Corrosion Science 47 (2005) 3187–3201

www.elsevier.com/locate/corsci

Impedance studies of the passive film

on aluminium
F.J. Martin a, G.T. Cheek b, W.E. OÕGrady c,
P.M. Natishan c,*
a
GEO-Centers, Inc., Fort Washington, MD 20749, USA
b
Chemistry Department, US Naval Academy, Annapolis, MD 21402, USA
c
US Naval Research Laboratory, Code 6314, 4555 Overlook Avenue SW,
Washington, DC 20375-5343, USA

Available online 9 September 2005

Abstract

Impedance spectroscopy has been used to study the characteristics of aluminium in 0.10 M
NaCl solution at potentials below the pitting potential (
1.50VSCE to
0.70VSCE). Multiple
capacitive (or pseudocapacitive) impedance features have been identified and analyzed over this
range of potentials. Higher frequency resistances have been compared with surface analytical
data, establishing a correlation between higher frequency impedances and the concentration
of chloride in the passive film. The relationship shows that both charging resistance and
chloride concentration reach a maximum immediately prior to passive film breakdown. At
the onset of metastable pitting, both the impedance and chloride concentration decrease from
the measured maxima. In contrast to X-ray photoelectron spectroscopy studies, the impedance
results do not reveal oxide thinning prior to the onset of metastable pitting. Modelling from
impedance data indicates that the thickness of the oxide layer sampled by impedance is much
thinner than the oxide measured with surface analytical techniques, suggesting that the imped-
ance method senses only the space charge layer in the oxide and not the total film thickness.
Ó 2005 Published by Elsevier Ltd.

Keywords: Aluminium; Passivity; Pitting corrosion; Impedance; Space charge layer; Oxide film

*
Corresponding author. Tel.: +1 202 767 9255; fax: +1 202 767 4642.
E-mail address: natishan@nrl.navy.mil (P.M. Natishan).

0010-938X/$ - see front matter Ó 2005 Published by Elsevier Ltd.

doi:10.1016/j.corsci.2005.05.058
3188 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

1. Introduction

An understanding of the breakdown of the passive film on Al has long been

sought [1–10]. Prior work at the US Naval Research Laboratory involved polarizing
pure aluminium in deaerated 0.10 M NaCl solution and examining the oxide film
formed at various potentials below the pitting potential using X-ray photoelectron
spectroscopy (XPS) and X-ray absorption near-edge structure (XANES) techniques
[11,12]. Comparison of metallic and ionic aluminium (+3) XPS intensities showed
the thickness of the oxide layer was 5 nm between
1.030 V to
0.80 V, and
4 nm between
0.75 V to
0.70 V [11]. The XPS data also revealed the presence
of two chloride species in the oxide layer. The total amount of chloride in the oxide
film, determined from fluorescence XANES data, reaches a maximum at
0.80 V
[12]. This information leads to further questions regarding the overall structure of
the layer, and the contribution of the chloride to electrochemical changes in the pas-
sive film prior to pitting. To investigate these questions an electrochemical imped-
ance study has been performed on this well-analyzed passive film system.
Several prior studies of the impedance behaviour of aluminium in chloride solu-
tions have been published [13–16]. In 0.5 M NaCl at
0.90 V (passive region), Bes-
sone et al. [13] obtained two capacitive loops linked by an inductive response. These
results were modelled as a single barrier layer on aluminium. With onset of pitting
(
0.74 V), only one capacitive loop was observed, whereas active pitting (
0.72 V)
produced a closed low-impedance loop, characteristic of corroding systems. Similar
results were obtained in 0.5 M NaCl by other investigators [14], although impedance
results were reported only in the passive region. These authors observed a lower-
frequency capacitive loop which was attributed to film dissolution, in which a soluble
aluminium salt was formed. In 0.01 M NaCl solutions, Lee [15] also reported two
capacitive loops, with no intervening inductive response, in the passive region. In
contrast to Bessone et al.Õs work [13] the two-loop behaviour persisted into the pit-
ting region, undoubtedly reflecting the much lower chloride concentrations in the
Lee study [15]. The high-frequency loop was taken as evidence of chloride incorpo-
ration into the oxide film; however, this loop does not appear to be present in studies
at higher chloride concentrations [13,14]. A study by Brett et al. [16] in solutions of
0.4 M K2SO4 and 0.10 M KCl clearly shows the effect of chloride in promoting pit-
ting corrosion; however, the impedance data were not shown for frequencies below
0.01 Hz. The presence of sulphate also raises some concerns regarding its participa-
tion in the film structure. Our previous spectroscopic studies on aluminium surfaces
were carried out in 0.10 M NaCl solutions, prompting our present investigations of
the impedance behaviour of aluminium in the same medium.
Despite this wealth of impedance information on aluminium in chloride solutions,
the experimental conditions employed by other investigators cited above vary con-
siderably and are not those used in our previous surface analytical work. By using
the same solution conditions (0.10 M NaCl) in both our previous surface analytical
work and our present work, a correlation has been obtained between the impedance
of the passive film on aluminium and the concentration of chloride in the passive
film.
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3189

