electrochemical behaviour of aluminium in ... - Science Direct

ELECTROCHEMICAL BEHAVIOUR OF ALUMINIUM SODIUMPOLYSULPHIDE MELTS AT 330°C R.

and S.

KNOEDLER

MENNICKE

Brown. Boveri & Cie AG. Central Research Laboratory, (Rccewd

2 Auyusr

IN

Heidelberg,

W. Germany

1982)

Abstract-An experimental set-up was developed which allows the investigation of electrodes in pure sodiumpolysulphide melts. It could be shown that aluminium forms a passivating layer of aluminiumsulphlde ar an anodic potential of about 1 V zs a sodium reference electrode. Below this value, aluminium is active. With a copper containmgaluminium alloy the passivating potent&4 I$ increased to about 1.2 V. Pining corrosion was observed after operating the alummium electrode at current densities of several hundred mA cm-‘. Reaction mechanisms which could exulain the acme, passlvc and trampassive rcgmns observed, are discussed.

INTRODUCTION

The Beta Battery candidates

(Na/S) is one of the most promising for electric vehicle traction and off-peak

energy storage[l]. IL can be expected that the system, which is working at 300-3SO’C yields an energy density four times higher than the lead-acid battery. A sodium sulphur cell. in principle, consists of a liquid sodium anode and liquid sulphur cathode (soaked in graphite felt). These electrodes are separated by a solid separator tube consisting of sodium ion conducting beta-alumina. During discharge sodium migrates through the separator and with the sulphur forms sudiumpolysulphide Na,S, (x depends on the state of discharge and goes usually from x = 20

to x = 3). The open circuit voltage is 2.08 V from Na,S,, to Na,S, (“two-phase region”‘) and drops to 1.78 V for Na,SI (“one-phase region”), The current cell design at Brown Boveri requires that the cell casmg is in contact with Na,S,. It turned out that the sodium polysulphides corrode most construction materials, like steel and even stainless steel[2]. Therefore, various coating@, 41 and linings [S] had been investigated. Aluminlum, which would be the ideal casing material as far as cost, weight, fabrication properties and conductivity are concerned,

of preparation. The easiest way to obtain such conditions is to produce the polysulphides directly in an Na/S ccl1 and to carry out the electrochemical measurements in the same cell without handling the polysulphide. Figure 1 shows the experimental arrangement. A sodium/sulphur cell with a relatively large /7-Al,O, tube (34 mm outer diameter, 260 mm length) was built. Inside the tube a perforated titanium tube was placed

Working electrode

~~

Reference

eiectrode

,evacuated

.-m-Glass

unfortunately forms an insulating Al,S, layer in Na,S,. Therefore, coatings have to be applied, which in some cases[2,6] turned out to be successful. However, although the practical results of coated

alumimum are rather satisfactory so far, It is still unclear how uncoated aluminium behaves electrochcmically. This is important to know because coated layers always contain some impcrfcctlons. Therefore, the purpose of the work presented here was to obtain cxperimcntal data on the electrochemical properties of aluminium in molten polysulphide. EXPERIMENTAL From prevtous experiments with polysulphides it turned out that, because of the hygroscopic nature of these substances, it is essential to exclude air in all steps 1033

Fig. 1. Experimental arrangement for voltammetric measurements in a sodiumpolysulphide melt. Working electrodes: Al or Al-alloy, reference electrode: Na/&Al,O,, counter electrode: Ti-tube.

R.

1034

KNOEDLERAND~.

