O18 tracer study of porous film growth on aluminum in phosphoric acid

Article abstract of DOI:10.1149/1.3490640

^O18 tracer is used to investigate the development of porous anodic films at constant current in phosphoric acid on electropolished aluminum. A barrier layer and porous region form initially with the pore size related to the surface texture of the substrate. Subsequently, major pores emerge, with their sizes related to the anodizing voltage. The evolution of the film is accompanied by increases in growth rate and formation efficiency. The ^O ions of a preformed oxide are retained in the film during anodization in a nonenriched electrolyte, with ^O18 being partitioned among (i) the surface region of texture-dependent porosity, (ii) the walls of major pores, and, in diminishing amounts, (iii) the inner region of the barrier layer.

Full text of DOI:10.1149/1.3490640

Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
C399
0013-4651/2010/157͑11͒/C399/9/$28.00 © The Electrochemical Society
18O Tracer Study of Porous Film Growth on Aluminum
in Phosphoric Acid
,z
A. Baron-Wiecheć,a J. J. Ganem,b S. J. Garcia-Vergara,c P. Skeldon,a,
*
a,**
*
b,
G. E. Thompson,
and I. C. Vickridge
aCorrosion and Protection Centre, School of Materials, The University of Manchester,
Manchester M17 9PL, United Kingdom
bSAFIR, Institut des Nano Sciences de Paris, Université de Pierre et Marie Curie, 75252 Paris Cedex 05,
France
cGrupo de Investigación en Corrosión, Escuela de Ingeniería Metalúrgica y Ciencia de Materiales,
Univesidad Industrial de Santander, Bucaramanga, Colombia
18O tracer is used to investigate the development of porous anodic films at constant current in phosphoric acid on electropolished
aluminum. A barrier layer and porous region form initially with the pore size related to the surface texture of the substrate.
Subsequently, major pores emerge, with their sizes related to the anodizing voltage. The evolution of the film is accompanied by
increases in growth rate and formation efficiency. The 18O ions of a preformed oxide are retained in the film during anodization
in a nonenriched electrolyte, with 18O being partitioned among ͑i͒ the surface region of texture-dependent porosity, ͑ii͒ the walls
of major pores, and, in diminishing amounts, ͑iii͒ the inner region of the barrier layer.
© 2010 The Electrochemical Society. ͓DOI: 10.1149/1.3490640͔ All rights reserved.
Manuscript submitted July 8, 2010; revised manuscript received August 26, 2010. Published September 29, 2010.
The use of 18O as a tracer species in studies of ionic transport
during thermal1-3 and anodic oxidation4,5 is well established. The
experiments are carried out by sequentially oxidizing a specimen,
for example, first in an 18O-enriched environment and then in one of
the normal isotopic concentration, the natural abundance of 18O be-
ing only ϳ0.2%. The final location of 18O can indicate the mecha-
nism of oxygen transport in the growing oxide. The 18O distribution
can be determined nondestructively by nuclear reaction analysis
͑NRA͒.6-9 Secondary-ion mass spectrometry and time of flight glow
discharge mass spectrometry may also be employed with sputtering
of the oxide layers or imaging of oxide cross sections.10-16
topes, suggesting a short-circuit mechanism of transport, which is
contrary to the mechanism in uniform barrier films.4 The use of
phosphoric acid in the present experiments enables the formation of
cells that are larger than those in films from sulfuric acid, such that
the 18O distribution can be followed in more detail between pore
initiation and the establishment of the major pores.
Experimental
Specimen preparation.— Aluminum foil of 99.99% purity ͓30
Cu, 20 Fe, 20 Si ͑ppm͔͒ and 0.35 mm thickness ͑Toyo Aluminum
K.K.͒ with a cubic texture was used in the study. Specimens of
dimensions 3 ϫ 1.5 cm were cut from the foil and then electropol-
ished for 180 s in a mixture of 75% perchloric acid and ethanol ͑1:4
by volume͒ at 278 K. Following rinsing with ethanol and deionized
water, the specimens were coated with lacquer ͑Lacomit͒ to provide
a working area of 2 cm2. Subsequent anodizing was carried out at a
constant current density of 4.5 mA cm−2 in 0.4 M phosphoric acid
electrolyte ͑PAE͒ prepared with 85% H3PO4 ͑Fisher Scientific͒ and
either water enriched in 18O to 10.0% ͑CK Gas Products, Ltd.͒
͑electrolyte designated PAE18͒ or deionized water of the natural
isotopic composition ͑PAE16͒. pH of both electrolytes was 1.8. A
small cell was used to minimize the required volume of
O18-enriched water ͑30 mL͒, with the specimen as the anode and an
aluminum sheet of area 18 cm2 as the cathode. The current was
controlled using a dc power supply ͑Metronix model 6911͒ con-
nected to a digital multimeter ͑National Instruments͒ interfaced to a
computer for recording the voltage–time response.
In the present work, 18O tracer is used with NRA to investigate
the formation of porous films on aluminum under mild anodizing
conditions. Porous anodic films have been studied extensively due to
their importance in the protection of aluminum alloys17 and, more
recently, for various nanotechnological applications.18-21 The films
are formed by oxidation of aluminum under anodic polarization,
usually in an acid electrolyte. They consist of a thin barrier layer
next to the metal and a relatively thick overlying porous layer.22,23 In
ideal form, the porous layer comprises close-packed, columnar cells
of anodic alumina with uniform hexagonal cross sections; each cell
contains a central pore running from the barrier layer to the film
surface. In reality, the cell arrangement is usually less than the ideal.
The barrier layer thickness, pore diameter, and interpore distance are
proportional to the anodizing voltage23 with respective ratios of
23,24
ϳ1.0, 1.0, and 2.5 nm V−1
.
