Self-organized formation of hexagonal pore arrays in anodic alumina

Article abstract of DOI:10.1063/1.121004

The conditions for the self-organized formation of ordered hexagonal structures in anodic alumina were investigated for both oxalic and sulfuric acid as an electrolyte. Highly ordered pore arrays were obtained for oxidation in both acids. The size of the ordered domains depends strongly on the anodizing voltage. This effect is correlated with a voltage dependence of the volume expansion of the aluminum during oxidation and the current efficiency for oxide formation. The resulting mechanical stress at the metal/oxide interface is proposed to cause repulsive forces between the neighboring pores which promote the formation of ordered hexagonal pore arrays.

Full text of DOI:10.1063/1.121004

Self-organized formation of hexagonal pore arrays in anodic alumina
O. Jessensky, F. Müller, and U. Gösele
Citation: Appl. Phys. Lett. 72, 1173 (1998); doi: 10.1063/1.121004
View online: http://dx.doi.org/10.1063/1.121004
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v72/i10
Published by the American Institute of Physics.
Related Articles
Crystallographic and magnetic identification of secondary phase in orientated Bi5Fe0.5Co0.5Ti3O15 ceramics
J. Appl. Phys. 112, 073919 (2012)
Fabrication and performance optimization of the magneto-optical (Tb1−xRx)3Al5O12 (R=Y, Ce) transparent
ceramics
Appl. Phys. Lett. 101, 131908 (2012)
Domain evolution in lead-free thin film piezoelectric ceramics
J. Appl. Phys. 112, 052014 (2012)
Ferroelectric domain morphology and structure in Li-doped (K,Na)NbO3 ceramics
J. Appl. Phys. 112, 052005 (2012)
Room temperature ferroelectric and magnetic investigations and detailed phase analysis of Aurivillius phase
Bi5Ti3Fe0.7Co0.3O15 thin films
J. Appl. Phys. 112, 052010 (2012)
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 08 Oct 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

APPLIED PHYSICS LETTERS
VOLUME 72, NUMBER 10
9 MARCH 1998
Self-organized formation of hexagonal pore arrays in anodic alumina
O. Jessensky, F. M u¨ ller,a) and U. G o¨ sele
Max-Planck-Institute of Microstructure Physics, Weinberg 2, D-06120 Halle, Germany
͑
Received 7 July 1997; accepted for publication 14 January 1998͒
The conditions for the self-organized formation of ordered hexagonal structures in anodic alumina
were investigated for both oxalic and sulfuric acid as an electrolyte. Highly ordered pore arrays were
obtained for oxidation in both acids. The size of the ordered domains depends strongly on the
anodizing voltage. This effect is correlated with a voltage dependence of the volume expansion of
the aluminum during oxidation and the current efficiency for oxide formation. The resulting
mechanical stress at the metal/oxide interface is proposed to cause repulsive forces between the
neighboring pores which promote the formation of ordered hexagonal pore arrays. © 1998
American Institute of Physics. ͓S0003-6951͑98͒03510-4͔
Porous oxide growth on aluminum under anodic bias in
various electrolytes has been studied for more than 40 years.1
Because of their relatively regular structure with narrow size
distributions of pore diameters and interpore spacings, po-
rous alumina membranes are used for the fabrication of na-
which reach the oxide/electrolyte interface contribute to ox-
12,13
ide formation in the case of barrier oxide growth.
The
atomic density of aluminum in alumina is by a factor of two
lower than in metallic aluminum. A possible origin of forces
between neighboring pores is therefore the mechanical stress
which is associated with the expansion during oxide forma-
tion at the metal/oxide interface. Since the oxidation takes
place at the entire pore bottom simultaneously, the material
can only expand in the vertical direction, so that the existing
pore walls are pushed upwards. Under usual experimental
conditions the expansion of aluminum during oxidation leads
to less than twice the original volume, since Al3 ions are
mobile in the oxide under the electric field, so partly the
oxidized aluminum does not contribute to oxide formation.
Whereas relative cation transport numbers between 0.4 and
2
–4
nometer scale composites.
In 1970, O’Sullivan and
5
Wood presented a model to describe the self-regulating pore
growth. This model based on the electric field distribution at
the pore tips is able to explain why pores grow at all and why
their size distribution is quite narrow. This and further re-
6
–8
fined models
can give microscopic explanations for the
ϩ
dependence of, e.g., pore diameters and pore distances on
applied voltage or electrolyte composition.