as cathodic current collector (titanium is sufficiently corrosion resistant for these purposes). Between the Ti and the P-AI,O, tube graphite felt rings and a-Al,O, felt rings were inserted and were filled with sulphur. The space outside the P-A1,03 tube contains sodium. The Ti tube is extending out of the furnace area and is connected via a flange with a glass tube with feedthroughs for working and reference electrode. The working electrode consists of a 4 cm2 Al or Al-alloy piece, connected to a molybdenum wire (outside the melt). The sodium reference electrode consisted of a glass tube molten into a small piece of P-AI,03. Prior to insertion into the cell, this tube was filled with sodium by electrolysing an NaNO,/NaNO, melt. Both the sulphur compartment and the reference electrode were evacuated. After the cell was heated up to 33o”C, it was discharged about 20Ah, which corresponds to a discharge depth of about 30’;,. Thus, the melt around the working electrode was in the two-phase region, where Na,S, and S are present. With different electrodes cyclic voltammograms and impedance characteristics were measured. Voltammograms were taken with a potentiostat and a scanner mostly at the 0.5 mVs 1 rate. With this rate a stationary state is achieved in the active state, but in the passive state even a rate of 0.05 mV s- ’ is not sufficient for equilibrium. Further investigations into this behaviour are envisaged. The impedance measurements were carried out with a Solartron frequency response analyser, which allowed measurements with a superimposed dc current. Details ol this technique can be found elsewhere[7].

HESULTS

AND DlSCUSSlON

Figure 2 shows a voltammogram for 4 different electrode materials. The open circuit voltage (OCV) is

Fig. 2. Current/voltage

relationship

MENNICKE

2.08 V, corresponding to the two phase (Na,S,/S) region. The stainless steel electrode shows a simple ohmic behaviour in both anodic and cathodic direction with a resistance ofabout 0.5 Rem’. In contrast, with the aluminium electrodes clearly three potential regions can be distinguished: between about 1.0 and 2.08 V a passive region, below about 1.0 V an active region and beyond 2.08 a region which is called here the “transpassive” region (however, this has not the same meaning as with iron and chromium dissolution), Both in the activeand the transpassive region an ohmic behaviour is observed with resistances between S and 2 Rem’. The lowest value wasobserved with the Al-& alloy. There are differences in the potential where the transition active/passive occurs. The Al-Cu alloy shows with about 1.1 V the most anodic, whereas Al 99.5 with i 0.9 V shows the most cathodic “activalion potential” of the materials investigated. An impedance analysis was carried out in order to elucidate the condition of the materials in the active, passive and “transpassive” state. Figure 3 shows impedance plots at potentials in the passive and “transpassive” region. In the active region the resistance is almost independent of frequency about 10Rcm2. There are semicircles with their center above the abscissa. This effect generally is attributed to surface roughness. The interpretation of the semicircles may be tried in terms of an equivalent circuit, which obviously consists of an ohmic resistance R,, in series with another resistance R,,, shunted by a capacity Co. It is known from electrode kinetics that R,, is the resistance of the melt and can be measured at high frequencies. In the present case, R,‘is about 10 ncrn’ which corresponds well with the values measured in the voltammetric curves. R,, represents a polarization resistance (concentration, charge-transfer, rtc.) and C, is the double layer capacity which is affected by the presence of passivating layers or films. The values of curves in the R,, indicate that the voltammetric

of aluminium, aluminium alloys and stainless steel in molten Na,S,/S at 330 C. Scan rate: 0.5 mV s ’

Behaviour of aluminium in sodiumpolysulphide melts

1035

Z”/ R cm’

Fig. 3. Impedance plots (real part Z’ w imaginary part Z”) ofaluminlum in molten Na,S,/S at two different potentials (USNa-electrode; see Fig. 2) at 33O’C. Figures at the semicircles Indicate the frequency in l/s.

“transpassive” region are non-stationary (Rds is higher than determined from Fig. 2), but in the active region are stationary (there are no polarization effects in this region). The double layer capacity can be calculated by Co = 1/2rrfR, where f is the frequency at the maximum of the semicircle and R the imaginary part of the resistance at this point. In Fig. 4 these values together

0

Fig. 4. Double layer capacity

function

0.5

1.0

with the “resistance” R,,- R,, are plotted for different potentials. It can be seen that the capacity has two plateaus which is the general bchaviour of passive materials like, eg chromium in aqueous solutions[8]. In the active region the capacity is high and its value corresponds to a film-free double layer in the molten salt systems[9]. In the passive and transpassive region,

1.5

2.0

2.5

lo

VvsNa

and electrode resistance (Rdu- R,,) of alumniurn in molten Na,S,/S of the electrode potential, as calculated from impedance plots. Temperalure: 330°C.

as a

R.

