Assessment of proton transport in amorphous aluminum oxide by cathodic polarization. Time of flight-elastic recoil detection analysis method

Article abstract of DOI:10.1149/1.1578479

A study has been made of proton migration in an anodic aluminum oxide film during cathodic polarization in 1 M hydrochloric acid solution at 30°C. The charge transported by proton migration in the oxide film during the polarization was quantitatively determined using time-of-flight-elastic recoil detection analysis. An exponential increase in ionic current with increasing cathodic potential indicated that proton conduction within the barrier-type anodic oxide film follows the high field conduction model. The corresponding parameters were also determined.

Full text of DOI:10.1149/1.1578479

Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
B329
0013-4651/2003/150͑7͒/B329/7/$7.00 © The Electrochemical Society, Inc.
Assessment of Proton Transport in Amorphous Aluminum
Oxide by Cathodic Polarization
Time of Flight-Elastic Recoil Detection Analysis Method
Jong Hyun Seoa and Dong Nyung Leez
School of Materials Science and Engineering, College of Engineering, Seoul National University,
Seoul 151-742, Korea
A study has been made of proton migration in an anodic aluminum oxide film during cathodic polarization in 1 M hydrochloric
acid solution at 30°C. The charge transported by proton migration in the oxide film during the polarization was quantitatively
determined using time-of-flight-elastic recoil detection analysis. An exponential increase in ionic current with increasing cathodic
potential indicated that proton conduction within the barrier-type anodic oxide film follows the high field conduction model. The
corresponding parameters were also determined.
© 2003 The Electrochemical Society. ͓DOI: 10.1149/1.1578479͔ All rights reserved.
Manuscript submitted August 5, 2002; revised manuscript received January 29, 2003. Available electronically May 16, 2003.
It is well known that an aluminum/oxide film/electrolyte system
exhibits rectification. Anodic polarization of an aluminum electrode
covered with oxide films gives rise to low currents, whereas ca-
thodic polarization results in high currents. Kliem1 investigated the
in corrosion and the rectification of aluminum oxide film in anodiz-
ing have been carried out separately. Similarities exist between the
rectification of aluminum oxide film on cathodic polarization and
the deterioration of anodic oxide film by cathodic polarization in
aqueous solutions. Both phenomena occur at a relatively low electric
field strength of about 107 V/m. The deterioration of the oxide film
on cathodic polarization in aqueous solution may account for recti-
fication phenomena in relatively low cathodic potential regions.
However, it is still not known if the cathodic conduction in the
oxide film follows a high field conduction mechanism, like the an-
odic conduction, in which the ionic flow rate increases exponentially
with electric field across the film.
Some electrochemical methods including impedance spectros-
copy and the galvanostatic step method have been proved effective
in studies of the ionic conduction of the oxide films.6,9,15-17 Never-
theless, the techniques possess difficulties in interpretation of results
and direct separation of ionic current from measured total reduction
current. However, conventional hydrogen detection methods have
some advantages, especially in the study of proton conduction
within thin films. It is possible to quantify the ionic charge directly
from the hydrogen quantity detected in the sample. The change in
the hydrogen concentration profiles also gives further insights into
the hydrogen ionic conduction process. In the study of anodic alu-
minum oxide films, only a few researchers have reported the hydro-
gen distribution and concentration within the anodic barrier alumi-
num oxide depending on the anodizing conditions using secondary
ion mass spectroscopy18 and nuclear magnetic resonance analysis,19
but no attempt has been made to investigate the change in the hy-
drogen distribution and amount within the oxide film caused by
cathodic polarization using the conventional hydrogen detection
methods.
Among the conventional hydrogen detection techniques, we
chose time of flight-elastic recoil detection analysis ͑TOF-ERDA͒ to
determine the hydrogen content because it involves a relatively
simple preparation of target samples and a depth resolution good
enough to be applicable to the thin anodic oxide films used in this
work. TOF-ERDA has become a technique for many light element
profiling applications because it offers the possibility of simulta-
neous multielement profiling,20 and also offers fair sensitivity,21 and
enhanced depth resolution compared with conventional ERDA.22
TOF-ERDA is a technique that uses elastic collision phenomena
between target atoms and incident particles with high energies. Be-
cause the mass of the incident particles ͑in this case 35Cl5ϩ) is much
greater than that of hydrogen in the sample, the incident energy is
transferred primarily to the hydrogen atoms. Therefore, the detected
particles are not the incident ions but the hydrogen atoms recoiled
from the sample. The energy and mass of the recoiled hydrogen are
measured independently by combining a measurement of the ions’
flight time with that of the total energy.
dielectric relaxation of thin Al2O3 films in
a dry system
(Al/Al2O3/Au͒. He attributed the nonlinear current-voltage relation-
ship to a tunneling process of protons existing in the oxide film.
