard chlorine electrode. The oxygen evolution reaction of other oxide ion donors, O H - and H20, was examined subsidiarily.

REFERENCES 1 M. Takahashi, Y. Katsuyama and Y. Kanzaki, Extended Abstract of 17th CITCE Meeting, Tokyo, 1966. 2 M. Takahashi, Y. Katsuyama and Y. Kanzaki, Yoyuen, 8 (1965) 571. 3 R. Littlewood, J. Electrochem, Soc., 109 (1962) 525. 4 N. Masuko, M. Okada and Y. Hisamatsu, Yoyuen, 6 (1963) 569. 5 G. Delarue, J. Electroanal. Chem., 1 (1959-1960) 13. 6 P. Delahay (Ed.), Advances in Electrochemistry and Electrochemical Engineering, Vol. 4, Interscience, New York, 1966, p. 144. 7 N. S. Wrench and D. Inman, J. Electroanal. Chem.. 17 (1968) 319. 8 P. G. Zarnbonin and J. Jordan, J, Amer. Chem. Soc., 91 (1969) 2225. 9 P. G. Zambonin, J. Electroanal. Chem., 24 (1970) App. 25. 10 Y. Kanzaki and M. Takahashi, J. Electroanal. Chem., 58 (1975) 349. 11 M. Takahashi, Denki Kagaku, 25 (1957) 432. 12 H. A. Laitinen and C. H. Liu, J. Amer. Chem. Soc., 80 (1958) 1015. 13 J. E. B. Randles, Trans. Faraday Soc., 44 (1948) 327. 14 A. Sevcik, Collect. Czech. Chem. Commun., 13 (1948) 349. 15 H. Matsudaland,Y. Ayabe, Z. Elektrochem., 59 (1955) 494. 16 J. M. Savdant and E. Vianello, Electrochim. Acta, 12 (1967) 1545. 17 Y. Kanzaki, K. Yoshizumi and M. Takahashi, Yoyuen, 10 (1967) 105.