Formation of aluminum nanodot array by combination of nanoindentation and anodic oxidation of aluminum

Article abstract of DOI:10.1016/S0039-6028(03)00433-3

Porous alumina films formed by aluminum anodic oxidation have been intensively studied to use them as molds to form nanowires or dots. Recently we established the formation of porous alumina on solid substrates such as Si, and found a formation of an ordered aluminum hexagonal dot array after finishing of anodic oxidation on a SiO₂ substrate. We investigated AFM nanoindentation to control the initial position of nanoholes during anodic oxidation, in order to realize nanoholes and dots with various sizes and densities. Arrays of nanoholes with a nearest neighbor distance from 50 to 120 nm were successfully formed. Al dot tetragonal and hexagonal arrays were formed after selective wet chemical etching of porous alumina film. Electron transport through Al nanodots array at low temperature exhibited a non-linear characteristic with a suppression of current around zero bias, which strongly suggests existence of Coulomb blockade.

Full text of DOI:10.1016/S0039-6028(03)00433-3

Surface Science 532–535 (2003) 317–323
www.elsevier.com/locate/susc
Formation of aluminum nanodot array by combination
of nanoindentation and anodic oxidation of aluminum
*
Shoso Shingubara , Yusuhiko Murakami, Kazunori Morimoto,
Takayuki Takahagi
Graduate School of Advanced Sciences and Matters, Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima 739-8530, Japan
Abstract
Porous alumina films formed by aluminum anodic oxidation have been intensively studied to use them as molds to
form nanowires or dots. Recently we established the formation of porous alumina on solid substrates such as Si, and
2
found a formation of an ordered aluminum hexagonal dot array after finishing of anodic oxidation on a SiO substrate.
We investigated AFM nanoindentation to control the initial position of nanoholes during anodic oxidation, in order to
realize nanoholes and dots with various sizes and densities. Arrays of nanoholes with a nearest neighbor distance from
50 to 120 nm were successfully formed. Al dot tetragonal and hexagonal arrays were formed after selective wet chemical
etching of porous alumina film. Electron transport through Al nanodots array at low temperature exhibited a non-
linear characteristic with a suppression of current around zero bias, which strongly suggests existence of Coulomb
blockade.
Ó 2003 Published by Elsevier Science B.V.
Keywords: Electrochemical methods; Electrical transport measurements; Aluminum; Aluminum oxide; Electrical transport
(conductivity, resistivity, mobility, etc.); Self-assembly; Surface structure, morphology, roughness, and topography; Solid–liquid
interfaces
1
. Introduction
mina nanoholes with distances of several tens of
nm can be formed under the self-organization
conditions of anodic oxidation. We recently suc-
ceeded in forming an Al dot hexagonal array on a
Porous alumina films formed by anodic oxida-
tion of aluminum (Al) have been intensively
studied for use as molds to form nanowires or dots
by depositing various metals or semiconductors in
them, since highly ordered nanohole arrays can be
formed [1–12]. By the use of oxalic acid, sulfuric
acid, or phosphoric acid, trigonal lattices of alu-
SiO /Si(1 0 0) substrate through adequate condi-
2
tions of anodic oxidation and the subsequent se-
lective removal of porous alumina film [13–15]. Al
dots measuring a few tens of nm were formed at an
interface between the porous alumina film and the
2
SiO film by a break-up of Al film due to com-
pletion of anodic oxidation. These Al dots are
useful for application to quantum effect devices
such as single electron memory nodes or quantum
cellular automata devices based on tunneling
phenomena within two-dimensional quantum dot
*
Corresponding author. Address: Department of Electrical
Engineering, Hiroshima University, 1-4-1 Kagamiyama, Hig-
ashi-Hiroshima 739-8527, Japan. Tel.: +81-824-24-7645; fax:
+
81-824-22-7195.
E-mail address: shingu@hiroshima-u.ac.jp (S. Shingubara).
0
039-6028/03/$ - see front matter Ó 2003 Published by Elsevier Science B.V.
doi:10.1016/S0039-6028(03)00433-3