However, there was no difference between the anodic current flow
and cathodic current flow, indicating that the rectification of the
aluminum oxide film does not occur in a dry system. Takahashi
et al.2 conducted the measurements of current flow through an an-
odic aluminum oxide film in two systems, dry and wet. In the dry
system, they also observed no rectification and suggested that the
phenomenon is due to a high rate of hydrogen reduction during
cathodic polarization in solution, based on proton conduction via
flaws of the film. Young3 and Vermilyea4 suggested that weak spots
such as fissures in the oxide film are responsible for the low cathodic
resistance and also for most of the observed leakage current during
anodic polarization. In cathodic polarization, hydrogen ions, because
of their small size, may travel easily through the fissures to be dis-
charged. In anodic polarization, small ions are not available in acid
solution.
There is a fundamental difference between dry, consisting of
aluminum/Al2O3/metal, and wet systems. In a dry system, the ap-
plied voltage measures the difference in Fermi level across the sys-
tem, and this is the driving force for the current. In a wet system,
charge is carried across the interface by an electrode process. If the
oxide film is an excellent ionic conductor, in the dry system, the
number of charge carriers within the oxide film is fixed, while the
quantity of the charge carriers within the film can be varied by the
addition of charge carriers such as oxygen ions or protons from the
electrolyte.
The proton migration through the aluminum oxide film strongly
influences deteriorations of the oxide film at cathodic polarization
such as compositional and structural changes of the film5-9 or me-
chanical rupture of the film.10 We have observed blistering of anodic
oxide film on aluminum in hydrochloric acid by cathodic polariza-
tion. Blister formation and rupture have also been observed in the
pitting of aluminum and its alloys.11-14 The oxide blistering is re-
garded to occur as a result of excessive generation of hydrogen
atoms at the metal/oxide interface, which is attributed to the proton
migration caused by an electric field that builds across the oxide
film.
The studies on the cathodic breakdown of aluminum oxide film
a Present address: Samsung Electronics Co., Ltd., LCD R and D Center, AMLCD
Division, Yongin 449-711, Korea.
z E-mail: dnlee@snu.ac.kr
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B330
Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
Figure 1. Schematic geometry for TOF-ERDA experiment. T1 and T2 are
carbon foil time detectors separated by flight path of 1 m, and E is silicon
surface barrier detector.
Because little is known about the kinetics of hydrogen transport
through barrier-type aluminum oxide films during cathodic polariza-
tion, this study focused on the direct measurement of the ionic
charge transported by the proton in the oxide film using TOF-
ERDA. The parameters for the ionic conduction of the proton and
the diffusion of hydrogen in barrier-type aluminum oxide film were
obtained.
Experimental
All specimens were cut from 0.8 mm thick cold-rolled aluminum
sheet of 99.98% in purity. Prior to anodizing, they were degreased in
acetone and then electropolished in a mixture of perchloric acid and
ethanol at 20 V for 5 min. The barrier-type anodic oxide films were
grown on the aluminum sheet by anodizing in ammonium adipate
solution at 60°C. Initially, anodizing was conducted at a constant
current of 5 A/m2 until the anodizing voltage reached a desired
value and then switched to a constant voltage mode. After anodiz-
ing, the edge was carefully coated with a silicon adhesive to prevent
current concentration on the edges of the sample during cathodic
polarization experiments. The area exposed to the electrolyte was 3
cm2. Polarization experiments of the anodized aluminum specimens
were performed in 1 M HCl solution at 30°C using an EG&G 273A
potentiostat. A saturated calomel reference electrode ͑SCE͒ was
used, and all potentials are referred to the SCE in this paper.
After the specimens were tested electrochemically, they were
rinsed thoroughly in deionized water. The specimens were examined
using TOF-ERDA right after the polarization experiments. TOF-
ERDA was used to measure the hydrogen contents in the samples.
The geometry of the experiment is shown in Fig. 1. A 10 MeV
35Cl5ϩ ion beam was incident on the sample at an angle of 75° to the
sample normal, and recoiling hydrogen atoms were detected with a
time-energy telescope. The scattering and solid angles were 30° and
0.001 msr, respectively. For hydrogen concentration profiles, the en-
ergy scale is converted into the depth scale using the stopping cross
sections tabulated by Ziegler23 and the yield scale is converted into
the concentration by a spectral scaling method.24 The hydrogen
Figure 2. TEM micrographs of ͑a͒ plane view of stripped anodic oxide film
along with electron diffraction pattern and ͑b͒ ultramicrotomed section of
anodic oxide film on aluminum formed at 30 V at 5 A/m2.
depth resolution of TOF-ERDA using the 10 MeV Cl5ϩ beam is
about 3 nm.