In contrast, under hard anodizing
conditions, associated with fast film growth and low electrolyte tem-
perature, the ratios also depend on the current density.25
The most commonly accepted model of porous film growth in-
volves the formation of the alumina due to the ionic migration of
Al3+ and O2− ions in the barrier layer26 under an electric field of
ϳ108 to 109 V m−1 and the generation of pores by dissolution of
the alumina at the pore base.27 The dissolution is accelerated by the
electric field and Joule heating compared with the usual chemical
dissolution in the electrolyte. However, experiments using a thin
layer of tungsten tracer, which is incorporated into the film from the
substrate, have suggested that flow of oxide rather than dissolution
is important for pore generation.28 Earlier 18O tracer studies,26,29
which employed sulfuric acid electrolyte, showed an anodizing ef-
ficiency of ϳ60% and an inversion of the order of the oxygen iso-
An anodic film was first produced by anodizing at 4.5 mA cm−2
up to 20 V in PAE18. The PAE18 was then replaced by PAE16 for
a second period of anodizing, either to a selected voltage or for a
selected time. In both stages, the current was switched on within 1
min of placing the specimen in the electrolyte. The electrolyte was
stirred during anodizing using a magnetic stirrer and was maintained
at a temperature of 295 Ϯ 0.5 K. After anodizing, the specimens
were rinsed in deionized water and dried in a cool air stream. A
self-supporting anodic film was also prepared by a previously devel-
oped method.30 A circular region of 7 mm diameter on one side of
an electropolished specimen was coated by lacquer. Following an-
odizing in PAE18 to 20 V and in PAE16 for 360 s, the lacquer was
dissolved in acetone and the exposed aluminum was removed by
electropolishing to leave a region of substrate-free film. The speci-
men was finally rinsed in ethanol and distilled water and dried in a
cool air stream.
*
Electrochemical Society Active Member.
Electrochemical Society Fellow.
**
z E-mail: p.skeldon@manchester.ac.uk
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C400
Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
Specimen characterization.— Cross sections of the films were
prepared by ultramicrotomy for examination by transmission elec-
tron microscopy ͑TEM͒ using a JEOL FX 2000 II instrument oper-
ated at 120 kV. The sections were ϳ15 nm thick. The films were
also examined by scanning electron microscopy ͑SEM͒ using a
Zeiss Gemini Ultra 55 instrument operated at 10 kV with cross
sections of the films prepared in an ultramicrotome using glass and
diamond knives.
Ion beam analysis of the films employed the Van de Graaff ac-
celerator of the Institut des Nano Sciences de Paris. The analyzed
region was usually 1 mm diameter. Rutherford backscattering spec-
troscopy ͑RBS͒ was carried out using 1.8 MeV He+ ions, with the
ion beam at normal incidence to the specimen surface and a scatter-
ing angle of 165°. The data were interpreted using SIMNRA
software.31 Oxygen contents were determined with the 18O͑p,␣͒15N
and 16O͑d,p͒17O reactions using 750 and 870 keV ion beams, re-
spectively. The beams were at normal incidence with the detector at
150° to the direction of the incident beam. A 13 ␮m thick mylar
film placed in front of the detector excluded the detection of elasti-
cally scattered ions. Quantification of data employed anodized tan-
talum references containing 255 Ϯ 8 ϫ 1015 O18 atoms cm−2 and
690 Ϯ 20 ϫ 1015 O16 atoms cm−2. The precision of the analyses
is typically ϳ3%, being determined by uncertainties in the solid
angle and the oxygen contents of the references and counting
statistics.4,6,28,29 Details of the methods have been described
elsewhere.6 18O depth profiling was performed using the exception-
ally narrow resonance of the 18O͑p,␣͒15N reaction at 151 keV.9 An
automatic energy scanning system was used to change the beam
energy. A 2 mm diameter beam was incident at 50° to the specimen
surface with an angle of 90° between the beam and the detector. A
3 ␮m thick mylar film excluded elastically scattered protons. 18O
profiles were extracted from excitation curves using the SPACES
code.32
Results
Formation of the anodic films.— Figure 1a shows
a typical
voltage–time response for a specimen anodized sequentially for the
longest time of the present study ͑ϳ375 s͒. From observations of
several specimens, the voltage surges to ϳ2 V at the start of anod-
izing in PAE18 due to the oxide on the electropolished aluminum
and then increases approximately linearly to 20 V at a rate of
1.25 Ϯ 0.05 V s−1. Upon changing to PAE16, the voltage rises rap-
idly to the final voltage in PAE18 and then steadily increases but at
a decreasing rate; thus, the average rates are 1.17 Ϯ 0.05 and
1.10 Ϯ 0.05 V s−1 from 0 to 40 V and from 0 to 100 V, respec-
tively. The voltage reaches a maximum between 135 and 145 V and
then decreases from ϳ105 to 107 V when anodizing was stopped.
The labels S20, S40, S100, Smax, and Sconst on Fig. 1 ͑used to
designate the specimens͒ indicate the anodizing times for particular
specimens; the anodizing data are given in Table I. S20, S40, and
S100 were anodized under a rising voltage, Smax to the maximum
voltage, and Sconst to the final voltage. Figure 1b compares the
voltage responses for specimens anodized with and without interrup-
tion, the differences being within the reproducibility of anodizing.
Figure 1. ͑a͒ Voltage–time response for aluminum anodized sequentially at
4.5 mA cm−2 in 0.4 M phosphoric acid at 295 K. ͑b͒ Comparing sequential
anodizing in PAE18 and PAE16 and anodizing in PAE16 only.
Figure 3a shows a 23–27 nm thick anodic film on S20 with the
typical appearance of amorphous anodic alumina. Gentle undula-
tions of the film with peaks separated from 50 to 110 nm follow the
texture of the aluminum substrate. Occasional voids were evident
along the metal/film interface with lengths up to ϳ30 nm and depth
of ϳ10 nm. The voids were common to all the anodized specimens;
their origin is uncertain, but they are not considered to be relevant to
present interests. Following anodizing at 40 V ͑S40͒, the undulations
of the film are more pronounced; the film thickness is increased to
47 Ϯ 2 nm at troughs in the film surface, as shown in Fig. 3b, and
further increased by ϳ10 nm near the peaks. With anodizing at 60
V ͑S60͒, the film surface became strongly scalloped, with film thick-
nesses at troughs and peaks of 65 Ϯ 4 and 95 Ϯ 8 nm, respec-
tively, as shown in Fig. 3c. The peaks were separated from 50 to 70
nm, similar to previous specimens. Figure 3d reveals a film formed
at 100 V ͑S100͒, showing features broadly the same as those of S60.
