Abnormal polarization change in anodic dissolution of aluminium at high current densities

Pm•p mov Tress L[d. 1981 . Fm

SHORT COMMUNICATION ABNORMAL POLARIZATION CHANGE IN ANODIC DISSOLUTION OF ALUMINIUM AT HIGH CURRENT DENSITIES A. R . DES iC,, D . M . DKAljt, S . K . ZECEVlt and R . T. ATANASOSKI Faculty of Technology and Metallurgy and Institute of Electrochemistry, ICPM, University of Beograd, Beograd, Yugoslavia (Received 29 April 1980) INTRODUCTION

forming oxidation resulting in layers of significant thickness, rather regular structure and with good adsorbing qualities (representing the basis of deTwo types of anodic oxidationofaluminiuxnbavebeen corative anodizationn processes) . Extensive investirecognized in the past[ly oxidation requiring a steady gations in our laboratory in recent years indicate that a increase of voltage third type of oxidation should berecognized, following across the interface up to a few hundred volts in omde> to-overcome the resistance of seemingly normal anodic dissolution kinetics, ie inincreasingly thick compact oxide layer and maintain a crease in polarization with increasing current density steady oxidation current, and (b) a porous oxide layer of usual type exemplified by the dashed line of Fig . 1 .

drop

Fig . 1 . Anodic polarization curves for Al in 2 M NaCl, measured (dashed line) and corrected (full line) for pseudo-ohmic overpotential This type of dissolution has been obtained in chloride[2] chlorate and perchlorate solution and in some other electrolytes[3] . In this third type, it has been found recently[4], however, that a careful elimination of pseudo-ohmic overpotential leads to unusual type of polarization behaviour, whereby after reachingg some current density range, the polarization of the aluminium-electrolyte interface alone, decreases with increasing current density in an approximately linear fashion as shown in Fig . I (full line) . It is the

purpose of this short communication to report in more details about this phenomenon which is tantamount to appearance of negative equivalent resistance at the surface .

EXPERIMENTAL High-purity aluminium (99 .999%) in the form of cylindrical electrode with the base of 0 .2 ran' open to 173



1 74

A . R . DESPIC, D . M . DRAZtt,

S. K.

2 M NaCI solution in tripply distilled water, has been polarized anodically in a cell having a provision of moving the tip of Luggin capillary to precisely measured distances from the electrode surface . Beside the polarization circuit, which enabled recording steadystate current-potential relationships, a standard electronic circuitry provided for recording galvauostatic transients of short duration by superimposed pulses of constant current . Measuring procedure consisted of: (a) polarizing the electrode by a certain constant anodic current density (cd) and measuring the steady-state value (established after less than I min) of electrode potential . E (us sce) at a fixed distance, between the electrode surface and the tip of the Luggin capillary, (b) superimposing galvanostatic pulses of short duration (0.1 s) and a fixed value of current, i,, at different distances land recording the initial potential jump on the oscilloscope, (c) recording, at the same fixed dis-

ZECEVIC AND

R. T .

ATANASOSKI

lance I as in (a) and at the given steady state cd, a series of galvanostatic transients at increasing values of i, . This has been done to a series of steady-state cd. Prior to measurement, the electrode has been submitted to high-current density (500 mA cm 2 ) anodic dissolution in order to remove any previously formed films and impurities accumulated in the course of preparation. Some sets of measurements have also been made in 2 M NaCIO, and 2 M NaClO 4 solutions . RESULTS Galvanostatic time-responses were all of the same type shown in Fig . 2 . They consisted of an initial potential jump (IPJ) and a subsequent potential change exhibiting (except at low i,-s) a maximum after a few ms .

Fig . 2 . Typical galvanostatic time-responses for different superimposed anodic pulses : 4 .5 mA/cm' : Y = 50m V,:div X = ms/div . Plots of IPJ Ins the distance 1, from measurements at a certain constant value of current in the pulse, were found to be linear, extrapolating to zero 1 . This indicated that any ohmic resistance in the surface oxide layer is a part of a complex impedance and hence, that IPJ reflects ohmic resistance of the electrolyte only . Thus, using such a value of the resistance at a given l, at which the steady-state polarization measurements have been made, corrections for pseudo-ohmic polarization could be made at every steady-state rd . In this way corrected polarization plots of Fig. 3 cxtending over several orders of magnitude of cd, have been obtained, Characteristic are three regions : (a) normal polarization change, (b) region of independence of polarization on cd and (c) region of decreasing polarization with increasing cd (negative resistance). Superimposing galvauostatic pulses and measuring Aq,,,a ,-the height of the overpotential maximum with

56,

31, 18 and

respect to overpotential at the beginning of the transient (at the end of IPJ) rendered plots of Fig . 4 . The idea behind using Agmox is that in a first approximation it reflects the resistance of the oxide layer at the given steady-state cd, before changes in its structure have time to take place (while those changes caused by the fact that now the total current is larger than the steady-state one, are reflected in the change of overpotential after the maximum), it is seen (i) that at a given steady-state cd the plots are linear with increasing i„the slope depending on the steady-state cd, i, and (ii) at a constant i,Ag me7 depends linearly on the inverse of i . Similar dependences have been found in chlorate and perchlorate solutions, as shown in Fig . 3 . However, in those solutions chloride ions have also been discovered. (When, in a related piece of work, quantity of electricity pertaining to the dissolved metal