Results
Characterization of anodic oxide film.—Figure 2a shows a trans-
mission electron microscopy ͑TEM͒ image and a electron diffraction
pattern of an anodic oxide film formed at 30 V in ammonium adi-
pate solution. The electron diffraction pattern consists of four or five
very broad diffuse rings, suggesting an amorphous structure. The
TEM image at high magnification shows a fine grain-like morphol-
ogy of the order of 2 nm. Figure 2b shows an ultramicrotomed
section of a 44 nm thick barrier-type oxide film on aluminum.
Cathodic polarization of anodic oxide film.—Figure 3 shows the
hydrogen evolution current density transients of aluminum anodized
to have 44 nm thick oxide film when applied in 1 M HCl solution at
30°C under cathodic potentials of Ϫ2 to Ϫ3 V. At potentials of
Ϫ2.75 and Ϫ3 V, the abnormal hydrogen evolution is clearly seen in
the current transient curves after induction times. Hassel and
Lohrengal25 also observed similar cathodic current transients for an-
odized aluminum above a certain potential in an acetate solution.
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
B331
Figure 4. Hydrogen concentration profiles obtained from TOF-ERDA for
aluminum specimen anodized to form 44 nm thick oxide after polarization
under various cathodic potentials in 1 M HCl electrolyte at 30°C; ͑᭿͒ with-
out polarization, ͑᭡͒ cathodic potential of Ϫ2.5 V for 300 s, and ͑᭹͒ ca-
thodic potential of Ϫ3 V for 300 s.
rupture of the oxide film like the blisters at high electric fields. The
mechanical breakdown of the oxide films tends to occur at the pe-
riphery of the blisters, where the maximum strain develops during
the blister growth.11 The strong increase in the total hydrogen cur-
rent density reflects the active hydrogen evolution reaction via the
easy path of cracks or completely exposed bare aluminum surface at
the peripheries of blisters. During the growth of the blisters, both
breakdown and repassivation occur simultaneously. The rupture of
the oxide film takes place when the hydrogen pressure within the
blister cause the strain of the oxide film to exceed the fracture strain
of the film. As a result of the breakdown, the bare aluminum surface
is exposed to the electrolyte. The rate of hydrogen evolution is ac-
celerated on the bare aluminum surface, and it causes the pH to shift
enough to form the passive oxide film on the bare aluminum surface.
The reported incremental growth of the blister can be understood in
terms of breakdown and subsequent healing of the oxide film.
We investigated the change of the microstructure of the oxide
film before and after the cathodic polarization using TEM and
Rutherford backscattering spectroscopy. However, we did not detect
any change of the structure and stoichiometry of the oxide film after
the cathodic polarization.11
Figure 3. ͑a͒ Hydrogen evolution current density transients under varying
cathodic potentials in 1 M HCl at 30°C for aluminum anodized to form 44
nm thick oxide. ͑b͒ Magnified figure of current at Ϫ2.5 V.
Parts of the oxide film must be broken to enable such high currents.
The optical observations of the sample surface after polarization
revealed that the breakdown of the oxide film occurred accompanied
by the formation of blisters. The blisters generally had a circular
shape. The average radius of the blisters ranged from 10 to 15 ␮m.
The blisters were also observed in the sample subjected to Ϫ2.5
V, even though the abnormal hydrogen evolution was not steady.
The evidence of the oxide breakdown is apparent from the spikes of
cathodic current in Fig. 3b. The decay with time following the steep
rise in the current-time curve indicates that bare aluminum spots
exposed by the breakdown of the oxide film are immediately repas-
sivated. There are some similarities between the spikes in cathodic
current and those in dissolution current found in metastable pitting
of aluminum below the pitting potential range.26 The repassivation
of bare aluminum spots occurred within 1 or 2 s comparable to the
repassivation times for metastable pits. Bargeron and Givens27 ob-
served continuous oscillations in the current for anodized aluminum
maintained above the pitting potential. They interpreted it as a sign
of incremental growth of blisters in the oxide film. In our experi-
mental results, we conclude that the incremental growth of the blis-
ter is caused by simultaneous cracking and healing processes of the
blister.
Hydrogen concentration profiles after cathodic polariza-
tion.—Figure 4 shows hydrogen concentration profiles obtained
from TOF-ERDA for aluminum specimens anodized to form a 44
nm thick oxide after polarization under various cathodic potentials
in 1 M HCl electrolyte at 30°C. The concentration of hydrogen in
the original 44 nm thick oxide film before cathodic polarization is
3.7 ϫ 1027 atoms/m3, which lies within the range of hydrogen con-
centrations detected in various aluminum oxide films.19 The hydro-
gen content in the oxide film results from both the surface hydrogen
due to humidity in the air24 and adsorption of water molecules on
the flaws within the film.28
High hydrogen concentration peaks emerge near the metal/oxide
interface after cathodic polarization. The peak height shows a strong
dependence on the applied cathodic potential, indicating that the
proton migration causes enrichment of hydrogen at the metal/oxide
interface. The hydrogen content increases rapidly with increasing
applied cathodic potential, whereas the hydrogen concentration dis-
tribution near the specimen surface does not vary with potential.