2. Experimental

Electrochemical impedance studies were carried out on 99.99% pure alumin-

ium rods (Alfa-Aesar; 0.95 cm nominal diameter), in deaerated (with Ar) 0.10 M
NaCl solutions. All specimen surfaces were prepared by dry polishing the
aluminium to 600 grit, and degreasing in acetone, ethanol, and methanol followed
by rinsing with de-ionized water and drying with nitrogen gas. The working area of
each specimen was bounded by 3 M 1490 anti-crevice electrochemistry tape (3 M,
St. Paul, MN), shielding both the air–water interface from contributing to the elec-
trochemical impedance signal. Rod specimens were machined to create a hemi-
spherical end having a radius of curvature similar to the cylindrical portions of
the rod. Tape was placed circumferentially around the stem of the rod exposing
3.9 cm2.
The reproducibility of measured impedance information followed the reproduc-
ibility of polarization currents. Polarization currents and impedance values mea-
sured in the course of this work were consistently reproducible to within 5%
experiment-to-experiment.
Solutions were prepared using type I water and reagent-grade sodium chloride
(ACROS 42429). Solutions were deaerated by bubbling with argon for a minimum
of 1 h prior to immersing the specimen into the solution. A PAR Green Cell (Ametek
Instruments, Oak Ridge, TN) was employed for all experiments. A saturated calomel
reference electrode (SCE) in a Luggin–Haber capillary and platinum counter elec-
trode were used. All potentials in this paper are reported with respect to the SCE.
Open-circuit potentials were monitored for 12 hours prior to any polarization or
impedance measurements. Linear polarization resistance and impedance at open-
circuit potential were then measured prior to polarizing the specimens away from
open-circuit potential. Impedance scans were acquired under potentiostatic condi-
tions, starting at the lowest potential (
1500 mV) and progressing in increments
of 25, 50, or 100 mV to potentials above well above the pitting potential of

690 mV. The specimens were maintained at each applied potential for one hour
prior to recording the impedance spectra. The impedance characteristics of the
pre-pitting behaviour of the passive film on aluminium at potentials between

1500 mV and
700 mV are reported in the present body of work.
Impedance characterization was performed at each of the potentiostatically-main-
tained potentials using a PAR 273A potentiostat (Ametek Instruments, Oak Ridge,
TN) coupled with a FRA1250 frequency response analyzer (Solartron Analytical,
Hampshire, UK). Considerable care was taken to optimize the potentiostat/FRA
interface by using the shortest possible BNC cables for connection between the
instruments. In addition, inductive coupling between the cell cables was minimized
by using spiral wrap ribbon to bind the cables together. Impedances were
characterized over the frequency range from 65 kHz to 1 mHz, using a 10 mV
RMS potential stimulus. Potentials were maintained without solution resistance
compensation. Impedance-derived solution resistance compensation was, however,
performed during data analysis. The low currents associated with passive region
polarizations resulted in a negligible difference between the raw potentials and the
3190 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

iR-compensated potentials. Equivalent circuit modelling was performed using

ZView2 software (Scribner Associates, Southern Pines, NC) version 2.80.

