Ekcmchbnica Am, Vol. 37, No. 12. pp. 22834290. I!392 0013486/%?2s5.00 + 0.

00
Rioted in Great Britain. 0 1992. FergamollRem Ltd.

AN EIS STUDY OF ALUMINIUM BARRIER-TYPE OXIDE

FILMS FORMED IN DIFFERENT MEDIA
J. B. Bmsomr,* D. R. SALINAS,* C. E. MA=,+ M. Ennrtv and W. J. LoRENzt
*Institute of Electrochemical Engineering and Corrosion, Universidad National de1 Sur, Bahia Blanca,
Argentina
tInstitute of Physical Chemistryand Electrochemistry,University of Karlsruhe,Karlsruhe,Germany

(Received 30 Jamury 1992)

Abstract-Impedance measurementswere Performedon slutniniurnin differentaggressive(0.5 M NaCl)

and non-aggressive(0.5 M Na,SO,, 0.16 M NH,-tartrate)electrolytes.Similarimpedanceresponsesof the
formedpassive film were found. The capacitive behaviourobservedat high frequenciesis relatedto the
thickness and dielectric propertiesof the barrieroxide 6lm. The impedance spectra obtained at low
frequencieswereinterpretedin termsof a diffusion-controlledprocesswithin the oxide Clm.The Farad&
impedancegiven by the metal dissolution at the oxide f&n/electrolyte interphase is strongly alkted by
the anion type. This fact and the local pH developed at the interphase passive film/electrolyte determine
the chemical stability of the film. The increased impedance observed in chloride solution under rotation
conditions is explained by a local change of pH at the hhn/solution interphase. EIS can also be used to
study pitting phenomena. In the stabilixed active pitted area an ohmic control given by the presence of
a salt fihn is responsible for the capacitive and low frequency inductive behaviour obtained.

Key words: impedance spectroscopy, altinium, pitting corrosion, oxide film, passivation.

1. INTRODUCTION The aim of the present work is to clear up this

The passive behaviour of aluminium in aqueous
solution is strongly affected by the nature of the
electrolyte. The passive film structure and thickness
depend on both the electrode potential and the
electrolyte composition where it is formed. In sol-
utions containing aggressive ions such as chloride
ions, pitting corrosion as localized metal dissolution
can be initiated at active sites for the film breakdown
and at a certain potential[l, 21. Electrochemical
impedance spectroscopy (EIS) has been used success-
fully to examine the behaviour of relatively thick
barrier-type aluminium oxide films in non-aggres-
sive[3,4] and aggressive[5-7] forming solutions. The
properties and growth kinetics of thin aluminium
barrier films formed in ammonium tartrate at a
potential resulting in a film thickness of d Q 100 A,
have also been studied. In this case, standard poten-
tiostatic and galvanostatic measurement tech-
niques[l] as well as EIS[9] were used. The Faraday
impedance would correspond to a metal dissolution
process at the oxide !ilm/electrolyte interphase. More
recently[lO], the EIS results obtained at high and
medium frequency range in neutral solutions contain-
ing chloride ions below the pitting potential were
explained in terms of oxide film thickness and its
internal dielectric relaxation according to Taylor
and Dignam’s model[l 11. However, the explanation
for the results obtained at low frequencies seems to
be not without doubt. According to Nguyen and
Foley[l2], the relaxation time for the oxide 6hn
dissolution by the formation of chloride containing
aluminium complexes, largely exceeds the low fre-
quency range to which this reaction was ascribed.
apparent controversy and to assess the influence of
different anion aggressiveness on the stability of
aluminium passive films, in unstirred solution or
under welldetined hydrodynamic conditions, by EIS.
EIS was chosen as the principal experimental tech-
nique because of its capability of characterizing
different relaxation phenomena as well as active
pitting processes.
2. EXPERIMENTAL
Cylinidrical electrodes (r&r) were made from high
purity aluminium (Al 99.99%; main impurities: Cu,
0.002%, Fe, 0.004% and Si, 0.003%). They were
mounted in a Teflon holder and embedded with
epoxy resin (Araldite D) leaving a 0.50cm2 area
exposed. Surface polishing was successively per-
formed by means of: (a) wet polished 600 grit (Sic);
(b) red felt polishing cloth embedded with 1 pm and
then 0.3 lrn of A&Or powder; (c) finally, the mechan-
ical polishing of the electrode surface to a mirror
finish was followed by a weak alkali attack by
immersing the electrode in 0.1 M NaOH for 4 min.
The electrode surface was then thoroughly washed
with bidistilled water and immediately immersed in
the test solution where the operating constant poten-
tial in the passive range was applied and stability of
the passive film was attained aEer 2 h.
Impedance measuremen ts were carried out in
different electrolytes: 0.5 M NaCl (pH 6.95),
0.5 M Na,SD, (pH 6.3), and 0.16 M NH,-tartrate
(pH 6.0), at room temperatum. The solutions were
prepared from p.a. reagents (Merck, Darmstadt) and