Though the hydrogen concentration gradually decreases as the dis-
tance increases from the interface into bulk metal due to the diffu-
sion process, it is clear from the highest peak at the metal/oxide
interface that most of the hydrogen atoms reside at the interface.
Such accumulation of hydrogen is attributed to the difference be-
As the rate of generation of hydrogen at the interface exceeds the
rate of hydrogen diffusing into the bulk metal substrate, hydrogen
atoms begin to accumulate under the oxide film. Eventually, the
increased hydrogen gas pressure at the interface causes a mechanical
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B332
Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
Table I. Summary of applied cathodic potential Eappl , hydrogen
content ␳, cumulative proton charge Qp , and total cathodic
charge Qt . Qt was obtained from integration of current density
transient.
␳
Eappl
͑V͒
Qp
Qt
Qp /Qt
(%)
(ϫ1020 atoms/m2)
͑C/m2)
(ϫ103 C/m2)
2.47
3.41
5.60
11.7
23.5
13.5
28.2
63.7
161
0.276
0.79
2.88
110
325
4.9
3.6
2.21
0.15
0.11
Ϫ2
Ϫ2.25
Ϫ2.5
Ϫ2.75
Ϫ3
351
tween generation of H by reaction and the sum of diffusion of H into
the film as well as into the substrate. The probability of hydrogen
accumulation at the metal/oxide interface is particularly high for
aluminum, because the diffusion coefficient of hydrogen in alumi-
num is on the order of 10Ϫ14-10Ϫ16 m2/s,29,30 which is much lower
than that of 10Ϫ9 m2/s for other metals.31 The frequently reported
blisters in aluminum can be understood by such a high probability of
the hydrogen accumulation at the metal/oxide interface.
Effect of field strength on proton migration in Al2O3.—The cu-
mulative charge transported by protons through the oxide film dur-
ing cathodic polarization was calculated from hydrogen areal densi-
ties obtained from TOF-ERDA by the following equation
nF ␳ Ϫ ␳ ͒
͑
o
Qp ϭ
͓1͔
Na
where n is the number of electrons in the reaction, F is Faraday’s
constant, ␳ is the detected hydrogen areal density of the specimen
after cathodic polarization, ␳o is the initial hydrogen areal density of
anodic oxide film, and Na is Avogadro’s number. For example, the
detected hydrogen areal density ␳ in the sample after a cathodic
polarization of Ϫ2.75 V for 300 s was 1.17 ϫ 1021 atoms/m2. This
yields an ionic charge of 161 C/m2 with a ␳ value of 1.63 ϫ 1020
o
atoms/m2 for the 44 nm thick oxide film.
The ionic flow through the oxide film is clearly dependent on the
electric field across the film; therefore, it is proper to use the electric
field strength instead of the applied cathodic potential for examining
the conduction behavior of protons. The electric field strength in this
study was calculated by the difference between the open-circuit po-
tential and the applied cathodic potential divided by the film thick-
ness, neglecting potential drops at the interfaces of the oxide layer.
The electric field necessary for the proton movement across the film
under cathodic polarization is an order of magnitude smaller than
that for ionic movements of the aluminum and oxygen ions under
anodic polarization.32 This low driving force for proton migration
can be ascribed to the relatively small size of the proton compared
with those of aluminum and oxygen ions33 and the relatively easy
migration path. Protons are likely to migrate through the oxide film
via the interstitial sites instead of the exchange of vacancies.
The areal densities of hydrogen, cumulative proton charges, and
total cathodic charges at applied cathodic potentials are summarized
in Table I. The cumulative ionic charge occupies less than 5% of the
total charge passed over the cathodic polarization time. As the ap-
plied potential increases negatively from Ϫ2 to Ϫ2.5 V, a meager
increase in the total charge is observed. At potentials above Ϫ2.5 V,
the breakdown of the oxide film takes place, which in turn brings
about a strong increase in the total current density, but does not give
rise to an abrupt change in the ionic conductivity of the oxide film.
This implies that the breakdown of the anodic oxide film is not
caused by the compositional changes of the oxide film, which would
supposedly decrease the resistance of the film to ionic transport.5,6,8
Figure 5 shows the cumulative proton charge as a function of
electric field strength. The charge transported by proton migration
Figure 5. ͑a͒ Cumulative ionic charge Qp and ͑b͒ logarithmic plot of Qp ͑᭹͒
and total charge Qt ͑᭿͒ as a function of electric field strength across anodic
oxide film of aluminum.
increases exponentially with increasing applied electric field, as
shown in Fig. 5a. It indicates that the proton conduction in the
aluminum oxide film also follows the high field conduction mecha-
nism.