3. Results and discussion

The potentiodynamic curve shown in Fig. 1 demonstrates the overall polarization

behaviour of aluminium in 0.10 M NaCl as a function of applied potential. The ano-
dic currents begin at
1.27 V, consistent with the open-circuit potentials observed
for aluminium samples in this work. The manifestation of passivity persists as the
potential is increased to a value exceeding
0.7 V. The relative insensitivity of cur-
rent to increasing potential (characteristic of the passive region) changes abruptly
as passivity breaks down at
0.69 V, where a strong potential dependence of the cur-
rent emerges and stable pitting ensues. The present paper is concerned with the po-
tential region from
1.50 V to
0.70 V and deals with the characterization of the
passive film.
Potentiostatically-measured impedance data for this potential range is shown in
Figs. 2 and 3. Fig. 2 shows two well-defined semicircles in the Nyquist plot at

1500 mV, indicating a two time-constant response. The impedance increases mono-
tonically as the potential is stepped to more positive values over the potential range

-0.6

-0.7

99.99% Aluminum Foil

-0.8 deaerated 0.1M NaCl
scan rate 0.6V/hr
Potential (V SCE)

-0.9

-1.0

-1.1

-1.2

-1.3
1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02
Current Density (A/cm²)

Fig. 1. Potentiodynamic polarization of aluminium in deaerated 0.1 M NaCl solution.

 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3191

-30000
-1100mV
-1200mV
-1300mV
-1400mV
-1500mV
-20000
Z" (ohms)

-10000

0
0 10000 20000 30000
Z' (ohms)

Fig. 2. Nyquist plot of Al impedance in deaerated 0.1 M NaCl, measured over the potential range

1500 mV to
1100 mV; frequency range 65 kHz to 1 mHz; electrode area 3.9 cm2. Specimens polarized
at 25 mV increments, from
1500 mV to
625 mV. Total time at each potential including impedance
measurement: 4 h.

shown in Fig. 2. A mere hint of the lower frequency semicircle remains at
1200 mV,
and is completely absent beyond the lower frequency limit of 1 mHz as the potential
approaches
800 mV (Fig. 3). Persistence of the lower frequency process has been
observed in measurements extending to 0.1 mHz, and representative curves are
shown for
750 mV and
1000 mV in Fig. 3. The higher frequency time constant
observed on Al at
1500 mV takes on a compound character as a secondary curva-
ture clearly emerges at potentials above
1000 mV. Thus the two time-constant re-
sponse observed at
1500 mV becomes a three time-constant response at potentials
above
1000 mV. Observance of the low frequency time constant at potentials above

1300 mV requires the use of frequencies well below 1 mHz.
Analysis of this low-frequency feature is not the intent of the present communica-
tion. Several of the authors referenced above have attributed these low frequency
features to transport phenomena in the oxide film [13,14,23]. Anecdotally it has been
noted in our experiments that this feature is strongly dependent on the DC polariza-
tion current, and can thus be strongly influenced by dissolved oxygen in the 0.1 M
NaCl solution.
Fig. 4 is a plot of imaginary impedance versus frequency for the data already
shown in Figs. 2 and 3. Plotted in this manner, changes in capacitance, time con-
stant, and whole cell impedance are readily visible. The capacitive portion of the
impedance spectrum—the imaginary impedance at frequencies 1 Hz and higher—
indicates that capacitance highest for
1400 mV and decreases as the potential
3192 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

-150000
- 750mV (0.1mHz)
- 750mV
- 800mV
- 900mV
- 1000mV (0.1mHz)
-100000 - 1000mV
Z" (ohms)

-50000

0
0 50000 100000 150000
Z' (ohms)

Fig. 3. Nyquist plot of measured Al impedances in deaerated 0.1 M NaCl solution over the potential
range
1000 mV to
750 mV. Continuation of experimental data shown in Fig. 2, measured over the
potential range
1100 mV to
750 mV. Frequency range: 65 kHz to 1 mHz. Supplemental curves are
shown for
750 mV and
1000 mV, showing impedance response to a lower frequency limit of 0.1 mHz.