2283
2284 J. B. BDSONEet al.
fourfold distilled water and deaerated by purified
nitrogen. A conventional three-compartment Pyrex
cell was used with a large platinum sheet and
saturated calomel electrode (see) as counter and
reference electrode, respectively. When appropriate,
a saturated Hg/Hg,SO, reference electrode was
used. All potentials in this work are referred to
the saturated calomel electrode [E&e) = 0.241 V
us nhe]. After pitting propagation, the attack
morphology was observed and analysed by scanning
electron microscopy (Cambridge Stereoscan Mk IIa)
with an EDX device attached (EDAX 711).
The electronic equipment consisted of a Solartron
Frequency Response Analyser 1170 remotely con-
trolled by an HP 9845B computer in connection with
an HP 9872A plotter. EIS spectra were obtained
under potentiostatic conditions by a Wenking poten-
tiostat LB75L after the system had reached a steady
state, indicated by a constant passive state current.
Impedance data were found to be independent of the
perturbation signal amplitude, AE, in the range
5 mV < AE < 15 mV. The experimental value chosen
was BE = 10 mV. The results presented were
measured at increasing electrode potential. The
measurements at each potential were performed at
least twice. In the passive region, a new electrode
surface was used for each run. In the pitting region,
propagation was left to evolve until a quasi-steady-
state current was achieved (> 5 h), otherwise a large
amount of scattering was observed at lower frequen-
cies. After reaching the steady state, successive runs
at different potentials within the active range were
measured, as well as the E-I stationary polarization.
The frequency range studied was 5 x lo-* <
f d lo4 Hz. The integration time was 10 cycles for
each frequency to provide higher precision, especially
in the lower frequency range. Under these experimen-
tal conditions, a long time stability of the system
under test and good reproducibility of the impedance
data were obtained.
3. RESULTS AND DISCUSSION
3.1. Steady-state behaviour
The stationary E-Z curve of Al in 0.5 M NaCl
through the whole potential range studied is shown in
Fig. 1, where two different regions are clearly ident-
ified (passive region A and pitting region C), separ-
ated by a narrow one (B). This region, the transition
or induction range, is characterized by current fluctu-
ations, whose amplitude and frequency increase as
the potential increases until pit propagation is
reached. Pit propagation is marked by a steep current
increase. A linear E-Z response is obtained for range
C, if the propagation is left to evolve until a constant
current is achieved. When polarization is reversed
from the pitting region C, at a certain potential the
slope of the linear E-Zcurve decreases again to match
the behaviour of the previous passive polarization,
but now the current is one order of magnitude or
more higher, corresponding to an increased area due
to pit surface irregularities. This clearly defined po-
tential might be the point where a critical chloride
concentration is reached and propagation initiates.
This, we may suggest, is the actual pitting potential
(E,) and its real meaning. Recently, Kaesche[l3]
explained this ohmic control by the presence of a thin
stabilized Al-halogenide layer produced from an
oversaturated salt solution present within the pit,
which gives some kind of salt passivity. The conse-
quence is a microtunnelling process which allows a
crystallographic attack following the (100) planes,
where Al dissolution is thought now to be under Al’+
ion mass transport control through this salt film. In
our case, after several EIS runs the electrode surface
was viewed by SEM and analysed by EDX. The same
crystallographic attack as that found by Kaesche
following (100) planes was observed (Fig. 2a) and
EDX denoted the presence of chloride inside the
pits. Also, the mud structure observed after drying
(Fig. 2b) suggests the presence of a film. We followed
Vetter and Strehblow’s procedure[ 141to avoid wash-
ing the electrode and the salt film, ie the electrode was
slowly removed from the aqueous solution, passing
through a benzene layer placed on it. In the passive
region A, the E-I behaviour is not well defined,
although a quasi-linear response appeared. The
stationary state polarization curves obtained in
0.5 M Na,SO, and 0.16 M NH,-tartrate are also
shown in Fig. 1.
3.2. EIS data
The impedance behaviour of this system in the
three potential regions (A, B and C) is depicted in
Fig. 3. Three loops were observed in the passive
region A, as previously described[lO]: (a) a capacitive
loop at high frequencies; (b) an inductive loop at
medium frequencies; and (c) a second capacitive
A
!
El-
7- A
POSSIW stat.
yg 6
1 5-
._
I-
.i
3- I
2- /li
.
lj;;;z-z;Jji:;
ES,-- /m”’ -
Fig. 1. Steady-state polarization curve for aluminium in
different electrolytes: (0) 0.5 M NaCI; (x) 0.5 M Na,So,;
(A) 0.16 M NH,-tartrate.
 An EIS study of Al barrier-type oxide films 2285