According to the high field conduction model, the ionic current ip
can be expressed as Eq. 2, as determined empirically by Gunters-
chulze and Betz34 and rationalized by Verwey.35 That is
ip ϭ ipo exp ␤E͒
͓2͔
͑
where ␤ is the cathodic high field conduction coefficient, E is the
electric field strength across the oxide film, and ipo is the ionic
current at a field strength of zero. For a given polarization time, both
sides of Eq. 2 are integrated to obtain the cumulative ionic charge
Qp
Qp ϭ Qpo exp ␤E͒
͓3͔
͑
or
log Qp ϭ log Qpo ϩ 0.4343␤E
͓4͔
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
B333
Figure 6. Effect of temperature of hydrochloric acid solution on proton
migration through 44 nm thick oxide film polarized for 300 s at cathodic
potential of Ϫ2.25 V.
The cathodic conduction parameters, Qpo and ␤, are determined
from a straight line in Fig. 5b to be 0.508 C/m2 and 1.44 ϫ 10Ϫ7
m/V, respectively.
Temperature dependence of proton migration.—Figure 6 shows
the temperature dependence of proton migration through a 44 nm
thick aluminum oxide film at a given electric field strength. The
activation energy for proton migration is obtained from the slope in
Fig. 6.
In Eq. 3, Qpo includes the activation energy for ionic conduction
W
Qpo ϭ A exp Ϫ
͓5͔
ͩ
ͪ
kT
where A is a constant, W is the activation energy for proton trans-
port, and k is the Boltzmann constant (8.617 ϫ 10Ϫ5 eV/K͒. As-
suming that ␤ is constant irrespective of the electrolytic temperature,
the activation energy for the proton conduction within the aluminum
oxide film can be determined from the double-logarithm plot of
accumulated proton charge and electrolytic temperature.
Figure 7. Hydrogen concentration profiles for sample cathodically charged
at constant current density of 5 A/m2 for 30 s ͑a͒ before and ͑b͒ after diffu-
sion for 2 ϫ 106 s at 25°C. Symbols are the experimental data, and solid line
curves are results simulated assuming DH ͑aluminum oxide film͒ ϭ 8
ϫ 10Ϫ20 m2/s.
The calculated activation energy is 0.46 Ϯ 0.12 eV. Compared
with reported values for aluminum and oxygen ions within the an-
odic oxide film, which lies within 0.7-1.7 eV32 the obtained value is
much lower.
gen concentration profile, CH(x,t), is established by TOF-ERDA at
time t, was adopted. The comparison of the measured concentration
profile with the appropriate solution of Fick’s second law yielded the
Discussion
Hydrogen diffusion through aluminum oxide film.—In this paper,
we tried to determine the absolute amount of proton migration
through an aluminum oxide film during cathodic polarization from
the detected hydrogen quantity in the sample using ERDA. During
the cathodic polarization, a hydrogen concentration gradient devel-
ops at the electrolyte/oxide and oxide/metal interfaces. It may be
caused by the unexpected entrance and exit of hydrogen atoms by a
diffusion process. Only when the quantity of hydrogen detected in
the sample is not influenced by the hydrogen diffusion, can it be
used to determine the proton migration quantity. The diffusion rate
of hydrogen in aluminum oxide film should be much lower than that
of the electrolyte and the bulk aluminum substrate. In other words,
the barrier-type aluminum oxide film should act as a hydrogen dif-
fusion barrier. Some hydrogen permeation results revealed that the
anodic oxide films on aluminum hindered the hydrogen diffusion
process.36,37 However, they did not evaluate the diffusion coefficient
of hydrogen in the amorphous aluminum oxide films. In this paper,
among the experimental methods for the determination of diffusion
coefficient, a direct profile measuring method, in which the hydro-
diffusion coefficient DH͑Al2O3͒
.
Figure 7 shows the hydrogen profiles for a sample cathodically
charged at a current density of 5 A/m2 for 30 s before and after
diffusion for 2 ϫ 106 s at 25°C. A sharp peak exists in the surface
region, followed by a high hydrogen concentration spreading from
the metal/oxide interface to the inside of the metal before the diffu-
sion. The existence of high hydrogen concentration near the metal/
oxide interface is an evidence of proton movement through the ox-
ide film during cathodic polarization. Many protons migrated into
the oxide film due to the electric field imposed during cathodic
polarization and reduced into hydrogen atoms or molecules at the
metal/oxide interface.