The film is 118 Ϯ 4 and 175 Ϯ 7 nm thick at troughs and peaks,
respectively. From Fig. 3a-d, the films are thicker at the peak regions
of the film surface than at the trough regions, with the difference
increasing from ϳ10 to ϳ57 nm between 40 and 100 V. Further,
the peak regions correlate with peaks in the metal/film interface. The
barrier regions of the films in Fig. 3a-d increase in thickness with
increase in voltage, with formation ratios at troughs of ϳ1.19, 1.08,
1.22, and 1.17 nm V−1 for S20, S40, S60, and S100, respectively,
giving an average of 1.17 Ϯ 0.03 nm V−1. Following anodizing at
Substrate morphology.— Figure 2 shows the scanning electron
micrograph of the aluminum surface with regions of light appear-
ance corresponding to ridges that form due to local differences in the
electropolishing rate. The cellular pattern of the ridges arises from
the cubic texture of the aluminum with ͑100͒ planes parallel to the
surface, which will be discussed later. The average cell size was
64 Ϯ 6 nm. The NRA of the surface revealed ϳ19 ϫ 1015 oxygen
atoms cm−2, which is equivalent to an ϳ3 to 5 nm thick film of
aluminum oxide/hydroxide.
Film morphologies.— Bright-field transmission electron micro-
graphs of the films attached to the aluminum substrate are shown in
Fig. 3 for S20, S40, S60, S100, and Smax. S20 and S60 were an-
odized in PAE18 only; other specimens were anodized sequentially.
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
C401
P/Al
Table I. Summary of results from NRA and RBS analysis.
O RBS
16O NRA
18O NRA total 18O NR − natural 18O
Al
͑1015
O16 + O18 NRA
Anodizing
time
͑1015
͑1015
͑1015
͑1015
͑1015
Potential
͑V͒
Specimen
͑s͒
atom cm−2
͒
atom cm−2
͒
atom cm−2
͒
atom cm−2
͒
atom cm−2
͒
atom cm−2
͒
S20
S40
S100
Smax
Sconst
Sconst-ss
21
41
101
140
107
107
16
35
91
149
385
385
126.9
270.7
738.9
1360.1
3809.0
––
127.2
257.9
708.3
1289.2
3406.9
3403.5
11.93
12.05
12.90
14.00
17.91
18.66
11.69
11.79
11.74
11.67
11.35
12.11
79.9
165.4
448.7
821.7
2305.0
––
139.1
270.0
721.2
1303.2
3424.8
3422.2
0.052
0.050
0.051
0.057
0.058
––
the maximum voltage of 140 V ͑Smax͒, regions of increased film
thickness are evident, associated with pores up to ϳ200 nm deep
and a 154 Ϯ 4 nm thick barrier layer, as shown in Fig. 3e. Else-
where, the barrier layer thickness is in the range of 165–195 nm,
with the film being thicker at peaks by up to ϳ90 nm. Figure 3f
shows the scanning electron micrograph of Sconst formed at 107 V
͑see Fig. 1͒, revealing well-developed pores with a barrier layer
thickness of 117 Ϯ 5 nm. The total film thickness is 850 Ϯ 20 nm,
with the outer region disclosing fine pores. Small circular voids
separated by ϳ100 nm were present in rows on the walls of the
major pores, as shown in the inset of Fig. 3f. The formation ratios
for the barrier layers beneath the deep pores of Smax and Sconst are
ϳ1.10 nm V−1. The dependence of the total film thickness on the
anodizing time is shown in Fig. 4, which includes results from
specimens additional to the main ones of the study. The data are
1 µm
Figure 2. Scanning electron micrograph of the surface of the electropolished
aluminum.
(e)
(a)
200 nm
50 nm
(b)
(f)
Figure 3. Transmission electron micro-
graphs of aluminum anodized in 0.4 M
phosphoric acid at 295 K to ͑a͒ 20 ͑S20͒,
͑b͒ 40 ͑S40͒, ͑c͒ 60 ͑S60͒, ͑d͒ 100 ͑S100͒,
and ͑e͒ 140 V ͑Smax͒. ͑f͒ Scanning elec-
tron micrograph of Ϫ107 V ͑Sconst͒. Inset
shows secondary pores at walls of major
pores.
50 nm
(c)
200 nm
50 nm
(d)
100 nm
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C402
Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
(a)
50 nm
Figure 4. Dependence of total thickness of the anodic film on the time of
anodizing.
(b)
approximated by linear regions that intersect at 75 s with slopes of
ϳ1.6 and 2.4 nm s−1, the lower rate corresponding to shorter
times.
Figure 5a shows the transmission electron micrograph of a plan-
view of the film stripped from S100. A cellular network indicates
regions of increased film thickness with cell dimensions from ϳ40
to 165 nm and an average of ϳ70 nm. The scanning electron mi-
crograph in Fig. 5b shows a similar cellular network for Sconst ͑cell
dimensions from ϳ30 to 140 nm and average of ϳ60 nm͒. The cell
dimensions correspond with the morphology of the films in Fig.
3a-e, with the pores developing between the ridges of the textured
aluminum ͑Fig. 2͒. Dark regions of Fig. 5b correspond to major
pores within cells, the pore mouths sometimes being constrained in
shape by cell edges. The interpore distance and the dimensions of
major pores are typically within the ranges of 150–500 nm and
30–120 nm, respectively. From the measurements of the pore area,
the texture-dependent region has a pore volume of ϳ60%, whereas
that of the major pores is ϳ10%. Observation of the film base fol-
lowing dissolution of the substrate revealed an average cell size of
ϳ267 nm, as shown in Fig. 5c, which is reasonably consistent with
the interpore distances at the film surface.
1 µm
(c)
Compositions of films from RBS.— Figure 6 shows the RBS data
for specimens anodized according to Fig. 1, together with an ex-
ample of a spectrum fitted using the SIMNRA program. The spectra
reveal edges due to the scattering of He+ ions from aluminum, oxy-
gen, and phosphorus at the film surface and from aluminum the
substrate, the locations of the latter edges indicating the changing
film thickness. The data were fitted by adjusting the ratios of alumi-
num, oxygen, and phosphorus in the film and the number of atoms
per unit area; the results are presented in Table I. The accuracy of
data ͑ϳ5%͒ is determined mainly by the uncertainty in background
signals, the solid angle, and the counting statistics. For S20 and S40,
the P:Al atom ratio in the phosphorus-containing film region was
derived assuming that phosphorus is present in the outer ϳ0.6 of the
film, as found by Takahashi et al.,33 because the thickness of the
phosphorus-containing region was too low for reliable measurement.