increases. Following Z00 MAX from
1500 mV through
750 mV, emergence of the
secondary curvature is clearly observed as this point moves to lower frequencies
(longer time constants).
Equivalent circuit analysis confirms the trends observed in Fig. 4. Equivalent cir-
cuits used to rationalize the impedance data are shown in Fig. 5. These circuits con-
tain constant phase elements. Figs. 6 and 7 detail frequency selections for fitting
Models 1 and 2, respectively. It should be noted that Model 2 (the two time-constant
circuit) is only applicable to impedance data at potentials
1000 mV and above.
Fig. 8 contains collected information on equivalent circuit components elements.
It should be noted that the CPE coefficient ranged between 0.907 and 0.925 for
the CPEÕs shown in Fig. 8. A decisive decrease in resistance occurs between

725 mV and
700 mV—potentials just below the pitting potential. Potentiostatic
currents at
700 mV (not shown) showed measurable metastable pitting events,
but no metastable pitting events were detectable at
750 mV, and only rarely at

725 mV.
The impedance information in Fig. 8 was used to calculate the effective oxide layer
thickness using the method of Frers et al. [14], as follows: Capacitance values ob-
tained in the modelling procedure are assumed to be partly due to the oxide layer,
and partly due to the Helmholtz layer. The oxide and Helmholtz layers are treated
as serialized capacitors, where the equivalent capacitance of the pair of layers can be
expressed by
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3193

105
-750mV
-800mV
Time constants -900mV
increase as -1000mV
potential -1100mV
104 increases -1200mV
-1300mV
-1400mV
-1500mV

103
Z'' (ohms)

102

Capacitance decreases
as potential increases

101

100
10 -3 10 -2 10 -1 10 0 10 1 10 2 10 3
Frequency (Hz)

Fig. 4. Plot of measured imaginary electrode impedance (Al in deaerated 0.1 M NaCl) versus frequency
over the potential range
1500 mV to
750 mV. Note the decreasing frequency of surface charge
processes, while the oxide transport time constant disappears from view at low frequencies. Frequency
range 65 kHz to 1 mHz; electrode area 3.9 cm2.

Modell 1 Model 2
Ru R Ru R1 R2

CPE CPE2
CPE2

CPE1

Fig. 5. Diagram of equivalent circuit models employed in analysis of electrochemical impedance data.

1 1 1
¼ þ ð1Þ
C DL C H C SC
ee0
C SC ¼ ð2Þ
d
Here CDL is the effective double layer capacitance, CH is the Helmholtz layer capac-
itance, e0 is permittivity of free space, and CSC is the effective oxide layer capacitance
(all capacitances are normalized for area, with units of lF/cm2). The nominal area of
the aluminium rods was 3.9 cm2. Assuming a Helmholtz layer capacitance of 50 lF/
cm2 [14], the oxide layer capacitance is then isolated using the double layer capaci-
tance rationalized from the equivalent circuit in Fig. 5. Considering that the oxide
3194 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

--20000

Z''(ohms)
--15000

--10000

-750mV
-
Model 1 (-750mV)
--5000 -1000mV
-1
Model 1 (-1000mV)
--1500mV
Model 1 (-1500mV)
0
0 5000 10000 15000 20000
a Z' (ohms)
105

104

103
Z''(ohms)

102

101

-750mV
100 Modell 1 (-750mV)
-1000mV
Model 1 (-1000mV)
10-1 -1500mV
Model 1 (-1500mV)
(-
10-2
10-3 10-2 10-1 100 101 102 103 104 105
b Frequency (Hz)

Fig. 6. Summary of equivalent circuit analysis for equivalent circuit Model 1: this equivalent circuit was
employed in modelling intermediate to high frequency impedance data at all potentials. Open markers
indicate the portion of impedance spectrum analyzed using circuit Model 1: (a) Nyquist plot of impedance
data for
750 mV,
1000 mV, and
1500 mV, (b) imaginary impedance versus frequency for these same
potentials.

layer is probably at least partially hydrated, the thickness of the oxide layer (d) is
obtained from Eq. (2) as a range of values, using a relative dielectric constant (e)
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3195

-100000

-7
-750mV
Model 2(-750mV)
-75000 --1000mV
Model 2(-1000mV)
Z" (ohms)

- 50000

-25000

0
0 25000 50000 75000 100000
a Z' (ohms)