Ignoring the variation of diffusion coefficient with concentration,
the governing diffusion equation, Fick’s second law, becomes
ץCH
ץt
ץ2CH
ץx2
ϭ DH
͓6͔
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B334
Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
sion zones. The path length from the real surface to a given depth
relative to the apparent surface plane is variable. As a result, recoil
energies from a given depth are dispersed.41
Here CH is the hydrogen concentration, t is the diffusion time, DH is
the hydrogen diffusion coefficient, and x is the apparent thickness
from the sample surface.
In this diffusion experiment, hydrogen diffusing out was con-
ducted by maintaining the charged sample under a pressure of 10Ϫ7
Torr at 25°C for 2 ϫ 106 s. Mathematically, this procedure provides
the boundary condition
Ionic conductivity of barrier type aluminum oxide
film.—Although the aluminum oxide film is known to be an excel-
lent ionic conductor in anodizing solutions such as neutral borate,
phosphate, and tartrate electrolytes, where a barrier-type anodic film
is formed, there is no evidence that the anodic aluminum oxide film
behaves as a good ionic conductor in hydrochloric acid solution. The
grown anodic aluminum oxide film is thought to be an insulator
having high activation energy ͑1-2 eV͒ for electrical conduction.
According to our experimental results, Qpo in Eq. 5 is calculated
to be 0.508 C/m2. It can be converted into the average ionic current
density for a fixed polarization time of 300 s. This yields an ex-
change proton current density within the oxide film of the order of
1.7 ϫ 10Ϫ3 A/m2. It is reasonable that the ionic current in the oxide
film by anodic polarization in anodizing solution is much higher
than that by cathodic polarization. Johansen et al. reported that dur-
ing anodizing the value of the exchange ionic current density in the
oxide film was 2.1 ϫ 10Ϫ2 A/m2.42
In an inorganic compound such as an aluminum oxide film, the
total electrical conductivity is given by the sum of the electronic and
ionic conductivities. It is concluded from the low ratio of proton
charge to total cathodic charge that electrons are major charge car-
riers in the aluminum oxide film during cathodic polarization.
Schmalzried et al.43 found from solid-state electromotive force cell
studies that Al2O3 is an electronic conductor above 1300°C,
whereas it is a mixed ionic and electronic conductor at lower tem-
peratures. Electronic transport in barrier oxide films may occur as
resonance tunneling or direct elastic tunneling.44 The tunneling pro-
cess is strongly related to the barrier layer thickness. Resonance
tunneling is likely to occur in a relatively thick oxide film above 10
nm. However, if the oxide film thickness is thinner than 10 nm,
direct elastic tunneling is preferred. It was reported that at an electric
field of 2.2 ϫ 108 V/m, the room temperature tunneling current den-
sity in Al2O3 was about 2 ϫ 10Ϫ2 A/m2.45
Ͼ
C
H ϭ 0, x ϭ 0, and x → ϱ,
t
0
͓7͔
Within the bulk aluminum region, a reported hydrogen diffusion
coefficient DH͑Al͒ of 1.9 ϫ 10Ϫ14 m2/s29 was used. The diffusion
coefficient of hydrogen within the oxide film DH͑Al2O3͒ is not avail-
able. An abrupt change in diffusivity may occur at the metal/oxide
interface. Because DH͑Al2O3͒ is expected to be far less than DH͑Al͒
,
the diffusivity at the metal/oxide interface was assumed to be the
harmonic mean value of DH͑Al͒ and DH͑Al2O3͒ as follows38
Ϫ1
0.5
0.5
DH(interface)
ϭ
ϩ
͓8͔
ͩ
ͪ
DH(Al2O3)
DH(Al)
The value of DH͑Al2O3͒ can be obtained using Eq. 6-8 and the ex-
perimental data in Fig. 7. The initial hydrogen concentration distri-
bution, CH(x,0), is obtained by a polynomial fit of the experimental
data of Fig. 7a. Arbitrary value for DH͑Al2O3͒ and the value for
CH(x,0) are substituted into Eq. 6. Equation 6 is solved to obtain the
concentration distribution at a time of 2 ϫ 106 s using the finite
difference method taking the boundary condition of Eq. 7 and 8 into
account. The value of DH͑Al2O3͒ is varied until the calculated con-
centration distribution best fits the measured data in Fig. 7b. In the
present case, the value of DH͑Al2O3͒ was tried from 10Ϫ15 to 10Ϫ25
m2/s. The best result was obtained at DH͑Al2O3͒ ϭ 8 ϫ 10Ϫ20 m2/s.