The results for all the specimens indicate a P:Al atomic ratio of
0.05:0.06 in the phosphorus-containing region, independent of the
anodizing time. The analysis is insensitive to hydrogen, for example
due to hydroxyl species, water molecules, mono- or dihydrogen
phosphate ions, or phosphoric acid molecules in the film material,
hydration products, or adsorbed layers. However, hydrogen is con-
sidered to be a minor film component. Lanford et al.34 found Ͻ0.4
atom % H in films formed in phosphate electrolyte at 368 K. More
1 µm
Figure 5. Plan-views of the film formed at ͑a͒ 100 ͑S100: transmission
electron micrograph͒ and ͑b͒ 107 V ͑Sconst: scanning electron micrograph͒.
͑c͒ Base of film on Sconst ͑scanning electron micrograph͒.
recent thermogravimetric analysis has also suggested a low hydro-
gen content.35 Further, the present films were formed for short times,
which minimizes hydration.
Amount of 18O in aluminum specimens sequentially anodized for
.— In estimation of the 16O and 18O contents of film,
various times
the exchange of 18O ions between water molecules and phosphate
ions in the electrolyte and also between the film and the electrolyte
is presumed to be negligible, following the findings of previous
studies.26,29 The yields from the 16O͑d,p1͒17O reaction for the speci-
mens S20, S100, Sconst, and the self-supporting film of Sconst-ss
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
C403
Figure 6. ͑a͒ RBS spectra of S20, S100,
Smax, and Sconst. ͑b͒ Example of spec-
trum fitted using the SIMNRA program
͑S100͒.
͑see Fig. 1͒ are presented in Fig. 7, which also shows yields from the
16O͑d,p0͒17O and 12C͑d,p0͒13C reactions. Carbon contents were es-
timated using the cross-section ratios for the 12C͑d,po͒13C and
16O͑d,p1͒17O reactions measured by Lennard et al.,36 showing an
increase in carbon with an increase in the anodizing time from
ϳ1.3 ϫ 1016 to ϳ2.7 ϫ 1016 carbon atoms cm−2 between S20 and
Sconst. The carbon is considered to be due mainly to the adsorption
of hydrocarbons and to increase with pore surface area. Examples of
yields for the 18O͑p,␣͒15N reaction are shown in Fig. 8a for S20,
S100, and Sconst; for specimens with thicker films ͑Smax and
Sconst͒, account was taken of the reduction of the cross section as
protons lose energy in the film, following Amsel and Samuel.6 The
energy loss was estimated using the measured film thicknesses and
the stopping power of amorphous alumina with a density of
37
3.1 g cm−3
,
calculated by SRIM.38 The results of the oxygen
measurements are presented in Table I. The dependence of the oxy-
gen content on the anodizing time is shown in Fig. 9. The data are
approximated by two linear regions with slopes ϳ7.13 ϫ 1015 and
9.13 ϫ 1015 atoms cm−2 s−1 for times Ͻ75 and Ͼ75 s, respec-
tively. Figure 9 also shows the aluminum contents of the films de-
termined by RBS, which are similarly approximated by linear re-
gions with slopes of ϳ4.40 ϫ 1015 and 6.56 ϫ 1015 atoms
cm−2 s−1
.
The oxygen contents of the films determined by RBS ͑see Table
I͒ are within 5% of the values determined by the NRA; the latter
values are preferred because NRA is more reliable for measurement
of elements of low atomic number in the presence of heavier ele-
ments. The amounts of 18O do not change significantly with increas-
ing time of anodizing in PAE16 and are similar to the level in the
film formed in PAE18 only ͑S20͒. The 18O enrichment in S20 is
Figure 7. Proton spectra from the 16O͑d,p1͒17O and 12C͑d,p͒13C reactions
for S20, S100, Sconst, and S const-ss; deuteron energy 870 keV, charge
10 ␮C.
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C404
Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
(a)
(b)
Figure 8. Alpha particle spectra from the
18O͑p,␣͒15N reaction for ͑a͒ S20, S100,
and Sconst; proton energy 750 keV,
charge 10 ␮C. ͑b͒ S20, Sconst and
Sconst-ss; proton energy 750 keV, charge
30 ␮C.
for fitting all of the present data. In comparison, a stopping power of
0.149 keV nm−1 was obtained for 150 keV protons from SRIM in a
film with a density of 3.1 g cm−3 containing units of Al2O3 and
AlPO4 with a P:Al atomic ratio of 0.05. Figure 10a also shows the
fitting of the data using SPACES based on a film containing a con-
stant concentration of 18O, as depicted in the schematic distribution
below the excitation curve. The fitting was achieved with a 24 nm
thick film, which satisfactorily agrees with the thickness from TEM
͑see arrow in Fig. 10͒, considering also that the TEM thickness may
include 2–3 nm of oxide retained from electropolishing. For S40, the
majority of 18O is located in the inner half of the film with roughly
one-third in the outer half, as shown in Fig. 10b, in which the 18O
concentration is scaled with respect to S20. The film thickness from
SPACES agrees with the result from TEM. For S100, 18O is en-
hanced in a ϳ20 nm thick region at the film surface and an
ϳ60 nm thick region near the metal/film interface ͑Fig. 10c͒. The
outer, middle, and inner regions contain ϳ0.25, 0.20, and 0.55 of
the total 18O, respectively. The film thickness agrees roughly with
the result from TEM but may be underestimated due to the signifi-
cant thickness ͑ϳ57 nm, constituting ϳ33% of the film͒ of the
porous region. The films of Smax and Sconst were too thick for the
measurement of complete profiles but revealed a near-surface distri-
bution of 18O similar to those of thinner films, as shown in Fig. 11.
Figure 12 compares the excitation curves for the self-supporting
film ͑Sconst-ss͒, with the film analyzed from the metal side, and
S20. 18O is present in Sconst-ss in an ϳ25 nm thick layer near the
film base with an enrichment of ϳ13%. The thickness may be over-
estimated because the base is scalloped. About 17% of the total 18O
in the film is located in this region. A similar comparison was made
using the nonresonant 18O͑p,␣͒15N reaction, with Sconst and
Sconst-ss analyzed with the beam incident on the original surface
and the original film base respectively, as shown in Fig. 8b. The
relative position of the peaks reveals that 18O is mainly buried
within Sconst-ss, with most being located nearer to the film surface
than the film base.