(a) (a) 6000

EScE = -900mV

“E LOW
:
G A
r 2000
‘;

o2ooooo
1500

“E
: 750
2 B
t!
E
‘T 0

0 750 1500 2250 3000

0
Re (Z) /R cm-*

-4
z23
E
j2
1

-03 -2 -1 0 1 2 3 4 5
log (f/Hz 1
Fig. 3. Characteristic impedance response of aluminium in
Fig. 2. SEM micrograph of Al in 0.5 M NaCl after pitting 0.5 M NaCl for the passive range A, the pitting induction B
at 0.680 V. (a) 1kx; (b) 10 kx. and the pitted active range C. (a) Nyquist plot; (b) Bode
plot.

response at low frequencies. The EIS data corre-

sponding to the transition region B show the same
behaviour but with large scattering in the low fre-
quency range. Here, current fluctuates markedly and
pitting initiation takes place. For region C, a quite
different EIS spectrum is observed. The capacitance
increases with potential, in accordance with Mansfeld
and Shih’s data[ 151which can be accounted for by an
increased active area as well as the presence of the salt
film detected within it. For s +O, the impedance
shown in the Nyquist plot intersects the real axis at
a point which nearly matches the Solution resistance.
(Here s =jo is the complex Laplace variable, and
cu = 2nf is the angular velocity of a perturbation
signal with frequency f.) This represents the polariz-
ation resistance 4 of the corrosion process, whose
value is coincident with that obtained from the slope
of the stationary E-Zcurve (Fig. 1) for this region. It
can be seen that the absolute value of the system
impedance (the impedance modulus) is reduced dras-
tically as we pass from the passive range to the pitting
range. Also, the onset of pitting can be clearly
determined by considering the low frequency range in
the Bode representation (1ogJZI us logf, Fig. 3b). It
presents capacitive behaviour when local defects re-
passivate, an indetermined behaviour when pitting
initiates (current fluctuations), and inductive be-
haviour when pitting propagates and reaches an
active state.
Figure 4 shows a set of Nyquist and Bode plots
obtained at different potentials in chloride solution.
2286 J. 9. -NE et al.
.............
.............
.............
.............
.............

.............
... ..a.......