318
S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
arrays [16]. However, the size and geometrical ar-
rangement of the ordered porous alumina nano-
hole array are restricted by the self-organization
conditions of each acid species. For example, the
nearest neighbor distances of nanoholes at the self-
organization condition are 475, 90 and 60 nm for
10–15 nm/min, and substrate was tightly attached
to water-cooled copper block, and furthermore,
sputtering was carried out by alternation of 1 min
deposition and 1 min rest so as not to raise sub-
strate temperature. By this method, aluminum film
surface roughness could be controlled within 2 nm
in root mean square roughness (RMS) even for Al
films which thickness were larger than 10 lm.
When an anodic oxidation of Al film is carried
out with the typical self-organization conditions of
40 V at 2 °C in 0.15 M oxalic acid solution, well-
ordered nanohole array is obtained when Al film
thickness is larger than 10 lm. By a so-called two
step anodic oxidation in which a properly long first
anodization and selective removal of porous alu-
mina film were carried out, a well ordered nano-
hole array is formed in the second anodic
oxidation [7]. Poly-crystalline domain structure of
nanohole trigonal lattice is formed, and the do-
main size increases with an increase in the first step
anodic oxidation time. It is not clear whether do-
main boundary corresponds to the crystalline
grain boundary. The domain size might be limited
by the grain size in the cases of self-organized
formation of nanohole arrays.
phosphoric acid (anodization voltage V ¼ 195 V)
a
[
(
10], oxalic acid (V ¼ 40 V) [7,8], and sulfuric acid
a
V
a
¼ 28 V) [9], respectively. Furthermore, the
arrangement of nanoholes is limited to trigonal
lattices only. It is desirable to form an ordered
nanohole array with a desired dimension and geo-
metry, and Masuda et al. proposed pretexturing
of aluminum surface using SiC molds that had
lithographically delineated periodic convexes
[
17,18].
In this study, at first, detail characterizations of
Al dots formed on SiO /Si substrate were made by
2
atomic force microscopy (AFM) and transmission
electron microscopy (TEM) observations. Then a
control of a geometrical arrangement of a nano-
hole array as well as Al dots by the use of AFM
nanoindentation were investigated. We focused on
the effect of indentation interval on a regularity of
nanohole array in this experiment. Finally, elec-
trical conduction characteristics between Al dots
were examined by attaching electrodes using fo-
cused ion beam induced tungsten deposition.
When a long time anodic oxidation is carried
out, most of the Al film is converted to porous
alumina. We found a formation of Al dot at the
2
interface between porous alumina and SiO . Initial
sputtered Al thickness was 20 lm, and anodic
oxidation was carried out with 40 V by 0.15 M
oxalic acid. Cross-sectional views of Al dots are
shown in Fig. 1. Scanning electron microscopy
(SEM) micrographs of the nanohole bottoms
shown in Fig. 1(a) and (b) were taken at different
locations on the same sample formed by 10 h an-
odic oxidation. These micrograph was taken after
cutting the Si substrate, and the Al cross-section
was slightly elongated. In some areas, Al film was
still continuous while periodic roughness reflecting
hemispherical nanohole bottoms existed as Fig.
1(a). However, Al film broke-up to individual dots
in other areas as seen in Fig. 1(b). A non-unifor-
mity of the rate of anodic oxidation would be as-
cribed to distribution of electric field because
electrode was contacted at the limited area of the
Al surface. With an enough long anodic oxida-
tion of 11 h, the entire Al dots were converted to
2
. Results and discussion
2
In order to form porous alumina film on SiO /Si
substrate, we deposited pure aluminum (99.999%)
film by sputtering on the substrate. The essential
point to get well-ordered nanohole array for
sputtered Al film by anodic oxidation is to keep
aluminum surface flat. The regularity of nanohole
arrangement is delicately dependent on the alu-
minum surface roughness, and usually regularity is
poor when the surface roughness is large. For this
reason, electropolishing with a diluted perchloric
acid is carried out in the most cases of anodic
oxidation for bulk Al. In the present study using
sputtered Al films, it is difficult to deposit thick Al
film to undergo electropolishing, hence we opti-
mized sputtering conditions to keep surface as flat
as can. The typical sputtering rate of Al is about