9.7 Ϯ 0.4%, which is slightly below the enrichment of the water,
due mainly to the presence of phosphate ions in the films.
To evaluate the efficiency of film growth in the early period of
anodizing, the oxygen contents were compared for films formed at
16–91 s in 0.1 M ammonium pentaborate electrolyte ͑ABE͒ and
PAE16. The anodizing times are equivalent to those employed for
anodizing S20 to S100. The former electrolyte results in an effi-
ciency close to 100%.39 The ratios of oxygen contents of films
formed for the same times in PAE16 and ABE ͑Table II͒ suggest an
increase in efficiency for PAE16 from ϳ0.55 to ϳ0.63.
Depth profiles of 18O species in films
.— The 18O distributions
through the films were determined using the resonance probe. Initial
experiments measured the stopping power of the alumina. Figure
10a shows the excitation curve for S20 ͑PAE18 only͒. The yield of
alpha particles increases steeply near the resonance energy. A shift
of 0.5 Ϯ 0.2 keV in the leading edge is due to adsorbed hydrocar-
bon and any residual oxide/hydroxide from electropolishing. The
yield subsequently achieves an approximate plateau and then de-
clines to negligible values. The ratio of the width at half-maximum
of the excitation curve and the path length of the beam in the film
indicated a stopping power of ϳ0.127 keV nm−1, which was used
Discussion
Substrate influences on initial film growth and pore forma-
tion.— Electropolishing of aluminum usually creates a surface tex-
ture arising from local height differences40 with a pattern related to
the grain orientation, e.g., cells, asperities, and furrows for ͑100͒,
͑111͒, and ͑110͒ grains. The origin of the texture is uncertain, with
suggestions of effects of a mosaic structure within the aluminum41
and the metal purity,40 among other factors. Asahina et al. have also
shown that the texture is influenced by the voltage used for
electropolishing.42 The ͑100͒ orientation of the present aluminum
provides a uniform texture in comparison with substrates of variable
orientation. The growth of the anodic films in PAE proceeds through
a sequence of film morphologies. Anodizing at 20 V produces a
barrier film with a negligible porosity. With an increase in voltage
up to 100 V, the film thickens noticeably above ridges of the alumi-
num surface, leading to pore arrangements determined by the tex-
ture. As the voltage rises further, regions of emerging major pores
Figure 9. Dependence of the oxygen and aluminum contents films deter-
mined by NRA and RBS upon the time of anodizing.
Table II. Comparison of oxygen contents determined by NRA of
specimens anodized in ABE and PAE16.
Time
͑s͒
O NRA ABE
O NRA PAE16
␩ = PAE16/ABE
͑%͒
͑1015 atom cm−2
͒
͑1015 atom cm−2
͒
16
35
91
244
497
1155
133
275
725
55
55
63
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
C405
(a)
Figure 11. Comparison of the excitation curves determined using the reso-
nance at 151 keV of the 18O͑p,␣͒15N reaction for the near-surface regions of
S40, Smax, and Sconst.
(b)
become increasingly significant and predominant as the voltage de-
clines toward a final steady value. The metal/film interface is shal-
lowly scalloped in the formation of the initial pores, with the bot-
toms of scallops coincident with the regions of thinner oxide.
Deeper scallops develop with the major pores of the film. The gen-
eral behavior is similar to that reported by O’Sullivan and Wood23
for the early stages of film growth in PAE at slightly increased
current density and temperature. Further, previous work42 has also
shown the relation between the surface texture and pore develop-
ment for films formed in oxalic acid, with cell sizes and scallop
depths increasing during the initial stages of the oxide growth and
with relatively smooth surfaces promoting the initiation of more
numerous pores.
Kinetics of film growth.— Figure 4 reveals a slower rate of film
growth in the initial period of anodizing than at later times. The
average rates, calculated using the difference between the thick-
nesses of the anodic film and the oxide on the electropolished alu-
minum ͑ϳ2 nm͒, were ϳ1.44, 1.57, 1.60, and 1.89 nm s−1 for
S20, S40, S60, and S100, respectively, suggesting an accelerating
rate. A final rate of ϳ2.4 nm s−1 ͑Fig. 4͒ correlated with the for-
mation of the major pores. The barrier regions of the films increased
in thickness with formation ratios of ϳ1.17 and ϳ1.10 nm V−1
(c)
Figure 10. Experimental ͑points͒ and simulated ͑solid line using SPACES͒
excitation curves determined using the resonance at 151 keV of the
18O͑p,␣͒15N reaction for ͑a͒ S20, ͑b͒ S40, and ͑c͒ S100. The arrows in the
schematic diagrams of the 18O distributions denote the thicknesses of the
anodic films determined by TEM.
Figure 12. Comparison of the excitation curves determined using the reso-
nance at 151 keV of the 18O͑p,␣͒15N reaction for S20 and Sconst-ss. The
latter specimen was analyzed from the metal side.
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C406
Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
during the formation of the texture-dependent and major pores, re-
spectively. Because the oxidation of aluminum is the principal an-
odic reaction, the kinetics indicates that the ratio of the thickness
anodic film relative to the thickness of the oxidized aluminum in-
creases as the film evolves. Aluminum oxidizes at a rate of
1.56 nm s−1 during anodizing at 4.5 mA cm−2 at 100% current
efficiency. Thus, in the initial and final regions of Fig. 4, the film
grows at ϳ0.90 and ϳ1.54 times the rate of metal oxidation, re-
spectively. However, secondary porosity has been observed in films
formed in PAE,43,44 leading to pores along cell walls similar to those
of Fig. 3f; such pores may result from oxygen generation associated
with copper impurity in the aluminum.44 Thus, in reality, a small
amount of the current may be used in forming oxygen, leading to an
overestimation in the previously calculated rate of metal oxidation.
Film compositions and efficiency of film growth.— Ono et al.43
used controlled crystallization of cell walls of films formed in PAE
to disclose an amorphous region next to the electrolyte, a partially
crystallized intermediate region with both regions containing phos-
phorus and a highly crystallized phosphorus-free innermost layer.
The amorphous region was suggested to be stabilized by protons.
Ono and Masuko showed that the phosphorus-containing region in-
creased in thickness with an increase in the anodizing voltage.45
Formation at 100 V resulted in phosphorus in the outer ϳ0.78 of the
pore wall, with an average P:Al atomic ratio of ϳ0.043 for the film.