: : solution ::
.............
.............
.............
.............
.............
.............
.............
.............
.............
mn fh
I II

2000 moo 6000 6ooo
Re (2)/n cm*
log (f/Hz)
Fig. 4. hpcdance diagrams for aluminium in 0.5 M NaCl
obtained at different potentials in passive region A. (V)
E= -0.9OV; (0) E= -0.85v; (0) E= -0.76V.
(a) Nyquist plot; (b) Bode plot.
As shown by the Bode representation, the capaci-
tance observed at high frequencies slightly increases
as the potential shifts to more positive values. Never-
theless, the same &, is reached as s +O. In general,
the same features appeared from the aluminium
passive state developed in different electrolytes
(Fig. 5). Increasing impedance was found in the
overall potential range for each system studied
(ZMlcl- < ZII,~,- < ZMlunnre). These results were
interpreted according the following model.
Fig. 6. Schematic model for the aluminium passive state.
3.3. Model description
3.3.1. Passive state. It is always expected that an
oxide Mm/aqueous solution interphase (f/s inter-
phase) will be more hydrated than a metal/oxide fihn
(m/f) interphase. Brock’and Wood found some evi-
dence in this sense when they studied relatively thick
ahuninium oxide film in different non-aggressive elec-
trolytes[3]. Nevertheless, under our working con-
ditions, although some degree of hydration has to be
present at the film/solution interphase, no evidence
for such a sandwich oxide/film emerged from our EIS
and steady-state polarization data.
Schematically, Fig. 6 gives the physical model
proposed to interpret the electrochemical impedance
spectra obtained from the aluminium passive state.
The oxide film dissolves via a chemical reaction at
interphase II:
AI( + mX- -+Al(OH),,& + mOH-. (1)
Under constant applied potential, depending on
anion aggressiveness, equation (1) will determine the
tilm thickness d. It is proposed that the transfer
function will follow the EIS model shown in Fig. 7.
Accordingly, the total impedance (Z) will be related
to the Faradaic impedance (Z,) by:
1
-=sc++ (2)
Z F
z,==z,+z, (3)
z,++-g$ 2

6 90 and
z _B tanh&
5- 75 T- T’ (5)
JtTs’

j k- 60
\ 3- 45 F
B \
2- 30 Q
s” 0
l- 15 ’
O- 0
-1-
-L -3 -4 -1 u I L .Y 4

log (f/Hz)
Fig. 5. Frequency dependence of (Z( and phase angle 4.
(A) O.SNacI, (0) 0.5MNa2504; (Cl) O.l6MNH,-
tartrate. Fig. 7. Impedance model for the aluminium passive state.
 An EIS study of Al barrier-type oxide tilms 2281

6 90
I

5 75 5- 75

Ns L 60 - L- 60
z
%3 45 F ; 3- 45
= 3:
cn2 30 e G 2- 30 L
9 f)
1 15 - l- 15 '

0 0 O- 0

-11
-4 -3 -2 -1 0 1 2 3 L
-15
- - _
--143-15
log(f/Hzl logIf/Hz1
Fig. 8. Simulated (-) and experimental (A) impedance Fig. 9. Rode plot for A&&)/O.5 M NaCl system, pH = 6.95,
diagrams for aluminium in 0.5 M NaCl solution at E = -0.85 V. (A) v,, = OS-‘; (0) v, = 16.7 s-‘; (0)
E = - 0.850 V. Vrol = 41.7 s-‘; (0) v,, = 58.3 s-‘.