105

104

103
Z" (ohms)

102

101

100
--750mV
-1 Model 2 (-750mV)
10 -1000mV
Model 2 (-1000mV)
10-2
10-3 10-2 10-1 100 101 102 103 104 105
b Frequency (Hz)

Fig. 7. Summary of equivalent circuit analysis for equivalent circuit Model 2: this equivalent circuit was
employed in modelling low to high frequency impedance over a limited range of potential. Open markers
indicate the portion of impedance spectrum analyzed using circuit Model 2: (a) Nyquist plot of impedance
data for
750 mV and
1000 mV, (b) imaginary impedance versus frequency for these same potentials.

of 9.0 [13] for the dry oxide layer (lower bound) and 40 for a partially hydrated layer
(upper bound). The latter value is suggested by work by Hiemstra et al. on hydrated
alumina layers [24]. The thickness range of the oxide layer as a function of potential
3196 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

60 250

50
200
Effective Capacitance (µF/cm²)

Resistance (Kohms-cm²)
40
150

30

100
20
Model 1 - CPE
Model 2 - CPE1
Model 1 - R 50
10
Model 2 - R1

0 0
-1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700
Potential (mV SCE)

Fig. 8. Passive film resistance and capacitance obtained by equivalent circuit analysis using Models 1 and
2 shown in Fig. 5 above. Note: Model 2 is applicable to a smaller range of potentials than Model 1.

is shown in Fig. 9. where the average values undergo a slight increase from 0.21 nm
at
1.00 V to 0.45 nm at
0.70 V.
This ‘‘electrochemical oxide thickness’’ information is compared to oxide thick-
nesses measured by Yu et al. [12] in Fig. 9. Even assuming some oxide layer hydra-
tion, there is an obvious disparity between electrochemically-calculated oxide
thicknesses and those obtained by XPS. XPS measures the total oxide thickness
and gives a value 5 nm that is somewhat potential independent up to about

0.80 V. The oxide film thins between
0.80 V and
0.75 V to a thickness
4 nm.The nature of the thinning process has not been determined. The decrease
in thickness is either the result of mass loss (oxide dissolution), or a reordering of
the oxide (compaction) and is a topic of ongoing research.
The large difference between the passive film thicknesses determined by XPS and
the passive film thicknesses derived from the impedance data, Fig. 9, raises the ques-
tion of the true meaning of the capacitance data. Treatment of capacitance data in
the semiconductor literature utilizes the concept of the space charge layer [17]. This
treatment suggests that the passive film is essentially a defect-rich oxide that passes
charge via vacancy motion. Young [17] proposed that a depletion region develops in
the passive film at the film/solution interface. This was reiterated by Myamlin and
Pleskov [18] and more recently in a review by Gomes and Van Maekelbergh [19].
The charge depletion layer develops as oxygen vacancies (for n-type semiconductors)
or metal ion vacancies (for p-type semiconductors) transfer charge across the film/
solution interface [18–20]. Thus treatment of the capacitance from impedance mea-
surements provides the thickness of the charge depletion region or space charge
layer.
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3197

5
Film thickness (nm)

XPS data (Yu et al. ref 12)

3
e = 9 (CH 50uF/cm2 )
e = 40 (CH 50uF/cm2 )
e=9
2
e = 40

0
-1000 -900 -800 -700
Potential (mV vs SCE)

Fig. 9. Comparison of impedance-measured electrochemical oxide thickness, computed from capacitance

using various approaches, with XPS-derived oxide thickness [12]. Regardless of the dielectric constant or
treatment of Helmholtz capacitance, the space charge layer thickness is a fraction of the total oxide
thickness.