S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
319
alumina as shown in Fig. 1(c). Thus anodic oxi-
dation gradually proceeds even after breaking-up
of Al film to dots, and this phenomena was sup-
posed to be carried out by ionic conduction within
anodic alumina film. A cross-sectional TEM mi-
crograph of Al dots (Fig. 1d) clearly showed
crystalline phase of Al dot.
AFM observations of the remaining Al after
selective wet chemical etching of porous alumina
using a mixture of chromic acid and phosphoric
acid are shown in Fig. 2. The observed sample was
formed by 10.5 h and 40 V anodic oxidation by
0
2
.15 M oxalic acid with the initial Al thickness of
0 lm. It is clear that Al hexagonal nanodot array
is formed on a flat surface of SiO . The diameter
2
and height of the Al dots are 40 and 15 nm, re-
spectively, and the nearest-neighbor distance of
the Al dots is 65 nm. The diameter of the hexagon
is equal to the cell size of the nanoholes. This is
one typical example of Al dot array, however,
there is a serious problem in realizing large area Al
dot array because non-uniformities existed in the
dot size as well as in the inter-dot spacing. They
Fig. 1. Cross-sectional view of Al dots formed by anodic oxi-
dation of sputtered Al film on SiO (50 nm)/Si substrate. Initial
2
sputtered Al thickness was 20 lm, and anodic oxidation was
carried out with 40 V by 0.15 M oxalic acid. (a–c) SEM cross-
sectional micrographs of the nanohole bottoms. (a and b) Ob-
servations at different locations on the same sample with 10 h
anodic oxidation. (c) Sample with 11 h anodic oxidation. (d)
Cross-sectional TEM micrograph of Al dots after 10.5 h anodic
oxidation.
Fig. 2. AFM observations and its height analysis of Al hexa-
2
gonal dot array on SiO , that remained after selective wet
chemical etching of porous alumina film.

320
S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
arose from non-uniformity of anodic oxidation
speed with a very thick initial Al thickness of 20
lm. If we use a thinner Al film with a shorter first
step anodization time, regularity of nanohole ar-
ray as well as Al dot array is poor.
version from tetragonal array to trigonal array
proceeds. We used anodic voltage between 30
and 40 V; in these conditions, nanohole nearest
neighbor distance is between 65 and 90 nm. We
observed tetragonal arrays were stable when the Al
film thickness was smaller than 0.5 lm. In the
following experiments, we used Al films with 0.1
lm thickness.
Then, we tried to control the size and arrange-
ment of a nanohole array, by making an array of
concavities using AFM nanoindentation prior to
Al anodic oxidation [15]. As a matter of fact, it
turned out that the use of a thin Al film was es-
sential in the control of nanoholes and Al dots by
pretexturing techniques using AFM nanoindenta-
tion. The fabrication procedure of the nanohole
and Al dot array is shown schematically in Fig. 3.
We formed an array of concavities at a constant
interval from 50 to 150 nm by an AFM nanoin-
dention on the surface of a sputtered Al film with a
thickness of 100 nm (Fig. 3a). Next, a nanohole
array was formed by the anodic oxidation using
We examined a dependence of alumina nano-
hole arrangement on the indentation strength F .
When the indentation strength F was lower than
À5
2.0 Â 10 N, depths of concavities formed by in-
dentation were equivalent or less to the average
roughness of the Al film, and the porous alumina
film surface morphology did not show any clear
periodicity. In fact, the RMS of the 100 nm thick
Al film was about 2.0 nm, and the depth of
concavity formed by AFM nanoindentation dis-
tributed between 1.0 and 2.5 nm. When the in-
À5
0
.15 M oxalic acid at a DC voltage of 30–40 V
Fig. 3b). Finally, the alumina film was selectively
dentation strength was increased to 4.0 Â 10 N,
(
etched away. Al dots that existed at an interface
the concavities depth formed by indentation dis-
tributed between 15 and 18 nm, and a very
clear periodic tetragonal array of nanoholes was
formed after anodic oxidation. Scanning speed of
nanoindentation is another important factor to
form clearly separated nanoholes. When scanning
speed was too high, concavities made by nanoin-
dentation were connected each other and finally
nanoholes formed by anodic oxidation were con-
nected. We adopted a scanning speed slower than
10 dots/s for nanoindentation experiments with the
between the porous alumina and SiO films then
2
appeared on the surface (Fig. 3c). We used
MMAFM of Beeco Corp. equipped with a dia-
mond tip cantilever, which was specially designed
for nanoindentation, with a curvature of 10 nm.
At first we examined the effect of the indentation
strength on the nanohole arrangement in the case
of tetragonal arrays. Tetragonal hole arrays are
known to be unstable when they are created by the
periodic concavity formed by indentation using a
SiC mold prior to anodic oxidation [17,18]. Te-
tragonal hole arrays tend to rearrange into trigo-
nal ones during growth of nanoholes by anodic
oxidation. If Al film thickness is too large, con-
À5
indentation force of 4.2 Â 10 N. This rate is rel-
atively very slow, and it takes pretty long time
when a large area is nanoindented.
Next we investigated the effect of an indentation
interval (dint) on the spatial order of nanoholes
Fig. 3. Schematic diagram of Al nanodot array fabrication procedures.