where Zr is a function derived for a diffusion con-
trolled process within a finite length[l6], with
(6)
where 1, is the diffusion length, c. the concen-
tration of the diffusing species z. P’”or x = 0, D the
diffusion coefficient of species i, and the other vari-
ables have their usual meaning. The time constant rr
is given by:
l2
t+
Following the proposed model and the transfer func-
tion derived, good agreement between experimental
and simulated impedance data is observed (Fig. 8).
Notice that according to equation (2)-(6), the limit-
ing value of Z, for s +O is given by R,+ RT,which
should be equal to the polarization resistance Rp.
From the slope of the steady-state curve in Fig. 1,
R, = 12000 ohm cm2 was calculated, in good agree-
ment with R, + RT values given in Table 2.
When different rotation speeds (0 Q v,,,,< 58 s-i)
were applied to the rde, no meaningful change in the
diffusion length was observed (cJ Table 1; assumed
D = lo-’ cm2 s-i), and the impedance increased sig-
nificantly instead of decreasing (Fig. 9). As a conse-
quence, a possible Nernstian diffusion can be ruled
out.
If we consider instead a point defect model for
aluminium passive film growth and conductivity de-
veloped earlier[lTJ, it is possible to explain the pres-
ence of a diffusion controlled process within the oxide
film. The model contains these basic features:
(i) The oxide film (barrier layer) reaches a crys-
talline structure which contains a high concentration
of point defects, mainly V,, and V,-.
(ii) These cation and anion vacancies, as well as any
charged species, are in equilibrium at the m/f and f/s
interphases. This situation defines the concentration
gradient for each species within the oxide film.
(iii) The kinetics of oxide 6lm growth is under
transport control of vacancies across the film (inter-
action between point defects is assumed negligible).
Under these considerations the following reactions
are in equilibrium at each interphase (according to
Kroeger-Vink notation; the meaning of the symbols
is given elsewhere[ 11):
I m=M, + x/2Vo.. + xe’ (8)
m/f { m + V,,*M, + xe’ (9)
Vo..+H20=2H+ +00 (10)
M,G=V,,. + M=+ (11)
II Mx+ + xH,O=M(OH), + xH+ (12)
f/s For chloride solutions:
M(OH), + mCl-~M(OH),Cl,
1 (soluble) + mOH-. (13)
Apparently, anion vacancies (V,.) are produced at
interphase I and consumed at interphase II, equations
(8) and (IO), while cation vacancies (VM.,.) diffuse
from interphase II to I; equation (9) and (11). From
this set of reactions one can also envisage that in
unstirred solutions, combination of equations (12)
and (13) leads to a local acidification at the f/s
interphase, therefore the oxide growth can be sus-
tained by reactions at both interphases. The net
diffusion of Vo.. (or 0,) results in oxide growth at the
m/f interphase and depends on the concentration of
V,. On the other hand, the diffusion of V,, (or MM)
may result in metal dissolution or film growth at
this interphase depending on the type of electrolyte

Table 1. Rotation frequency dependence of the fitted parameters for Al/OS M NaCl at E = - 0.8 V

v,l s-’ Rnln~* C,/pFCm-2 R,lknCn+ &ILnCm2 L/kH cm2 RTlknCld ITIs LA
0 8.45 5.08 2.0 2.61 2.3 14.6 -246 9.7
8.3 7.06 5.57 13.20 61.4 292.0 6681 50.4
16.7 7.11 5.45 g 17.30 117.0 328.0 5516 45.8
33.3 7.38 5.52 21:o 18.60 196.0 362.0 5280 44.8
2288 J. B. BESONEet 01.

Table 2. Potential dependence of the fitted parameters for Al/O.5 M NaCl in passive region A
EIV wQ~z C,//lFCn+ R, /kQ cm* R2/knm2 L/kH cm2 RT/kficm2 h/S hIA
-0.80 3.02 8.34 0.79 1.47 1.46 10.48 262 9.9
-0.85 2.67 9.25 1.10 1.63 1.85 11.68 319 11.0
-0.90 3.51 9.61 1.66 1.81 2.12 12.08 344 11.4
-0.95 2.35 9.93 1.97 1.77 2.13 10.28 289 10.5
-1.00 2.66 10.53 2.44 1.87 2.54 10.61 3.07 10.8
-1.05 2.65 11.01 3.03 1.96 3.04 10.72 324 11.1