The hydrogen atoms produced at the metal/oxide interface dif-
fuse more easily into defects of bulk aluminum such as vacancies
and dislocations than lattice interstices.39 Compared to a hydrogen
diffusion coefficient in the order of 10Ϫ9 m2/s in the electrolyte,31
such a low value of hydrogen diffusion coefficient implies that the
aluminum oxide film behaves as a diffusion barrier of hydrogen
during cathodic polarization. Therefore, our previous assumption
that the hydrogen diffusion does not affect the detection of quantity
of proton migration using ERDA is reasonable. For comparison, the
self-diffusion coefficient of aluminum in polycrystalline ␣-Al2O3 is
reported to be in the order of 6.2 ϫ 10Ϫ23 m2/s.40
Conclusions
The direct measurement of the ionic charge transported by pro-
tons in a 44 nm thick anodic oxide films ͑amorphous to 2 nm in
grain size͒ of aluminum after cathodic polarization in 1 M HCl
solution was undertaken using the TOF-ERDA method. The key
findings obtained in this study are as follows.
1. The hydrogen concentration peaks emerge near the metal/
oxide interface and increase with increasing applied cathodic poten-
tial ͑from Ϫ2 to Ϫ3 V͒, whereas the hydrogen concentration distri-
bution near the specimen surface does not vary with potential. The
hydrogen concentration gradually decreases with increasing distance
from the interface due to diffusion into bulk metal as well as the
film.
2. The ionic charge transported by protons occupies approxi-
mately 5% of the total charge passed through the aluminum oxide
film during cathodic polarization. The proton conduction across the
oxide film apparently follows the high field conduction mechanism.
3. Even after the breakdown of the anodic oxide film during
cathodic polarization, the ionic conductivity of the oxide film does
not give rise to an abrupt change while the total current density
substantially increases. This implies that the breakdown of the oxide
film is not caused by compositional changes of the film, but by the
pressure of hydrogen gas generated at the oxide/metal interface.
4. The activation energy for proton conduction within the oxide
film is estimated to be 0.46 Ϯ 0.12 eV, which is much lower than
reported activation energies for aluminum and oxygen ions.
5. The diffusivity of hydrogen atoms in the aluminum oxide film
is estimated to be 8 ϫ 10Ϫ20 m2/s.
In determining the proton migration quantity using the quantity
of the hydrogen detected in the sample, we may have to consider the
amount of the discharged hydrogen that could escape the sample
when the hydrogen bubble collapses. Ryan and McCafferty14 calcu-
lated the amount of hydrogen required to rupture the blisters on
aluminum. According to their results, the calculated value of the
hydrogen atoms was 1.3 ϫ 102 mol/m3, which is negligible com-
pared with the quantity of the hydrogen detected in the sample, 104
to 106 mol/m3.
The decay of the profile from the metal/oxide film interface to-
ward the oxide surface at Ϫ3 V shown in Fig. 4 is caused by two
sources. One is the diffusion process of hydrogen, and another is the
artifact in depth profiling related to the roughness of the metal/oxide
film interface. The oxide precipitates generated during repassivation
process eventually roughen the metal/oxide interface. The rough
metal/oxide interface causes the artifacts in hydrogen concentration
profiling. Lanford et al.19 also pointed out the same problem in
depth profiling that the hydrogen concentration profile spreads out
broadly at the rough metal/oxide interface. Note that the profiling
technique does not give the true thickness of the oxide film but gives
the apparent thickness calculated from the recoiling energy and stop-
ping power. The rough metal/oxide interface after cathodic polariza-
tion produces sloping edges in the signals, as do uniform interdiffu-
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑7͒ B329-B335 ͑2003͒
B335
22. K. Kimura, K. Nakajima, H. Imura, Nucl. Instrum. Methods Phys. Res. B, 140, 397
͑1998͒.
23. J. F. Ziegler, The Stopping and Ranges of Ions in Matter, Pergamon, New York
͑1977͒.
24. J. R. Tesmer and M. Nastasi, Handbook of Modern Ion Beam Materials Analysis,
Materials Research Society, Pittsburgh, PA ͑1985͒.
25. W. Hassel and M. M. Lohrengel, Electrochim. Acta, 40, 433 ͑1995͒.
26. S. T. Pride, J. R. Scully, and J. L. Hudson, J. Electrochem. Soc., 141, 3028 ͑1994͒.
27. C. B. Bargeron and R. B. Givens, J. Electrochem. Soc., 129, 340 ͑1982͒.
28. K. Shimizu, G. M. Brown, H. Habazaki, K. Kobayashi, P. Skeldon, G. E. Thomp-
son, and G. C. Wood, Electrochim. Acta, 44, 2297 ͑1999͒.
29. T. Ishikawa and R. B. McLellan, Acta Metall., 34, 1091 ͑1986͒.
30. H. Hashimoto and T. Kino, J. Phys. F: Met. Phys., 13, 1157 ͑1983͒.
31. J. O’M. Bockris and A. K. N. Reddy, Modern Electrochemistry, Vol. 2, p. 1328,
Chapman & Hall, London ͑1973͒.
Acknowledgment
This study was supported by Texture Control Laboratory ͑NRL͒,
Seoul National University.
Seoul National University assisted in meeting the publication costs of this
article.