The latter result is reasonably consistent with the present P:Al ratio
of ϳ0.05:0.06 and a ratio of ϳ0.05 found previously in the
phosphorus-containing region of barrier films formed in phosphate
electrolytes at low pH.33 Assuming a P:Al ratio of 0.05, the average
composition of the present films can be expressed as
Al2O3·0.106AlPO4. The average charge number for oxygen is 1.85
due to the incorporation of phosphate ions. The charge number
would be reduced by the presence of any HPO24−, H2PO4−, OH− ions,
or water molecules. In reality, the phosphorus species may be inte-
grated into the amorphous alumina structure rather than being
present in a separate phase.
Figure 9 shows the dependences of the oxygen and aluminum
contents of films on the anodizing time, which were approximated
by linear regions for times below and above ϳ75 s. The products of
the gradients for the oxygen lines and the charge of oxygen ions
using a charge number of 1.85 divided by the anodizing current
density indicate respective efficiencies of film growth of ϳ47 and
ϳ60%. Similarly, the aluminum lines indicate efficiencies of ϳ47
and ϳ70%. Using the results of the efficiencies derived from the
measurements of both the oxygen and aluminum contents, the aver-
age efficiencies for the two linear regions are ϳ47 and ϳ65%;
although in practice, the anodizing efficiency may increase gradu-
ally with time over the two regions.
During the growth of barrier-type films at high efficiency, the
Al3+ ions migrate outward and O2− ions migrate inward, resulting in
the formation of alumina at the film surface and film base, respec-
tively. The details of ionic migration in anodic oxides are still un-
certain with diffusion of point defects,46 place exchange of cations
and anions,47 liquid droplet formation,48 and lattice relaxation
around oxygen vacancies,49 among others being considered. How-
ever, irrespective of the precise mechanism, short-circuit transport of
oxygen has been unimportant under conditions of uniform ionic
current.4 The transport numbers of Al3+ and O2− ions in anodic
alumina have been determined from marker experiments, the values
being affected by the anodizing conditions, the nature of the marker,
and the method of measurement.50-54 The transport number of Al3+
ions has usually been in the range from 0.35 to 0.5.51,55-57 Thus, an
efficiency of ϳ50 to 65% would be anticipated under conditions in
which all outwardly mobile Al3+ ions are ejected to the electrolyte,
a condition considered critical for pore initiation.55 For anodizing in
phosphate electrolyte, Takahashi et al.33 reported an efficiency of
close to ϳ50% using phosphate electrolyte at low pH. Under such
conditions, an alumina film with a density of 3.1 g cm−3 would be
ϳ0.83 times the thickness of the oxidized metal, i.e., the product of
the Pilling–Bedworth ͑PBR͒ ratio of the alumina ͑1.65͒ and the
efficiency ͑0.50͒. The experimental value in the present work
͑ϳ0.90͒ may differ from the prediction due to the accuracy of the
measurement and uncertainties in the PBR, transport numbers, and
film density used to derive the predicted value.
.— The amounts of 18O tracer
Distributions of oxygen tracer
species in films formed by sequential anodizing did not change sig-
nificantly during the evolution of the film between the barrier and
porous stages. A similar finding for films formed in sulfuric acid was
attributed to a decomposition process of the alumina at the pore
base, which resulted in cation transfer only to the solution.26,29 The
findings of the present work show that in the early stages of the
texture-dependent pore development, a large fraction of 18O is lo-
cated in a layer near the metal/film interface. However, toward the
end stages, 18O is distributed between an inner region, an interme-
diate region, and an outer region in proportions ϳ0.55:0.20:0.25.
The 18O distribution near the film surface does not change greatly
with an increase in the anodizing time and is clearly associated with
the texture-dependent porous region. Following the establishment of
the major pores in the 850 nm thick film of Sconst-ss, most 18O was
located in the outer half of the film, but 18O was also in a Յ25 nm
thick layer near the film base, possibly in the phosphorus-free re-
gion, which has a thickness of ϳ23 nm, assuming that it constitutes
20% of the barrier layer.45 In comparison with the S100, the propor-
tion of the total O18 in the barrier layer has reduced from ϳ55 to
ϳ17% due to the transfer of 18O to the cell walls.
Origin of porosity.— The authors’ previous work using tungsten
tracers has suggested that film growth under the present conditions
involves flow of film material during the formation of the major
pores.28,58 The flow prevents tungsten ions from migrating through
the barrier layer to the pore base. From studies using various elec-
trolytes, the flow appears to be promoted by incorporation of anion
species into the film.59 Mathematical modeling has indicated that the
material flows downward from beneath the pores and outward to-
ward the cell walls.60
The texture-dependent and major porosities differ significantly,
with interpore distances and pore volumes of ϳ75 nm and 60% and
ϳ300 nm and 10%, respectively. In a previous work that invokes
field-assisted dissolution,23 pore initiation has been proposed to in-
volve a first stage of enhanced film growth due to preferential oxi-
dation at certain sites of the aluminum substrate. The current subse-
quently becomes focused at the thinner film regions, where embryo
pores are stabilized by a balance between the formation and the
dissolution of oxide. However, the present findings ͑Fig. 3͒ show
less metal oxidation beneath the thickened film than elsewhere. Pre-
suming that isotopic order is conserved at preferentially oxidized
sites, 18O should become buried beneath the 16O-rich oxide, which
is contrary to the observation. The thickness differential of the oxide
is therefore suggested to be due to either the displacement of plas-
ticized film material from the thinner regions61 or the field-assisted
dissolution at the thinner regions, which is facilitated by the scal-
loped texture of the metal/film interface. Tungsten tracer studies
may distinguish which is the main process. Flow may be driven by
stresses from electrostriction and film growth. Initially displaced
material would contain 18O from PAE18 and, upon change to
PAE16, both 16O and 18O ions, with near-surface oxide providing
mainly 16O ions, which are incorporated preferentially at thinner
regions. At such regions, 18O ions migrate inward and a proportion
of these are near the metal. Most of the 18O is eventually transported
to the cell walls of the major pores, either by migration along the
approximately radial directions of the field lines in the barrier layer
or by flow of film material from the barrier layer. However, 18O is
still present at the film base of Sconst-ss, when the film is ϳ34
times thicker than the film formed in PAE18 and the length of major
pores is ϳ9 times the thickness of the barrier layer; the film base is
possibly a region of reduced flow.