present, the local pH at the f/s interphase and the
potential difference applied, as will be seen later.
Depending on the anion type, different stabilized or
non-stabilized aluminium compounds are expected,
which would give rise to a dissolution-precipitation
mechanism. Anions such as chloride, perchlorate,
and nitrate, which have a poor chemical affinity for
Al’+ ions, react giving highly basic soluble salts[l8].
However, sulphate ions facilitate a quick basic salt
precipitation because the SOi- anions join the first or
second coordination sphere, avoiding or retarding the
polymeric OH-Al reaction, and ultimate hydrolysis
reaction[l8,19]. The same effect is expected in the
presence of the tartrate anion, where in addition its
buffer action plays an important role in stabilizing the
Al compound at the f/s interphase[20]. Under the
experimental conditions used[9, lo], the maximum
film thickness obtained in 0.5 M NaCl or 0.16 M
NH,-tartrate solutions is about 30 A. If we consider
now the time constant value obtained for the diffu-
sion controlled process from the fitting procedure
applied to equations (2)-(7), it yields a diffusion
coefficient D x 10-l’ cm* s-l. This value indicates
that the diffusion controlled process takes place
within the oxide film. From the potential dependence
of the fitted parameters (Table 2) obtained for
D = 3.8 x lo-“cm* s-i in 0.5 M NaCl solution, one
can observe increasing C values for decreasing poten-
tials. This capacitance may be related to a series
connection between the double layer capacity (C,)
and the oxide layer capacity (C,,,):
(14)
where C,, = u,,/d, where q, is the permittivity of the
vacuum (E,,= 8.85 x lo-‘* As V-’ m-l), .Zthe dielec-
tric constant of the oxide and d its thickness.
Since the f/s interphase is polarizable, it is expected
that the potential drop at this interphase, (pcS, is a
function of the applied potential, AE, and solution
pH. From previous investigations(91, a linear plot of
l/C trs EH is expected, where
d=uAE (15)
with AE = EH_EIo and Ec,o is the potential of zero
thickness of the barrier layer, which can be con-
sidered as the equilibrium potential of the aluminium
oxide film on aluminium. Therefore:
Thus, equations (1 5)-( 17) represent the functional
dependence of pprlron AE and pH.
As was mentioned above, unexpected results ap
peared when different rotation speeds were applied.
In chloride solution, the system impedance increased
drastically as compared with unstirred solutions
(Fig. 9); this effect was attenuated for increasing
rotation speed. In sulphate and ammonium tartrate
solution practically no difference was observed be-
tween the impedance spectra obtained with and with-
out rotation (Fig. 10). Table 1 presents the fitted
parameter values obtained under constant potential
and for different rotation speeds in chloride solution.
Clearly, rotation increased the film thickness. If one
plots l/C US(EH -El,) (Fig. 11) and includes data
obtained in NH,-tartrate ([9], Fig. lo), practically the
same slope is observed but different initial forming
potentials result. These differences can be ascribed to
the different local pH developed at the f/s interphase.
Comparing the chloride with the tartrate solution,
without rotation, and applying equation (17), the pH
values at the f/s interphase are three and six, respect-
ively. As mentioned above, equations (12) and (13)
explain a local acidification at the interphase. In
unstirred chloride solution, this facilitates the chemi-
cal film dissolution via a chloride basic salt. Con-
versely, in the presence of a buffer solution, eg 0.16 M
NH,-tartrate[Zl], the acidification is hindered and
the oxide film formation favoured. Also, the tar-
trate anion stabilizes the oxide[l8] instead of dissolv-
ing it, and the local pH fixed at the f/s interphase
increases the applied potential difference
(AE = E,, - E,,). As a consequence, a more efficient
film forming electrolyte results and the highest
impedance modulus is obtained (Fig. 5). On the
other hand, when the electrode is under rotation

On the other hand, a linear pH dependence of Ef.o

exists:
Fig. 10. Bode plot for Al(r&)/O. 16 M NH,-
2.303RT tartrate, pH = 6.0, E = - 0.60 V. (A) v,, = 0 s-‘;
Ef,o= E; - F pH. (17) (0) v,, = 16.7 s-‘.
 An EIS study of Al barrier-type oxide tilms 2289

..
..
..
..
..
..
:::salt::: ..
.......... ..
..
: solution: ..
.......... ..
.......... ..
.......... ..
.......... ..
.......... ..
.......... ..
.......... ..
.......... ..
al/ a1 ,.fJa
I II
Fig. 12. Schematic model for aluminium pitted active state.
ESCE ‘”
Fig. 11. l/C us E plot in different electrolytes. 0.5 M NaCI, increases many orders of magnitude with respect to
(A) “,t =Os-‘; (0) v,,= 16.7s-‘; (0) O.l6MNH,- the initial geometric area, which accounts for the new
tartrate, v,~ = 0 s-l.
impedance spectrum observed in region C. The physi-
cal model proposed for this case is shown in Fig. 12.
in chloride solution (Fig. 9), the local pH at the Figure 13 shows the EIS response for increasing
f/s interphase tends to match that of the bulk. This potentials in the pitting region. These EIS spectra
effect will reverse reaction (13), decreasing the correspond to the stationary state attained in region
chemical dissolution rate and increasing the film C (Fig. l), where practically the whole pitted active
thickness, therefore a higher impedance is observed. area matches that of the electrode surface and is
For increasing rotation speed, it is readily seen covered by a salt film (see Fig 2a and b). Pitting
that little additional variation, if any, is expected in phenomena generally are difficult to handle using
the impedance spectra. The presence of sulphate EIS techniques due to their evolutionary nature.
avoids or diminishes the hydrolysis reaction rate, However, when the propagation process has already
facilitating the basic salt precipitation and resulting
in a higher film forming efficiency than chloride
solution. Thus, rotation produces a small increase
in the sulphate impedance. For the NH,-tartrate
solution there is practically no difference in the
impedance spectra obtained with and without
rotation (Fig. 10).
3.3.2. Pitted active state. Pitting is a potential- and
time-dependent process. Evidence was reported of
local defects or flaws present at the thin oxide film:
aluminium surface decoration produced by galvanic
deposition of noble metal ions (Cu’+, Sri*+, In’+) and
surface amalgamation and film detachment in the
case of H$+ ions[2]; local H2 bubble evolution when
these electrodes are immersed in alkaline media[2];
transmission electron microscope observation of a
transverse aluminium/oxide film view[22]; and am- -so-
0 20 LO 60 80 100 120
plified hearing of oxide crack+].
These flaws may arise from surface irregularities or Re (Z)/Q cm2
impurities[22] or from accumulation of metal 3r 1
vacancies (“metal holes”) left by the diffusion of
metal cations from the m/f to the f/s interphases[23].
Under applied potential, mechanical stresses or elec-
trostriction effects may crack these flaws, relaxing the
film. Immediately the crack is healed by the fast
aluminium oxidation and reaction with the adsorbed
OH- ions. This is the case in passive region A.
At increasing potentials (region B), chloride ions
compete with OH- ions in the adsorption pro-
cess[8, 13,201 preventing quick repassivation of
cracked tlaws and resulting in a higher anodic current
flow, which is now detected (current fluctuations).
This is the process of pit incubation[24].
Once a certain chloride ion concentration is
reached within the pit embryo, repassivation is Fig. 13. Nyquist (a) and Bode (b) plot of aluminium
hindered and propagation follows. At this stage a in 0.5 M NaCl for different potentials in region C.
salt film has been established and the pitting area (0) E= -0.72v; (0) E= -0.7OV.
2290 J. B. BESSONE
taken place, a reproducible stationary state can be
reached. It is assumed here that no intact passive
surface area was left. The capacitive loop observed at
high frequencies defines a capacitance which may be
ascribed to a series connection between the double
layer capacity (C,) and the salt flm capacity (C,)
according to:
i=&+&. (18)
Increasing capacity values are obtained (16.3 Q
C < 53.0pF) for decreasing potential (Fig. 13).
Again, this may be explained by an increased salt film
thickness[25]. Within the potential region C, an
ohmic control given by the salt film developed is
effective[26] which defines an R, = 9.0 Qcm* (Fig. 1).
This value is in good agreement with that obtained
for s +O in Fig. 14, calculated as the difference
between the low and high frequency intercepts of the
real axis. Inductive behaviour is observed at low
frequencies. Increasing numbers of inductive loops
seem to be resolved as potential increases in region C.
They can be ascribed to surface area modulation or
salt film property modulations such as its density,
ionic conductivity or thickness. Another possibility is
the relaxation of an adsorbed intermediate species.
Further study is needed of this issue.
4. CONCLUSIONS
It is possible to interpret the EIS results of the
aluminium passive film formed in different electro-
lytes, especially in the low frequency range. Here, a
diffusion controlled process within the oxide film is
responsible for the EIS spectra observed. The passive
film is formed by reactions which take place at the
m/f and f/s interphases. The former is little influenced
by the type of electrolyte, while the latter is strongly
affected by the anion type, which will decide the
chemical stability of the compound formed and the
local pH. Both reactions, in turn, establish the film
thickness.
The increased impedance observed in chloride sol-
ution under rotation conditions is explained by the
increased pH produced at the f/s interphase by this
effect. The investigations in the pitting region show
that EIS can also be used as a powerful tool to study
this important corrosion process.
Acknowledgements-The authors thank the Consejo
National de Investigaciones Cientificas y Tecnicas
(C.O.N.I.C.E.T., Argentina) and the Stiftung Volkswagen-
werk (Germany) for financial support. Also, some ex-per-
imental results produced by Ings. M. M. Stefenel and S. E.
Frers are deeply acknowledged.
et al.
REFExENcEs
1. K. Videm, Kjeller Report, KR-149, Institutt for Atom-
energi, Kjeller, Norway (1974).
2. J. B. Bessone, Ph.D. thesis, UMIST, Manchester
(1983).
3. A. J. Brock and G. C. Wood, Electrochim. Acta 12,395
(1967).
4. H. J. de Wit, C. Wijenberg and C. Crevecoeur,
J. electrochem. Sot. 126, 779 (1979).
5. J. Hitzig, K. Jtittner, W. J. Lorenz and W. Paatsch,
J. electrochem. Sot. 133, 887 (1986).
6. W. Paatsch, W. J. Lorenz and K. Jiittner, Metall-
oberfldche 41, 6 (1987).
I. F. Mansfeld and M. W. Kending, J. electrochem. Sot.
136, 128 (1988).
8. J. B. Bessone, R. A. Suarez Baldo and C. E. Mayer, in
The anodic formation of barrier-type aluminium oxide
tilms, Proceedings of the 8th International Congress
on Metallic Corrosion, Mainz 1981. DECHEMA,
Frankfurt/M. (1981).
9. J. B. Bessone, C. Mayer, K. Jiittner and W. J. Lorenz,
Electrochim. Acta 28, 171 (1983).
10. S. E. Frets, M. M. Stefenel, C. Mayer and T. Chierchie,
J. appl. Electrochem. 20, 996, (19%).
11. D. F. Tavlor and M. J. Dianam. J. electrochem. Sot.
120, 1296(1973). - ’
12. T. H. Nguyen and R. T. Foley, J. electrochem. Sot. 129,
27 (1982).
13. H. Kaesche, Werkstofi Korros. 39, 152 (1988).
14. K. J. Vetter and H. H. Stmhblow, in Proceedings ofthe
R R. Evans Conference on Localized Corrosion,
Williamsburg, Virginia (Edited by R. W. Staehle, B. F.
Brown, J. Kruger and A. Agarwal), p. 240. NACE,
Houston (1974).
IS. F. Mansfeld and H. Shih, J. efectrochem. Sot. 135, 1171
(1988).
16. K. Jtittner, W. J. Lorenz, M. W. Kending and
F. Mansfeld, J. efectrochem. Sot. 135, 332 (1988).
17. C. Y. Chao, L. F. Lin and D. D. Macdonald,
J. electrochem. Sot. 128, 1187 (1981).
18. Pa Ho Hsu, Aluminimn hydroxides and oxyhydroxides,
in Minerals in Soil Environments (Edited by J. B. Dixon
and S. B. Weeds), pp. 119-122. Soil Science Society of
America, Madison, Wisconsin (1977).
19. H. De Hek, R. J. Stol and D. Bruyn, J. Colloid fnterfoce
Sci. 64, 72 (1978).
20. M. A. Bless, A. E. Regazzoni and A. J. Maroto, Mater.
Sci. Forum 29, 31 (1988).
21. A. E. Martell and R. M. Smith, in Critical Stability
Constants, Vol. 3. Plenum Press, New York (1977).
22. G. E. Thompson and G. C. Wood, Treatise on materials
science and technology, in Aqueous Process and Passive
Films (Edited by J. C. Scully), Vol. 23, Ch. 5, p. 220.
Academic Press, New York (1981).
23. L. F. Lin, C. Y. Chao and D. D. Macdonald,
J. electrochem. Sot. 128. 1194 (198 I).
24. T. R. Beck and R. C. Alkire, J: electrochem. Sot. 126,
1662 (1979).
25. T. R. Beck, Electrochtm. Acta 30, 725 (1985).
26. T. R. Beck, Electrochim. Acta 29, 485 (1984).