References
1. H. Kliem, IEEE Trans. Electr. Insul., 24, 185 ͑1989͒.
2. H. Takahashi, K. Fujiwara, and M. Seo, Corros. Sci., 36, 677 ͑1994͒.
3. L. Young, Trans. Faraday Soc., 55, 842 ͑1959͒.
4. D. A. Vermilyea, J. Appl. Phys., 27, 963 ͑1956͒.
5. R. Despic, D. M. Drazic, J. Balaksina, L. J. Gajic-Krstajic, and R. M. Stevanovic,
Electrochim. Acta, 35, 1747 ͑1990͒.
6. S.-M. Moon and S.-I. Pyun, J. Appl. Electrochem., 22, 151 ͑1992͒.
7. C.-F. Lin, M. D. Porter, and K. R. Hebert, J. Electrochem. Soc., 141, 96 ͑1994͒.
8. M.-H. Wang and K. R. Hebert, J. Electrochem. Soc., 143, 2827 ͑1996͒.
9. C.-F. Lin and K. R. Hebert, J. Electrochem. Soc., 141, 104 ͑1994͒.
10. H. Takahashi, K. Fujiwara, and M. Seo, Corros. Sci., 36, 689 ͑1994͒.
11. J. H. Seo, S.-Y. Jeong, H.-J. Shin, and D. N. Lee, in Proceedings of the 2nd
Aluminum Surface Science and Technology, University of Manchester Institute of
Science and Technology, Manchester, England ͑2000͒.
12. C. B. Bargeron and R. B. Givens, J. Electrochem. Soc., 124, 1845 ͑1977͒.
13. P. M. Natishan and E. McCafferty, J. Electrochem. Soc., 136, 53 ͑1989͒.
14. R. L. Ryan and E. McCafferty, J. Electrochem. Soc., 142, 2594 ͑1995͒.
15. L. L. Cabot, J. A. Garrido, E. Perez, and J. VirgiLi, Corros. Sci., 26, 357 ͑1986͒.
16. J.-S. Lee and H.-I. Yoo, Solid State Ionics, 68, 139 ͑1994͒.
17. R. Kirchheim, Electrochim. Acta, 32, 1619 ͑1987͒.
32. M. M. Lohrengel, Mater. Sci. Eng., R., R11, 243 ͑1993͒.
33. L. Young, Abnodic Oxide Films, Academic Press, London and New York ͑1961͒.
34. A. Guntherschulze and H. Betz, Z. Phys., 92, 367 ͑1934͒.
35. E. J. W. Verwey, Physica (Amsterdam), 2, 1059 ͑1935͒.
36. A. Csanady, K. Papp, and E. Pasztor, Mater. Sci. Eng., 48, 35 ͑1981͒.
37. W. Song, J. Du, Y. Xu, and B. Long, J. Nucl. Mater., 246, 139 ͑1997͒.
38. S. V. Patankar, in Numerical Heat Transfer and Fluid Flow, Taylor & Francis,
London ͑1980͒.
39. G. A. Young, Jr. and J. R. Scully, Acta Mater., 46, 6337 ͑1998͒.
40. P. Kofstad, in Nonstoichiometry, Diffusion, and Electrical Conductivity in Binary
Metal Oxides, Wiley-Interscience, New York ͑1972͒.
41. J. Tirira, Y. Serruys, and P. Trocellier, in Forward Recoil Spectrometry: Applica-
tions to Hydrogen Determination in Solids, Plenum Press, New York ͑1996͒.
42. H. A. Johansen, G. B. Adams, and P. Van Rysselberghe, J. Electrochem. Soc., 104,
339 ͑1957͒.
18. M. F. Abd Rabbo, J. A. Richardson, and G. C. Wood, Electrochim. Acta, 22, 1375
͑1977͒.
43. H. Schmalzried, Z. Phys. Chem. (Munich), 38, 87 ͑1963͒.
44. R. H. Bube, in Electronics in Solids, Academic Press, Inc., p. 239, New York
͑1992͒.
45. S. M. Sze, Physics of Semiconductor Devices, 2nd. ed., Wiley & Sons, New York
͑1981͒.
19. W. A. Lanford, R. S. Alwitt, and C. K. Dyer, J. Electrochem. Soc., 127, 405 ͑1980͒.
20. G. E. Thompson, P. Skeldon, G. C. Wood, X. Zhou, U. Kreissig, E. Wieser, H.
Habazaki, and K. Shimizu, Surf. Interface Anal., 27, 57 ͑1999͒.
21. J. K. Kim, Y. S. Kim, H. W. Choi, G. D. Kim, H. J. Woo, and C. N. Whang, Nucl.
Instrum. Methods Phys. Res. B, 129, 323 ͑1997͒.
Downloaded on 2014-08-12 to IP 128.118.88.48 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).