S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
321
Fig. 4. Porous alumina nanohole array formed by different indentation pitch and the subsequent anodic oxidation. Al surfaces were
À5
nanoindented by different intervals from 55 to 110 nm, at the force of 4.16 Â 10 N, and later they were anodically oxidized at 40 V,
1
min with 0.15 M oxalic acid.
that were formed by anodic oxidation after nano-
indentation. Fig. 4 shows SEM micrographs of
the surfaces of porous alumina films formed by the
anodic oxidation of 40 V at various indentation
intervals. At the interval of 55 nm (Fig. 4a),
nanoholes were connected to each other. On the
other hand, at the interval of 110 nm (Fig. 4d),
additional holes were formed randomly between
the holes which positions were determined by in-
dentation. The ordered arrays of nanoholes were
formed at the intervals of 65 and 80 nm, as shown
in Fig. 4b and c. The average interval of holes
obtained after anodic oxidation was almost the
same as the indentation interval, and there was a
linear relationship between them when the inden-
tation interval was smaller than 80 nm. However,
in the case of 110 nm interval nanoindentation, the
average interval of nanoholes was much smaller
than the indented interval. This is due to the for-
mation of additional holes besides those deter-
mined by nanoindentation. The spatial order of
nanoholes was good when the interval was 65–80
nm, while it got worse when the interval was either
smaller or larger than this range. If there is no
indentation, an alumina nanohole trigonal array is
formed self-organizingly. The nearest neighbor
distance of the self-organized nanoholes at 40 V is
about 90 nm [8]. If we assume that the cell size is
equivalent between triangular and tetragonal ar-
rays, the nearest neighbor nanohole distance in the
tetragonal array is estimated to be 78 nm, which is
almost the middle of 65 and 85 nm. So we can
conclude that if the cell size of the nanoholes dif-
fers too much from that of the self-organization
condition, the order of the nanohole arrangement
becomes poor.
We attempted to form tetragonal as well as tri-
gonal Al dot arrays by selective wet chemical
etching of porous alumina films. SEM plan-view
images and AFM images of the trigonal as well as
tetragonal Al dots at the 65 nm indentation in-
terval are shown in Fig. 5. The voltage for anodic
oxidation was 30 V. The height and the nearest
neighbor distance of the Al dots are 15 and 45 nm
for trigonal array, and 20 and 65 nm for tetragonal
array, respectively.
We further evaluated electron transport prop-
erty by attaching electrodes around Al nanodot
array. Tetragonal Al dot array was formed by
anodic oxidation at 40 V and indentation interval
with 65 nm. Two tungsten electrodes were formed
using focused ion beam induced deposition as
shown schematically in the inset of Fig. 6. The
distance of the gap between electrodes was about
1.5 lm and the width of each electrodes was 1.0
lm, thus 23 Â 15 dots array existed between the
electrodes. A current–voltage characteristic of Al
dot array are shown in Fig. 6. There is a non-linear
current–voltage characteristics at room tempera-
ture (Fig. 6a). This is considered to come from
tunneling current between the dots, in which alu-
mina barrier exited in between dots. When tem-
perature was lowered to 4.2 K, a suppression of
the current near zero bias that was supposed to

322
S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
Fig. 5. Trigonal (a) and tetragonal array (b) of nanoholes and Al dots formed by AFM nanoindentation and anodic oxidation.
Interval of nanoindentation was 65 nm, and anodic voltage was 30 V. SEM plan view micrographs of the nanohole, AFM image of
Al dots after selective wet chemical etching, and its height analysis are shown.
Fig. 6. Current–voltage characteristics of 23 Â 15 Al dot array measured at (a) 298 and (b) 4.2 K.