Just recently Masuda et al.9 reported self-organized
pore growth, leading to a densely packed hexagonal pore
structure for certain sets of parameters. But highly regular
polycrystalline pore structures occur only for a quite small
processing window, whereas an amorphous pore structure
can be obtained for a very wide range of parameters without
substantial change in morphology. The former models based
only on field distribution cannot easily explain this behavior.
The self-organized arrangement of neighboring pores in hex-
agonal arrays can be explained by any repulsive interaction
between the pores. In this work, the dependence of the struc-
tural properties on the anodization conditions have been in-
vestigated to gain more insight into the formation conditions
of ordered pore arrays and to find a microscopic explanation
of the repulsive forces.
,10
0
.5 have been reported in literature independently of experi-
12,13
mental conditions,
in our experiments the relative thick-
ness of the porous alumina layer compared to the consumed
aluminum was found to vary with voltage and electrolyte
composition.
Since no oxide formation takes place at the oxide/
electrolyte interface during porous oxide growth but all cat-
In order to explain the effect of self-organization, the
situation during steady state pore growth has to be consid-
ered ͑Fig. 1͒. Pores grow perpendicular to the surface with an
equilibrium of field-enhanced oxide dissolution at the oxide/
electrolyte interface and oxide growth at the metal/oxide
5
,8
interface. While the latter is due to the migration of oxy-
gen containing ions (O2 /OH ) from the electrolyte through
the oxide layer at the pore bottom, Al3 ions which simulta-
neously drift through the oxide layer are ejected into the
Ϫ
Ϫ
ϩ
11
solution at the oxide/electrolyte interface. The fact that
3
ϩ
Al ions are lost to the electrolyte has been shown to be a
3ϩ
prerequisite for porous oxide growth, whereas Al ions
FIG. 1. Expansion of aluminum during anodic oxidation. On the left the
level of the unoxidized metal surface is depicted.
a͒Electronic mail: fmuel@mpi-halle.mpg.de
0003-6951/98/72(10)/1173/3/$15.00
1173
© 1998 American Institute of Physics
Downloaded 08 Oct 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

1174
Appl. Phys. Lett., Vol. 72, No. 10, 9 March 1998
Jessensky, M u¨ ller, and G o¨ sele
ions reaching the interface are ejected into the electrolyte,
this observation can be explained by a variation of the rela-
tive transport numbers of Al3 and (O /OH ) ions with
voltage and electrolyte composition. Therefore we investi-
gated the influence of the mechanical stress on the structural
features by varying the forming voltage using two of the
major anodizing acids.
ϩ
2Ϫ
Ϫ
The volume expansion during oxide growth was ob-
tained by two different measurement techniques. On the one
hand, the step height between the aluminum surface and the
alumina surface at the edge of the anodized regions was mea-
sured with a mechanical profiler. The overall thickness of the
alumina was determined by investigating the cross section in
an optical microscope. The ratio of both values leads to rela-
tive alumina thickness. Since the alumina thickness varied
between the edge and center of the sample by 20%, the ac-
curacy of the measurement was estimated to be within 10%.
The second method was based on measuring the current ef-
ficiency for oxide formation from the mass of the produced
alumina and the charge flow during oxidation. Since the
charge transfer over the anode is carried entirely by electrons
that are generated by oxidation of Al to Al3 , the current
efficiency for oxide formation yields the ratio of aluminum
which is transformed to Al O to the total amount of oxi-
FIG. 2. SEM bottom view of a porous anodic Al O3 layer with ordered
hexagonal pore structure after opening the pore bottoms by chemical etching
2
͓
40 V, 0.3 M (COOH) , 1 °C͔.
2
and 4 days, layer thicknesses between 100 and 200 ␮m and
aspect ratios ͑ratios of diameter and depth͒ of the pores of
more than 1000 were obtained. After anodization, the re-
ϩ
maining aluminum was removed in a saturated HgCl solu-
2
tion. Subsequently, the pore bottoms were opened by chemi-
cal etching in 5 wt.% aqueous phosphoric acid.