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 157 ͑11͒ C399-C407 ͑2010͒
C407
͑1987͒.
The transport processes in porous alumina are probably complex
compared with those of barrier films due to nonuniform currents and
stresses. A recent theoretical analysis of ion migration by Hebert and
Houser62 highlighted the importance of mechanical stress, leading to
a dependence of transport numbers on current density. Another work
has suggested that stress may arise from density variations in the
barrier layer.63 Experimental studies have shown a significantly in-
12. C. Courty, H. J. Mathieu, and D. Landholt, Electrochim. Acta, 36, 1623 ͑1991͒.
13. J. B. Wilkinson, S. Brooks, and D. G. Lees, Mater. Sci. Technol., 4, 1114 ͑1988͒.
14. A. A. Alibhai, R. J. Chater, D. S. McPhail, and B. A. Shollock, Appl. Surf. Sci.,
203–204, 630 ͑2003͒.
15. F. Rouillard, C. Cabet, K. Wolski, and M. Pijolat, Corros. Sci., 51, 752 ͑2009͒.
16. A. Tempez, S. Canulescu, I. S. Molchan, M. Döbeli, J. A. Whitby, L. Lobo, J.
Michler, G. E. Thompson, N. Bordel, P. Chapon, et al., Surf. Interface Anal., 41,
966 ͑2009͒.
17. S. Wernick, R. Pinner, and P. G. Sheasby, The Surface Treatment and Finishing of
Aluminum and its Alloys, Finishing, Teddington ͑1996͒.
18. H. Masuda and K. Fukada, Science, 268, 1466 ͑1995͒.
19. W. Lee, R. Ji, U. Gösele, and K. Nielsch, Nature Mater., 5, 741 ͑2006͒.
20. D. Routkevitch, T. Bigioni, M. Moskovits, and J. M. Xu, J. Phys. Chem., 100,
14037 ͑1996͒.
21. A. Yamaguchi, K. Hotta, and N. Teramae, Anal. Chem., 81, 105 ͑2009͒.
22. F. Keller, M. S. Hunter, and D. L. Robinson, J. Electrochem. Soc., 100, 411
͑1953͒.
23. J. P. O’Sullivan and G. C. Wood, Proc. R. Soc. London, Ser. A, 317, 511 ͑1970͒.
24. G. C. Wood and J. P. O’Sullivan, Corros. Sci., 15, 1865 ͑1970͒.
25. K. Schwirn, W. Lee, R. Hillebrand, M. Steinhart, K. Nielsch, and U. Gösele, ACS
Nano, 2, 302 ͑2008͒.
creased efficiency of film growth in PAE between
5
and
64
30 mA cm−2
,
with similar findings by Vrublevsky et al. for sul-
furic acid.65 Such influences of current density may be relevant to
the kinetics and efficiency of film growth revealed in Fig. 4 and 9 as
the current density changes with changing dimensions of pores. It
may be speculated that the simultaneous ionic migration and flow
lead to self-regulating but nonuniform distributions of phosphorus
species, which are possibly critical to the transition between the
texture-dependent and major porosities that depend on the substrate
and anodizing voltage.
26. C. Cherki and J. Siejka, J. Electrochem. Soc., 120, 784 ͑1973͒.
27. T. P. Hoar and N. F. Mott, J. Phys. Chem. Solids, 9, 97 ͑1959͒.
28. P. Skeldon, G. E. Thompson, S. J. Garcia-Vergara, L. Iglesias-Rubianes, and C. E.
Blanco-Pinzon, Electrochem. Solid-State Lett., 9, B47 ͑2006͒.
29. J. Siejka and C. Ortega, J. Electrochem. Soc., 124, 883 ͑1977͒.
30. K. Shimizu, P. Skeldon, G. E. Thompson, and G. C. Wood, Surf. Interface Anal., 4,
208 ͑1982͒.
31. M. Mayer, in Proceedings of the 15th International Conference on the Application
of Accelerators in Research and Industry, J. L. Duggan and I. L. Morgan, Editors,
American Institute of Physics, p. 541 ͑1999͒.
32. I. C. Vickridge and G. Amsel, Nucl. Instrum. Methods Phys. Res. B, 45, 6 ͑1990͒.
33. H. Takahashi, K. Fujimoto, and M. Nagayama, J. Electrochem. Soc., 135, 1349
͑1988͒.
34. W. A. Lanford, R. S. Alwitt, and C. K. Dyer, J. Electrochem. Soc., 127, 405
͑1980͒.
35. F. Le Coz, L. Arurault, S. Fontorbes, V. Vilar, L. Datas, and P. Winterton, Surf.
Interface Anal., 42, 227 ͑2010͒.
36. W. N. Lennard, G. R. Massoumi, P. F. A. Alkemade, I. V. Mitchell, and S. Y. Tong,
Nucl. Instrum. Methods Phys. Res. B, 61, 1 ͑1991͒.
37. P. Skeldon, K. Shimizu, G. E. Thompson, and G. C. Wood, Surf. Interface Anal., 5,
247 ͑1983͒.
38. J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, SRIM, The Stopping and Range of
Ions in Matter, www.srim.org, last accessed June 2010.
39. A. C. Harkness and L. Young, Can. J. Chem., 44, 2409 ͑1966͒.
40. B. E. Cuff and J. Grant, J. Inst. Met., 87, 248 ͑1958͒.
41. M. S. Hunter and D. L. Robinson, Trans. Am. Inst. Min., Metall. Pet. Eng., 197,
717 ͑1953͒.
42. T. Asahina, H. Ishihara, H. Asoh, and S. Ono, J. Jpn. Inst. Light Met., 58, 375
͑2008͒.
43. S. Ono, H. Ichinose, and M. Masuko, Corros. Sci., 33, 841 ͑1992͒.
44. I. S. Molchan, T. V. Molchan, N. V. Gaponenko, P. Skeldon, and G. E. Thompson,
Electrochem. Commun., 12, 693 ͑2010͒.
Conclusions
The growth of anodic films on electropolished aluminum at
4.5 mA cm−2 in 0.4 M phosphoric acid electrolyte at 295 K occurs
with no significant loss of oxygen to the electrolyte between the
initial stage of the formation of a barrier film to the final stage of
establishment of the major pores of the porous film.