S. Shingubara et al. / Surface Science 532–535 (2003) 317–323
323
come from the Coulomb blockade due to single
electron charging effect was observed. The off-set
voltage (Voff ) was 0.13 V. If a current passes along
the most conductive path with N-junctions of Al
rous alumina based nanofabrication to various
nanoelectron devices.
dots, Voff is approximately equal to Nðe=2CÞ [19],
References
À17
and C is estimated to be 2.4 Â 10
F assuming
[
[
1] F. Keller, M.S. Hunter, D. Robinson, J. Electrochem. Soc.
00 (1953) 411.
2] J.P. OÕSullivan, G.C. Wood, Proc. Roy. Soc. London A
17 (1970) 511.
that N is 23. The control of tunneling gap between
Al dots is not so easy since anodic oxidation still
continues even after Al film was broken up to is-
lands. In other samples that were formed almost in
the same way, some were completely insulating
due to too large gap between Al dots, and for
others cases, Al dots were connected and linear
current–voltage characteristics were observed, and
further, we observed some kind of hopping con-
ductions in some other sample. Further delicate
control of the gap between dots would be needed
to observe tunneling effect more stably.
1
3
[3] C.G. Granqvist, A. Anderson, O. Hundrei, Appl. Phys.
Lett. 35 (1979) 268.
[
[
[
4] H.L. Hornyak, C.J. Patrissi, C.R. Martin, J. Phys. Chem.
B 101 (1997) 1548.
5] D. Routkevitch, T. Bigioni, M. Moskovits, J.M. Xu, J.
Phys. Chem. 100 (1996) 14037.
6] D. Routkevitch, A.A. Tager, J. Haruyama, D. Almawlawi,
M. Moskovits, J.M. Xu, IEEE Trans. Electron Devices 43
(
1996) 1646.
[
[
7] H. Masuda, K. Fukuda, Science 268 (1995) 146.
8] S. Shingubara, O. Okino, Y. Sayama, H. Sakaue, T.
Takahagi, Jpn. J. Appl. Phys. 36 (1997) 7791.
[9] H. Masuda, F. Kasegawa, S. Ono, J. Electrochem. Soc. 144
(
1997) L127.
10] H. Masuda, K. Yada, A. Osaka, Jpn. J. Appl. Phys. 37
1998) L1340.
3
. Conclusions
[
[
(
2
Porous Al film was formed on SiO /Si substrate,
11] S. Shingubara, O. Okino, H. Sakaue, T. Takahagi, Solid
State Electron. 43 (1999) 1143.
and after completion of anodic oxidation of Al
film, hexagonal Al dot array was confirmed by
cross-sectional SEM and TEM observations. By
the use of AFM nanoindentation prior to anodic
oxidation, tetragonal array of Al nanodot that
could not be realized by self-organization was
formed. The ordered array of alumina nanohole or
Al nanodot could be obtained at the indentation
interval, which was close to that of the self-orga-
nization condition at a given anodic voltage. The
results suggest that adjusting the anodic voltage
against to the indentation interval can form Al dot
array with a desired geometry. Furthermore, we
observed a non-linear current–voltage character-
istics in the Al dots array at 4.2 K, which strongly
suggests Coulomb blockade between dots. These
results strongly suggest a high potentiality of po-
[12] S. Shingubara, O. Okino, H. Sakaue, T. Takahagi, J. Vac.
Sci. Technol. B 19 (2001) 1901.
[13] Y. Murakami, S. Shingubara, H. Sakaue, T. Takahagi,
Digest papers of Microproc. Nanotechnol. 2000 (2000)
1
78.
[
[
[
[
[
[
14] S. Shingubara, Y. Murakami, H. Sakaue, T. Takahagi,
Jpn. J. Appl. Phys. 41 (2002) L340.
15] S. Shingubara, Y. Murakami, H. Sakaue, T. Takahagi,
Mat. Res. Soc. Symp. Proc. 705 (2002) 133.
16] I. Amlani, A.O. Orlov, R.K. Kummamuru, Appl. Phys.
Lett. 77 (2000) 738.
17] H. Masuda, H. Yamada, M. Satoh, H. Asoh, Appl. Phys.
Lett. 71 (1997) 2770.
18] H. Asoh, K. Nishio, M. Nakao, T. Tamamura, H.
Masuda, J. Electrochem. Soc. 128 (2001) B152.
19] H. Grabert, M.H. Devoret (Eds.), Single Charge Tunnel-
ing––Coulomb Blockade Phenomena in Nanostructures,
Nato ASI Series, Prenum Publishing, New York, 1992
(Chapter 7).