1 75

Abnormal polarization change in anodic dissolution of aluminium

Z

1.

i 10"

10' t./.A

10'

10

'

Fig, 3. Anodic polarization curves (corrected for pseudo-ohmic polarization) for AI in 2M solutions of NaCI(O). NaCIO I (n) and NaCIO 4 (Q) .

was calculated, from weight-loss of electrode polarized for some time by a constant current density, it was found that it checks with the sum of quantities of electricity pertaining to anodically dissolved metal, to hydrogen evolved and to chloride ions found in solution, assuming the latter to be formed by reduction of chlorate or perchloratc). One should note that similar polarization behaviour has been found for electrochemically active aluminium alloys, containing small amounts of gallium and indium (0 .2 % w/w) .

Hence, equations (1) and (2) can both hold only it' R(i)=

(3)

P qr

ri

where a, b, p, q are constants . The resistance R(i) calculated from the polarization plot (Fig. 1) using equations (2) and (3) is shown in Fig . 5. The plots of Hg . 4 substantiate the above relationships . One can maintain that (Arl,m,,li,); = R(1)

(4)

DISCUSSION

and hence, that it should obey the relationship (3). The plots are seen to be linear, ie

One could describe formally the cd-potential relationship by the Ohm's law, ie

(Aq,,,,,)1=RQ)ia

Ell) =E(o)+R(i)i

(1)

(5)

and

P - q1 _ry (Anmax)1,'r, ri

P 4 ri-r

a _1 y i-b

(6)

where E(o) is the open circuit potential and R(i) the equivalent interface resistance at current density 1. 1f the interface resistance R(i) is determined basically by the resistance to transport in the oxide film and if it remained constant, one would obtain an increase of potential in the positive direction, as it is obtained in the first part of the plot of Fig. 3 . The region of constant potential, however, signifies that the resistance of the oxide layer decreases in direct proportion with increasing cd, while the change in the negative direction indicates a larger decrease of the resistance than is the increase of the anodic cd . The polarization behaviour in that third region (Fig . 1) can be described by

The resistance R(i) calculated using (4) as a function of i is also shown in Fig. 5 . The discrepancy between the two lines can be understood if one takes into account that peaks in the galvanostatic transients are likely to deviate from values which would reflect the unchanged oxide layer resistance, the more so the closer is i, to i . I a attempting to understand the physical meaning of the above relationships one can assume the resistance to be determined by

E-E(o)=a-bi

where d,,, and rc o , are the thickness and the con-

(2)

R(i) = -

( 7)

176

A . R . DE6PIC, D. M. DRA2IC, S . K . ZEtEVIC AND R . T . A]Ammos I

5

Aeln'

100 120

250 450

tp/mAcm '

Fig. 4. Dependence of Aqm°" of superimposed anodic galvanostatic pulse on the pulse current il,(a) and inverse of the basic current,i(b}



177

Abnormal polarization change in anodic dissolution of aluminium

t

5

to

15

20

Fig 5 . Dependence of the interface resistance R(i) on the anodic current density calculated from Fig . 1 using (2) and (3) (a) and from Fig. 4 using (4) (b) .

ductivity of the oxide layer respectively . The simplest way of obtaining (3) is to assume dos = p

- qi

(8)

and K oy

=ri

(9)

Since the plots of Fig. 5 extrapolate close to zero this indicates small q values, ie that the major factor in preventing the expected increase in polarization of aluminium with increasing steady-state current density is the increase in conductivity of the oxide film . This must be due to increasing destabilization of the oxide film with increasing flux of anodic oxidation products, possibly concentration of Al' ions (which should also increase linearly with increasing current density), or increasing degree of disorder in the oxide . However, if it were for that factor only, the polarization would stay constant and independent of i . (One should note that if some power other than I was taken for the dependences (3) and (4) the linearities of Figs I and 4 could not be interpreted) . Hence, the negative resistance should be due to a

decrease in the oxide layer thickness with increasing current density of anodic dissolution . The finding that essentially the same behaviour occurs in chlorate and perchlorate solutions is likely to be due to the fact that chloride ions are always present there as well, under conditions of anodic dissolution of aluminium . The fundamental question, however, of the reasons behind the effect of chloride ions enabling continuous anodic dissolution remains open and will be discussed at another instance[3] .

REFERENCES 1 . J . W. Diggle, T . C . Downie, C . W . Goulding, Chem . Reu. 69, 365 (1969) 2 . G . Petrova, E. Berkman, E . Ivanov, V . Nikoloski, Sbor. robot po khim 1st_ toka 5, 183 (1970) and 5 . 187 (1970) 3 . D . M . Drafie, S . K . ZeLevic, R. T . Atanasoski, A. R . Despic, Short Communication (in preparation) 4 . R . T . Atanasoski, A . R. Despic, D . M . Draaic, S . K . Ze&vi6, Communication at the 1st Frumkin Memorial Symposium, Moscow, October (1979) .