1036

KNOEDLER

AND

however, the capacity decreases about a factor of 50. This means that a passivating layer is present which is about 100 A (order of magnitude) thick. The resistance R,,R,, is lowest in the active region (however, the value of stainless steel is still lower). reaches a maximum in the passive region and takes on a medium value in the transpassive region. This corresponds well with the behaviour of the capacity indicating the presence of passive layers. From the findings described above some suggestions can be made on the processes involved. It is possible that the passive region is governed by a corrosion type process: anodic reaction:

Al -+A13+ +3e-,

cathodic reaction:

312 S +3e-

--f 312 S2-.

The overall reaction at the aluminium electrode thus is simply the formation of Al,S, which protects the surface from rapid corrosion. However, this effect also means a very slow cathodic sulphur reduction which prevents the use of Al as current collector. It must be stressed that the above reactions can go on also without electric current (Like corrosion of some metals in acid solutions by reducing H’-ions). In the active region (below about I.0 V cs Na) the AI,S, layer will be reduced, thus exposing a free surface with low resistivity. In the “transpassive” region (beyond 2.08 V GSNa) both polysulphide and aluminium are oxidized:

The electrons required for these reactions are furnished by the reactions at the counter electrode. As the S2- ions will come from the melt as well as out of the AI,S, layer, it is conceivable that for this reason theAl,S, layer will undergo some kind of“activation”. Therefore, the resistance should bc similar to that in the active region, which in fact was observed. It is not possible with the present experiments to extract data on the actual corrosion rate of aluminium. However, from theconsiderations given above it seems obvious that corrosion can take place only in the “transpassive” region. Corrosion of aluminium in an Na/S cell will therefore be expected mainly during the charging process. Examinahon of the electrodes after termination of the experiments showed that the investigated aluminium and aluminium alloys corrode in this environment. Besides a more uniform roughening of the surface pitting was observed. This last form of corrosion will be the most dangerous one. because it

S. MENNICKE

can lead to holes in the cell casing within rather short

times. The present preliminary investigations allowed no distinction between Al-alloys in this direction. Also it will be necessary to investigate the corrosion rate in dependence of‘ discharge current, charge and composition of the melt (= state of discharge).

CONCLUSIONS It could be shown that aluminium in molten sodiumpolysulphide shows an active, a passive and a “transpassive” region dependent on the applied potcntial. The potential of transition from the active into the passive region can be influenced by alloying the aluminium. In contrast to previous observations under stationary conditions (without current flow), aluminium was observed to corrode when electric currents are applied. Therefore, either corrosion resistant (espccially against pitting) alloys or high-quality corrosion resistant coatings have to be developed for application in Na/S cells. If a corrosion resistant alloy can be found, the coating (which is necessary to provide good conduction) must not be as reliable as in the case of a corrosion susceptible substrate. However, more work must be done in order to elucidate the corrosion processes in more detail.

A~knowl~,dgfm~nrs~The work was sponsored m part by the German Ministry of Research and Technology. We would like to thank Dr. F. Harbach and Mr. W Bansemir for their valuable assistance with the impedance measurements.

REFERENCES 1. W. Fischer, Solid-Sr. lonics 3, 413 (198I). 2 B. Hartman. J. Power Suurct~ 3, 227 (1978). 3. A. Wicker, G. Desplanches and H. Saisse. TItin S&d Film\ 83,437 (1981). 4. D. S. Park and D. Chatterji, Thin Solrd Films 83,429 (1981). 5. B. Dunn, M. W. Breiter and D. S. Park, J. crppl.

Electrochem. 11, 10.1(1981). 6. A. R. Tilley and M. L. Wright, Proc. 16th IEL‘,N‘, Atlanta. GA. p. 84i. August (1981). 7. F. Harbach and W. Basemir, to be published. 8. R. KnGdler and K. E. Heusler, Elec;rochun. Acta 17. 197 (1972). 9. H. A. Laitinen and R. A. Osteryoung. in Fusrtl S&s. (Edited by 8. R. Sundheim) p. 271. McGraw-H111,I*lew

York (1964).