The space charge layer thickness (labelled electrochemical oxide thickness in

Fig. 9) is a small fraction (around 1/10 to 1/4) of the total oxide thickness. A small
increase is observed in the space charge thickness and this is due to the increased
penetration of the space charge layer into the oxide film over the potential range
shown in Fig. 9. The space charge layer does not appear to be affected by the
oxide film thinning that occurs prior to pitting. Thus, from a comparison of the
XPS and impedance results, it is apparent that the impedance measurements are
sensing only the space charge layer, and that the XPS experiments give the total
oxide thickness.
It should be noted here that the ‘‘point-defect’’ model often applied to oxide film
growth [25–27] apparently is not applicable here, since the actual oxide film thickness
in the present work does not continuously increase with increasing potential, as
would be required by the point-defect model. It is of some interest to note, however,
that the space charge layer thickness does involve a linear dependence on potential.
Equivalent circuit-derived capacitance values were also used to produce a Mott–
Schottky plot (Fig. 10), demonstrating n-type semiconductor behaviour in the pas-
sive region (positive slope) with an apparent flatband potential of
1.40 V. Note also
that the flatband potential predicted by extrapolating the linear region on the Mott–
Schottky plot coincides with the capacitance maximum shown in Fig. 8. N-type
semiconductor behaviour suggests that the primary charge carriers in the passive film
are oxygen vacancies.
3198 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

1800

1600

1400
R2 = 0.9943
1200
1/C² (mF/cm 2)-2

1000

800

600

400

200

0
-1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700
EFB = -1400mV Potential (mV SCE)

Fig. 10. Mott–Schottky plot of space charge layer capacitance for aluminium in 0.1 M NaCl, showing flat
band potential of
1400 mV.

The effect of chloride incorporation into the passive film on the impedance data
can be determined by referring to the X-ray fluorescence XANES data of Yu
et al. [12]. The penetration depth of the 2.3 keV X-rays permits determination of
the chloride present for the total thickness of the oxide and is reproduced in
Fig. 11. In order to obtain the chloride concentration in the passive film, the total
chloride content must be divided by the oxide film thickness, also reproduced from
the work of Yu et al. [12] and shown in Fig. 11. Normalizing the fluorescence data to
provide chloride concentration in the passive film produces the relationship shown in
Fig. 12, which is overlaid with the equivalent circuit resistance shown earlier in the
figure.
The data in Fig. 12 show an increasing amount of chloride in the passive film.
This can be related to (or is responsible for) the decreasing conductivity of the pas-
sive film at potentials prior to the pitting potential. The relationship between oxide
film chloride concentration and the space charge layer resistance is counterintuitive
but is entirely self-consistent. As we have already shown, the passive film on alumin-
ium is an n-type semiconductor; therefore, oxygen vacancies are responsible for
charge transport in the passive film. Incorporation of chloride into oxygen vacancy
sites [20] impedes the motion of these vacancies through the film. Restricted oxygen
vacancy motion results in a more resistive film, resulting in lower conductivity in the
space charge layer. Hence the resistivity of the film increases with increasing chloride
content up to
0.750 V. Above this potential the chloride concentration decreases,
the resistivity also decreases, and metastable pitting ensues at
0.700 V. This obser-
vation is supported by the work of Lee and Pyun [15], and Pyun et al. [21], who
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3199

2 6

1.9
5
1.8
Normalized Absorption Intensity1

1.7

Oxide Thickness (nm)

4
1.6

1.5 3

1.4
2
1.3

1.2 Oxide Cl- (Yu)

1
Oxide Thickness (Yu)
1.1

1 0
-1000 -950 -900 -850 -800 -750 -700 -650
Potential (mV SCE)

Fig. 11. Passive film chloride content and passive film thickness, as reported by Yu et al. [12]. Chloride
content is based on X-ray fluorescence XANES; oxide thickness is derived from XPS analysis.

0.40 250
Chloride Absorption Intensity/Oxide Thickness

Oxide Chloride Concentration

Space Charge Layer Resistance (Kohms-cm2)

Space Charge Layer Resistance
200
0.35

150

0.30

100

0.25
50
EPIT

0.20 0
-1600 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700 -600
Potential (mV SCE)

Fig. 12. Comparison of oxide film chloride concentration with space charge layer resistance. Refer to text
for discussion of oxide chloride concentration.

assert that chloride ions inhibit anodic dissolution of the passive film at potentials
below pitting and recent work by Wall et al. which shows that the resistivity of
3200 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201

aluminium oxide films that were ion implanted with chloride increased as the chlo-
ride concentration increased [22].