2
3
dized aluminum. A current efficiency of 100% therefore cor-
responds to an expansion of the initial aluminum during ox-
ide formation to double the volume. The weight of the
alumina has to be corrected for the film porosity, which is
approximately 10% as determined by SEM investigations on
cleaved edges and in agreement with other results obtained
While the pores nucleate at the surface at almost random
positions, ordered pore domains of ␮m size were observed at
the bottom of the layers ͑Fig. 2͒. The lattice constant within
the ordered domains was approximately proportional to the
voltage, in accordance with the literature, where proportion-
ality of both pore and cell wall diameters to the voltage has
6
Ϫ
for oxalic acid. Incorporation of acid anions and OH also
1
4,15
5,6
has to be considered as a potential source of error.
all the error is estimated to be about 10%.
Over-
been established. The quality of the ordering of this struc-
ture depends on the anodization voltage. The best results
were obtained for 40 V, which corresponds to the oxidation
High purity ͑99.999%͒ aluminum foils from Goodfellow
were used as the starting material. Prior to anodizing, the
aluminum was annealed under forming gas or nitrogen atmo-
sphere at 500 °C in order to enhance the grain size in the
metal and to obtain homogeneous conditions for pore growth
over large areas. Subsequently, the foils were electropolished
9
parameters given by Masuda and Fukuda. Perfect hexagonal
ordering is obtained within domains of ␮m size, which are
separated from neighboring domains with different orienta-
tion of the pore lattice by grain boundaries. Thus, a polycrys-
talline pore structure is observed. When the voltage is in-
creased to 60 V, ordered domains of comparable size, but
also nonordered regions are observed in the sample. Lower-
ing the voltage to 30 V leads to a much smaller size of the
ordered domains, so that moderate voltages seem to be pref-
erable in order to obtain highly ordered structures.
The volume expansion during oxidation was determined
by measuring the layer thicknesses as well as the current
efficiencies for oxide formation with consistent results. The
dependence of the current efficiency on the forming voltage
is given in Table I. The current efficiency is increasing with
rising voltage. Optimal conditions for the formation of or-
dered structures occur at a current efficiency which corre-
in a 4:4:2 by weight mixture of H PO , H SO , and H O.
3
4
2
4
2
While the surface roughness of the aluminum foils was in the
range of 5 ␮m before electropolishing, the roughness of the
electropolished surface was typically between 20 and 30 nm
on a lateral length scale of 10 ␮m. Porous alumina mem-
branes prepared from nonannealed or from nonelectropol-
ished aluminum foils did not exhibit any ordered pore do-
mains. A more detailed analysis of the influence of the
preparation conditions on the growth kinetics and structural
properties of the porous alumina membranes will be reported
1
6
elsewhere.
The aluminum foils were mounted on a copper plate
serving as the anode and exposed to the acid in a thermally
isolated electrochemical cell. The diameter of the exposed
area was 10 mm. While the anode temperature was kept
constant at 1 °C, the temperature of the electrolyte deviated
from that value by less than 0.1 °C. The electrolyte was rig-
orously stirred during anodization. In a first set of experi-
ments, 0.3 M oxalic acid, as used by Masuda and Fukuda,9
was used as the electrolyte. The anodization voltage was
varied between 30 and 60 V. With oxidation times between 2
TABLE I. Current efficiency for oxide formation in 0.3 M oxalic acid at
1
°C.
Voltage ͑V͒
Current efficiency ͑%͒
30
45
55
65
4
6
0
0
Downloaded 08 Oct 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Appl. Phys. Lett., Vol. 72, No. 10, 9 March 1998
Jessensky, M u¨ ller, and G o¨ sele
1175
measured current efficiency values are significantly lower
than those deduced from thickness measurements, especially
for the lowest anodizing voltages. A strong increase of the
volume expansion with the voltage can be observed. In ad-
dition to the sample prepared at 18.7 V, the samples anod-
ized with 19 and 20 V to a minor extent also exhibit ordered
domains. Again, conditions which lead to a moderate expan-
sion of the aluminum during oxide formation, are most suit-
able for obtaining hexagonal ordered pore arrays. With vol-
ume contraction or very strong expansion during oxide
formation, no ordered structures can be achieved. While in
the case of contraction no repulsive forces between the pores
are expected, large volume expansion may result in structural
defects in the alumina and irregular pore growth. A large
volume expansion is also associated with high anodizing
voltages and growth rates and therefore with reduced inter-
action between the neighboring pores.
FIG. 3. SEM bottom view of a porous alumina layer prepared in sulfuric
acid ͑20 wt.% H SO , 18.7 V, 1 °C͒.