The rate of film growth increases from ϳ1.4 to ϳ2.4 nm s−1
between the commencement of anodizing and the formation of the
major pores, which coincides with an increasing efficiency from
ϳ47 to ϳ65%, expressing the proportion of aluminum species re-
tained within the film.
The thickening of the initial barrier film is accompanied by the
development of an outer region of film with an ϳ60% porosity,
followed by the formation of the major pore region with an ϳ10%
porosity. The morphology of the former pores is determined by the
texture of the electropolished aluminum.
The 18O tracer is redistributed by the development of pores. Fol-
lowing the formation of an ϳ850 nm porous film, 18O is at the film
surface within the cell walls of the major pores and the inner part of
the barrier layer. 18O at the surface is associated with the texture-
dependent porous region and is relatively constant as the films
thicken. In contrast, 18O of the barrier layer decreases due to trans-
port to the cell walls.
Acknowledgments
45. S. Ono and N. Masuko, Corros. Sci., 33, 503 ͑1992͒.
46. L. F. Lin, C. Y. Chao, and D. D. Macdonald, J. Electrochem. Soc., 128, 1194
͑1981͒.
47. A. T. Fromhold, Jr., J. Electrochem. Soc., 127, 411 ͑1980͒.
48. N. F. Mott, Philos. Mag. B, 55, 117 ͑1987͒.
49. M.-H. Wang and K. R. Hebert, J. Electrochem. Soc., 146, 3741 ͑1999͒.
50. J. A. Davies, B. Domeij, J. P. S. Pringle, and F. Brown, J. Electrochem. Soc., 112,
675 ͑1965͒.
51. F. Brown and W. D. Mackintosh, J. Electrochem. Soc., 120, 1096 ͑1973͒.
52. K. Shimizu, G. E. Thompson, and G. C. Wood, Thin Solid Films, 88, 255 ͑1982͒.
53. N. Khalil and J. S. L. Leach, Electrochim. Acta, 31, 1279 ͑1986͒.
54. K. Shimizu, G. M. Brown, H. Habazaki, K. Kobayashi, P. Skeldon, G. E. Thomp-
son, and G. C. Wood, Electrochim. Acta, 44, 2297 ͑1999͒.
55. G. E. Thompson, Y. Xu, P. Skeldon, K. Shimizu, S. H. Han, and G. C. Wood,
Philos. Mag. B, 55, 651 ͑1987͒.
The authors are grateful to the Engineering and Physical Sci-
ences Research Council ͑U.K.͒ for their support to this work under
the Portfolio Award “Light Alloys for Environmentally Sustainable
Transport.” They also acknowledge the support of the European
Community’s Seventh Framework Programme FP7/2007-2013 un-
der grant agreement no. PIEF-GA-2008-219913. A.B.W. gives spe-
cial thanks to George Amsel from the Institut des Nano Sciences des
Paris for enlightening discussions and his endless kindness and Em-
rick Briand for his technical assistance with the ion beam analysis.
References
1. D. J. Neild, P. J. Wise, and D. G. Barnes, J. Phys. D, 5, 2292 ͑1972͒.
2. J. M. Calvert and D. G. Lees, J. Radioanal. Nucl. Chem., 48, 287 ͑1979͒.
3. E. Rosencher, A. Straboni, S. Rigo, and G. Amsel, Appl. Phys. Lett., 34, 254
͑1979͒.
4. G. Amsel and D. Samuel, J. Phys. Chem., 23, 1707 ͑1962͒.
5. G. Amsel, C. Cherki, M. Croset, G. Feuillade, J. P. Nadai, C. Ortega, and J. Siejka,
Collect. Czech. Chem. Commun., 36, 883 ͑1971͒.
56. K. Shimizu, K. Kobayashi, G. E. Thompson, G. C. Wood, and P. Skeldon, Philos.
Trans. R. Soc. London, Ser. A, 354, 213 ͑1996͒.
57. L. Iglesias-Rubianes, P. Skeldon, G. E. Thompson, H. Habazaki, and K. Shimizu,
Corros. Sci., 45, 2905 ͑2003͒.
58. S. J. Garcia-Vergara, P. Skeldon, G. E. Thompson, and H. Habazaki, Electrochim.
Acta, 52, 681 ͑2006͒.
6. G. Amsel and D. Samuel, Anal. Chem., 39, 1689 ͑1967͒.
7. G. Battistig, G. Amsel, E. d’Artemare, and I. C. Vickridge, Nucl. Instrum. Methods
Phys. Res. B, 61, 369 ͑1991͒.
8. G. Battistig, G. Amsel, E. d’Artemare, and I. C. Vickridge, Nucl. Instrum. Methods
Phys. Res. B, 66, 1 ͑1992͒.
9. I. C. Vickridge, D. Tromson, I. Trimaille, J.-J. Ganem, E. Szilágyi, and G. Battistig,
Nucl. Instrum. Methods Phys. Res. B, 190, 574 ͑2002͒.
10. M. J. Graham, S. Moisa, G. I. Sproule, X. Wu, D. Landheer, A. J. SpringThorpe, P.
Barrios, S. Kleber, and P. Schmuki, Corros. Sci., 49, 31 ͑2007͒.
11. J. S. Arlow, D. F. Mitchell, and M. J. Graham, J. Vac. Sci. Technol. A, 5, 572
59. S. J. Garcia-Vergara, P. Skeldon, G. E. Thompson, and H. Habazaki, Surf. Interface
Anal., 39, 860 ͑2007͒.
60. J. E. Houser and K. Hebert, Nature Mater., 8, 415 ͑2009͒.
61. D. H. Bradhurst and J. S. Llewelyn Leach, J. Electrochem. Soc., 113, 1245 ͑1966͒.
62. K. R. Hebert and J. E. Houser, J. Electrochem. Soc., 156, C275 ͑2009͒.
63. G. Patermarakis, J. Electroanal. Chem., 635, 39 ͑2009͒.
64. S. J. Garcia-Vergara, P. Skeldon, G. E. Thompson, and H. Habazaki, Nanotechnol-
ogy, 18, 415605 ͑2007͒.
65. I. Vrublevsky, V. Parkoun, V. Sokol, J. Schreckenbach, and G. Marx, Appl. Surf.
Sci., 222, 215 ͑2004͒.
Downloaded on 2013-12-16 to IP 129.97.180.147 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use