4. Conclusions

From a comparison of impedance and X-ray spectroscopic measurements of

potentiostatically polarized aluminium in NaCl solutions, it has been shown that a
relationship between chloride concentration and impedance characteristics of the
passive film exists. Increasing concentrations of chloride in the passive film are linked
to an increasing space charge layer resistance. This suggests that the presence of the
chloride hinders the motion of oxygen vacancies, the primary charge carriers in this
n-type semiconductor. Although related to the role of chloride in the passive film,
impedance data do not show a correlation with the measured thickness of the passive
film. The lack of correlation between impedance and oxide film thickness determined
by XPS is due to the fact that XPS allows the determination of the entire oxide layer
thickness, whereas the impedance technique senses only the space charge layer of the
oxide film. This space charge layer is of course, much thinner than the oxide film
itself.

Acknowledgements

The authors gratefully acknowledge support from the Office of Naval Research
and the NRL-USNA Cooperative Program for Scientific Interchange.

References

[1] T.P. Hoar, Nature 216 (1967) 1299.

[2] H. Bohni, H.H. Uhlig, J. Electrochem. Soc. 116 (1969) 906.
[3] U.R. Evans, Electrochim. Acta 16 (1971) 1825.
[4] Z. Szklarska-Smialowska, Corrosion 27 (1971) 223.
[5] R.T. Foley, Corrosion 42 (1986) 277.
[6] J. Kruger, Int. Mater. Rev. 33 (1988) 113.
[7] P.M. Natishan, E. McCafferty, G.K. Hubler, J. Electrochem. Soc. 135 (1988) 321.
[8] N. Sato, Corros. Sci. 31 (1990) 1.
[9] B. MacDougall, D.F. Mitchell, G.I. Sproule, M.J. Graham, J. Electrochem. Soc. 130 (1983) 543.
[10] G. Burstein, R. Souto, Electrochim. Acta 40 (1995) 1881.
[11] P.M. Natishan, W.E. OÕGrady, E. McCafferty, D.E. Ramaker, K. Pandya, A. Russell, J. Electrochem.
Soc. 146 (1999) 1737.
[12] S.Y. Yu, W.E. OÕGrady, D.E. Ramaker, P.M. Natishan, J. Electrochem. Soc. 147 (2000) 2952.
[13] J.B. Bessone, D.R. Salinas, C.E. Mayer, M. Ebert, W.J. Lorenz, Electrochim. Acta 37 (1992)
2283.
[14] S.E. Frers, M.M. Stefenel, C. Mayer, T. Chierchie, J. Appl. Electrochem. 20 (1990) 996.
[15] W.-J. Lee, S.-I. Pyun, Electrochim. Acta 44 (1999) 4041.
[16] C.M.A. Brett, I.A.R. Gomes, J.P.S. Martin, J. Appl. Electrochem. 24 (1994) 1158.
[17] L. Young, Anodic Oxide Films, Academic Press, New York, 1961, p. 150.
 F.J. Martin et al. / Corrosion Science 47 (2005) 3187–3201 3201

[18] V.A. Myamlin, Y.V. Pleskov, Differential capacitance, in: Electrochemistry of Semiconductors,
Plenum Press, New York, 1967, pp. 55–63.
[19] I.A.R. Gomes, D. Vanmaekelbergh, Electrochim. Acta 41 (1996) 967.
[20] E. McCafferty, Corros. Sci. 37 (1995) 481.
[21] S.-I. Pyun, S.-M. Moon, S.-H. Ahn, S.-S. Kim, Corros. Sci. 41 (1999) 653.
[22] F.D. Wall, C.M. Johnson, J.C. Barbour, M.A. Martinez, J. Electrochem. Soc. 151 (2004) B77.
[23] J.R. Scully, Electrochemical Impedance: Analysis and Interpretation, ASTM STP 1188 (1993) 26.
[24] T. Hiemstra, H. Yong, W.H. Van Riemsdijk, Langmuir 15 (1999) 5942.
[25] C.Y. Chao, L.F. Lin, D.D. MacDonald, J. Electrochem. Soc. 128 (6) (1981) 1187.
[26] C.Y. Chao, L.F. Lin, D.D. MacDonald, J. Electrochem. Soc. 129 (9) (1982) 1874.
[27] L.F. Lin, C.Y. Chao, D.D. MacDonald, J. Electrochem. Soc. 128 (6) (1981) 1194.