2
4
We conclude that in porous anodic alumina, ordered
hexagonal pore arrays formed by a self-organization process
during growth with oxalic as well as sulfuric acid as an elec-
trolyte. A systematic dependence of the volume expansion of
the aluminum during oxide formation on the anodizing volt-
age was observed. With both acids, optimal conditions for
the growth of ordered structures are accompanied by a mod-
erate expansion of the aluminum, whereas no ordered do-
mains can be observed in the cases of contraction or very
strong volume expansion. We suggest that the mechanical
stress, which is associated with the expansion of the alumi-
num during oxide formation is the cause of repulsive forces
between neighboring pores during the oxidation process,
which lead to self-organized formation of hexagonal pore
arrays.
sponds to a moderate expansion of the aluminum during oxi-
dation.
In order to check whether this condition for optimal or-
dering also holds for a different electrolyte, the voltage de-
pendence of the alumina thickness and the ordering of the
structures was investigated using 20 wt.% sulfuric acid at
1
°C. Although the applied voltages ranged only between 18
and 25 V, the anodizing current was on average 4 times
higher than in the case of oxalic acid. With oxidation times
between 1 and 2 days, alumina thicknesses of about 200 ␮m
were achieved. The most regularly ordered structure, ob-
tained with 18.7 V anodizing voltage, is shown in Fig. 3.
Corresponding to the lower voltage, the feature size is
roughly a factor of two smaller than with oxalic acid. Or-
dered domains with a hexagonal structure can be observed,
although nonordered domains are also visible. The lower de-
gree of perfection of the structure compared to that observed
with oxalic acid is probably due to the higher growth rate
and reduced interaction time for the self-organization process
as well as to not entirely optimized anodization conditions.
The relation between the anodizing voltage and the alumina
layer thickness in relation to the consumed aluminum ob-
tained by directly measuring the sample thickness, is shown
in Table II. Due to the higher rate for chemical etching and
the corresponding weight loss of the alumina, the directly
1
F. Keller, M. S. Hunter, and D. L. Robinson, J. Electrochem. Soc. 100,
411 ͑1953͒.
C. H. Martin, Chem. Mater. 8, 1739 ͑1996͒.
H. Masuda and M. Satoh, Jpn. J. Appl. Phys., Part 2 35, L126 ͑1996͒.
D. Routkevitch, A. A. Tager, J. Harujama, D. Almawlawi, M. Moskovits,
2
3
4
and J. M. Xu, IEEE Trans. Electron Devices 40, 1646 ͑1996͒.
J. P. O’Sullivan and G. C. Wood, Proc. R. Soc. London, Ser. A 317, 511
5
͑
1970͒.
6
7
8
G. Bailey and G. C. Wood, Trans. Inst. Metal. Finish. 52, 187 ͑1974͒.
G. E. Thompson and G. C. Wood, Nature ͑London͒ 290, 230 ͑1981͒.
V. P. Parkhutik and V. I. Shershulsky, J. Phys. D: Appl. Phys. 25, 1258
͑1992͒.
H. Masuda and K. Fukuda, Science 268, 1466 ͑1995͒.
H. Masuda, F. Hasegwa, and S. Ono, J. Electrochem. Soc. 144, L127
9
0
1
͑
1997͒.
1
1
1
2
TABLE II. Alumina layer thickness compared to the consumed aluminum
layer using 20 wt. % H SO at 1 °C.
J. Siejka and C. Ortega, J. Electrochem. Soc. 124, 883 ͑1977͒.
Y. Xu, G. E. Thompson, and G. C. Wood, Trans. Inst. Met. Finish. 63, 98
2
4
͑
1985͒.
13
Voltage ͑V͒
Layer thickness ͑% Al layer͒
K. Shimizu, K. Kobayashi, G. E. Thompson, and G. C. Wood, Philos.
Mag. A 66, 643 ͑1992͒.
G. E. Thompson and G. C. Wood, Corrosion: Aqueous Processes and
Passive Films, edited by J. C. Scully ͑Academic, London, 1983͒, p. 205.
A. Despi c´ and V. P. Parkhutik, Modern Aspects of Electrochemistry, ed-
ited by J. O. Bockris, R. E. White, and B. E. Conway ͑Plenum, New York,
1989͒, p. 401.
O. Jessensky, F. M u¨ ller, and U. G o¨ sele ͑unpublished͒.
2
2
1
1
1
5
0
9
8.7
8
162
134
141
122
86
14
1
1
5
6
Downloaded 08 Oct 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions